2023 Volume 71 Issue 1 Pages 1-8
Marine sponges are among the most primitive animals and often contain unique, biologically active compounds. Several of these compounds have played an important roles as pharmaceutical leads for anti-cancer drugs, such as halichondrin B, which led to the development of an anti-breast cancer drug. Some compounds with remarkable biological activities are accumulated in significantly high concentrations in the sponge. How and why the marine sponges produce and accumulate bioactive natural products are long-standing questions with both biochemical and ecological implications, since in sponges, the animal-microbe symbioses are presumed to be responsible for the biosynthetic machinery, consisting of efficient enzymes and regulatory systems for the specific biological activities of medicinally relevant natural products. In this review, I focus on the chemically rich Theonellidae family sponges and discuss the biosynthesis of bioactive peptides and polyketides. In particular, the biosynthetic pathway of calyculin A suggests that crosstalk between the sponge host and bacterial symbiont confers a chemical defense system on the immobile animal-microbe holobiont.
Among marine invertebrates, sponges in particular are known as reservoirs of bioactive natural products.1) They are the most primitive sessile animals, living by filter feeding on large volumes of seawater (several liters per gram of sponge per day) to ingest organic matter.2,3) Because of this lifestyle, they constantly associate with large numbers of bacteria containing symbiotic and non-symbiotic transient bacteria. The consortium comprising the host sponge and symbiotic microorganisms is called a “holobiont,” which has been perceived as a proficient chemical factory generating bioactive natural products with diverse and complex skeletons.4,5) However, the questions of why sponge holobionts produce such complex compounds, and their significance, remain to be answered.
Some sponge-derived compounds share common partial structures with natural products originating from other free-living microorganisms, leading to the hypothesis that most of the natural products derived from sponges are not produced by the sponges themselves, but are most likely produced by symbiotic microorganisms.5,6) Since sponges are immobile animals and rarely over-grown by fouling organisms in the surrounding seawater, many of the sponge-derived bioactive compounds are thought to chemically defend the holobiont against bacteria, fungi, algae and predators. From an ecological point of view, they have the broad range of bioactivities and are therefore expected to be promising for pharmaceutical lead compounds.7) Among them, halichondrin B represents the successful application of a sponge-derived bioactive compound developed as an anticancer drug, and it should be noted that its structure is similar to the dinoflagellate-derived polyketide, even though the producer organism has not been identified.8)
Recent innovative advances in sequencing technology have led to rapid progress in the analysis of microbial genomes and the identification of biosynthetic genes for various secondary metabolites in free-living microorganisms.9) The biosynthetic genes are blueprints of natural products and provides valuable insights into the biosynthetic machinery including cascade reactions involving multiple enzymes. Considering that limited amounts of sponge-derived bioactive compounds are available as pharmaceutical resources, the identification and acquisition of their biosynthetic genes are important because they are the foundation for not only identifying the true producers but also tackling the supply problem by constructing heterologous expression systems. However, the biosynthetic genes of most sponge-derived metabolites are poorly understood in comparison to those of free-living microorganisms due to the complexity of the symbiotic consortia and the difficulty of preparing the sponge metagenomic DNA.
This review describes recent research results on sponges of the Theonellidae family, especially the genera Theonella and Discodermia (Fig. 1), in which a wide variety of bioactive natural products have been reported.5,6) Theonellidae family sponges belonging to the former order Lithistida (now revised as Tetractinellida)10) have provided significant chemical diversity with more than 400 compounds according to MarinLit —a database of marine natural products (https://marinlit.rsc.org/). Despite the close relationship between Theonella and Discodermia, the specialized metabolites are exclusive to each genus, with little overlap (Fig. 2), providing the diversity of chemical compounds in the Theonellidae family sponges and also raising the question of why this family has such large varieties of specialized metabolites, such as polyketides and peptides (Figs. 3, 4).
Y and W indicate yellow and white chemotypes, respectively.
