This Article Abstract Full Text (PDF) Supplemental material Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Copyright Information Books from ASM Press MicrobeWorld Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Taylor, M. W. Articles by Wagner, M. Search for Related Content PubMed PubMed Citation Articles by Taylor, M. W. Articles by Wagner, M.

Sponge-Associated Microorganisms: Evolution, Ecology, and Biotechnological Potential

Department of Microbial Ecology, University of Vienna, Althanstr. 14, A-1090 Vienna, Austria

SUMMARY

INTRODUCTION

EVOLUTION AND DIVERSITY OF SPONGE-ASSOCIATED MICROORGANISMS

Known Diversity of Microorganisms from Sponges

Existing Evidence for Sponge-Specific Microorganisms

Census of Sponge-Associated Microorganisms

Sponge-Associated Microorganisms: Ancient Partners or Recent Visitors That Have Come To Stay?

Scenario 1: Ancient symbioses maintained by vertical transmission.

Scenario 2: Parental and environmental symbiont transmission.

Scenario 3: Environmental acquisition.

ECOLOGICAL ASPECTS: FROM SINGLE CELLS TO THE GLOBAL SCALE

Establishment and Maintenance of Sponge-Microbe Associations

Physiology of Sponge-Associated Microorganisms

Carbon.

Nitrogen.

Sulfur.

Other aspects of microbial metabolism in sponges.

The Varied Nature of Sponge-Microbe Interactions

Mutualism/commensalism.

Microorganisms as a food source for sponges.

Harming the host: pathogenesis, parasitism, and fouling.

The Big Picture: Temporal and Biogeographic Variability in Microbial Communities of Sponges

BIOTECHNOLOGY OF SPONGE-MICROBE ASSOCIATIONS: POTENTIAL AND LIMITATIONS

Biologically Active Chemicals from Marine Sponge-Microbe Consortia and Their Commercial-Scale Supply

Methods for Accessing the Hidden Chemistry of Marine Sponges

Cultivation of metabolite-producing microorganisms.

Sponge culture.

Metagenomics.

Cell separation and metabolite localization.

Other Biotechnologically Relevant Aspects of Sponges

CONCLUSIONS AND FUTURE DIRECTIONS

ACKNOWLEDGMENTS

REFERENCES

Top Next References

INTRODUCTION

Top Previous Next References

 
Marine sponges represent a significant component of benthic communities throughout the world, in terms of both biomass and their potential to influence benthic or pelagic processes (73, 74, 124, 220). Sponges (phylum Porifera) are among the oldest of the multicellular animals (Metazoa) and possess relatively little in the way of differentiation and coordination of tissues (26, 371). They are sessile, filter-feeding organisms which, despite a simple body plan, are remarkably efficient at obtaining food from the surrounding water (290, 308, 443). The more than 6,000 described species of sponges inhabit a wide variety of marine and freshwater (somewhat more restricted) systems and are found throughout tropical, temperate, and polar regions (167). Sponges have been the focus of much recent interest (Fig. 1) due to the following two main (and often interrelated) factors: (i) they form close associations with a wide variety of microorganisms and (ii) they are a rich source of biologically active secondary metabolites. This increasing research interest has greatly improved our knowledge of sponge-microbe interactions, and yet, as apparent throughout this article, many gaps remain in our knowledge of these enigmatic associations. For example, we still lack a clear picture of microbial diversity—and the factors which influence it—in these hosts. Similarly, the physiology of most sponge-associated microorganisms remains unclear, as do many fundamental aspects of sponge symbiont ecology. (Throughout this article, the terms "symbiont" and "symbiosis" are used in their loosest possible definitions, to refer simply to two [or more] different organisms that live together over a long period of time, similar to the original de Bary definition. No judgment is made regarding benefit to either partner.) Here we aim to provide a comprehensive review of the current knowledge of the evolution, ecology, and biotechnological potential of sponge-microbe associations.

View larger version (15K): [in this window] [in a new window]   FIG. 1. Increasing research interest in marine sponge-microorganism associations. (A) Number of publications retrieved from the ISI Web of Science database by using the following search string: (sponge* or porifera* or demospong* or sclerospong* or hexactinellid*) and (bacteri* or prokaryot* or microbe* or microbial or microorganism* or cyanobacteri* or archaeon or archaea* or crenarchaeo* or fung* or diatom* or dinoflagellate* or zooxanthella*) not (surgery or surgical). (B) Number of sponge-derived 16S rRNA gene sequences deposited in GenBank per year. The 2006 value includes the 184 sequences submitted to GenBank from this article. The search string used to recover sequences was as follows: (sponge* or porifera*) and (16S* or ssu* or rRNA*) not (18S* or lsu* or large subunit or mitochondri* or 23S* or 5S* or 5.8S* or 28S* or crab* or alga* or mussel* or bivalv* or crustacea*).

View larger version (104K): [in this window] [in a new window]   FIG. 2. Schematic representation of a sponge. Arrows indicate the direction of water flow through the sponge. (Adapted from reference 328 with permission of Brooks/Cole, a division of Thomson Learning.)

