High impact of bacterial predation on cyanobacteria in soil biocrusts

Tracing the symptomology of predation through macroscopic plaques

A culture bioassay (Expanded Microcoleus Mortality Assay, or EMMA) (Fig. 1 and see Materials and Methods) based on the capacity of a soil to induce complete mortality in the foundational biocrust cyanobacterium Microcoleus vaginatus helped us trace the pathogen detected in biocrust production facilities to the development of cm-sized plaques, or zones of cyanobacterial clearing, in natural biocrusts. These plaques were revealed to the naked eye (Fig. 2) when the soil was wet (i.e., after a rain event), as impacted areas would fail to green up by the migration of cyanobacteria to the surface²¹, enabling us to detect and quantify them with relative ease. Soil samples obtained from such plaques (n = 30) from different sites (n = 6; Table S1) in the US Southwest were invariably EMMA + , and the pathogens always filterable with pore sizes 0.45–1 µm but not larger, and always insensitive to the eukaryotic inhibitor cycloheximide, indicating the agent’s prokaryotic nature and small size, while paired samples from asymptomatic areas just outside the plaques were always EMMA- (Table S2). These end-point EMMA solutions never gave rise to cyanobacterial re-growth upon further incubation and maintained its infectivity of fresh cyanobacterial cultures for up to 6 months. A one-time, small-scale sampling across a plaque at intervals of 2 mm using microcoring²² showed that the boundary of the visible plaque demarcated exactly the end of infectivity, samples 0–2 mm outside the plaque proving non-infective. Further, inoculation of healthy, natural biocrusts with EMMA + suspensions resulted in the local development of biocrust plaques, and soil from these plaques was itself EMMA + , in partial fulfillment of Koch’s postulates. Yet, standard microbiological plating failed to yield any isolates that were EMMA + (we tested 30 unique isolates), even though standard plating with similar isolation efforts can successfully cultivate a large portion of heterotrophs from biocrusts²³.

Cultivation, identification, and salient genomic traits of the cyanobacterial pathogen

To study these organisms, we turned to enrichment of pathogen/prey co-cultures based on repeated passages through EMMA and differential size filtration combined with dilution-to-extinction approaches, followed by purification with flow cytometry/cell sorting. The process was monitored by 16S rRNA gene amplicon sequencing, and eventually yielded a highly enriched co-culture of the cyanobacterium with a genetically homogenous (one single Amplicon Sequence Variant) population that made up more than 80% of reads (Fig. 3 a, b). We name the organism represented by this ASV Candidatus Cyanoraptor togatus. That it corresponds indeed to the predator is supported by the fact that of the 17 ASV’s detected in the final enrichment, only 10 were consistently detected at all infectious stages in the process and, among these, only our candidate ASV steadily increased in relative abundance through the enrichment process (Fig. 3 a, b). This final enrichment of C. togatus, LGM-1, constitutes the basis for downstream biological and molecular analyses. Its ASV was most similar to little-known members of the family Chitinophagaceae in the phylum Bacteroidetes. LGM-1’s genome was sequenced and assembled into a single 3.3 Mb contig with 1,781 putative and 1,328 hypothetical genes (Table S3), though most proteins had low identity (<70%) to their homologs in Chitinophaga pinensis, the nearest relative with a fully sequenced genome, although not particularly closely related to it. The 16S rRNA gene sequence from LGM-1’s two identical copies, was only 90% similar to that of its closest isolate in culture according to BLAST (version BLAST + 2.11.0) searches. This sequence was used to narrow LGM-1’s phylogenetic placement (Fig. 3c) indicating its affinity with members of the Chitinophagaceae, but basally so, and distinct enough to make a definite assignment at the family level uncertain. Notably, no other reported predatory prokaryotes are related to LGM-1, although at least in one case, a fungal endosymbiont has been reported that shows phylogenetic affinities to the genus Chitinophaga²⁴. LGM-1 represents yet a new branch for the notoriously polyphyletic guild of predatory prokaryotes. Some genomic comparisons to other predatory bacteria can be made (Table S3). Unlike obligate symbionts, predatory bacteria undergo no reduction in genome size²⁵, possibly due to their reliance on an extracellular stage. Cyanoraptor’s genome size, if somewhat smaller, is not atypical of that of members of the Chitinophagaceae, which cluster around 4.5 Mb. Many bacterial predators lack complete amino acid biosynthetic pathways²⁶. Similarly, Cyanoraptor only has full biosynthetic pathways for glutamine and asparagine. This, together with our inability to obtain prey-independent cultures, suggests that Cyanoraptor is obligately predatory. Further, predators tend to contain a wide breadth of hydrolytic enzymes and no quorum sensing genes²⁶. Fitting this pattern, 3% of Cyanoraptor’s genes were assignable to hydrolases, and it also lacked quorum sensing genes. Finally, a common genomic trait of predatory bacteria is the presence of the mevalonate pathway for isoprenoid biosynthesis, which is rarely found in bacteria, but common in Eukaryotes. It has been suggested that predators are able to scavenge acetyl-coA, the initial molecule in the pathway, rather than using the energy expensive process to synthesize it from pyruvate and G3P. However, Cyanoraptor deviates from this and only contains the common bacterial pathway. Cyanoraptor is apparently non-motile in any of its life stages, lacking visible flagella, and lacking genes ascribable by similarity to known motility functions (flagellar or gliding) or taxes in its genome. This is unlike all other known predatory prokaryotes.