It is conceivable that structural similarities exist between some natural products from sponges and those from cyanobacteria. In addition to this structural resemblance, numerous filamentous bacteria with similar morphology to cyanobacteria are present in many Theonella sponges, inspiring the hypothesis that cyanobacteria are the actual producers of peptides and polyketides from the sponges.11) Faulkner and colleagues took on the investigation to identify the real producer associated with the sponges by using density gradient centrifugation to fractionate the Palauan sponge Theonella swinhoei into sponge cells, filamentous bacteria, single cell bacteria, and cyanobacteria fractions.12) The compounds contained in each fraction were then examined by HPLC and NMR. As a result, a cyclic peptide, theopalauamide, was detected in the filamentous bacteria fraction, while a polyketide, swinholide A, was detected in the single-cell bacteria fraction (Fig. 3). The localization of the compounds suggested the existence of a producer other than the sponge, and more importantly, the filamentous bacteria found are not cyanobacteria because they lack thylakoids, which are necessary for photosynthesis. Accordingly, it appreared that the producers of the specialized metabolites in the sponges of the genus Theonella were symbiotic bacteria other than cyanobacteria. Subsequently, in 2000, the same research group performed a 16S ribosomal RNA (rRNA) analysis of this novel bacterium with unique filamentous morphology, and named it ‘Candidatus Entotheonella palauensis.’13) Because of its extremely slow doubling time, this bacterium cannot be cultured in the laboratory and thus is not yet applicable for compound production by culture-dependent methodologies.
This unique bacterium was seemingly associated with the Theonella sponge as a candidate for a microbial producer, but the cloning of the biosynthetic genes of specialized metabolites was highly desirable to obtain solid evidence. Piel’s pioneering research focused on the pederin and onnamide skeletons, which are commonly found in both insects and sponges. First, he obtained the biosynthetic gene for pederin, a venom of the beetle Paederus fuscipes, and showed that it is a prokaryotic gene without introns.14) Subsequently, they succeeded in identifying the onnamide biosynthetic gene cluster from a yellow chemotype Theonella swinhoei from Hachijo Island, Japan. This was the first report of a biosynthetic gene cluster for a specialized metabolite from a sponge.15) They successfully isolated the biosynthetic gene for onnamide from a sponge metagenomic library, by searching for biosynthetic genes similar to pederin.
About a decade later, technological innovations in next-generation sequencers have made the direct sequencing of environmental DNA possible. Taking advantage of this, Piel and colleagues performed the direct sequencing of the filamentous bacterium-enriched fraction of the yellow chemotype T. swinohei from Hachijo Island. These efforts successfully yielded the first draft genome sequence of the microbial producer, which was named as ‘Candidatus Entotheonella factor’ belonging to the new candidatus Phylum ‘Tectomicrobia.’ Remarkably, the biosynthetic gene clusters of almost all natural products isolated from T. swinhoei yellow chemotype, including onnamide and polytheonamide (Fig. 3), were encoded in the genome of the single phylotype ‘E. factor.’16) Subsequently, the actin-binding dimeric polyketide, misakinolide and antifungal cyclic peptides, theonellamides (Fig. 3) contained in T. swinhoei white chemotype were also revealed to be produced by ‘Candidatus Entotheonella serta.’17,18) Most recently, the producer bacterium of aurantoside (Fig. 3), the remaining metabolite originating in yellow chemotype T. swinohei, was identified as the different phylotype ‘Candidatus Poriflexus aureus.’19)
The genus Discodermia, as well as Theonella, belongs to the Theonellidae family, and harbors a vast array of bioactive natural products. Of particular note is the microtubules stabilizer, discodermolide, a polyketide isolated from Discodermia dissoluta, which was collected from the deep waters of the Caribbean Sea.20) The gram-scale synthesis of this complex molecule, performed by the Smith group, set the stage for the preclinical trial as anti-cancer drug.21) An attempt was also made to identify the biosynthetic gene, but despite of detailed and comprehensive cloning attempts,22) the identification of the biosynthetic gene for discodermolide has proven to be a formidable task and not yet been achieved. In 2008, a research group from Harbor Branch Oceanographic Institution revealed that D. dissoluta also associates with filamentous bacterial symbionts similar to Theonella.23) The filamentous bacterium was identified as Entotheonella based on the 16S rRNA analysis. In the Discodermia sponges from the seas around Japan, discodermolide has not been reported, but discodermin,24) discokiolide,25) and lipodiscamide26) have been isolated from Discodermia kiiensis, and calyculin A,27) calyxamide,28) kasumigamide,29) and icadamides30) have been isolated from Discodermia calyx (Fig. 4).