View larger version (41K): [in this window] [in a new window]   FIG. 3. Sponges of diverse size, shape, and color. The encrusting sponge Tedania digitata (left), the branching sponge Axinella cannabina (center), and the giant barrel sponge Xestospongia testudinaria (right) are shown. The last two images were kindly provided by Armin Svoboda (Ruhr-Universität, Bochum, Germany).

Interactions between sponges and microorganisms occur in many forms. To a sponge, different microbes can represent food sources (290, 307, 308), pathogens/parasites (16, 171, 199, 455), or mutualistic symbionts (474, 477). Microbial associates can comprise as much as 40% of sponge tissue volume (427), with densities in excess of 10⁹ microbial cells per ml of sponge tissue (159, 453), several orders of magnitude higher than those typical for seawater. The diversity in types of interaction is matched by the phylogenetic diversity of microbes that occur within host sponges. It was already evident from early microscopy and culturing studies of sponge-associated microbes that high levels of morphological and metabolic diversity were present (62, 218, 336, 430, 471-473). The application of molecular tools over the past decade has greatly extended the known diversity of microorganisms within these hosts (100, 106, 146, 214, 294, 390, 458). Each of the three domains of life, i.e., Bacteria, Archaea, and Eukarya, are now known to reside within sponges. We now consider in detail this enormous diversity together with the evolutionary mechanisms driving its existence.

EVOLUTION AND DIVERSITY OF SPONGE-ASSOCIATED MICROORGANISMS

Top Previous Next References

 
Marine sponges are widely considered the most primitive of the metazoans, arising at least as early as the Precambrian, some 600 million years ago (206). According to molecular clocks, the divergence of sponges from the ancestors of other metazoans may have occurred even earlier, around 1.3 billion years ago (144). During subsequent periods of the Paleozoic era, sponges accounted for much of the biomass on marine reefs (167, 491). Today, they remain important members of both shallow- and deep-water communities, occupying as much as 80% of available surfaces in some areas (74). Such sustained evolutionary and ecological success is probably due, at least in part, to their intimate associations with microbial symbionts. However, unlike many other studied host-microbe associations, in which only a very small number of participants are involved (e.g., squid-Vibrio fischeri [258], amoeba-Chlamydiae [168], and Bugula-"Endobugula" symbioses [142, 210]), it is apparent that sponge-associated microbial communities can be highly diverse, with a range of different microorganisms consistently associated with the same host species. In this section, we describe the extent of this diversity, providing in-depth phylogenetic analyses of all known sponge-associated microorganisms. We summarize current evidence for the existence of sponge-specific microorganisms and conclude by considering whether sponge-microbe associations are evolutionarily ancient or are, instead, recently initiated relationships involving microorganisms which are present in the surrounding seawater.

With a few exceptions in the Euryarchaeota (164, 456), archaea reported from marine sponges are members of the phylum Crenarchaeota (164, 200, 226, 294, 454, 456). Lipid biomarkers also suggested the presence of both crenarchaeotes and euryarchaeotes in a deep-water Arctic sponge, though no phylogenetic information was provided in that study (272). The group I.1A Crenarchaeota are extremely prevalent in marine environments (180), and almost all sponge-derived archaeal sequences are affiliated with this group. The best-studied sponge-associated archaeon is the psychrophilic crenarchaeote "Candidatus Cenarchaeum symbiosum," which comprises up to 65% of prokaryotic cells within the Californian sponge Axinella mexicana (135, 294, 343, 345).

Eukaryotic microbes also occur in sponges. Sponge-inhabiting dinoflagellates (120, 152, 153, 338, 339, 355, 382, 454, 477) and diatoms (16, 47, 51, 53, 65, 113, 305, 390, 409, 454) have been reported, with the latter seemingly most prevalent in polar regions (16, 51, 53, 113, 305, 409, 454). Freshwater sponges often contain endosymbiotic microalgae, primarily zoochlorellae (30, 108, 109, 331, 333, 475, 488). Two previous reports of cryptomonads in sponges were noted by Wilkinson (477), while marine sponge-derived fungi are receiving increasing attention due to their biotechnological potential (44, 163, 191). Interestingly, of 681 fungal strains isolated worldwide from 16 sponge species, most belonged to genera which are ubiquitous in terrestrial habitats (e.g., Aspergillus and Penicillium) (163). It thus remains unclear in most cases whether such fungi are consistently associated with the source sponge, or even whether they are obligate marine species. Compelling evidence for symbiosis of a yeast with sponges of the genus Chondrilla was obtained by extensive microscopy studies of both adult sponge tissue and reproductive structures, with strong indications of vertical transmission of the yeast symbiont (221).

Little is known about viruses in sponges, although virus-like particles were observed in cell nuclei in Aplysina (Verongia) cavernicola (432). It was suggested that these particles could be involved in sponge cell pathology. Infection of an Ircinia strobilina-derived alphaproteobacterium by a bacteriophage isolated from seawater has also been demonstrated (211), although the propensity of this siphovirus to infect the bacterium in nature is not known.

