Selective cell lysis pressure on rare and abundant prokaryotic taxa across a shelf-to-slope continuum in the Northern South China Sea

ABSTRACT

Microbial growth and mortality are fundamental to community assembly and drive the elemental biogeochemical cycles in the ocean. Virus-induced host lysis contributes, on average, half of prokaryotic mortality and has a substantial effect on composition and diversity of marine microbes. Nevertheless, virus-mediated taxon-specific cell lysis is few studied to date. In the present study, we investigated the taxon-specific cell lysis and estimated its contribution to the variations of community composition in rare and abundant microbial taxa. The dominant taxa Prochlorococcus, Synechococcus, SAR11, and Rhodobacteraceae displayed lower cell lysis index (CLI, the rate of extracellular to intracellular rRNA) in surface seawater. Meanwhile, Alteromonas, Pseudomonas, and Halomonas had high CLI values in the bottom seawater. Cell lysis contributed a larger percentage of variation in rare taxa (5.0%–9.4%) than in abundant taxa (0.1%–1.7%). Furthermore, linear regression analysis indicated that rare taxa were more likely to experience higher viral lysis pressure relative to abundant taxa. Our findings provide insight into the impact of virus-mediated cell lysis on prokaryotic community structure and diversity and further improve our understanding of the various abiotic and biotic factors contributing to community assembly in the ocean.

IMPORTANCE

Virus-induced host lysis contributes up to 40% of total prokaryotic mortality and plays crucial roles in shaping microbial composition and diversity in the ocean. Nonetheless, what taxon-specific cell lysis is caused by viruses remains to be studied. The present study, therefore, examined the taxon-specific cell lysis and estimated its contribution to the variations in the rare and abundant microbial taxa. The results demonstrate that taxon-specific mortality differed in surface and bottom of the coastal environment. In addition, active rare taxa are more susceptible to heightened lytic pressure and suggested the importance of viral lysis in regulating the microbial community composition. These results improve our understanding of bottom-up (abiotic environmental variables) and top-down (viral lysis) controls contributing to microbial community assembly in the ocean.

INTRODUCTION

Microorganisms are present in large numbers, possess great diversity, and show expansive metabolic activity in the ocean, which drives the cycling of essential biogeochemical elements (1, 2). The composition of microbial communities is highly dynamic across different aquatic environments, and microbial growth and mortality are fundamental to the microbial food web by assimilating, transferring, and releasing nutrients within their environs (3). Viral lysis is thought as an essential factor in determining the composition and diversity of microbial communities in ecosystems, as it is estimated to be responsible for 10%–50% of total prokaryotic mortality, as well as up to 80% of heterotrophic production removal in the natural environments (2, 4, 5). On a global scale, virus-mediated host lysis is estimated to release 145 Gt C year⁻¹ in the tropical and subtropical oceans (6) and 0.37–0.63 Gt C year⁻¹ in the deep sea ecosystem (7). While the importance of viral lysis on an ecological scale is mostly based on the bulk estimates, its impacts on individual microbial species in natural environments have been studied to a lesser extent.

Environmental microbial communities primarily comprised a relatively small number of abundant taxa and a much larger proportion of rare species that serve as reservoirs or “seed banks” (6–8). Abundant microorganisms contribute to the majority of biomass and corresponding carbon cycling in the ocean (9). Nonetheless, rare taxa have been found to be more metabolically active and essential for maintaining microbial diversity (10–12). Some rare taxa are thought to play keystone roles in environmental-microbial interactions (8, 13). Here, the resource availability (bottom-up control) and viral lysis (top-down control), to some extent, shape distinct ecological strategies of abundant and rare taxa (11, 14, 15). The abundant taxa, referred to as “defense specialists” devote their resources and energy to be against viral infection or grazing and, thus, have relatively slow growth rates (11, 15). The rare taxa, also called “growth specialists”, may actually be the winners of the competition for limited resources with high growth rates, but they are extremely vulnerable to top-down control pressure due to lack of defensive strategies (11, 15, 16). The effects of environmental variables and viral lysis on variations in abundant and rare taxa remain yet to be fully explored.

Evaluating virus-induced cell lysis for each individual microbial taxon in the environment has been a conundrum for a long time. Recently, Zhong et al. proposed that sequencing extracellular 16S rRNA (rRNA_ext) primarily released by virus-mediated host mortality can provide insight into taxon-specific cell lysis in seawater (17). Microbial cells collected on the 0.22 µM filters could be used to investigate the total and active microbial community structure based on cellular 16S rDNA (rDNA_cell) and rRNA (rRNA_cell), respectively. The filtered seawater through 0.22 µM contains a great amount of rRNA_ext that was sourced from microbial cell lysis (17). The concentration of rRNA_ext could increase 100-fold in Vibrio sp. culture filtrate after adding its phage compared to without virus (17), while in the presence of grazers, the concentration of rRNA_ext decreased relative to that of control culture, suggesting that grazers consume rRNA_ext (17, 18). This implies that the release of bacterial rRNA is primarily caused by virus-mediated cell lysis, rather than the result of flagellate grazing. The rRNA_ext released by viral lysis is usually present in the form of ribosome complexes, which protect RNA from quick degradation by nucleases and make it relatively stable in seawater for a short period of time (17, 19). Thus, sequencing rRNA_ext can give an indication of the species that recently underwent cell lysis due to viral infection.

