Sponge samples were collected near the mouth of the Amazon River (latitude: 1.299817; longitude: -46.778867) on September 27, 2014. Samples from sponges of the Agelas and Geodia genera were collected, including the species Agelas dispar, Agelas clathrodes, Agelas clathrodes, Geodia sp., Geodia cf. corticostylifera, and Geodia neptuni. The main criterion for sponge selection was the availability of at least three individuals per genus. After being sampled on the RV, the sponges were immediately stored in liquid nitrogen. Each sponge’s tissue was dried and dissected with a scalpel to remove associated macroscopic organisms. Approximately 1 g of sponge tissue was frozen in liquid nitrogen and subsequently pulverized. DNA from a single individual of a selected sponge species was extracted and purified with 4 M guanidine hydrochloride, 50 mM Tris-HCl (pH 8.0), 0.05 M EDTA, 0.5% sodium N-lauroylsarcosine, and 1% β-mercaptoethanol. This was followed by phenol/chloroform extraction, isopropanol precipitation, and resuspension in 50 µL of ultrapure water (Trindade-Silva et al., 2012).

A DNA library was constructed for each sample with Illumina® TruSeq Nano DNA Preparation Kit. Sequencing was performed on the Novaseq system (Illumina®), generating 150 bp paired-end reads at 20 GB sequence coverage (Supplementary Data Set S1). Sickle v1.33 (https://github.com/najoshi/sickle) trimmed the reads, while BBtools v35 (https://sourceforge.net/projects/bbmap/) identified and removed phiX sequences and Illumina adapters. A total of six metagenomes were generated, one for each sponge species. The metagenomes were assembled with metaSPAdes v3.13.0 (Nurk et al., 2017), removing scaffolds shorter than 1 kb. All programs were used with default settings.

Metagenome-assembled genomes (MAGs) were recovered using MaxBin2 v2.2.14 with 40 markers for bacteria and archaea, and 107 markers for bacteria only (Wu et al., 2016). DASTool v1.1.2 aggregated the MAGs (Sieber et al., 2018). uBin v0.9.14 (Bornemann et al., 2023) selected MAGs based on GC content, coverage, and taxonomy. Dereplication was executed with dRep v3.2.2 (Olm et al., 2017) with a threshold of 95%. Default parameters were used in all of these programs. A total of 1,054 MAGs were recovered from the sponges’ microbiome. To exclude low-quality MAGs, coverage and contamination were calculated, respectively, by CoverM v0.6.1 (https://github.com/wwood/CoverM) and CheckM v1.0.13 (Parks et al., 2015). Genomes with <70% completeness or >10% contamination were discarded.

The remaining 205 MAGs underwent additional analysis. These MAGs were annotated with Prokka v1.14.5 (Seemann, 2014) and taxonomically assigned using the SILVA database v138 of 16S rRNA genes (Quast et al., 2013). MAGs with inconclusive 16S rRNA classification results were classified using phylogenetic marker genes trees. Based on these preliminary classifications, 10 MAGs were selected for further investigation as they were all assigned to the same under-described group, Latescibacterota. This phylum was selected because it is abundantly found in GARS sponges with low microbial abundance (Menezes et al., 2022), and its members may represent up to 2% of the GARS sponges’ metagenomes (Pinto et al., 2022) (Supplementary Data Set S2). Moreover, Latescibacterota members cannot be cultivated yet; therefore, one method of learning more about this candidate is through analysis of their genomes.

The quality of selected genomes was more accurately assessed based on criteria established by Bowers et al. (2017), including completeness, contamination, presence of 5S, 16S, and 23S rRNA genes, and the quantity of tRNA. CheckM, CoverM, Prokka, and GenomeQC web server (Manchanda et al., 2020) were used to obtain these genomic metrics. Additional assembly statistics that help define MAG quality, such as the number of scaffolds, assembly size, N50, L50, NG50, LG50, and GC content, were acquired with GenomeQC.