In addition to the realization of high microbial diversity per se, we are now beginning to recognize more subtle patterns of host-symbiont distribution. For example, it appears that a given species of sponge contains a mixture of generalist and specialist microorganisms (390) and that the associated microbial communities are fairly stable in both space and time (105, 390, 391, 454). One particularly interesting pattern to emerge is the apparent widespread existence of sponge-specific bacterial clusters, i.e., closely related groups of bacteria which are found only in sponges (146). In the following section, we examine the published evidence for such clusters.

The hypothesis of widespread, sponge-specific microbial communities put forward by Hentschel and colleagues (146) was compelling and was constrained only by the limited data set available at that time. They performed phylogenetic analyses with the 190 publicly available sponge-derived 16S rRNA gene sequences, the majority of which were from Aplysina aerophoba, Rhopaloeides odorabile, and Theonella swinhoei. These three sponges are phylogenetically only distantly related and were collected from the Mediterranean Sea, the Great Barrier Reef, and Micronesia/Japan/Red Sea, respectively, yet they contained largely overlapping microbial communities. Together with the earlier work of Wilkinson and contemporaries (e.g., see reference 483), these remarkable results suggested that even unrelated sponges with nonoverlapping geographic ranges might share a common core of bacterial associates. Indeed, subsequent studies have lent further weight to this notion, with numerous reports of similar (and in some cases sponge-specific) bacteria found in different sponge species (100, 151, 154, 198, 235, 342, 404, 407). Furthermore, both cultivation-based and molecular methods have provided evidence for distinct microbial communities between sponges and the surrounding seawater (151, 265, 334, 391, 472). Taken together, these results appear to indicate that sponge-associated microbial communities are indeed unique and at least partially sponge specific, and the existence of sponge-specific microorganisms has consequently become something of a paradigm in this field.

A total of 14 monophyletic, sponge-specific sequence clusters were identified in the original study of Hentschel et al. (146). These occurred in the Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Cyanobacteria, Nitrospira, and Proteobacteria (Alpha, Delta, and Gammaproteobacteria) and, in most cases, were strongly supported by bootstrap analyses (in all cases, the clusters were found with three different tree construction methods). Three further clusters—each sponge specific, with the exception of a single nonsponge sequence—were also identified in the Acidobacteria and in a lineage of uncertain affiliation (later recognized as Gemmatimonadetes (146, 499). Overall, 70% of the 190 sponge-derived sequences available at the time fell into one of these monophyletic clusters or the other. Interestingly, within-cluster 16S rRNA sequence similarities ranged down to as low as 77% (146), often considered indicative of phylum-level differences (170). Several subsequent, mostly cultivation-independent studies have also led to the recovery of apparently sponge-specific sequences. Approximately 50% of 16S rRNA gene sequences in a gene library obtained from the unidentified Indonesian sponge 01IND 35 were most closely related to sequences derived from other sponges (154). These included members of the Acidobacteria, Nitrospira, Bacteroidetes, and Proteobacteria, as well as several sequences in a group of uncertain affiliation (our analyses indicate that these may be deltaproteobacterial sequences [see Fig. 8]). A similar situation was reported for Discodermia dissoluta, whereby three-quarters of 160 retrieved 16S rRNA sequences were most similar to other sponge-derived sequences (342). Conversely, of 21 unique sequences (each representing a unique restriction fragment length polymorphism [RFLP] type) obtained from the Caribbean sponge Chondrilla nucula, only 5 retrieved other sponge-derived 16S rRNA sequences during BLAST searches (although with the advantage of our larger data set, we found indications that several more of the C. nucula sequences are in fact from members of sponge-specific clusters) (151). Perhaps the most impressive sponge-specific cluster to be reported so far is the candidate phylum "Poribacteria" (100). Fieseler and colleagues found members of this lineage, which is moderately related to the Planctomycetes, Verrucomicrobia, and Chlamydiae (446), in several sponges from geographically diverse locations, but never in adjacent seawater or sediment samples (100). It will be especially interesting to see whether "Poribacteria" sequences are recovered from other environments in the future.

View larger version (44K): [in this window] [in a new window]   FIG. 8. 16S rRNA-based phylogeny of sponge-associated Deltaproteobacteria organisms. Details are the same as those provided for Fig. 5 to 7.

We began, using the ARB program package (217), by establishing an encompassing database that contains all sponge-derived 16S rRNA sequences which were available in GenBank on 28 February 2006. In addition to these 1,499 sequences (plus 11 18S rRNA sequences amplified from eukaryotic microbes in sponges), we contributed a further 184 bacterial and archaeal sequences from three hitherto unstudied sponges, namely, Agelas dilatata, Plakortis sp. (both from the Bahamas; kindly provided by U. Hentschel), and Antho chartacea (from southeastern Australia). Preliminary phylogenetic analyses identified members of putative sponge-specific clusters, and for each cluster, the most similar nonsponge sequences were retrieved by BLAST searches (from both regular NCBI and environmental genome databases) and imported into ARB for subsequent alignment (automatic and manual). The resulting ARB database, containing an alignment of all sponge-derived sequences and their nearest relatives (together with annotated information [e.g., host species and collection location] for the sponge sequences), is available upon request. Extensive phylogenetic analyses (see the supplemental material for details) were conducted, taking all (n = 1,694) sponge-derived sequences into account. In order to rigorously test the existence of monophyletic, sponge-specific sequence clusters, we employed multiple tree construction methods (maximum likelihood, neighbor joining, and maximum parsimony), together with the use of various sequence conservation filters and correction parameters.

