Search

Background Cyanobacteria from the genus Microcystis can form large mucilaginous colonies with attached heterotrophic bacteria—their microbiome. However, the nature of the relationship between Microcystis and its microbiome remains unclear. Is it a long-term, evolutionarily stable association? Which partners benefit? Here we report the genomic diversity of 109 individual Microcystis colonies—including cyanobacteria and associated bacterial genomes—isolated in situ and without culture from Lake Champlain, Canada and Pampulha Reservoir, Brazil. Results We identified 14 distinct Microcystis genotypes from Canada, of which only two have been previously reported, and four genotypes specific to Brazil. Microcystis genetic diversity was much greater between than within colonies, consistent with colony growth by clonal expansion rather than aggregation of Microcystis cells. We also identified 72 bacterial species in the microbiome. Each Microcystis genotype had a distinct microbiome composition, and more closely related genotypes had more similar microbiomes. This pattern of phylosymbiosis could be explained by co-phylogeny in only two out of the nine most prevalent associated bacterial genera, Roseomonas and Rhodobacter. These phylogenetically associated genera could enrich the metabolic repertoire of Microcystis, for example by encoding the biosynthesis of complementary carotenoid molecules. In contrast, other colony-associated bacteria showed weaker signals of co-phylogeny, but stronger evidence of horizontal gene transfer with Microcystis. These observations suggest that acquired genes are more likely to be retained in both partners (Microcystis and members of its microbiome) when they are loosely associated, whereas one gene copy is sufficient when the association is physically tight and evolutionarily long-lasting. Conclusions We have introduced a method for culture-free isolation of single colonies from nature followed by metagenomic sequencing, which could be applied to other types of microbes. Together, our results expand the known genetic diversity of both Microcystis and its microbiome in natural settings, and support their long-term, specific, and potentially beneficial associations.

Cyanobacteria from the genus Microcystis can form large mucilaginous colonies with attached heterotrophic bacteria – their microbiome. However, the nature of the relationship between Microcystis and its microbiome remains unclear. Is it a long-term, evolutionarily stable association? Which partners benefit? Here we report the genomic diversity of 109 individual Microcystis colonies – including cyanobacteria and associated bacterial genomes – isolated in situ and without culture from Lake Champlain, Canada and Pampulha Reservoir, Brazil. We found 14 distinct Microcystis genotypes from Canada, of which only two have been previously reported, and four genotypes specific to Brazil. Microcystis genetic diversity was much greater between than within colonies, consistent with colony growth by clonal expansion rather than aggregation of Microcystis cells. We also identified 72 bacterial species in the microbiome. Each Microcystis genotype had a distinct microbiome composition, and more closely-related genotypes had more similar microbiomes. This pattern of phylosymbiosis could be explained by co-phylogeny in two out of the nine most prevalent associated bacterial genera, Roseomonas and Rhodobacter, suggesting long-term evolutionary associations. Roseomonas and Rhodobacter genomes encode functions which could complement the metabolic repertoire of Microcystis, such as cobalamin and carotenoid biosynthesis, and nitrogen fixation. In contrast, other colony-associated bacteria showed weaker signals of co-phylogeny, but stronger evidence of horizontal gene transfer with Microcystis. These observations suggest that acquired genes are more likely to be retained in both partners (Microcystis and members of its microbiome) when they are loosely associated, whereas one gene copy is sufficient when the association is physically tight and evolutionarily long-lasting.