Furthermore, eukaryotic-like proteins (ELP), which contain domains important for symbiosis, were searched using Pannzer2 and CD-Search web servers in default settings. Pannzer2 uses the Uniprot database, and these proteins were manually identified using the description of annotated genes (Törönen and Holm, 2022). CD-Search utilizes a specific conserved domain database (CDD v3.20), and ELPs were manually curated according to the domain name (Marchler-Bauer and Bryant, 2004).

GTDB-Tk (Chaumeil et al., 2020; version 2.3.0) assigned the selected MAGs to a taxonomic group using GTDB (Parks et al., 2020; version 214). A codon tree was generated using resources present in PATRIC v3.6.12 (Davis et al., 2020). The “Similar Genome Finder” function found genomes similar to the MAGs, while the “Phylogenetic Tree” function identified homologous groups of single-copy genes, aligned those sequences, and concatenated the alignments. The tree was constructed using maximum likelihood estimation with RAxML (Stamatakis, 2014). FigTree v1.4.4 (https://github.com/rambaut/figtree) was used to visualize the tree, making aesthetic changes for improved visualization. The codon tree was composed of Latescibacterota members, members from a similar phylum (Gemmatimonadetes), and an outgroup of Firmicutes that was used to root the tree.

Additionally, average nucleotide index (ANI) and digital DNA-DNA hybridization (dDDH) were used to compare the 10 MAGs and determine their relatedness. While ANI was calculated on JspeciesWS v.3.9.1 (Richter et al., 2016), dDDH was calculated on GGDC v2.1 (Meier-Kolthoff et al., 2013). Default settings were applied to both web servers.

Metabolic reconstruction was performed by verifying the existence of genes involved in reactions of specific metabolic pathways in KEGG v97.0 (Kanehisa et al., 2017) and Metacyc v26.0 (Caspi et al., 2020) databases. BlastKOALA v2.2 (Kanehisa et al., 2016) and KEGG Mapper v5.0 (Kanehisa and Sato, 2020) were used to reannotate genomes and access KEGG data. The software Pathway Tools v23.0 was used to reconstruct pathways present in Metacyc (Karp et al., 2021). Graphics and tables were generated with LibreOffice Calc and LibreOffice Writer, while the R package “pheatmap” was utilized to create heat maps.

Genes that encode industrial enzymes were searched for in annotated genomes. The enzymes were selected based on previously reported applications in scientific papers, using the BRENDA database (Jeske et al., 2019) to aid in the selection (Supplementary Data Set S3). AntiSMASH v6.0 (Blin et al., 2021) was used to identify clusters of genes related to the production of secondary metabolites. DbCAN2 web server (Zhang et al., 2018) searched for carbohydrate-active enzymes (CAZyme) present in CAZy database v08062022 (Drula et al., 2022) and dbCAN HMMdb v11 using three tools: Hmmer, Diamond, and eCAMI. Only the genes identified by these three methods were considered valid.

Additionally, biologically active peptides (BAP) were searched for. Peptide sequences were extracted from annotated genomes using FaBox tools v1.61 (Villesen, 2007). Due to software limitations, only sequences with <100 amino acid residues were retrieved. PeptideRanker calculated the chance that a BAP would be present in a protein sequence. Sequences with a probability >0.5 were selected for functional prediction. AntiCP 2.0 (Agrawal et al., 2021), Antifp server (Agrawal et al., 2018), antiAngioPred server (Ramaprasad et al., 2015), and Toxinpred server (Gupta et al., 2013) predicted, respectively, anticancer, antifungal, anti-inflammatory, and toxin activities. Antimicrobial activity was also predicted using antiBP2 server (Lata et al., 2010) and CAMPr3 (Waghu et al., 2016). A predicted function was considered valid when the calculated probability value was >0.5. All programs were run with default settings.

According to the criteria defined by Bowers et al. (2017), a high-quality genome draft must have 5S, 16S, and 23S rRNA genes, at least 18 tRNA genes, completeness >90%, and contamination <5%. Medium-quality drafts only need completeness ≥50% and contamination <10%. Finally, low-quality drafts have completeness <50% and contamination >10%. Therefore, only three of the selected metagenome-assembled genomes (MAGs) are considered high-quality genome drafts (Supplementary Data Set S4). The others are medium-quality genomes due to contamination above 5%, completeness less than 90%, or the absence of certain rRNA genes. Additional metrics used to define MAG quality were also obtained and are described in Supplementary Data Set S4.