In total, sequences representing 16 bacterial phyla and both major archaeal lineages (Crenarchaeota and Euryarchaeota) have been recovered from sponges (Fig. 4). In addition to those known prior to this study (Acidobacteria, Actinobacteria, Bacteroidetes, Chloroflexi, Cyanobacteria, Deinococcus-Thermus, Firmicutes, Gemmatimonadetes, Nitrospira, Planctomycetes, "Poribacteria," Proteobacteria [Alpha-, Beta-, Delta-, and Gammaproteobacteria], Spirochaetes, Verrucomicrobia, and Chlorobi FISH signals), we report for the first time the presence in sponges of 16S rRNA sequences affiliated with the phylum Lentisphaerae and the candidate phylum TM6. The number of sequences representing each phylum varied widely, from single sequences for the Lentisphaerae and TM6 to more than 250 sequences for each of the Actinobacteria, Alphaproteobacteria, and Beta/Gammaproteobacteria (Table 1). The proportions of sequences derived from cultivated versus noncultivated microorganisms also varied greatly among phyla.

View larger version (72K): [in this window] [in a new window]   FIG. 4. 16S rRNA-based phylogeny showing representatives of all bacterial and archaeal phyla from which sponge-derived sequences have been obtained. Sponge-derived sequences are shown in bold, with additional reference sequences also included. The displayed tree is based on a maximum likelihood analysis. Bar, 10% sequence divergence.

View larger version (48K): [in this window] [in a new window]   FIG. 5. 16S rRNA-based phylogeny of sponge-associated cyanobacteria and chloroplasts. Sponge-derived sequences are shown in bold. The displayed tree is a maximum likelihood tree constructed based on long (1,000 nucleotides) sequences only. Shorter sequences were added using the parsimony interactive tool in ARB and are indicated by dashed lines. Shaded boxes represent sponge-specific monophyletic clusters, as defined by Hentschel et al. (146), i.e., a group of at least three sponge-derived 16S rRNA gene sequences which (i) are more similar to each other than to sequences from other, nonsponge sources, (ii) are found in at least two host sponge species and/or in the same host species but from different geographic locations, and (iii) cluster together irrespective of the phylogeny inference method used (all clusters shown here also occurred in neighbor-joining and maximum parsimony analyses). Names outside wedges of grouped sequences represent the sponges from which the relevant sequences were derived; the number in parentheses indicates the number of sequences in that wedge. Filled circles indicate bootstrap support (maximum parsimony, with 100 resamplings) of 90%, and open circles represent 75% support. The outgroup (not shown) consisted of a range of sequences representing several other bacterial phyla. Bar, 10% sequence divergence.

View larger version (38K): [in this window] [in a new window]   FIG. 15. 16S rRNA-based phylogeny of sponge-associated archaeal organisms. Details are the same as those provided for Fig. 5 to 7.

Another bacterial phylum containing many sponge-specific sequence clusters is the Chloroflexi (Table 1; Fig. 6). Of the 109 sponge-derived sequences analyzed, 62% comprised such clusters, while the occurrence of a further 13% of sequences in clusters was weakly supported. In the new analyses, all but one of the members of a sponge-specific cluster described by Hentschel and coworkers (146) remained in a cluster, although these sequences were now dispersed over four different clusters. Such movement of sequences was frequently observed and is not surprising given the much larger data set at our disposal now (i.e., many new related sequences, both sponge- and non-sponge-derived, were included in the phylogenetic analyses described here). None of the sponge-derived sequences were closely related to the few described Chloroflexi species, although many were similar to sequences from uncultivated organisms, particularly from marine environments (Fig. 6).

View larger version (60K): [in this window] [in a new window]   FIG. 6. 16S rRNA-based phylogeny of sponge-associated Chloroflexi organisms. Details are the same as those provided for Fig. 5, with the following additions. Shaded boxes contained within dotted lines represent sponge-specific clusters supported by only two tree construction methods (ML, maximum likelihood; MP, maximum parsimony; and NJ, neighbor joining), and new sequences from our laboratory have the prefix "AD" (for the sponge Agelas dilatata), "AnCha" (Antho chartacea), or "PK" (Plakortis sp.).

View larger version (50K): [in this window] [in a new window]   FIG. 7. 16S rRNA-based phylogeny of sponge-associated Acidobacteria organisms. Details are the same as those provided for Fig. 5 and 6, with the following two additions. Open boxes represent monophyletic clusters containing sponge-derived sequences and a single, nonsponge origin sequence, and open boxes with asterisks outside them signify clusters containing only sponge- and coral-derived sequences (the number of asterisks corresponds to the number of coral-derived sequences within the cluster).

View larger version (26K): [in this window] [in a new window]   FIG. 9. 16S rRNA-based phylogeny of sponge-associated Gemmatimonadetes organisms. Details are the same as those provided for Fig. 5 to 7.