Calyculin A was isolated as the dominant specialized metabolite of Discodermia calyx by Fusetani et al. in 1986, with a remarkably high isolation yield (0.15% wet weight of sponge).27) Calyculin A exhibits particularly potent cytotoxicity against several cancer cell lines with pM range IC50 values, which is attributable to the specific inhibition of protein phosphatases 1 and 2A, in a similar manner to okadaic acid and microcystin.31) Given that the highly potent cytotoxin is accumulated in the animal tissue, the sponge D. calyx likely has unique storage systems that facilitate the compatibility between self-resistance and chemical defense. Calyculin A, consisting of a polyketide moiety containing tetraene and nitrile, and a peptide moiety consisting of two γ-amino acids linked by an oxazole, is biosynthetically interesting, since its structure implies the hybrid biosynthetic pathway between polyketide and peptide. To obtain the clues toward identifying biosynthetic genes of calyculin A, we comprehensively searched for ketosynthetase (KS) fragment sequences in the PCR amplicons, using degenerate primers specific for the KS domain of polyketide synthase (PKS) with the sponge metagenomic DNA. Based on the sequence of the PCR product that was considered to be part of the calyculin PKS, we screened it from a separately prepared fosmid library of the sponge metagenome. As a result, we obtained a biosynthetic gene cluster of around 150 kb in length. The gene cluster is mainly composed of modules of non-ribosomal peptide synthetase (NRPS) and trans-AT PKS (Fig. 5a), and the structure of the metabolite predicted from the domain organization was in good agreement with the chemical structure of calyculin A.32)
a) The ORFs encoded in calyculiln BGC. The ORFs associated with PKS and NRPS are highlighted in red. The kinase CalQ and phosphatase CalL are highlighted in green and blue, respectively. b) Bioconversion process of calyculin A.
To obtain the functional evidence of the biosynthetic gene cluster in relation to the calyculin biosynthetic pathway, we focused on the upstream modification enzymes, and found three open reading frames (ORFs), calM, calP, and calQ with predicted functions as phosphotransferases. They were heterologously expressed in Escherichia (E.) coli and investigated by in vitro enzymatic reactions, using several calyculin analogues as putative substrates. Remarkably, the protein encoded by one of the three ORFs, CalQ catalyzed the phosphorylation reaction with calyculin A as the substrate to generate a new analog, phosphocalyculiln A (Fig. 5b). The reaction did not proceed with other calyculin derivatives, such as geometric isomers of calyculin A and hemicalyculin A, suggesting its strict substrate specificity. These data support that the obtained gene cluster is actually responsible for calyculin biosynthesis.32,33)
Calyculin A was thought to be the end product of biosynthesis, but it was further enzymatically phosphorylated to produce a pyrophosphate, phosphocalyculin A, which had never been isolated or reported. Indeed, when previously frozen or raw sponge specimens were extracted using various organic solvents and buffers, phosphocalyculin A could not be detected in the extracts. On the other hand, the specimens were flash frozen in liquid nitrogen immediately after collection and lyophilized to remove water, and then the freeze-dried sponge tissue was extracted with alcohol. Surprisingly, the major compound of this extract was no longer calyculin A, but phosphocalyculin A. This finding suggested that phosphocalyculin A can be enzymatically converted to calyculin A in the sponge tissue. Indeed, when phosphocalyculin A was treated with the crude enzyme solution prepared from the raw sponge, dephosphorylation facilely proceeded, and phosphocalyculin A completely disappeared within a few minutes. This is the reason why calyculin A was obtained as the major metabolite in the conventional extraction method, because the tissue damage caused by mixing in the presence of water which induced the dephosphorylation of phosphocalyculin A.