Molecular detection and abundance of C. togatus

LGM-1’s 16S rRNA sequences also allowed us to probe biocrust communities for its presence in plaques and healthy crusts surrounding them. Candidatus Cyanoraptor togatus or ASVs closely allied to it, were found on all plaques tested (n = 15 from 7 different sites), although we could also detect it molecularly in apparently healthy areas surrounding them (n = 12, from 4 sites tested; Table S4), albeit in significantly lower proportions (Fig. 4). Given the lack of infectivity in such areas, as discussed above, we must ascribe this detection to the presence of Cyanoraptor’s relic DNA, a phenomenon that is typical of soil environments²⁷. Overall, however, Cyanoraptor-like ASVs were never dominant (<3.6% of reads; Table S5), even within plaques. Similarly, we could detect the presence of Cyanoraptor-like sequences in archived data from all available molecular surveys of biocrusts in many geographical areas beyond our physical survey (Table S6), indicating that its incidence is probably global. In these, Cyanoraptor-like sequences were even rarer, perhaps expectedly, because the datasets were not designed to capture plaques, but rather to survey biocrust diversity at large.

Life cycle and ultrastructural features

We studied Cyanoraptor’s life cycle by following the infection dynamics within typical EMMA through transmission electron and confocal microscopy. End-point EMMAs invariably revealed the absence of healthy filaments of Microcoleus (only carcass-like ghost filaments lacking cytoplasm could be seen; Fig. 1b), and a large number of small (0.8–1 µm diam), clearly internally compartmentalized, non-flagellated, Gram-negative cocci (Fig. 5a–d), which we interpret as Cyanoraptor’s extracellular propagules because of their (i) small size, (ii) high numbers and (iii) proximity to dead cyanobacteria. No dividing cells were detected among these propagules (n = 250; surveyed from TEM photographs). The inner, electron-dense compartment of the propagules contained a typical fibrillar nucleoid (Fig. 5b, Fig. S1) and was circumscribed by two membranes separated by an electron transparent region, as is typical of a gram-negative tegument (Fig. 5c). Two-membrane internal compartmentalization in bacteria was known previously only in the planctomycete Gemmata obscuriglobus²⁸. The outer compartment, or “toga”, 0.1–0.4 µm thick, was electron-light and separated from the extracellular space also by two membranes with a clear interspace (Fig. 5c). Upon infection of fresh prey, the cocci in close proximity to M. vaginatus developed tegumentary structures reminiscent of docking zones (Fig. 5d), and as infection proceeded, predatory cells gained entry into the cyanobacterial cytoplasm. Once inside, they lost all trace of compartmentalization, starting to grow into pseudo-filamentous forms (Figs. 5e and 6), with large numbers of cytoplasmatic inclusions, as M. vaginatus cells were degraded showing patent loss of macromolecular structures typical of healthy cells like thylakoids and carboxysomes²⁹, bulging cellular teguments due to loss of cell wall strength, and loss of cytoplasmatic contents (Fig. 5e, h). The damage, and the infection, often spread to multiple cells adjacent to the entry cell in the cyanobacterial filament (Fig. 5h). Multiple cell division occurred during Cyanoraptor’s intracellular phase only at the late pseudo-filamentous stage (Fig. 5f), accompanied by a loss of reserve polymers, the formation of a cocoon of fibrillar nature around it (Figs. 5f and 7), and by the excretion of large numbers of 10–20 nm sized membrane-bound extracellular vesicles (Figs. 5g and 8). After full degradation of M. vaginatus cells, propagules were released (Fig. 5j, k), remaining held in groups. We posit that the outer compartment acts as a repository for hydrolytic enzymes destined for the prey, a common strategy in predatory bacteria³⁰; that most of LGM-1’s genes (3%) annotated as polymer hydrolases are endowed with signal peptides for excretion supports this contention. This outer compartment seems to be formed by fusion of the extracellular vesicles trapped between cells and fibrillar cocoon. Given its lack of motility, it would appear that Cyanoraptor uses a strategy of ambush predation that likely relies on cyanobacterial motility for encounters, docking, and possibly dispersal. This is consistent with the prey range established in the lab, where all cyanobacterial strains sensitive to it are motile by gliding (Table S7). Thus, Cyanoraptor appears to be an obligatory, endocellular predatory bacterium, the first of this type as a predator of cyanobacteria, although a variety of Proteobacteria, Bacteroidetes and Firmicutes can lyse cyanobacterial cells extracellularly³¹.