Haemophilus influenzae causes common and sometimes severe adult and pediatric disease including chronic obstructive respiratory disease, otitis media and infections of the central nervous system. Serotype b strains, with a b-type capsule, have been the historical cause of invasive disease, and the introduction of a serotype b-specific vaccine has led to their decline. However, unencapsulated or non-b-type H. influenzae infections are not prevented by the vaccine and appear to be increasing in frequency. Here we report two pediatric cases of severe central nervous system H. influenzae infection presenting to the same hospital in San Diego, California during the same week in January 2016. Due to good vaccine coverage in this part of the world, H. influenzae cases are normally rare and seeing two cases in the same week was unexpected. We thus suspected a recent transmission chain, and possible local outbreak. To test this hypothesis, we isolated and sequenced whole genomes from each patient and placed them in a phylogenetic tree spanning the known diversity of H. influenzae. Surprisingly, we found that the two isolates (SD2016_1 and SD2016_2) belonged to distantly related lineages, suggesting two independent transmission events and ruling out a local outbreak. Despite being distantly related, the two isolates belong to two different lineages that have exchanged capsule loci in the recent past. Therefore, as in other bacterial pathogens, capsule switching by horizontal gene transfer may be an important evolutionary mechanism of vaccine evasion in H. influenzae.

Haemophilus influenzae causes common and sometimes severe adult and pediatric disease including chronic obstructive respiratory disease, otitis media and infections of the central nervous system. Serotype b strains, with a b-type capsule, have been the historical cause of invasive disease, and the introduction of a serotype b-specific vaccine has led to their decline. However, unencapsulated or non-b-type H. influenzae infections are not prevented by the vaccine and appear to be increasing in frequency. Here we report two pediatric cases of severe central nervous system H. influenzae infection presenting to the same hospital in San Diego, California during the same week in January 2016. Due to good vaccine coverage in this part of the world, H. influenzae cases are normally rare and seeing two cases in the same week was unexpected. We thus suspected a recent transmission chain, and possible local outbreak. To test this hypothesis, we isolated and sequenced whole genomes from each patient and placed them in a phylogenetic tree spanning the known diversity of H. influenzae . Surprisingly, we found that the two isolates (SD2016_1 and SD2016_2) belonged to distantly related lineages, suggesting two independent transmission events and ruling out a local outbreak. Despite being distantly related, the two isolates belong to two different lineages that have exchanged capsule loci in the recent past. Therefore, as in other bacterial pathogens, capsule switching by horizontal gene transfer may be an important evolutionary mechanism of vaccine evasion in H. influenzae .

Genome recombination is a major source of genotypic diversity and contributes to adaptation and speciation following interspecies hybridization. The contribution of recombination in these processes has been thought to be largely limited to the nuclear genome because organelles are mostly uniparentally inherited in animals and plants, which prevents recombination. Unicellular eukaryotes such as budding yeasts do, however, transmit mitochondria biparentally, suggesting that during hybridization, both parents could provide alleles that contribute to mitochondrial functions such as respiration and metabolism in hybrid populations or hybrid species. We examined the dynamics of mitochondrial genome transmission and evolution during speciation by hybridization in the natural budding yeast Saccharomyces paradoxus. Using population-scale mitochondrial genome sequencing in two endemic North American incipient species SpB and SpC and their hybrid species SpC*, we found that both parental species contributed to the hybrid mitochondrial genome through recombination. We support our findings by showing that mitochondrial recombination between parental types is frequent in experimental crosses that recreate the early step of this speciation event. In these artificial hybrids, we observed that mitochondrial genome recombination enhances phenotypic variation among diploid hybrids, suggesting that it could play a role in the phenotypic differentiation of hybrid species. Like the nuclear genome, the mitochondrial genome can, therefore, also play a role in hybrid speciation.