Phosphocalyculin A was undetectable by conventional extraction methods, suggesting that crushing the sponge tissue in aqueous solvents instantly converts phosphocalyculin A to calyculin A. This reaction could be reproduced using the sponge D. calyx crude enzyme solution and the dephosphorylation activity is absent in the heat-denatured crude enzyme solution, confirming that this process is an enzymatic conversion reaction. We next sought to identify the phosphatase responsible for the conversion of phosphocalyculin A to calyculin A (Fig. 5b). First, we attempted to purify the enzyme directly from the unbiassed sponge holobiont, because at that time there was no conclusive evidence for the biogenic origin of the phosphatase.
Fractionations of the sponge crude enzyme solution by various chromatographic methods were attempted, by monitoring the dephosphorylation activity with a malachite green assay. Gel filtration, anion-exchange column, cation-exchange column, and affinity column, as well as isoelectric electrophoresis, were examined. The major dephosphorylation activity fraction was not adsorbed on the anion-exchange resin, but was retained on the cation-exchange resin. Therefore, the target, phosphocalyculin phosphatase, was expected to be a basic protein. The active fraction eluted from the cation exchange column was further separated on a Phenyl Sepharose column, and the resulting active fraction was subjected to gel filtration on a Sephacryl S200. Finally, the active fraction was obtained by Mono S cation exchange column chromatography. The sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis revealed that the active fraction contained a protein with a molecular weight of approximately 43 kDa.34)
With the physico-chemical properties of the phosphatase in hands, we re-inspected the ORFs of modified enzymes encoded by the calyculin biosynthesis gene cluster. Among them, only one ORF, CalL, is annotated as a phosphatase with properties similar to those of the phosphocalyculin phosphatase purified from the sponge crude enzyme solution (Fig. 5b). Therefore, to determine whether the obtained phosphatase is identical to CalL, the corresponding band on the SDS-PAGE was subjected to the peptide mass fingerprinting. After trypsin digestion of the dissected approx. 43 kDa band in the active fraction, an LC-tandem mass spectrometry (LC-MS/MS) analysis of the peptide fragment revealed an amino acid sequence identical to that of CalL. Furthermore, an additional peptide fragment corresponding to the N-terminal sequence was unexpectedly detected, indicating that the start codon of CalL was wrongly assigned in the predicted reading frame based on the original sequence data. The reanalysis of the DNA sequence of the corresponding region uncovered a sequencing error and provided the new reading frame of CalL in the corrected sequence, which unveiled a unique signal peptide sequence at the N-terminus.
To confirm this finding, we performed a functional analysis of the recombinant CalL. CalL consists of a metallophosphoesterase domain responsible for the catalysis and a signal peptide sequence at the N-terminus, which is likely recognized by the SecYEG translocase for transport it into the periplasm. Since this N-terminal signal sequence can be cleaved by peptidases in the heterologous host E. coli, recombinant protein CalL (CalL-strep) with a Strep purification tag attached to the C-terminus was heterologously expressed in E. coli. Notably, soluble CalL-strep was purified from the periplasmic fraction of the host, suggesting that CalL is localized in the periplasmic space of the producing microorganism for compartmentalization from the protoxin, phosphocalyculin A. The recombinant CalL-strep exhibited dephosphorylation activity with kinetic efficiency comparable to that of native CalL. Based on these results, we concluded that the phosphocalyculin phosphatase is CalL, which is encoded by the calyculin biosynthesis gene cluster34) (Fig. 1).
Since most of the sponge symbionts, but not all, were expected to be difficult to culture, a culture-independent single-cell analysis method was employed to identify the producing bacteria. First, we performed a CARD-FISH (catalyzed reporter deposition-fluorescence in situ hybridization) analysis using biosynthetic gene cluster-specific probes on cell fractions derived from sponges. The results showed that filamentous bacteria with a characteristic morphology were selectively detected by fluorescence.32) For further confirmation, a number of single filaments were obtained by laser microdissection and used as templates for PCR. The 16S rRNA analysis revealed that this filamentous bacterium was ‘Candidatus Entotheonella sp.’ as previously reported from T. swinhoei and D. dissoluta.13,16,22,32) Together, these results suggest that sponge symbiotic bacteria of the genus Candidatus Entotheonella are the major producers of specialized metabolites, at least in Theonellidae family sponges.