Prey specificity in culture and in the field

Out of 70 cyanobacterial strains (in 14 genera) tested using EMMA but substituting the prey, only 14 (in 4 genera) were susceptible to attack (Table S7), all filamentous and non-heterocystous, and typical pioneer biocrusts formers. The common denominator of susceptible strains was their ability to form crowded, large bundles of filaments and to move by gliding motility, consistent with basic epidemiological principles, since this will promote prey mortality and contagion, respectively³². Comparisons of 16S rRNA-based community composition complemented with absolute quantification via quantitative PCR in paired samples (inside vs. outside of plaques), showed that field epidemics involved a significant population decline (p = 0.002; Wilcoxon paired ratio test; Fig. 4, Table S4) in cyanobacteria. While there was a tendency to cause absolute increases in other bacteria (i.e., non-cyanobacteria) this was not statistically significant (Fig. 4). No other single bacterial phylum showed significant absolute benefits or declines from the epidemic. Among the 29 cyanobacterial taxa identified in our field dataset, only Microcoleus vaginatus, Allocoleopsis sp., Potamolinea sp. and Xeronema sp. suffered demonstrably significant and consistent loses (p < 0.05; Table S8), in line with the prey range demonstrated in culture. The affinity of Cyanoraptor for M. vaginatus, likely the most abundant terrestrial cyanobacterium³³, and to some of members of the Coleofasciculaceae, of wide distribution across continents³⁴, affords it a veritable worldwide buffet.

Functional consequences of C. togatus epidemics

The ecological consequences of Cyanoraptor epidemics were assessed by comparing how relevant biocrust properties were affected across plaque boundaries. A result compilation of the tests carried out is in Fig. 4 and the full sets of data and statistics are in Tables S9 and S10. The most severe effect was on net primary productivity, which was invariably and fully obliterated in crossing from healthy biocrusts into plaques, turning them into net respiratory systems, the inflexion point coinciding spatially with the plaque boundary (Fig. 9a, b). Functional effects on oxygenic photosynthesis were much more severe than one could have surmised from losses of cyanobacterial 16S rRNA genes, again indicating the likelihood of relic cyanobacterial DNA blurring the full extent of morbidity by Cyanoraptor. Total Organic Carbon (TOC) and Total Organic Nitrogen (TON) content were also consistently lowered (by 13% and 38%, in average, respectively), concurrently increasing soil levels of Dissolved Inorganic Nitrogen (DIN) by 300% in average, which in absolute terms was roughly commensurate with TON loses (Table S10). The biocrust content of extracellular polysaccharide (EPS), responsible for some of the hydrological and dust-trapping character of biocrusts³⁵, was also consistently reduced in plaques (by 53% in average), possibly as a consequence of the release of DIN, which would render EPS a better substrate for growth for heterotrophs when N is available. Not surprisingly, important functional properties of biocrusts like moisture retention capacity during desiccation, and dust trapping ability were also negatively impacted (Figs. 4, 9c), by 67 and 36% respectively. It can perhaps be surprising that a bacterium that is relatively rare can elicit such devastating, cascading effects. However, Cyanoraptor’s feeding seems to be quite inefficient, destroying much more than it can reap, and making cyanobacterial biomass available to a range of other adventitious bacteria. A rough estimate of this biomass transfer efficiency from microscopic images points to values well below 1% (Table S11).