View larger version (24K): [in this window] [in a new window]   FIG. 10. 16S rRNA-based phylogeny of sponge-associated Nitrospira organisms. Details are the same as those provided for Fig. 5 to 7.

We report for the first time the recovery of Lentisphaerae (Fig. 11) and candidate phylum TM6 (Fig. 12A) sequences from sponges. Each phylum was represented by a single 16S rRNA sequence, from the marine sponges Plakortis sp. and Antho chartacea, respectively, and it cannot be ruled out that these represent contaminating sequences from the surrounding environment (although arguably this also applies to many, more commonly recovered sequence types). The Lentisphaerae phylum comprises part of the so-called Planctomycetes-Verrucomicrobia-Chlamydiae (PVC) superphylum (446), with sponge-derived sequences from the superphylum additionally being found in the Verrucomicrobia, Planctomycetes, and "Poribacteria" (Fig. 11). Members of the superphylum are frequently associated with eukaryotes. There is also a group of uncertain affiliation which falls near the PVC superphylum (but without strong bootstrap support) during phylogenetic analyses. This group includes sequences from many sponges, such as Agelas dilatata, Aplysina aerophoba, Discodermia dissoluta, and Theonella swinhoei. Those sequences most closely related to the sponge sequences are also from marine environments.

View larger version (47K): [in this window] [in a new window]   FIG. 11. 16S rRNA-based phylogeny of sponge-associated Verrucomicrobia, Planctomycetes, Lentisphaerae, and "Poribacteria" organisms and of a lineage of uncertain affiliation. These and associated lineages comprising the PVC superphylum (446) are shown. Details are the same as those provided for Fig. 5 to 7.

View larger version (32K): [in this window] [in a new window]   FIG. 12. 16S rRNA-based phylogeny of sponge-associated members of the candidate phylum TM6 (A), Deinococcus-Thermus organisms (B), and Spirochaetes organisms (C). Details are the same as those provided for Fig. 5 to 7. (B) In our analyses, the position of clone Dd-spU-11 (from the sponge Discodermia dissoluta) was not stable, and we are not certain of its phylogenetic affiliation.

View larger version (46K): [in this window] [in a new window]   FIG. 13. 16S rRNA-based phylogeny of sponge-associated Actinobacteria organisms belonging to the family Acidimicrobiaceae. Other sponge-derived actinobacteria are shown in Fig. S3 in the supplemental material. Details are the same as those provided for Fig. 5 to 7.

View larger version (35K): [in this window] [in a new window]   FIG. 14. 16S rRNA-based phylogeny of sponge-associated Alphaproteobacteria organisms affiliated with the genus Pseudovibrio and its relatives. Other sponge-derived alphaproteobacteria are shown in Fig. S4 and S5 in the supplemental material. Details are the same as those provided for Fig. 5 to 7.

All sponge-derived 16S rRNA sequences available on 28 February 2006 were analyzed phylogenetically, but for practical reasons the larger trees are available only in the supplemental material. Broadly speaking, the results of our analyses are consistent with the earlier study by Hentschel et al. (146), with, for example, the Actinobacteria, Nitrospira, and Acidobacteria still well represented by sponge-specific microorganisms. As could be expected, some sponge-specific clusters from the 2002 study now form parts of several new clusters, while others do not comprise clusters at all in the new data set. Conversely, the addition of more sequences meant that many formerly single sequences are now in specific clusters with other sponge-derived sequences.

Very few sequences were available from sponge-associated eukaryotic microbes at the time of database establishment (since then, some 45 18S rRNA sequences derived from sponge-associated fungi have been deposited in GenBank). Those that are included in our database include 9 16S rRNA sequences derived from diatom chloroplasts (Fig. 5) and 11 18S rRNA sequences obtained from diatoms and dinoflagellates. All but one of the 18S rRNA sequences were obtained from Antarctic sponges (454), with the remaining sequence representing a zooxanthella (Symbiodinium sp.) from the Palauan sponge Haliclona koremella (49).

We endeavored to be as thorough and as careful as possible throughout our analyses, yet there remain some caveats to our results. Despite extensive BLAST searches using members of all putative sponge-specific clusters, it is not inconceivable that we failed to include some key sequences which would have broken up otherwise specific sponge clusters. Another factor relates to the short lengths of many sponge-derived 16S rRNA sequences. We constructed our trees using only sequences longer than 1,000 bp, but more than two-thirds of all sponge-derived sequences are shorter than this (Table 1), and we added these via the parsimony interactive tool in ARB. In principle, this method allows the insertion of short sequences without changing tree topology (217). However, when many short sequences are added at once, they can influence each other's positioning (and potentially bias the analysis towards the formation of sponge-specific clusters). We attempted to gauge the severity of this problem by (for a selection of sequences) sequentially adding and removing individual short sequences and comparing their placement to the outcome when they were all added at once. The results were highly consistent, but it should not be assumed that this will always be the case. The alternative is to perform the entire phylogenetic analysis with short sequences and to truncate longer sequences to leave only the homologous region; this results in the loss of much phylogenetic information and is not recommended under any circumstances (216). Again, we reiterate the importance of obtaining at least one near-full-length sequence for each operational taxonomic unit obtained. This is not possible in some cases (e.g., excised DGGE bands) but is feasible in many others.