Both Pannzer2 and CD-Search detected several copies of eukaryotic-like proteins (ELPs) genes in all MAGs (Figure 1). The following domains were found in CD-Search: ankyrin repeat (ANK); leucine-rich repeat (LRR); NCL-1, HT2A, and Lin-41 repeat (NHL); SEL1-like repeat (SEL1); tetratricopeptide repeat (TPR); and WD40 repeat (WD40). Pannzer2 identified the same domains found by CD-Search, with the addition of fibronectin type III domain (FN3), pyrroloquinoline quinone repeat (PQQ), and NipSnap family. For both programs, the tetratricopeptide repeat (TRP) was the predominant ELP.

All 10 MAGs showing inconclusive taxonomic affiliation based on 16S rRNA genes were assigned to the candidate phylum Latescibacterota by GTDB-Tk (Supplementary Data Set S4). The codon tree shows the relation between the selected MAGs, other individuals from the candidate phylum Latescibacterota, and members from the close-related phylum Gemmatimonades (Figure 2).

In addition, ANI and dDDH were used to assess how similar the genomes were to one another. ANI values between 95% and 96% or dDDH values ≥70% indicate that the pair of analyzed genomes probably belong to the same species (Chun et al., 2018). Thus, the pair of MAGs Latescibacteria FT14_2_151_60/Latescibacteria FT41_2_97_61 (ANI >96%; dDDH =73.6%) and Latescibacteria FT41_80_57/Latescibacteria FT14_40_044 (ANI >95%; dDDH =72.3%) are likely from the same species (Table 1).

Supplementary Data Set S5 contains a list of all the genes and potentially expressed proteins that were identified during metabolic reconstruction, including transporters and enzymes. All Latescibacterota MAGs contained complete heterotrophic aerobic processes like the tricarboxylic acid (TCA) cycle, the electron transport chain (complexes I–V), and ATP production (Figure 3). As for anaerobic processes, a gene related to the dissimilatory sulfate reduction to hydrogen sulfide (sat) was present in seven MAGs, whereas genes related to dissimilatory reduction of nitrate to ammonium (narI and nirD) were present in six MAGs. Genes for pyruvate fermentation to acetoin (alsS and budC), ethanol (adh), or lactate (ldh) were found in all MAGs. Glycolysis and the pentose phosphate pathway were among the complete catabolic processes observed in three MAGs. Moreover, all of the analyzed Latescibacterota have the gene that encodes the enzyme carbon monoxide dehydrogenase, linked to energy generation from carbon monoxide. Furthermore, the gene for hydrogenase, an enzyme used to generate energy from hydrogen, was found in the Latescibacteria FT14_40_044 MAG. Finally, genes for carbon fixation pathway enzymes were found in all MAGs, though all those pathways are incomplete and genes for key enzymes are missing (Figure 3).

In addition, metabolic reconstruction results indicate that GARS sponge-associated Latescibacterota may degrade a variety of organic and inorganic molecules, including carbohydrates, lipids, amino acids, and other sulfur, phosphorous, and nitrogen compounds (Supplementary Data Set S5). The findings also suggest that they can synthesize important molecules such as carbohydrates, lipids, phospholipids, essential amino acids, vitamins, and secondary metabolites (Figure 3). Complete pathways for the production of phosphatidylethanolamine, myo-inositol, vitamins B6 and B7, and the essential amino acids isoleucine, leucine, lysine, phenylalanine, threonine, thryptophan, and valine were identified.