H. influenzaecauses common and sometimes severe pediatric disease including chronic obstructive respiratory disease, otitis media, and infections of the central nervous system. Serotype b strains, with a b-type capsule, have been the historical cause of invasive disease, and the introduction of a serotype b-specific vaccine has led to their decline. However, unencapsulated or non-b-typeH. influenzaeinfections are not prevented by the vaccine and appear to be increasing in frequency. Here we report two pediatric cases of severe central nervous systemH. influenzaeinfection presenting to the same hospital in San Diego, California during the same week in January 2016. Due to good vaccine coverage in this part of the world,H. influenzaecases are normally rare and seeing two cases in the same week was unexpected. We thus suspected a recent transmission chain, and possible local outbreak. To test this hypothesis, we isolated and sequenced whole genomes from each patient and placed them in a phylogenetic tree spanning the known diversity ofH. influenzae. Surprisingly, we found that the two isolates (H1 and H2) belonged to distantly related lineages, suggesting two independent transmission events and ruling out a local outbreak. Despite being distantly related, H1 and H2 belong to two different lineages that appear to engage in frequent horizontal gene transfer (HGT), suggesting overlapping ecological niches. Together, our comparative genomic analysis supports a scenario in which an f-type ancestor of H2 arrived in North America around 2011 and acquired an a-type capsule by recombination (HGT) with a recent ancestor of H1. Therefore, as in other bacterial pathogens, capsule switching by HGT may be an important evolutionary mechanism of vaccine evasion inH. influenzae.OUTCOMETwo cases of severe central nervous systemH. influenzaeinfection occurred during the same week in the same hospital in San Diego, California – a region where such infections are usually very rare due to vaccine coverage. We thus suspected a local outbreak of anH. influenzaeclone not covered by the vaccine. Using whole genome sequencing and phylogenetic analysis of two isolates (H1 and H2, one from each patient), we found that they were distantly related, rapidly ruling out a local outbreak and suggesting independent transmission events. This result highlights the potential for rapid global spread of non-vaccineH. influenzaestrains. In this case, both H1 and H2 both encoded a-type capsules, whereas the vaccine targets b-type capsules. We also present comparative genomic evidence that a recent f-type ancestor of H2 acquired an a-type capsule locus from a recent ancestor of H1, and that this horizontal gene transfer (HGT) event likely happened in the past decade in North America, but probably not in the San Diego hospital. These results highlight the potential importance of HGT in the capsule locus in allowingH. influenzaeto escape vaccine coverage.DATA SUMMARYH. influenzaeH1 and H2 genome sequences have been deposited in NCBI under BioProject PRJNA534512.CONFLICT OF INTEREST STATEMENTThe authors declare that they have no conflict of interest to report.

Cholera is a severe, water-borne diarrhoeal disease caused by toxin-producing strains of the bacterium Vibrio cholerae. Comparative genomics has revealed ‘waves’ of cholera transmission and evolution, in which clones are successively replaced over decades and centuries. However, the extent of V. cholerae genetic diversity within an epidemic or even within an individual patient is poorly understood. Here, we characterized V. cholerae genomic diversity at a micro-epidemiological level within and between individual patients from Bangladesh and Haiti. To capture within-patient diversity, we isolated multiple (8 to 20) V. cholerae colonies from each of eight patients, sequenced their genomes and identified point mutations and gene gain/loss events. We found limited but detectable diversity at the level of point mutations within hosts (zero to three single nucleotide variants within each patient), and comparatively higher gene content variation within hosts (at least one gain/loss event per patient, and up to 103 events in one patient). Much of the gene content variation appeared to be due to gain and loss of phage and plasmids within the V. cholerae population, with occasional exchanges between V. cholerae and other members of the gut microbiota. We also show that certain intra-host variants have phenotypic consequences. For example, the acquisition of a Bacteroides plasmid and non-synonymous mutations in a sensor histidine kinase gene both reduced biofilm formation, an important trait for environmental survival. Together, our results show that V. cholerae is measurably evolving within patients, with possible implications for disease outcomes and transmission dynamics.