The northern South China Sea is a complex ecosystem that has been shaped by the influence of the Pearl River plume, coastal upwelling, and Kuroshio Current (20, 21). This system includes a broad shelf and a steep slope (22), making it an ideal transitional area between land and open ocean, with highly complicated microbial-environmental interactions (23). Understanding the contributions of environmental variables and virus-induced lysis can be critical in assessing microbial community composition, particularly for abundant and rare prokaryotic taxa. A hypothesis is proposed as rare and abundant taxa exhibit differential sensitivity to cell lysis and abiotic environment factors, respectively (17, 24). To figure this hypothesis out, the present study aims to (i) explore the taxon-specific cell lysis that occurs in the shelf-slope continuum using rRNA_ext sequencing, (ii) compare the cell lysis pressure on rare and abundant taxa, and (iii) evaluate the contribution of environmental variables and cell lysis to microbial community assembly in the northern South China Sea.

RESULTS

The environmental characteristics of the sampling region

All 11 stations were classified into three groups based on environmental and biological variables, nearshore (S01, S14, and S15), offshore shelf (S05, S05, S10, S12, and S17), and slope (S07, S08, and S21) (Fig. 1). In the surface seawater (5 m), temperature ranged from 27.5 to 29.1°C, with the average salinity increasing from 33.39 to 33.69 and 33.72 psu along the nearshore-offshore shelf-slope continuum, respectively. Dissolved oxygen (DO) levels rose from 6.30 to 6.40 and 6.49 mg/L, while inorganic nutrients (PO₄³⁻, NO₃⁻-N, NH₄⁺-N, and DSi) decreased from nearshore to slope. Chlorophyll a (Chl a) concentrations averaged 0.41, 0.15, and 0.11 mg/L for the nearshore, offshore, and slope environments, respectively (Table 1). In the surface seawater, the average abundances of Prochlorococcus, Synechococcus, pico-eukaryotes, and heterotrophic bacteria were 5.49 ± 4.54 × 10³, 1.30 ± 1.24 × 10⁴, 9.76 ± 7.48 × 10², and 4.56 ± 1.18 × 10⁵ cells/mL, respectively (Table 1). In the bottom seawater (ranging from 30 to 950 m), temperature decreased from 28.7 to 4.6°C with increasing depth while salinity increased from 33.67 to 34.47 and 34.48 psu. DO decreased from 6.16 to 5.21 and 3.04 mg/L along the nearshore-slope continuum (Table 1).

Fig 1.

Sampling stations and clustering of environmental and biological variables. Sampling stations in the northern South China Sea (a). Cluster analysis based on environmental and biological variables of surface (b) and bottom (c) seawaters, respectively.

TABLE 1.

Characteristics of the sampling sites, including temperature (ºC), salinity, dissolved oxygen (DO) (mg L⁻¹), Chl a concentration (μg L⁻¹), NO₃⁻-N (μmol L⁻¹), NO₂⁻-N (μmol L⁻¹), NH₄⁺-N (μmol L⁻¹), PO₄³⁻ (μmol L⁻¹), DSi (μmol L⁻¹), and the abundance of heterotrophic bacteria, Prochlorococcus, Synechococcus, and Pico-eukaryotes

	Surface											Bottom
	Nearshore			Offshore					Slope			Nearshore			Offshore					Slope
	S01	S14	S15	S03	S12	S17	S05	S10	S07	S08	S21	S01	S14	S15	S03	S12	S17	S05	S10	S07	S08	S21
Temperature (°C)	28.78	29.11	28.73	28.19	28.55	28.85	28.30	28.13	27.51	27.92	28.46	28.70	28.75	28.68	21.92	24.52	23.44	20.50	20.63	6.71	4.64	4.55
Salinity (PSU)	33.46	33.66	33.05	33.58	33.61	33.77	33.62	33.86	33.88	33.65	33.64	33.50	33.65	33.84	33.47	34.28	34.48	34.56	34.60	34.42	34.51	34.51
Depth (m)	5.0	5.0	5.0	5.0	5.0	5.0	5.0	5.0	5.0	5.0	5.0	33.0	38.0	30.0	90.0	74.0	81.0	139.0	104.0	650.0	950.0	920.0
DO (mg L−1)	6.26	6.23	6.42	6.18	6.54	6.40	6.42	6.49	6.61	6.51	6.34	6.35	6.05	6.10	5.26	5.77	5.07	5.40	4.56	3.14	3.03	2.94
Chl a (μg L−1)	0.37	0.35	0.52	0.18	0.13	0.13	0.14	0.16	0.12	0.09	0.11	0.54	0.50	0.59	0.30	0.65	0.73	0.10	0.29	0.00	0.00	0.00
NO3—N (μmol L−1)	2.12	0.26	3.73	0.14	0.35	0.23	0.31	0.10	0.29	0.28	0.32	0.90	0.42	0.31	6.66	1.54	2.49	10.11	8.58	34.72	38.07	38.25
NO2−-N (μmol L−1)	0.02	0.04	0.37	0.00	0.01	0.00	0.03	0.01	0.04	0.02	0.01	0.05	0.21	0.07	0.13	0.92	1.12	0.12	0.06	0.08	0.06	0.02
NH4+-N (μmol L−1)	0.13	0.07	0.11	0.05	0.04	0.14	0.11	0.02	0.02	0.02	0.01	0.45	0.37	0.23	0.27	0.04	0.06	0.10	0.03	0.03	0.08	0.04
PO43− (μmol L−1)	0.02	0.03	0.03	0.01	0.01	1.34	0.00	0.01	0.01	0.00	0.00	0.08	0.08	0.04	0.43	0.23	0.32	0.67	0.66	2.20	2.57	2.14
Dsi (μmol L−1)	2.75	3.62	3.01	1.05	1.28	0.71	1.23	1.04	1.25	1.36	1.01	2.76	4.54	4.06	7.75	4.47	8.97	10.21	11.52	68.69	75.34	89.13
Heterotrophic bacteria (×105 cells mL−1)	5.76	5.91	8.97	3.78	4.37	3.35	3.65	3.32	4.56	3.34	3.65	3.20	3.08	4.67	2.85	4.07	4.64	1.15	2.21	0.24	0.24	0.27
Prochlorococcus (×102 cell mL−1)	7.41	44.72	44.33	2.50	61.29	24.82	4.17	7.31	201.9	89.90	116.2	5.56	40.74	70.74	50.27	115.74	142.03	17.68	46.76	1.39	0.74	0.76
Synechococcus (×103 cell mL−1)	39.1	30.47	37.88	6.40	2.48	8.89	4.58	3.30	5.39	3.47	1.15	26.06	25.29	25.26	1.35	3.61	3.55	0.15	0.50	0.03	0.03	0.10
Pico-eukaryotes (×102 cell mL−1)	8.33	32.03	10.46	3.33	20.28	4.17	4.63	17.41	3.24	1.57	1.85	17.13	65.56	48.51	15.47	21.76	32.56	2.22	19.07	0.09	0.09	0.46