Furthermore, the MAGs contain genes for enzymes involved in specific steps of the nitrogen, sulfur, and phosphorus cycles. Regarding the nitrogen cycle, genes for hydroxylamine oxidation to nitrite (hao), nitrate reduction to nitrite (narI), and conversion of nitrite to ammonia (nirD) were detected (Figure 4A). These genes indicate participation in nitrification, interconversion of nitrate and nitrite, and ammonification. Two genes required for the incorporation of ammonia into amino acids during glutamine synthesis (glnA and gltB) were observed as well. Concerning the sulfur cycle, the results suggest participation in assimilatory sulfate reduction (sat, cysC, cysH, and sir), dissimilatory sulfate reduction (sat), thiosulfate disproportionation (glpE), and incorporation of sulfide into cysteine (cysK) (Figure 4B). As for the phosphorus cycle, metabolic reconstruction indicates a contribution to phosphorus remineralization due to the presence of genes related to organic phosphorus compound degradation, such as aminoethyl phosphonate (phnW and phnY), methyl phosphonate (phnPP), and phosphonoacetate (phnA). Additionally, all genes for the polyphosphate cycle (ppx, ppk, and ppk2), a pathway involved in phosphorus storage, were verified in two MAGs (Figure 3).

Moreover, MAGs analysis shows that GARS sponge-associated Latescibacterota might be involved in heavy metals metabolism (Figure 3). All genes for arsenic compound bioremediation (arsC and gap) and narrow mercury resistance (merA and merP) were detected in eight and five MAGs, respectively. These microorganisms might also participate in the degradation of harmful organic compounds such as naphthalene, 4-chloronitrobenzene, 1,2-dichloroethane, fluoroacetate, and polybrominated diphenyl ether (PBDE), as verified respectively by the identification of genes nagAB, cnbZ, dhmA, dehH2, and bphA. Additionally, all MAGs contain the gene for glycine oxidase (thiO), which acts in the bioremediation of the pesticide glyphosate, whereas two MAGs have the gene for 2,4-dichlorophenol 6-monooxygenase (tdfA), which is involved in the degradation of the herbicides 2,4-dichlorophenoxyacetic acid (2,4-D) and 2-methyl-4-chlorophenoxyacetic acid (MCPA).

Importantly, the following prokaryotic defense mechanisms were observed in the GARS sponge-associated Latescibacterota MAGs: clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated protein (Cas) system (types I and III); restriction-modification (R-M) system (types I, II, III, and IV); toxin-antitoxin system (type II); and DNA phosphothiolation system (Figure 3). Five MAGs also contain the catalase (katG) and superoxide dismutase (sodA) genes, which encode enzymes that degrade reactive oxygen species (ROS).

Lastly, genes that encode enzymes related to the production of antibiotics were present in all Latescibacterota sponge-associated MAGs (Supplementary Data Set S5). However, all pathways were incomplete. The enzymes found were associated with the production of aurachin (auaCH), anthracycline (dauE), bacilisin (bacCG), beta-lactams (axeA, acyII, cefD, G7AC, PENDE), fosfomycin (fom1), kanamycin (kanJ), lividomycin (livQ), macrolide (pikAII), novobiocin (novN), pentalenolactone (ptlD), prodigiosin (pigC), pyocyanin (phzF), pyrrolnitrin (prnD), tetracycline (oxyT), and chlortetracycline (ctcP). Antibiotic resistance genes were also found in the 10 MAGs, indicating resistance to multiple drugs, including aminoglycosides, macrolides, vancomycin, penicillin, and triclosan (Supplementary Data Set S5).

The GARS sponge-associated Latescibacterota MAGs contain genes that encode industrial enzymes, including lipase, phospholipase, esterase, protease, aminopeptidase, amidase, dioxygenase, monooxygenase, dehalogenase, nitrile hydratase, alginate-lyase, alcohol dehydrogenase, and glycerol dehydrogenase (Table 2).

Gene clusters involved in secondary metabolites production were identified by AntiSMASH in all GARS sponge-associated Latescibacterota MAGs, with the MAG Latescibacteria FT41_80_57 as an exception (Figure 5). The clusters are linked to the production of terpene, non-ribosomal peptide synthase-like fragment (NRPS-like), linear azole(in)e-containing peptide (LAP), polyketide synthase type I (T1PKS), ranthipeptide, cluster containing RiPP recognition element (RRE), and other unspecified ribosomally synthesized and post-translationally modified peptide product (RiPP-like).