Cholera is a severe, waterborne diarrheal disease caused by toxin-producing strains of the bacteriumVibrio cholerae. Comparative genomics has revealed “waves” of cholera transmission and evolution, in which clones are successively replaced over decades and centuries. However, the extent ofV. choleraegenetic diversity within an epidemic or even within an individual patient is poorly understood. Here, we characterizedV. choleraegenomic diversity at a micro-epidemiological level within and between individual patients from Bangladesh and Haiti. To capture within-patient diversity, we isolated multiple (8 to 20)V. choleraecolonies from each of eight patients, sequenced their genomes and identified point mutations and gene gain/loss events. We found limited but detectable diversity at the level of point mutations within hosts (zero to three single nucleotide variants within each patient), and comparatively higher gene content variation within hosts (at least one gain/loss event per patient, and up to 103 events in one patient). Much of the gene content variation appeared to be due to gain and loss of phage and plasmids within theV. choleraepopulation, with occasional exchanges betweenV. choleraeand other members of the gut microbiota. We also show that certain intra-host variants have phenotypic consequences. For example, the acquisition of aBacteroidesplasmid and nonsynonymous mutations in a sensor histidine kinase gene both reduced biofilm formation, an important trait for environmental survival. Together, our results show thatV. choleraeis measurably evolving within patients, with possible implications for disease outcomes and transmission dynamics.Author SummaryVibrio choleraeis the etiological agent of cholera, a severe diarrheal disease endemic to Bangladesh and responsible for global outbreaks, including one ongoing in Haiti. Certain bacterial pathogens can evolve and diversify within the human host, often altering virulence and antibiotic resistance. However, most examples of within-host evolution have come from chronic infections, in which the pathogen has sufficient time to mutate and diversify, and little attention has been paid to more acute infections such as the one caused byV. cholerae. The goal of this study was to measure the extent of within-host evolution ofV. choleraewithin individual infected patients. By sequencing multiple bacterial isolates_from each of eight patients from Bangladesh and Haiti, we found that cholera patients can harbor a diverse population ofV. cholerae. As expected for an acute infection, this diversity is limited, ranging from zero to three point mutations (single nucleotide variants) per patient. However, gene gain/loss events are more prevalent than point mutations, occurring in every single patient, and sometimes involving the transfer of dozens of genes on plasmids. Even if rare, point mutations and gene gain/loss events may be maintained by natural selection, and can alter clinically-and environmentally-relevant phenotypes such as biofilm formation. Therefore, within-patient evolution has the potential to impact clinical and epidemiological outcomes. Together, our results demonstrate that within-patient evolution may be a general feature of both acute and chronic infections, and that gene gain/loss may be an important but under-appreciated feature of within-host evolution.

The Athabasca oil sand deposit is one of the largest single oil deposits in the world. Following surface mining, companies are required to restore soil-like profiles that can support the previous land capabilities. The objective of this study was to assess whether the soil prokaryotic alpha diversity (α-diversity) and β-diversity in oil sand soils reconstructed 20 to 30 years previously and planted to one of three vegetation types (coniferous or deciduous trees and grassland) were similar to those found in natural boreal forest soils subject to wildfire disturbance. Prokaryotic α-diversity and β-diversity were assessed using massively parallel sequencing of 16S rRNA genes. The β-diversity, but not the α-diversity, differed between reconstructed and natural soils. Bacteria associated with an oligotrophic lifestyle were more abundant in natural forest soils, whereas bacteria associated with a copiotrophic lifestyle were more abundant in reconstructed soils. Ammonia-oxidizing archaea were most abundant in reconstructed soils planted with grasses. Plant species were the main factor influencing α-diversity in natural and in reconstructed soils. Nitrogen deposition, pH, and plant species were the main factors influencing the β-diversity of the prokaryotic communities in natural and reconstructed soils. The results highlight the importance of nitrogen deposition and aboveground-belowground relationships in shaping soil microbial communities in natural and reconstructed soils. IMPORTANCE Covering over 800 km 2 , land disturbed by the exploitation of the oil sands in Canada has to be restored. Here, we take advantage of the proximity between these reconstructed ecosystems and the boreal forest surrounding the oil sand mining area to study soil microbial community structure and processes in both natural and nonnatural environments. By identifying key characteristics shaping the structure of soil microbial communities, this study improved our understanding of how vegetation, soil characteristics and microbial communities interact and drive soil functions.