Community composition from three fractions

Across 22 seawater samples, a total of 6,447 amplicon sequence variants (ASVs), belonging to 36 bacterial phyla and 4 archaeal phyla, were detected across 3 fractions (rRNA_cell, rRNA_ext, and rDNA_cell) and a total of 122 and 794 ASVs were considered to be abundant and rare taxa, presenting 62.95%–71.79% and 9.41%–13.65% of the total relative abundance in two cellular fractions, respectively. The most dominated phyla in bacteria were Cyanobacteriia, Bacteroidia, Gamma- and Alpha-proteobacteria, Planctomycetes, Acidimicrobiia, Marinimicrobia, and Actinobacteria, while Nitrososphaeria and Thermoplasmata were dominant members in archaea (Fig. 2a through c). Of the total ASVs, a substantial proportion (~51.1%, 3,297 ASVs) was detected in the rRNA_ext pool, representing 28 bacterial and all 4 archaeal phyla.

Fig 2.

Comparison of bacterioplankton communities among three groups Cellular 16S rDNA and rRNA (rDNA_cell and rRNA_cell) and extracellular rRNA (rRNA_ext) at the Class level (a–c). Principal Coordinate Analyses (PCoA) based on the Bray-Curtis similarity to visualize the beta-diversity of rDNA_cell, rRNA_cell, and rRNA_ext (d–f).

The community composition in the rRNA_ext fraction generally resembled that of the two cellular fractions at the phylum level, with Synechococcales, Pseudomonadales, Alteromonadales, Rhodobacterales, Flavobacteriales, SAR11, Sphingomonadales, Piscirickettsiales, Staphylococcales, Propionibacteriales, Corynebacteriales, Fusobacteriales, Chitinophagales, Pirellulales, and Halobacterales exhibiting relative abundance >1% (Fig. 2c).

Seawater samples from surface and bottom were characterized by distinct prokaryotic communities, as revealed by their total (rDNA_cell) and active (rRNA_cell) components (Fig. 2d and e) though the rRNA_ext fraction showed highly similar compositions between surface and bottom (Fig. 2f). Furthermore, community compositions across environmental gradients were significantly different (Adonis, P < 0.05) for three fractions except for the rRNA_cell and rRNA_ext data sets between offshore and slope (Table S1).

Linear discriminant analysis Effect Size (LEfSe) identified differentially occurring taxa between surface and bottom seawaters in the three fractions. In the rDNA_cell and rRNA_cell fractions, these taxa were mainly Synechococcus, SAR11, Flavobacteriales, Rhodobacterales, Actinomarinales, and Marine Group II archaea (Fig. S1a and b). In the rRNA_ext fraction, the differentially occurring taxa included SAR11, Flavobacteriales, Rhodobacterales, Actinomarinales, and SAR324 for the surface seawater, and Pseudomonadales, Oceanospirillales, and Sphingomonadates for the bottom seawater (Fig. S1c). Notably absent from this comparison was the abundant Synechococcus and Marine Group II.

Cell lysis across samples

According to the presence/absence of ASVs among three fractions (rRNA_cell, rRNA_ext, and rDNA_cell) in each sample, all ASVs could be assigned into seven groups (Fig. 3a). These four groups (III, IV, VI, and VII) with the presence of ASVs in the rRNA_ext pool represented lytic groups, accounting for 12.7%–68.5% of taxa among three fractions (Fig. 3b), with higher proportions in surface (Avg. 45.3% of taxa) than bottom (Avg. 34.0% of taxa) seawater, indicating more diverse microorganisms were infected in the upper ocean (Fig. 3c). Additionally, higher cell lysis proportions ranging from 36.8% to 59.8% were observed for the bottom seawater of three slope stations, likely facilitated by their low temperatures (4.5–6.7°C) promoting rRNA_ext stability (Fig. S2).

Fig 3.