It is prudent to consider whether the apparent occurrence of sponge-specific sequence clusters could have a more dubious origin, namely, laboratory contamination. Theoretically, a 16S rRNA gene-containing plasmid or PCR product could, if accidentally spread to DNAs from several sponges in the same laboratory, appear to form its own sponge-specific cluster. However, the available evidence strongly suggests that this is not the case, since many or most clusters contain sequences originating from several independent laboratories.

With almost 1,700 sponge-derived 16S rRNA sequences falling into some 16 or more bacterial and archaeal phyla, we sought to address the following question: how well sampled are marine sponge-associated microbial communities? If current studies are recovering mainly sequences which were previously obtained from sponges (as the presence of sponge-specific clusters might imply), then we may have already uncovered most of the microbial diversity in these hosts, suggesting that our current descriptive phase might be nearing its logical conclusion. Unfortunately, the available data are insufficient to satisfactorily address this issue for sponges. In a recent article in this journal, Schloss and Handelsman (348) used the program DOTUR to estimate richness at different levels of phylogenetic relatedness for each bacterial phylum represented in the Ribosomal Database Project (61). To perform an analogous study with the sponge symbiont data set, we were restricted to sequences which met the following criteria: (i) they were part of attempts at extensive microbial community surveys using general 16S rRNA gene primers for the construction of clone libraries; (ii) they overlapped a sufficient distance to be useful (Escherichia coli positions 100 to 500 would have been appropriate for a reasonable portion of the sponge data set); and (iii) they were not obtained from prescreened gene libraries (e.g., by RFLP analysis), as this would heavily bias results—thus, all collected sequences must have been deposited in GenBank. After applying these (in our eyes) minimal criteria, only 317 sequences (of 1,694) were deemed suitable for use with DOTUR or similar programs. For many phyla, only a few sequences were retained (e.g., for Cyanobacteria, 8 of 119 sequences were kept, and for Alphaproteobacteria, 21 of 311 sequences were kept), precluding meaningful analyses. Furthermore, even if 50 or more sequences were suitable (as in the case of the Beta/Gammaproteobacteria), these were not necessarily representative of the known (sponge-derived) diversity within that phylum, again preventing the drawing of meaningful conclusions. Although statistically robust analyses are therefore not possible at this stage, data from two recent studies can add greatly to this discussion. In the first, Lopez and colleagues at the Harbor Branch Oceanographic Institution (213) obtained more than 700 sequences from 20 different sponge-derived gene libraries by using general 16S rRNA primers, with the vast majority of these belonging to phyla already obtained from sponges, such as Chloroflexi, Cyanobacteria, Nitrospira, Planctomycetes, and Spirochaetes. Of the recovered sequences, Epsilonproteobacteria was the only major taxonomic group not previously obtained from sponges. In another study, examining the Adriatic sponges Chondrilla nucula and Tethya aurantium, Thiel and coworkers recovered representatives of only known sponge-associated phyla (Acidobacteria, Actinobacteria, Bacteroidetes, Cyanobacteria, Gemmatimonadetes, Planctomycetes, Proteobacteria [Alpha-, Beta-, Delta-, and Gammaproteobacteria], Spirochaetes, and Verrucomicrobia) (404; V. Thiel, T. Staufenberger, and J. F. Imhoff, presented at the 11th International Symposium on Microbial Ecology, Vienna, Austria, 20 to 25 August 2006). The lack of new phyla in these data allow one to speculate that the majority of sponge-associated microorganisms may have already been encountered in gene libraries, at least at the phylum level. However, two major caveats exist. Although we may arguably be nearing the point of diminishing returns with respect to using current techniques to recover novel lineages from sponges (i.e., gene libraries constructed using general 16S rRNA primers), it is highly likely that the use of phylum-specific primers and/or metagenomic (i.e., PCR-independent) approaches will reveal phyla previously unknown to exist in these hosts or even unknown to science (e.g., "Poribacteria") (100). To our knowledge, there is no example of a sponge for which the results of general versus specific 16S rRNA gene libraries have been compared. A second point is that few gene libraries, including those from sponges, are sequenced to full coverage, and it is possible that the recurring sequences obtained from different sponges are merely those that are most abundant (or those that PCR is most biased toward) in each sponge, with the unsequenced remainder of the library potentially contributing new sequence types. The advent of high-throughput sequencing technologies (e.g., see reference 227) offers the potential to sequence gene libraries to much greater depth, illuminating the rare biosphere within sponges (376).