Cyanobacterial blooms occur in lakes worldwide, producing toxins that pose a serious public health threat. Eutrophication caused by human activities and warmer temperatures both contribute to blooms, but it is still difficult to predict precisely when and where blooms will occur. One reason that prediction is so difficult is that blooms can be caused by different species or genera of cyanobacteria, which may interact with other bacteria and respond to a variety of environmental cues. Here we used a deep 16S amplicon sequencing approach to profile the bacterial community in eutrophic Lake Champlain over time, to characterize the composition and repeatability of cyanobacterial blooms, and to determine the potential for blooms to be predicted based on time-course sequence data. Our analysis, based on 135 samples between 2006 and 2013, spans multiple bloom events. We found that bloom events significantly alter the bacterial community without reducing overall diversity, suggesting that a distinct microbial community -- including non-cyanobacteria -- prospers during the bloom. We also observed that the community changes cyclically over the course of a year, with a repeatable pattern from year to year. This suggests that, in principle, bloom events are predictable. We used probabilistic assemblages of OTUs to characterize the bloom-associated community, and to classify samples into bloom or non-bloom categories, achieving up to 92% classification accuracy (86% after excluding cyanobacterial sequences). Finally, using symbolic regression, we were able to predict the start date of a bloom with 78-92% accuracy (depending on the data used for model training), and found that sequence data was a better predictor than environmental variables.

International audience; Cone snail venoms are considered an untapped reservoir of extremely diverse peptides, named conopeptides, displaying a wide array of pharmacological activities. We report here for the first time, the presence of high molecular weight compounds that participate in the envenomation cocktail used by these marine snails. Using a combination of proteomic and transcriptomic approaches, we identified glycosyl hydrolase proteins, of the hyaluronidase type (Hyal), from the dissected and injectable venoms ("injectable venom" stands for the venom variety obtained by milking of the snails. This is in contrast to the "dissected venom", which was obtained from dissected snails by extraction of the venom glands) of a fish-hunting cone snail, Conus consors (Pionoconus clade). The major Hyal isoform, Conohyal-Cn1, is expressed as a mixture of numerous glycosylated proteins in the 50 kDa molecular mass range, as observed in 2D gel and mass spectrometry analyses. Further proteomic analysis and venom duct mRNA sequencing allowed full sequence determination. Additionally, unambiguous segment location of at least three glycosylation sites could be determined, with glycans corresponding to multiple hexose (Hex) and N-acetylhexosamine (HexNAc) moieties. With respect to other known Hyals, Conohyal-Cn1 clearly belongs to the hydrolase-type of Hyals, with strictly conserved consensus catalytic donor and positioning residues. Potent biological activity of the native Conohyals could be confirmed in degrading hyaluronic acid. A similar Hyal sequence was also found in the venom duct transcriptome of C. adamsonii (Textilia clade), implying a possible widespread recruitment of this enzyme family in fish-hunting cone snail venoms. These results provide the first detailed Hyal sequence characterized from a cone snail venom, and to a larger extent in the Mollusca phylum, thus extending our knowledge on this protein family and its evolutionary selection in marine snail venoms.

Natural compounds have been utilised for medicinal purposes since ancient times. The toxic secretions of venomous animals have evolved over millions of years of evolutionary time to induce paralysis and systematic breakdown of homeostasis in prey, in addition to aiding in defence against predators. In the process, a remarkable arsenal of biochemical components has been developed, including toxins with astonishing specificity and affinity for various types of cells and receptors, many of which participate in essential physiological pathways. Hence, venoms and toxins have been perceived as rich sources of novel biochemical compounds for use in drug design and development. Although biased towards a narrow range of taxa, modern toxinological research has focussed on utilising these fascinating biochemical compounds not only in elucidating the intricate mechanisms of life but also in aiding the development of life-saving pharmaceutical drugs. Understanding the complex evolutionary mechanisms shaping different venom components is essential for the development of highly effective drugs. A multidisciplinary approach, spanning methods such as transcriptomics, proteomics, bioactivity testing and molecular evolution analyses, has the potential to unravel the true biodiversity of toxins and provide invaluable information for drug design and development.