Detection of cell lysis in sampling stations as inferred from rRNA_ext group. Venn diagram showed seven possible lysis groups including non-cell lysis (I, II, and V), ongoing lysis (III, IV, and VI), and prior lysis (VII) groups (the numbers in parentheses represent the average number of ASVs) (a). Each panel showed the proportion of ASVs in the rDNA_cell, rRNA_cell, and rRNA_ext pools across the seven lysis groups in all sapling stations (b). Proportion of three lysis categories (non-cell lysis, ongoing lysis, and prior lysis) from surface and bottom (c).

Group VII represents those taxa only found in the rRNA_ext fraction and comprised the majority of cell lysis from 10.0% to 58.4% of taxa per sample (Fig. 3b). The cell lysis in Group VII likely was not detected in the cellular fractions (0.22–20 μM) which could have occurred prior to sampling or may have derived from attached microbes on large particles (>20 µM) since the corresponding sequences were not detected in the cellular fractions. Groups IV and VI, detected in both the rRNA_ext and rDNA_cell fractions, suggest ongoing or recent lysis. ASVs in the Group IV were detected in all three fractions, consisting of 1.4%–10.9% of total taxa across samples. ASVs in the group VI (detected only in rDNA_cell and rRNA_ext fractions) accounted for 0.12%–5.9% of taxa. The absence in rRNA_cell fraction indicates these taxa may be low growth or activity at the time of sampling. ASVs in the group III were only detected in rRNA_cell and rRNA_ext fractions, implying that they were in very low abundance but undergoing lysis, and comprised 0%–11.5% of taxa (Fig. 3b).

Cell lysis index

The relative amount of cell lysis in each taxon can be estimated using the ratio of rRNA_ext to rRNA_cell, namely, the CLI with CLI >1 indicating higher and <1 lower lysis rates, respectively. The number of ASVs in group IV was highly variable across samples, ranging from 9 to 47 in each sample, and their CLI values displayed a wide range from 0.01 to 441 (Fig. S3). However, even the same ASV exhibited distinct CLI in different environments, especially between surface and bottom seawaters.

Relative abundance of ASVs in group IV from three fractions was analyzed. Generally, the moderately positive correlation (R² = 0.4, P < 0.001) between rDNA_cell and rRNA_cell was observed (Fig. S4a), which implied the microbial activity closely associated with its relative abundance in total community. Moreover, a poor relationship (R² = 0.1, P < 0.001) between rDNA_cell and rRNA_ext was detected (Fig. S4b). Nevertheless, a negative correlation between CLI and relative abundance of ASVs in rDNA_cell fraction in group IV was found according to the regression analysis (Fig. 4b). In particular, ASVs with higher CLI value (>1) and lower relative abundance (<0.1%) were observed more frequently in bottom seawater (n = 62) than surface (n = 39).

Fig 4.

Linear regression (red line) for the relative abundance of individual ASVs detected in group IV between the active index (the ratio of the relative abundance of rRNA_cell to rDNA_cell) and rDNA_cell (a), and the lysis index (the ratio of the relative abundance of rRNA_ext to rRNA_cell) and rDNA_cell (b). Purple line indicates the 95% confidence interval for the slope of the regression line. The density plots showed the distribution of values for the surface (blue) and bottom (yellow).

Cell lysis index of different taxa

The dominant bacterial groups Prochlorococcus, Synechococcus, SAR11 clades I and II, and members of Rhodobacteraceae, as well as SAR86, SAR324, and SAR116, exhibited relatively low CLI values across samples, especially in the surface seawater (Fig. S3a). Nevertheless, certain Synechococcus ASVs had high CLI in bottom seawater. Additionally, Gammaproteobacterial such as Alteromonas, Pseudomonas, and Halomonas also had an elevated CLI in bottom seawaters (Fig. S3b).

Some taxa displayed variable CLI values with different environments (surface vs bottom seawaters) and different ecotypes. For example, abundant Synechococcus ASV2 had much lower average CLI values (0.77 vs 4.34) in the surface seawater than in the bottom, and ASV 9 (5.99 vs 2.16), 40 (2.88 vs 2.73), and 76 (4.55 vs 2.70) display high CLI values in both surface and bottom seawaters (Fig. S3 and S5). Diverse SAR11 ecotypes also had different habitat preferences, as reflected by their varying lysis rates. This can be seen in samples, the abundant SAR11 clade I taxa (ASVs 11, 12, and 29) showed a low lysis rate with the average CLI values of 0.72, 0.75, and 0.43, respectively, across samples. However, the SAR11 clade II taxon (ASV16) displayed much lower CLI values in the surface seawater (0.44 vs 3.60) than in the bottom (Fig. S5).

Some ASVs exhibit distinct CLI in different environments. Specifically, Alteromonas (ASV3, ASV4) and Oceanospirillales (ASV7) demonstrate higher CLI in the nearshore region, while Actinobacteria (ASV23) and Flavobacteria (ASV352) exhibit higher CLI in the offshore area. Additionally, Alphaproteobacteria (ASV35) display a high CLI in the slope region (Fig. S3).