Scaling impacts to the ecosystem level

All the effects discussed above pertain to concentrated epidemics that span scales of cm. However, scaling such effects to the ecosystem level requires an assessment of the incidence and distribution of individual plaques at much larger scales. While plaques were found in all sites inspected, to scale-up more quantitatively, we carried out photographic field surveys of the incidence of plaques larger than 3 mm in 195, 1-m² quadrats along 9 linear transects at 3 geographical locations, chosen for their high biocrust cover and carried out during rainy days to make plaques conspicuous (Fig. 10, Supplementary Data 1). Plaque density in single transects ranged from 1 to 23 m⁻², averaging 9.0 ± 8.6 m⁻², and affecting 8.3 ± 14.3% of the biocrust area surveyed. Single quadrats ranged from 0 to 263 plaques m⁻² and from 0 to 98% areal infection. Plaques strongly followed aggregated distributions across scales (Fig. 10), even within quadrats (Nearest Neighbor test; p = 0.01). Scaling single-plaque functional effects on an area basis by simple arithmetic, it follows that the ecosystem-level consequences of Cyanoraptor infections are also significant, tithing primary productivity in the order of 10%. This must be considered an underestimate, as only fully formed plaques larger than 3 mm were counted, and Cyanoraptor was detected molecularly also in areas without conspicuous plaques.

Bacterial predation as loss factor and consequences for soil restoration

Since losses to viral infection or protistan/meiofaunal grazing remain to be quantified in biocrusts, prokaryotic predation stands, for now, as the single most important biological loss factor in these microbiomes. In the light of our work, prokaryotic predation should be considered, along with other biological loss factors, as a potentially important determinant of population dynamics in microbiomes. Further, biocrust restoration practices that are based on using whole communities for production of inoculum run the risk of spreading infection and should be performed with caution. Since our original report of catastrophic failures due to cyanobacterial mortality¹⁸, we have been able to successfully avoid disease spread by simply testing for diagnostic plaques in the starting material, and discarding any containing plaques in the downstream production.

Description of Candidatus Cyanoraptor togatus, genus novum et species nova

Candidatus Cyanoraptor, gen. nov., Cy.a.no.rap’.tor, Latinized Gr. m.n. Cyanos, blue-green and L. raptor m.n, plunderer, M.L. Cyanoraptor m. n a plunderer of the blue-green (cyanobacteria).

Prey-dependent intracellular Gram-negative predatory bacteria in the Bacteroidetes with affinities to the family Chitinophagaceae, showing conspicuous cell differentiation into distinct intra- and extracellular stages in its life cycle. Extracellular propagules are non-dividing, coccoid, internally compartmentalized by a double membrane. The inner compartment holds the nucleoid. Intracellular stages lack an outer compartment, grow into rods and eventually pseudo-filaments, to undergo simultaneous, multiple cell division. No motile phases exist. Flagella are absent. Likely auxotrophic for many amino acids.

Candidatus Cyanoraptor togatus. sp. nova. To.ga´.tus, L. togatus m. adj., dressed in a robe or toga, in reference to the outer compartment in the propagules.

Extracellular cocci 0.93 ± 0.15 µm, intracellular cells 1.02 ± 0.41 µm wide and more than 5 µm long. Preys on non-heterocystous, motile filamentous terrestrial cyanobacteria. Strain LGM-1^T, in enrichment form, and its genome, are the type material. Its genome is 3.3 Mb with a 42% G + C content. Isolated from biological soil crust in Arizona, USA. Maintained as a co-culture with M. vaginatus PCC 9802 as prey.

Source: Ecology - nature.com