Although snake venoms have been the subject of intense research, primarily because of their tremendous potential as a bioresource for design and development of therapeutic compounds, some specific groups of snakes, such as the genus Atractaspis, have been completely neglected. To date only limited number of toxins, such as sarafotoxins have been well characterized from this lineage. In order to investigate the molecular diversity of venom from Atractaspis aterrima—the slender burrowing asp, we utilized a high-throughput transcriptomic approach completed with an original bioinformatics analysis pipeline. Surprisingly, we found that Sarafotoxins do not constitute the major ingredient of the transcriptomic cocktail, rather a large number of previously well-characterized snake venom-components were identified. Notably, we recovered a large diversity of three-finger toxins (3FTxs), which were found to have evolved under the significant influence of positive selection. From the normalized and non-normalized transcriptome libraries, we were able to evaluate the relative abundance of the different toxin groups, uncover rare transcripts, and gain new insight into the transcriptomic machinery. In addition to previously characterized toxin families, we were able to detect numerous highly-transcribed compounds that possess all the key features of venom-components and may constitute new classes of toxins.

A response to Novel venom gene discovery in the platypus by Whittington CM, Papenfuss AT, Locke DP, Mardis ER, Wilson RK, Abubucker S, Mitreva M, Wong ESW, Hsu AL, Kuchel PW, Belov K, Warren WC. Genome Biol 2010, 11:R95.

Comparative mitochondrial genomics of arbuscular mycorrhizal fungi (AMF) provide new avenues to overcome long-lasting obstacles that have hampered studies aimed at understanding the community structure, diversity, and evolution of these multinucleated and genetically polymorphic organisms. AMF mitochondrial (mt) genomes are homogeneous within isolates, and their intergenic regions harbor numerous mobile elements that have rapidly diverged, including homing endonuclease genes, small inverted repeats, and plasmid-related DNA polymerase genes (dpo), making them suitable targets for the development of reliable strain-specific markers. However, these elements may also lead to genome rearrangements through homologous recombination, although this has never previously been reported in this group of obligate symbiotic fungi. To investigate whether such rearrangements are present and caused by mobile elements in AMF, the mitochondrial genomes from two Glomeraceae members (i.e., Glomus cerebriforme and Glomus sp.) with substantial mtDNA synteny divergence, were sequenced and compared with available glomeromycotan mitochondrial genomes. We used an extensive nucleotide/protein similarity network-based approach to investigate dpo diversity in AMF as well as in other organisms for which sequences are publicly available. We provide strong evidence of dpo-induced inter-haplotype recombination, leading to a reshuffled mitochondrial genome in Glomus sp. These findings raise questions as to whether AMF single spore cultivations artificially underestimate mtDNA genetic diversity. We assessed potential dpo dispersal mechanisms in AMF and inferred a robust phylogenetic relationship with plant mitochondrial plasmids. Along with other indirect evidence, our analyses indicate that members of the Glomeromycota phylum are potential donors of mitochondrial plasmids to plants.