Environmental variables impact on the community composition in three fractions

In the surface seawater, the environmental variables that significantly correlated with total and active prokaryotic community compositions included salinity, Chl a, dissolved inorganic nitrogen (DIN), DSi, NO₃⁻-N, NO₂⁻-N, and Synechococcus and heterotrophic bacterial abundance (Fig. S6a). In the bottom seawater, latitude, depth, temperature, DO, Chl a, PO₄³⁻, DIN, DSi, NO₃⁻-N, and four microbial abundances were significant predictors of total prokaryotic community composition, while active prokaryotic community composition was only corresponding to temperature, Chl a, DIN, NO₃⁻-N, and three microbial abundances (Fig. S6b). The strength of correlation between environmental variables and total prokaryotic community assemblages (r = 0.43–0.81) was greater than those between environmental variables and active prokaryotic community assemblages (r = 0.24–0.59).

No significant correlations (Adonis, P ≥ 0.05) between community composition from the rRNA_ext fraction, and measured variables were apparent from Mantel test results (P ≥ 0.05) (Fig. S6b). In Canonical Correspondence Analysis (CCA) models, environmental variables accounted for 45%, 34%, and 12% of total variations of prokaryotic community composition in rDNA_cell, rRNA_cell, and rRNA_ext fractions, respectively. Furthermore, Hierarchical Partitioning Analysis showed that measured environmental variables had lowest explanatory power over pattern of community composition from rRNA_ext fraction (Fig. S7).

The total and active prokaryotic community similarity at geographic distances significantly declined (Mantel tests, P < 0.05) in both surface and bottom seawaters. Notably, the community similarity was much higher in surface seawater than in bottom (Fig. S8a and b). Moreover, the total prokaryotic community similarity had a more substantial negative correlation with geographic distance (r = 0.19 and 0.19 in surface and bottom seawater, respectively, P < 0.001) compared to the active prokaryotic community (r = 0.08 and 0.05 in surface and bottom seawater, respectively, P ≤ 0.05). Conversely, the prokaryotes in the rRNA_ext fraction did not demonstrate a distance decay pattern in either surface or bottom seawater (Fig. S8).

Co-occurrence networks in three fractions

Based on the spatial pattern of the community composition in three fractions, their co-occurrence networks were constructed to infer ecosystem functioning in the surface and bottom seawaters, respectively. Generally, compared to their counterparts in the surface seawater, the connectivity (average degrees) between taxa in the bottom was obviously higher in all three fractions (Fig. 5a through f). The rRNA_ext network had fewer involved ASVs and lower average degrees (Fig. 5c and f). In both the surface and bottom co-occurrence networks, members of Flavobacteriales and SAR11 from cellular fractions were closely associated with lysed groups (e.g., Synechococcales, Alteromonadales, and Rhodobacterales), indicating their quick responses or potential succession among different microbial groups (Fig. S9).

Fig 5.

Network analysis based on the Spearman correlation of three groups cellular 16S rDNA and rRNA (rDNA_cell and rRNA_cell) and extracellular rRNA (rRNA_ext) with r² ≥ 0.7 and P  < 0.01 between ASVs. Co-occurrence networks of rDNA_cell (a and d), rRNA_cell (b and e), and rRNA_ext (c and f) groups in the surface and bottom seawaters, respectively. Relative abundance of the first six main ecological clusters of the three groups at the class level in the networks (g).

Assembly processes of prokaryotic communities in three fractions

The contribution of deterministic and stochastic processes to the assembly of prokaryotic communities was investigated by utilizing the Neutral Community Model (NCM) and Modified Stochasticity Ratio (MST) model. Results showed that the variation explained by stochastic processes was notably higher in the surface (R² from 0.55 to 0.62) than in the bottom seawater (R² from 0.09 to 0.15) (Fig. S10a through f). In addition, a similar pattern was seen in the estimated migration rate “m.” The assembly of both total and active prokaryotic communities was mainly determined by stochastic and deterministic processes in the surface and bottom seawaters, respectively, with the community composition in the rRNA_ext fraction primarily affected by stochastic processes and a relatively high R² value (between 0.61 and 0.62) compared to other fractions (Fig. S10c and f).

The MST model showed a similar result to the NCM model. For surface seawater, there was a considerable increase in the mean values of MST moving from total and active communities (average MST values 0.57 and 0.55, respectively) to rRNA_ext fraction (0.69) (Wilcoxon test, P < 0.05) (Fig. S10g). A similar pattern was observed in the bottom seawater, where values increased significantly from total community (0.34) and active communities (0.36) to rRNA_ext fraction (0.57) (Wilcoxon test, P < 0.05) (Fig. S10h). Additionally, the total and active communities had significantly higher MST values in the surface seawater compared to the bottom (Wilcoxon test, P < 0.01).

Environmental and cell lysis factors associated with microbial community

It is worth noting that microbial community in the rRNA_ext fraction had distinctive features in terms of community assembly, environmental regulation, geographic distance relation, and co-occurrence networks. Here, rRNA_ext complex from viral lysis represented non-living organic matter, distinguishing it from the cellular community. However, it can reflect the corresponding living cells that recently experienced lysis in the environment.

To further understand the effects of environmental variables and cell lysis on prokaryotic community assembly, environmental factors and extracellular microbial community factors were used to explain variations in total and active bacterial composition. VPA results revealed that both environmental variables and cell lysis explained 35.8% and 28.9% of total and active microbial composition variations in the surface seawater (Fig. 6; Fig. S11), respectively, and was much higher than that (17) of 18.6% and 21.4% of variations in the bottom seawater (Fig. 6; Fig. S11). Cell lysis alone was estimated to account for 2.92%–4.89% of cellular microbial community variations, with a higher contribution to active community (4.11%–4.89%) than total community (2.92%–3.07%). In addition, shared environmental and extracellular microbial community factors described >15.0% of variations in the surface seawater communities, whereas they only explained <1.0% of variations in the bottom (Figure S11). Furthermore, VPA illustrated that environmental and cell lysis factors explained higher proportions of community variation of abundant taxa (42.2%–57.0%) than rare taxa (19.0%–34.8%) in both surface and bottom seawater (Fig. S11). Moreover, the contribution of purely cell lysis to overall abundant taxa variation was modest (0.08%–1.70%), and it was substantial for rare taxa (4.96%–9.37%) (Fig. S11), indicating that virus-induced host selective mortality is important in influencing prokaryotic community structure, particularly in regards to rare taxa.