A total of 124 genes for CAZymes were annotated by DbCAN2 in the 10 MAGs (Figure 6). The following modules were identified: glycoside hydrolase (GH); glycosyltransferase (GT); polysaccharide lyase (PL); carbohydrate esterase (CE); and carbohydrate-binding modules (CBM). The auxiliary activity module (AA) was absent. It is worth noting that 13 sequences also have a signal peptide.

A total of 425 peptides were identified in all sponge-associated Latescibacterota MAGs. Of those peptides, 166 were categorized as biologically active peptides (BAP) by PeptideRanker (Figure 7A). In 126 of those BAPs, there was no biological function predicted. Nevertheless, antimicrobial, antifungal, anti-inflammatory, and anticancer activities were predicted in 40 BAPs (Figure 7B). For eight peptides, more than one type of activity was predicted. In contrast, Toxinpred did not detect toxin activity in any peptide.

GARS MAGs provide valuable insights into the functional potential of uncultured Latescibacterota associated with the sponge holobiont. These MAGs encode enzymes that degrade complex substrates such as proteases and lipases, releasing nutrients for easier absorption by the sponge (Selvin et al., 2010; Pita et al., 2018). These bacteria also possess transporters that may transfer nutrients and other compounds between the host and symbionts (Fan et al., 2012; Menezes et al., 2022). Additionally, they can synthesize myo-inositol, which serves as an osmolyte or part of cell structures, and phosphatidylethanolamine, which is a structural component of sponges and bacterial membranes (Murzyn et al., 2005; Roberts, 2006; Genin et al., 2008). Moreover, these microorganisms could supply the host’s demand for essential nutrients like vitamins and some amino acids (Pita et al., 2018; Robbins et al., 2021).

Another role potentially performed by the GARS sponge-associated Latescibacterota is host defense. Their MAGs contain genes involved in the production of antibiotics, biologically active peptides, and other secondary metabolites with antibiotic activity, which can prevent pathogen colonization (Selvin et al., 2010). Secondary metabolites may also contribute to chemical defense, predator evasion, and physiological improvement, influencing the sponge’s susceptibility to predation (Pita et al., 2018). Additionally, the 10 MAGs have genes related to stress proteins, defensive enzymes like phospholipases, and defense systems such as CRISPR-Cas, R-M system, toxin-antitoxin system, and DNA phosphotiolation. These genes, which are generally enriched in sponge symbionts, may protect the sponge against harmful substances and foreign organisms (Selvin et al., 2010; Horn et al., 2016). Moreover, five MAGs contain genes for antioxidant enzymes, which may improve the host’s defense against reactive oxygen species by increasing antioxidant activity and neutralizing capability (Regoli et al., 2000).

Furthermore, GARS sponge-associated Latescibacterota could aid in the elimination or mitigation of toxic substances. They may oxidize and assimilate ammonia, one of the main wastes generated by sponges, preventing its accumulation during moments of low water pumping and the consequential death of host tissues (Webster and Taylor, 2012; Pita et al., 2018). These bacteria may also regulate the concentration of hydrogen sulfide, which can be toxic in aquatic environments, by incorporating sulfide into cysteine, similar to the process observed in plants (Bagarinao, 1992; Birke et al., 2012). In addition, GARS sponge-associated Latescibacterota potentially convert mercury and arsenic into less toxic forms, safeguarding the host from their harmful effects. This protection could be crucial as heavy metal exposure impacts the sponge’s microbial community, physiology, and overall health (Selvin et al., 2010; Webster and Taylor, 2012).

To establish a symbiotic relationship, sponges can recognize and differentiate symbionts from consumable microbes through immune signaling (Thomas et al., 2010). ELPs, which are usually enriched in sponges and their symbionts, may play a role in this recognition as they are linked to phagocytosis evasion (Pita et al., 2018). Different ELPs were identified in the GARS sponge-associated Latescibacterota, consistent with findings in other sponge-associated microorganisms (Reynolds and Thomas, 2016; Díez-Vives et al., 2017; Frank, 2019).