Statistical comparisons of microbial community compositions allow for the inclusion of more sequences (relative to species richness estimates via DOTUR) due to less stringent selection criteria. We thus used the so-called parsimony test, implemented in the program TreeClimber (347), to compare our three new gene libraries (from the sponges Agelas dilatata, Antho chartacea, and Plakortis sp.) with selected sponge-derived libraries from the literature and those deposited in GenBank. The parsimony test compares phylogenetic trees rather than sequence data per se (228, 347), and various tree construction algorithms can be employed. Our criteria for sequence inclusion were that (i) general 16S rRNA gene primers were used and (ii) at least 25 sequences were available from each library. The main caveats are that prescreening of clones (e.g., by RFLP analysis) with subsequent representation of each operational taxonomic unit by a single sequence prevents strict application of the parsimony test (347), while low sequencing coverage of some libraries may obscure true similarities or differences among libraries by missing overlapping or distinct sequences, respectively. With these considerations in mind, we compared the three libraries obtained from this study with those from the marine sponges Theonella swinhoei (146), Aplysina aerophoba (146), Rhopaloeides odorabile (458), Cymbastela concentrica (Longford et al., unpublished data), Discodermia dissoluta (342), and Chondrilla nucula (151) and the freshwater sponge Spongilla lacustris (123). An initial analysis comprising all 10 libraries yielded a highly significant (P < 0.001) result (i.e., the differences in sequence composition among the various libraries were not due to chance). Likewise, comparisons of the marine versus freshwater (S. lacustris) libraries, as well as comparisons among the marine libraries and among broad geographic locations, were all highly significant. The usefulness of such analyses should increase as more 16S rRNA gene libraries are sequenced from sponges (and with greater sequencing coverage), including multiple species from the same location and/or from the same genus or family.

In the 22 years since the Wilkinson study, a wealth of molecular data has become available for sponge-associated microorganisms, ranging from sequences of single genes to an entire genome (for the archaeon "Candidatus Cenarchaeum symbiosum") (134). Here we ponder whether such data can be exploited to address the issue of the ancientness (or otherwise) of sponge-microbe associations. First, we consider some of the many possible evolutionary scenarios (summarized in Fig. 16), as follows.

View larger version (19K): [in this window] [in a new window]   FIG. 16. Summary of various evolutionary scenarios for sponge-microorganism associations.

It is likely that any ancient sponge-microbe symbiosis would be obligate for one or both partners, potentially involving a reduction in microbial genome size if the symbiont has developed a nutritional dependence on its host. This has been demonstrated for many obligate insect endosymbionts (e.g., see references 252, 437, and 501), but it is unknown whether such tight host-symbiont coupling occurs in sponges. Integration of host and symbiont genomes was discussed in the sponge context by Sara and colleagues (337), while a recent paper offers evidence for lateral gene transfer from a fungus to the mitochondrion of its host sponge (327). Such gene transfer would not be without precedent among marine invertebrates, as it is believed that the ascidian Ciona intestinalis laterally acquired a cellulose synthase gene from a bacterium (253). Future genome sequencing of sponges and their microbial associates should offer valuable insights into the nature of these symbioses.

As noted earlier, not all sponge species harbor abundant microbial communities, and it is worthwhile to take a moment to consider these organisms. Freshwater sponges, for example, typically contain a paucity of microbial associates, and it has been suggested that this is due to an obligate requirement for sodium ions by the symbiotic bacteria (469). When freshwater sponges colonized their new habitat from the sea some 20 to 50 million years ago, it is presumed that existing symbionts were lost. Many marine sponges also harbor only relatively small numbers of microorganisms. These so-called low-microbial-abundance sponges (148) often cooccur with the high-microbial-abundance bacteriosponges, so habitat variation cannot be invoked as an explanation for these differences. Whether these sponges once contained, but later lost, large communities of microbial symbionts is unknown. It is also unknown whether the (comparatively few) microorganisms in low-microbial-abundance sponges are phylogenetically similar to those in their high-microbial-abundance counterparts.

Based on sequence information already at hand, the nearest we can come to addressing these and the following hypotheses is to consider estimated rates of 16S rRNA evolution for members of given sponge-specific clusters and to attempt to infer when the last common ancestor of sponge-specific microbes from different sponges might have occurred. If one assumes equal mutation rates in different bacterial lineages and asserts that a 1 to 2% 16S rRNA sequence difference corresponds to approximately 50 million years of evolution (259), then sequence differences of at least 10% would be required to place a common ancestor of these organisms back in the late Precambrian (600 million years ago). Here we consider two examples, the cluster representing the cyanobacterium "Candidatus Synechococcus spongiarum" (426) and the "Poribacteria" (100). The "Ca. Synechococcus spongiarum" cluster is one of the largest of all sponge-specific sequence clusters, is well supported by all tree construction methods, and contains 52 sequences from 21 sponges located around the world (Fig. 5). We chose three of these sequences as an example, derived from the sponges Theonella conica (sampled from east Africa; GenBank accession number AY701309), Aplysina aerophoba (from the Mediterranean Sea; GenBank accession number AJ347056), and Antho chartacea (from southeastern Australia; GenBank accession number EF076240). The minimum pairwise 16S rRNA similarity among these sequences (after correcting for different sequence lengths) is 97.9%. This is a very minor difference when one considers the phylogenetically disparate hosts (the last two sponges are in different orders, while T. conica is in a different subclass) and their geographically distinct locations. Even if one assumes that cyanobacteria evolve very slowly, we argue that greater sequence divergence would be expected if these bacteria had indeed been living (separately) within their host sponges for 600 million years. This should be especially true for endosymbiotic microorganisms, which are believed to evolve more rapidly due to increased fixation of mutations within their small populations (259). These members of the "Ca. Synechococcus spongiarum" cluster may therefore have a much more recent common origin, reflecting a role of horizontal (i.e., environmental) transmission consistent with scenario 2 or 3 in Fig. 16. However, consideration of other sequences within the same cluster can yield a quite different result. The two least similar sequences within the "Ca. Synechococcus spongiarum" cluster are only 93% similar, suggesting a much older separation of these particular strains. A comparable degree of similarity is seen in comparing sequences from the "Ca. Synechococcus spongiarum" cluster with those from free-living relatives. We suggest that a combination of vertical and horizontal symbiont transmission (scenario 2) could explain the observed data. Possible vectors responsible for horizontal symbiont transmission could include sponge-feeding animals, such as fishes and turtles (205, 274), analogous to the coral-feeding fireworm Hermodice carunculata, which acts as a vector for the coral pathogen Vibrio shilonii (386).