Fig 6.

The purely abiotic (environmental variables) and biotic (cell lysis) factors influenced the abundant and rare taxa in the surface and bottom seawater, respectively, based on Variance Partitioning Analysis (VPA) analysis. The purple and orange numbers represented the proportion (%) of both factors explaining on total (rDNA_cell) and active (rRNA_cell) microbial community, respectively. T and S in the environment variables represent temperature and salinity, respectively.

DISCUSSION

Viral infection and lysis play crucial roles in determining microbial community composition and diversity in the ocean; however, the exact mechanisms underlying this regulation are still not well understood (14). Previous studies have made some rough estimation of virus-mediated microbial mortality rates and related organic carbon turnover but lacked a detailed understanding of virus-associated effects on individual bacterial populations. This study aimed to investigate the influence of cell lysis across prokaryotic taxa in a shelf-to-slope continuum on abundant and rare taxa.

Virus-induced prokaryotic selective mortality

Protistan predation and virus-mediated mortality are two distinct mechanisms for influencing microbial community composition. Whereas protistan predation is thought to involve cell size-based selection of grazing (25–27), virus-mediated mortality is characterized by host-specificity and population-selective impact (16, 28, 29). Previous studies indicated that viral lysis plays a significant role relative to protist predation in the release of cellular rRNA (17). Random dispersal is an important community assembly strategy for viruses (30, 31), and the virus-induced rRNA_ext fraction is also primarily affected by stochastic processes.

Cyanobacteria are important primary producers in the ocean, and their abundance, especially Prochlorococcus, could be comparable to or even exceed heterotrophic bacteria in the oligotrophic ocean (32, 33). Cyanophages infecting cyanobacteria were frequently isolated and primarily comprised members of the tailed phages (34, 35). Though diverse cyanophages were detected in the aquatic environment, low Prochlorococcus cyanophages infection and lysis rates were commonly observed (36, 37). Our result showed that two abundant Prochlorococcus ASVs had average cell lysis rates with the average CLI <1 both in surface and bottom seawater. However, the Trichodesmium was detected in half (11) samples in the rRNA_ext fractions which was only observed in one and four samples in rDNA_cell and rRNA_cell fractions with a lower relative abundance, respectively. This indicates these low abundance, but active Trichodesmium cells were subject to high viral lysis pressure in the seawater, resulting in rapid turnover in the ocean. In contrast, virus-mediated Synechococcus mortality varied widely between habitats (32, 36). These results implicate that virus-mediated lysis can affect not only cyanobacterial abundance but also their biogeography in the ocean (38).

The SAR11 clade is one of the most abundant members in the ocean and consists of up to 25% of total bacteria in the euphotic zone (39–42). As a member of extremely successful pelagic specialist in the oligotrophic ocean, they were thought to be immune to viral infection for a long time (43, 44). Contrarily, SAR11 phages were isolated and found to be widely distributed in the ocean in 2013 (45). Moreover, variable growth rates of SAR11 taxa were also detected in Delaware coastal waters as revealed by the ratio of rRNA_cell to rDNA_cell (11). Taken together, these suggest that SAR11 taxa may have diverse genomic and metabolic capacities that enable distinct ecological strategies.

SAR324, a member of the Deltaproteobacterial group, is widely distributed in the ocean and is capable of performing carbon fixation and C1 utilization mixotrophy lifestyle (46–48). Cell lysis for the SAR324 marine group B (ASV25) was only observed in surface seawater with a low average CLI of 0.15 (n = 6) despite high relative abundance in two cellular fractions in the bottom (Fig. S5g and h). This indicates that environmental conditions (e.g., light, pressure, temperature) might act as trigger factors for some bacterial taxa lysis.

Members of Gammaproteobacteria (e.g., Alteromonas, Pseudomonas, and Halomonas) frequently displayed high CLI, especially in the bottom seawater. Multi-prophages were found in their genomes (49–52), indicating lysogenic lysis may account for their high CLI in the bottom seawater. In addition, the dark and relatively stable environment of bottom seawater may favor rRNA_ext complexes storage. Except for Gammaproteobacteria, members of Bacteroidia appeared abundantly in the rRNA_ext fraction, but their corresponding taxa were absent or rare in two cellular fractions. Other processes such as bacterial predation (53) or programmed cell death (54) may also contribute to their cell lysis, but these processes have not yet been reported to account for substantially cell lysis in the seawater. Viral lysis is estimated to remove roughly 20%–30% of the daily standing stock of bacteria in the oceans (2, 55), and it is highly likely that virus-mediated cell lysis is the major source of rRNA_ext.