Besides aiding their host, GARS sponge-associated Latescibacterota can contribute to the reef by participating in nitrogen, sulfur, and phosphorus cycles. Concerning the nitrogen cycle, genes related to the interconversion of nitrate and nitrite, ammonification, nitrification, and incorporation of ammonia into amino acids were identified. Nitrifying bacteria and ammonia-oxidizing bacteria are frequently found in sponges from different environments (Pajares and Ramos, 2019). About the sulfur cycle, the analyzed Latescibacterota might be involved in assimilatory sulfate reduction, dissimilatory sulfate reduction, and thiosulfate disproportionation. However, this last reaction may serve a different purpose as it is catalyzed by rhodanese, an enzyme used to convert hydrogen cyanide into less toxic forms (Cipollone et al., 2007). Nonetheless, these Latescibacterota are likely sulfate-reducing bacteria, a type of microorganism regularly found in association with sponges (Jensen et al., 2017; Lavy et al., 2018). Regarding the phosphorus cycle, the described Latescibacterota may participate in phosphorus storage through its sequestration and deposit in polyphosphate granules, which are produced during the polyphosphate cycle. This process affects phosphorus availability as well as marine microbial metabolism and diversity (Dyhrman et al., 2007; Zhang et al., 2015; Pita et al., 2018). There is also possible involvement in remineralization via phosphonate degradation, a pathway commonly found in invertebrates’ microbiomes (Villarreal-Chiu et al., 2012; Podell et al., 2020).

The investigated microorganisms might also contribute to bioremediation. GARS sponge-associated Latescibacterota MAGs contain genes related to the degradation of pollutants such as naphthalene, 4-chloronitrobenzene, 1,2-dichloroethane, organohalides, fluoroacetate, PBDE, glyphosate, 2,4-D, and MCPA. However, none of these contaminants directly impact sponges. Conversely, PBDE is produced by a few species of sponges (Agarwal et al., 2017). Nonetheless, some of these pollutants, including naphthalene, glyphosate, and certain types of PBDE, are known persistent organic pollutants in the ocean, disturbing other aquatic organisms (Ipen, 2018; Vagi et al., 2021). Moreover, all MAGs contain genes for bioremediation enzymes like oxygenases, nitrile hydratase, and dehalogenase, which can break, transform, or remove toxic molecules such as aromatics, nitriles, and organohalogens (Karigar and Rao, 2011; Cheng et al., 2020; Oyewusi et al., 2020).

Furthermore, metabolic reconstruction of GARS sponge-associated Latescibacterota revealed an energy metabolism typical of heterotrophic bacteria, with coexisting aerobic and anaerobic processes. Heterotrophy is a feature often found in bacteria associated with sponges, typically through aerobiosis. However, facultative anaerobes and sulfate-reducing bacteria can be found in sponges’ microbiomes, possibly due to pumping fluctuations that create moments of anaerobiosis (Osinga et al., 2001; Taylor et al., 2007). In the GARS, heterotrophic processes may be important for sponges, particularly the ones with low microbial abundance, as they can obtain a carbon supply from organic particles in the Amazon River plume (Calegario et al., 2021; Menezes et al., 2022). Carbon fixation is another process occurring in the GARS turbid areas, albeit at a minor level (Pinto et al., 2022). These pathways are frequently seen in sponge symbionts and the candidate phylum Latescibacterota (Feng and Li, 2019; Tran et al., 2021), but were absent or incomplete in the 10 MAGs. Carbon monoxide dehydrogenase, an enzyme frequently present in litohererotrophics, was detected in all MAGs, which may allow the use of carbon monoxide as a carbon source or for energy conservation (Robb and Techtmann, 2018; Burgsdorf et al., 2022).