International audience; Although cone snail venoms have been intensively investigated in the past few decades, little is known about the whole conopeptide and protein content in venom ducts, especially at the transcriptomic level. If most of the previous studies focusing on a limited number of sequences have contributed to a better understanding of conopeptide superfamilies, they did not give access to a complete panorama of a whole venom duct. Additionally, rare transcripts were usually not identified due to sampling effect. This work presents the data and analysis of a large number of sequences obtained from high throughput 454 sequencing technology using venom ducts of Conus consors, an Indo-Pacific living piscivorous cone snail. A total of 213,561 Expressed Sequence Tags (ESTs) with an average read length of 218 base pairs (bp) have been obtained. These reads were assembled into 65,536 contiguous DNA sequences (contigs) then into 5039 clusters. The data revealed 11 conopeptide superfamilies representing a total of 53 new isoforms (full length or nearly full-length sequences). Considerable isoform diversity and major differences in transcription level could be noted between superfamilies. A, O and M superfamilies are the most diverse. The A family isoforms account for more than 70% of the conopeptide cocktail (considering all ESTs before clustering step). In addition to traditional superfamilies and families, minor transcripts including both cysteine free and cysteine-rich peptides could be detected, some of them figuring new clades of conopeptides. Finally, several sets of transcripts corresponding to proteins commonly recruited in venom function could be identified for the first time in cone snail venom duct. This work provides one of the first large-scale EST project for a cone snail venom duct using next-generation sequencing, allowing a detailed overview of the venom duct transcripts. This leads to an expanded definition of the overall cone snail venom duct transcriptomic activity, which goes beyond the cysteine-rich conopeptides. For instance, this study enabled to detect proteins involved in common post-translational maturation and folding, and to reveal compounds classically involved in hemolysis and mechanical penetration of the venom into the prey. Further comparison with proteomic and genomic data will lead to a better understanding of conopeptides diversity and the underlying mechanisms involved in conopeptide evolution.

Crustacean species have not been examined in great detail for their transposable elements content. Here we focus on galatheid crabs, which are one of the most diverse and widespread taxonomic groups of Decapoda. Ty1/copia retrotransposons are a diverse and taxonomically dispersed group. Using degenerate primers, we isolated several DNA fragments that show homology with Ty1/copia retroelements reverse transcriptase gene. We named the corresponding elements from which they originated GalEa1 to GalEa3 and analyzed one of them further by isolating various clones containing segments of GalEa1. This is the first LTR retrotransposon described in crustacean genome. Nucleotide sequencing of the clones revealed that GalEa1 has LTRs (124 bp) and that the internal sequence (4,421 bp) includes a single large ORF containing gag and pol regions. Further screening identified highly related elements in six of the nine galatheid species studied. By performing BLAST searches on genome databases, we could also identify GalEa-like elements in some fishes and Urochordata genomes. These elements define a new clade of Ty1/copia retrotransposons that differs from all other Ty1/copia elements and that seems to be restricted to aquatic species.

Cyanobacterial blooms occur in lakes worldwide, producing toxins that pose a serious public health threat. Eutrophication caused by human activities and warmer temperatures both contribute to blooms, but it is still difficult to predict precisely when and where blooms will occur. One reason that prediction is so difficult is that blooms can be caused by different species or genera of cyanobacteria, which may interact with other bacteria and respond to a variety of environmental cues. Here we used a deep 16S amplicon sequencing approach to profile the bacterial community in eutrophic Lake Champlain over time, to characterise the composition and repeatability of cyanobacterial blooms, and to determine the potential for blooms to be predicted based on time course sequence data. Our analysis, based on 135 samples between 2006 and 2013, spans multiple bloom events. We found that bloom events significantly alter the bacterial community without reducing overall diversity, suggesting that a distinct microbial community—including non-cyanobacteria—prospers during the bloom. We also observed that the community changes cyclically over the course of a year, with a repeatable pattern from year to year. This suggests that, in principle, bloom events are predictable. We used probabilistic assemblages of OTUs to characterise the bloom-associated community, and to classify samples into bloom or non-bloom categories, achieving up to 92% classification accuracy (86% after excluding cyanobacterial sequences). Finally, using symbolic regression, we were able to predict the start date of a bloom with 78–92% accuracy (depending on the data used for model training), and found that sequence data was a better predictor than environmental variables.