Two archaeal class Nitrososphaeria and Thermoplasmata were abundant in the rDNA_cell fraction of bottom seawater but had a low relative abundance (<1%) in the rRNA_ext fraction (Fig. S1). The relatively low activities as revealed by the low ratios of their rRNA_cell to rDNA_cell likely account for the low CLI. However, Danovaro et al. revealed that up to one-third of total microbial biomass mortality in deep-sea surface sediments is caused by virus-induced archaeal lysis through quantification of released 16S rRNA gene copies in rRNA_ext fraction (27). It would be enlightening to investigate different archaeal taxa lysis in varying conditions and their contribution to labile organic matter pool.

Members of Flavobacteriales and SAR11 were closely associated with cell lysis bacterial groups, Synechococcales, Alteromonadales, and Rhodobacterales as revealed by co-occurrence networks. Flavobacteria could quickly respond to viral lysate of Synechococcus, showing high abundance throughout incubation (56). Moreover, Flavobacteria can break down high-molecular-weight organic matter, while SAR11 primarily assimilate low-molecular-weight organic substrates (57–59). These two kinds of bacteria, thus, synergistically metabolize viral lysates in the environment, converting relatively labile organic matter into refractory substances (60). In addition, some of them have overlapping niches (SAR11 and Cyanobacteria, Flavobacteria, and Altermonadales), and all of these bacteria likely exchange vitamins and vitamin precursors (61, 62).

Cell lysis between surface and bottom seawaters

The rRNA_ext complexes-containing detritus produced by lysed hosts (especially for the dominant groups) tends to sink from surface seawater to the bottom in the shallow shelf regions; thus, no significant difference in the relative abundances of rRNA_ext taxa was found across depths in our samples, which was also observed in samples collected from five depths of the Strait of Georgia (17). The virus-to-bacterium ratio (VBR) was higher in the deeper seawater than in the surface in our studied region, and viral sinking from surface water could be responsible for the relatively high VBR (63–65). Additionally, the natural and microbial degradation of sinking cells from surface seawater could also partly account for the community composition in rRNA_ext fraction in the bottom seawater.

Cell lysis on abundant and rare taxa

Rare prokaryotic taxa usually exhibit high metabolic activity and CLI (Fig. 4). Our results consisted with previous finding that rare microorganisms comprise a major part of the microbial diversity (11) and could potentially be the competitive winners for limited resources (14–16). Dilution experiments have also revealed that rare taxa can become dominant under reduced viral lysis and protistan predation pressures (43). Abundant prokaryotic taxa with relatively low activity and lower CLI values (Fig. 4) supported their high presence in the environment. Recent studies demonstrated that lytic virus tends to switch life strategy to lysogeny, terms as Piggyback-the-Winner, when the host abundance is high (66, 67). This phenomenon may partially account for the observation that abundant taxa exhibited lower CLI values. Consistently, our VPA results indicate virus-induced prokaryotic lysis explained a much higher proportion of the variation in rare taxa than in abundant taxa. This suggests that virus-mediated host mortality might be considered a potential mechanism in balancing prokaryotic species evenness and diversity (7, 29).

Conclusions

Our investigation of microbial individual cell lysis along a shelf-to-slope continuum highlighted the influence of cell lysis on variations in both abundant and rare taxa. Although environmental factors were the more critical factors influencing the microbial community composition, viral lysis also played a non-negligible role in shaping community assembly, especially for rare taxa. The dominant taxa, including Prochlorococcus, Synechococcus, SAR11, and Rhodobacteraceae, showed lower cell lysis indices in surface seawater, while Alteromonas, Pseudomonas, and Halomonas exhibited higher CLI values in the bottom seawater. Overall, these results improve our understanding of bottom-up (abiotic environmental variables) and top-down (viral lysis) controls contributing to microbial community assembly in the ocean.

MATERIALS AND METHODS

Seawater sampling and filtration

A total of 22 samples were collected from 11 stations (surface and bottom seawater) in the northern South China Sea (113.29–115.81°E and 19.59–22.21°N) in September 2019 (Fig. 1; Table 1). Furthermore, rDNA_cell, rRNA_cell, and rRNA_ext were collected as described by Zhong et al. (17). Briefly, 100 mL of seawater was pre-filtered through a 20 µM mesh size Nitex screen to remove large particles, followed by gentle vacuum filtered onto a 47 mM 0.22 µM pore-size PVDF filter (Millipore, GVWP). Both the filter and filtrate were flash-frozen in liquid nitrogen and kept at −80°C within 3 months for further DNA and RNA extraction.

Nucleic acids extraction and cDNA synthesis

Microbial cellular nucleic acids (both DNA and RNA) were extracted from 0.22 µM filters using ZymoBIOMICS DNA/RNA Miniprep Kit (ZYMO) according to the manufacturer’s protocol. To isolate cellular DNA and RNA, half of the nucleic acids were treated with RNase A (5 µg/mL, Epicentre, 37°C for 30 min) and DNase I (1 U/µL, Invitrogen, 26°C for 15 min), respectively. The rRNA_ext was obtained from the extraction of the filtrate samples using a PureLink Viral RNA/DNA mini Kit (Invitrogen) and with DNase I treated as described by Zhong et al. (17). The purified cellular RNA and rRNA_ext were subsequently reverse-transcribed to cDNA using random hexamer primers via Super-Script III Reverse Transcriptase (Invitrogen) following the manufacturer’s instructions.