It is worth mentioning that the MAGs’ quality varied from medium to high. The assembly quality is essential for annotation and may affect the quantity and quality of identified genes (Florea et al., 2011), which could explain the predominance of incomplete pathways in the studied MAGs. Another hypothesis that could explain this incompleteness of metabolic pathways is the presence of alternative genes encoding unknown enzymes that may be involved in the missing reactions (Lannes et al., 2019). In some pathways, cooperation with other organisms in the community may be necessary to complete metabolic processes. Through the metabolic division of labor, different populations execute complementary reactions of a particular pathway, as seen in nitrification (Embree et al., 2015; Tsoi et al., 2018). Microorganisms can also interact with the host through cosynthesis or cometabolism. In sponges, integration between symbiont and host genes may be required to complete a pathway, or the product of one organism’s reactions may act as an input in the other’s pathway (Selvin et al., 2010). It is important to highlight that inferences were made using DNA sequences and did not include information about gene expression. Having a specific gene in a MAG does not guarantee that a particular process is being executed. Instead, it suggests potential for its execution.

In addition to their ecological role, GARS sponge-associated Latescibacterota may serve as a significant source of natural products as they potentially produce relevant metabolites. For example, these bacteria may produce myo-inositol, which is used in the food, pharmaceutical, and cosmetic industries as a beverage additive as well as to produce uronic acid and other inositol derivatives (You et al., 2020). Moreover, nine MAGs contain clusters of genes involved in the generation of secondary metabolites, including terpenes, ranthipeptides, LAPs, T1PKS, and NRPS-like proteins, as well as clusters with RRE, a conserved domain that interacts with peptides and is used to identify novel compounds (Burkhart et al., 2015). Sponge symbionts are a known source of NRPS and PKS (Pimentel-Elardo et al., 2012; Trindade-Silva et al., 2013). Except for ranthipeptides, whose function remains unknown, these metabolites may have exploitable activities. For instance, terpenes are employed as a repellent, pigment, flavoring agent, and medical component for infections and cardiovascular diseases (Schempp et al., 2018; Cox-Georgian et al., 2019).

Additionally, genes for commercial enzymes that are applied in many areas were identified in GARS sponge-associated Latescibacterota MAGs. For instance, proteases are used in the food industry to process meat, milk, and beverages, as well as in the production of pharmaceuticals, agrochemicals, and textiles (Adrio and Demain, 2014; Raveendran et al., 2018). Esterases, phospholipases, and aminopeptidases are also used in the food industry to enhance food flavor and fragrance (Raveendran et al., 2018; Nandan and Nampoothiri, 2020). Moreover, five classes of CAZymes, enzymes that act on the synthesis, breakdown, or modification of complex carbohydrates and other glycoconjugates, were found in the 10 MAGs (Cantarel et al., 2009). CAZymes from marine organisms are used in several areas, including bioenergy, food industry, and chemical industry, and may possess unique properties (Oliveira et al., 2020; Contesini et al., 2021).

Furthermore, GARS sponge-associated Latescibacterota showed great potential for generating bioactive peptides, containing possible peptides with antimicrobial, antifungal, anticancer, and anti-inflammatory properties predicted. BAPs are a group of peptides with less than 50 amino acid residues that exhibit a biological function (Macedo et al., 2021; Akbarian et al., 2022). These peptides also have significant pharmaceutical and nutraceutical potential, with commercialized peptide products derived from sponges such as the antibacterial polydiscamide A and the anticancer phakellistatin (Brinkmann et al., 2017; Macedo et al., 2021).

In conclusion, the 10 sponge-associated Latescibacterota may play a significant role in the sponge’s success in the GARS. These symbionts may participate in the host’s nutrition and protection against pathogens, heavy metals, and other harmful chemicals. Also, the presence of genes related to biogeochemical cycles and bioremediation indicates that these Latescibacterota contribute to nutrient flux in the reef and mitigation of damage from detrimental compounds. Additionally, the predominance of heterotrophic processes is in agreement with a recent hypothesis that GARS organisms might feed from sediments and other particles present in the Amazon River discharge (Menezes et al., 2022). However, it is essential to acquire transcriptional data to confirm the expression of found genes and endorse the execution of discussed functions. Finally, GARS sponge-associated Latescibacterota have great biotechnological value. Their genomes contain genes for industrial enzymes, secondary metabolites, and biologically active compounds that could be explored in future works.