The intracellular localization of CalL was expected to be periplasmic according to the N-terminal signal peptide sequence, and was corroborated by the periplasmic localization of the recombinant CalL in the heterologous E. coli host. For further confirmation, the Entotheonella cells were stained with a malachite green reagent after the lyophilized filamentous cells were rehydrated with artificial sea water. As expected, the Entotheonella cells were exclusively colored green, indicating that the CalL-mediated reaction proceeds in the Entotheonella cells to generate calyculin A and free phosphoric acids34) (Fig. 6).
Calyculins specifically inhibit protein phosphatasea (PP1 and PP2A) present in eukaryotes and exhibit potent cytotoxicity. In contrast, calyculin A shows minimal antibacterial activity against Gram-negative and Gram-positive bacteria. Therefore, the toxicity of calyculin A to the microbial producer Entotheonella itself is considered to be low. Nevertheless, the fact that Entotheonella produces the less toxic phosphocalyculin A as an end product of biosynthesis appears to be rather beneficial in lowering toxicity for the eukaryotic host. Furthermore, the activation of phosphocalyculin A for the operation of the chemical defense is triggered by the disruption of sponge tissue. Whereas the deactivating enzyme, CalQ, is cytoplasmic, the activating enzyme, CalL phosphatase, exists in the periplasmic space of Entotheonella, and therein is a mechanism by which the precursor phosphocalyculin A is dephosphorylated specifically at the site of tissue injury, to generate calyculin A. This bioconversion mechanism is regared as an activated chemical defense mechanism, which is well known in higher plants; for example, in the biosynthesis of cyanogenic glycosides.35) This finding supports the proposal that calyculin A plays a key role in the chemical defense of the sponge holobiont.36,37) The whole biosynthetic system is valuable and necessary to develop the sustainable production of medicinally important natural products through synthetic biology.
Since the identification of onnamide biosynthetic gene cluster in 2004, research groups around the world have searched for biosynthetic genes of sponge-derived natural products. Besides the metabolically talented bacterial phylotype called Entotheonella associated with Theonellidae sponges, the cyanobacterial symbionts Hormoscilla spongeliae (formerly Oscillatoria spongeliae) associated with Desidea sponges has been identified as the producers of halogenated diphenyl ethers.38) Candidatus Endohaliclona renieramycinifaciens was also identified as the producer symbiont of renieramycin from Haliclona sp.39) However, obtaining the biosynthetic genes of sponge-derived specialized metabolites remain challenging. The main reasons for this are the difficulties in culturing the symbiotic microorganisms that are thought to be responsible for production and preparing a sponge metagenomic library. Recent rapid and remarkable renovations of metagenomic or single-cell sequencing using long and error-free reads and de novo assembly technology will overcome these obstacles and enable the identification of biosynthetic gene clusters directly from the sponge-microbe association. The next questions to answer are whether such systems are also present in other sponges, how diverse are the chemically talented sponge symbionts, and what is the origin of the biosynthetic genes? The mysteries continue to deepen, and we still have a long way to go. In the future, further clarification of the chemical and biosynthetic spaces of as-yet-uncultured microorganisms originating from sponges will bear fruits as the chemical factories of numerous medicinally important natural products.
These studies have been performed at the Graduate Schools of Pharmaceutical Sciences of The University of Tokyo and Hokkaido University. The author expresses his sincere appreciation to Professors I. Abe, N. Fusetani and S. Matsunaga of The University of Tokyo, and Professor A. Takai of Asahikawa Medical University, and Professor J. Piel of ETH Zürich for fruitful discussions and encouragement. The author thanks past and present staffs and students: Drs. Y. Egami, T. Jomori, K. Matsuda, T. Kuranaga, A. Yoshimura and A. Uria, as well as excellent students whose contributions are cited in the text. This work has been financially supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (Japan Society for the Promotion of Science KAKENHI Grant Nos. 16703511 and 18056499).
The author declares no conflict of interest
This review of the author’s work was written by the author upon receiving the 2021 Pharmaceutical Society of Japan Award for Divisional Scientific Promotion.