In our second example, we consider the "Poribacteria." At first glance, there appears to be a strong case for an evolutionarily ancient relationship between these bacteria and their sponge hosts. The members of this monophyletic, exclusively sponge-specific bacterial lineage differ in their 16S rRNA sequences by up to 15% and are some 20% dissimilar to their nearest nonsponge relative (derived from Antarctic sediment) (Fig. 11). Such high divergence within the cluster, together with the low similarity to the next most similar known organism, is suggestive of an ancient symbiosis with sponges. However, the two least similar "Poribacteria" sequences were taken from closely related (same family) sponges collected at the same Bahamas location, perhaps indicating horizontal symbiont transfer between these hosts. If the associations were ancient and involved strict coevolution of host and symbiont, then their respective phylogenies would be more congruent, with the least similar microbes being hosted by the least similar sponges. Furthermore, the long naked branch leading to the "Poribacteria" in the 16S rRNA tree could potentially be explained by faster rates of evolution in these bacteria. "Poribacteria" are a sister phylum to the Planctomycetes (446), which are sometimes believed to exhibit higher rates of evolution than other lineages (392). Like the case for "Ca. Synechococcus spongiarum," a combination of vertical and horizontal symbiont transmission is thus the most likely scenario here, although the acquisition of these bacteria exclusively from the environment also cannot be ruled out.

Perhaps the most convincing evidence for a long-standing, symbiotic relationship between sponges and at least some microorganisms comes from demonstrations of coevolution. Despite difficulties in addressing this issue due to the phylogenetic complexity of sponge-associated microbial communities, several authors have now shown coevolution between sponges and microbes. In the first study, the mitochondrial cytochrome oxidase subunit 1 (CO1) gene and its bacterial homolog were amplified from several halichondrid sponges and their associated bacteria (98). A CO1-based phylogenetic tree of six putatively alphaproteobacterial symbionts was largely congruent with a tree containing sequences from the corresponding host sponges, suggesting that cospeciation had occurred (although there also appeared to have been a host switch event at one point). Subsequent studies of the filamentous cyanobacterium Oscillatoria spongeliae indicated a high degree of host specificity for various dictyoceratid sponges, with evidence of cospeciation as well as indications of some host switching (316, 396). Ongoing studies of this system by Thacker and coworkers (R. W. Thacker, personal communication) should further elucidate the complex evolutionary relationships among these tropical sponges and their cyanobacterial associates. Coevolution requires that the host and symbiont maintain close associations over evolutionary time, and as mentioned above, the mechanism by which this presumably occurs in sponges is vertical transmission of microorganisms in host eggs or larvae. An additional point to consider at this stage is that the phylogeny of sponges themselves is not fully resolved (40). Molecular data are often incongruent with traditional sponge taxonomy, which is based largely on morphological properties, such as growth form and spicule characteristics (8, 37, 190). Accordingly, our understanding of symbiont evolution in sponges will continue to develop only in parallel with improvements in our knowledge of host phylogeny. A recently initiated CO1 sequencing project for taxonomically diverse sponges (www.spongebarcoding.org) is a step in the right direction for achieving the latter goal.

The final type of evidence for ancient, close associations between sponges and microorganisms comes from the fossil record (43, 261, 377). Reef mounds constructed by siliceous sponges and cyanobacterial mats, with the latter represented in part by stromatolites still found today, flourished in (sub)tropical marine waters as far back as the early Cambrian (43). The fact that sponges and microbes closely coexisted hundreds of millions of years ago is thus clear, but the nature of that interaction (e.g., whether microbes lived within sponge tissues) remains less certain.

Scenario 2: Parental and environmental symbiont transmission. Demonstrated vertical transmission is generally considered a strong indicator of symbiosis, yet this does not rule out the possibility of horizontal (e.g., environmental) transmission of the same microbe as an additional mechanism. Indeed, this phenomenon has already been shown for insect-bacterium symbioses (reviewed in reference 67), and here we borrow from the well-developed literature on this topic. In aphids, the primary (obligate) bacterial symbiont Buchnera aphidicola is vertically transmitted