High-throughput sequencing and analysis

The bacterial 16S rRNA gene between the V4 and V5 regions of the DNA and cDNA samples was amplified with the forward primers 5′-GTGCCAGCMGCCGCGGTAA-3′ and reverse primers 5′-CCGYCAATTYMTTTRAGTTT-3′ using the polymerase chain reaction (PCR) procedure (68). The PCR program included an initial denaturation step (3 min at 95°C) followed by 34 cycles of denaturation (45 s at 95°C), annealing (45 s at 50°C), and extension (90 s at 68°C), concluding with a final elongation step at 68°C for 5 min to ensure complete amplification (69). Subsequently, the PCR product was purified using magnetic Agencourt AMPure XP beads (Beckman Coulter) at a 1:1 ratio of beads to product, effectively removing fragments less than 200 bp (e.g., dimers) (17). The preparation of 16S rRNA gene amplicon libraries was adapted from the online Illumina protocol with several modifications as described in reference (17). Quantified amplicons were sequenced using the Illumina MiSeq platform (Illumina, San Diego, CA, USA). All obtained sequences from three fractions were processed and analyzed using the QIIME pipeline version 2 (qiime2.2021.08). The raw paired-end reads were denoised and assembled by using DADA2 v1.1.3 (70). The high-quality reads were clustered as amplicon sequence variants (ASVs) at 99% sequence identity. The ASVs were taxonomically classified based on the SILVA database (version 138) (71). The resulting sequences contained 35,171 ± 15,865 merged reads of ~412 bp per library. These sequences were rarified to 15,500 reads for further analyses after removing reads number less than 10.

Statistical analysis

Environmental and biological variables were analyzed by hierarchical clustering to classify the nearshore, offshore, and slope regions with the ape package (72). Cell lysis index (CLI) of each ASV was calculated using the ratio of the relative abundance of rRNA_ext to rRNA_cell (17), and the P value was obtained by Benjamini–Hochberg (BH) correction from the multiple comparisons. Principal coordinates analysis (PCoA) was used to characterize the β-diversity pattern of bacterial community based on Bray-Curtis distance matrices (73). Permutational multivariate analysis of variance (Adonis) was applied to identify significant differences between microbial community groups. Linear discriminant analysis Effect Size (LEfSe) (74) (https://huttenhower.sph.harvard.edu/galaxy/) was conducted to differentiate between microbial taxa of differing depths (75). Correlations between microbial community of rDNA_cell, rRNA_cell, and rRNA_ext and environmental variables were assessed with Mantel tests (P < 0.05). Distance-decay analysis was conducted to explore the association between geographical distance and phylogenetic dissimilarity by the “lm” function of R. Co-occurrence networks based on Spearman’s rank correlation (r² ≥ 0.7, P < 0.01) were established to illustrate the relationships between different ASVs (occurrence frequency >1/3 of all samples). Corresponding R codes were derived from GitHub (https://github.com/ryanjw/co-occurrence) (76), and network visualizations were conducted in Gephi V0.9.2 and Cytoscape v3.8.2. To assess the relative importance of stochasticity and determinism for the assembly of three fractions (rRNA_cell, rRNA_ext, and rDNA_cell), the fit of a Sloan neutral community model was first evaluated (77). Output plots of neutral community model (NCM) mainly show (i) the fit of neutral model (R²) and the migration rate (m); (ii) neutral model predictions and its corresponding 95% confidence intervals (lines) and actual distributions (points) for different fractions and depth. Then, the modified stochasticity ratio (MST) was calculated for the overall data set (78). This MST index was established with 50% defined as the boundary point between more deterministic (<50%) and more stochastic (>50%) assemblies. Furthermore, abundant and rare taxa were identified based on two criteria: (i) ASVs present in both the rRNA_cell and rRNA_ext fractions, or in the rDNA_cell and rRNA_ext fractions; (ii) the average relative abundance of 0.1% in the rDNA_cell fraction used as a threshold for distinguishing between abundant and rare taxa. Variation partitioning analysis (VPA) was used to evaluate how much of the variation in cellular microbial community was purely explained by environmental abiotic variables and biotic variables cell lysis (rRNA_ext), for the explanatory data (environmental variables and rRNA_ext), as VPA breaks down when the number of factors (i.e., environmental variables/ASVs) is larger than the number of sampling station, a two-step approach was adopted (79, 80). First, PCoA was applied to explain data and a few PCoA axes which accounted for at least 90% of environmental variables and 70% of rRNA_ext community composition variations were retained as explanatory variables (81). Second, variance inflation factors (VIFs) were calculated to examine the presence of collinearity between the environmental and rRNA_ext variations, while variables with VIF >10 were removed to avoid the impact of collinearity. Subsequently, a forward selection procedure was used to select PCoA axes as explanatory data with significant explaining factors (P < 0.05) for further analyses. In the end, three environmental factors (PCoA axes 1–3) and 3–4 extracellular microbial community factors (PCoA axes 1–3 in surface seawater, and 1–4 in bottom seawater, respectively) were used to explain variations in cellular microbial community. To generate unbiased estimates of the variation partitioning based on RDA, cellular microbial community were Hellinger-transformed before VPA analysis and the overall importance of single predictors in multivariate models was estimated by rdacca.hp R package (82).

All statistical analyses and visualization were performed by R v4.0.4 with the packages phyloseq (83), vegan (84), and ggplot2 (85).

DATA AVAILABILITY

Sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under the BioProject accession number PRJNA907164 (SRA accession numbers SRR22476578 to SRR22476643).

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01393-23.

Supplemental material. aem.01393-23-s0001.docx.

Additional experimental details, Supplemental figures and tables.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.