The polyamine norspermidine is one of the major polyamines synthesized by Vibrionales and has also been found in various aquatic organisms. Norspermidine is among the environmental signals that positively regulate Vibrio cholerae biofilm formation. The NspS/MbaA signaling complex detects extracellular norspermidine and mediates the response to this polyamine. Norspermidine binding to the NspS periplasmic binding protein is thought to inhibit the phosphodiesterase activity of MbaA, increasing levels of the biofilm-promoting second messenger cyclic diguanylate monophosphate, thus enhancing biofilm formation. V. cholerae can also synthesize norspermidine using the enzyme NspC as well as import it from the environment. Deletion of the nspC gene was shown to reduce accumulation of bacteria in biofilms, leading to the conclusion that intracellular norspermidine is also a positive regulator of biofilm formation. Because V. cholerae uses norspermidine to synthesize the siderophore vibriobactin it is possible that intracellular norspermidine is required to obtain sufficient amounts of iron, which is also necessary for robust biofilm formation. The objective of this study was to assess the relative contributions of intracellular and extracellular norspermidine to the regulation of biofilm formation in V. cholerae. We show the biofilm defect of norspermidine synthesis mutants does not result from an inability to produce vibriobactin as vibriobactin synthesis mutants do not have diminished biofilm forming abilities. Furthermore, our work shows that extracellular, but not intracellular norspermidine, is mainly responsible for promoting biofilm formation. We establish that the NspS/MbaA signaling complex is the dominant mediator of biofilm formation in response to extracellular norspermidine, rather than norspermidine synthesized by NspC or imported into the cell.
Citation: Wotanis CK, Brennan WP III, Angotti AD, Villa EA, Zayner JP, Mozina AN, et al. (2017) Relative contributions of norspermidine synthesis and signaling pathways to the regulation of Vibrio cholerae biofilm formation. PLoS ONE 12(10): e0186291. https://doi.org/10.1371/journal.pone.0186291
Editor: Dongsheng Zhou, Beijing Institute of Microbiology and Epidemiology, CHINA
Received: April 6, 2017; Accepted: September 28, 2017; Published: October 18, 2017
Copyright: © 2017 Wotanis et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The raw data used to generate the graphs in the figures and supporting information is included in the resubmission as an Excel file named supporting document 10. Hence, all relevant data are within the paper and its Supporting Information files. The strains and plasmids used in this study are stored in the Biology Department at Appalachian State University. All the strains and plasmids are available without restrictions and can be requested from the corresponding author of the study.
Funding: Funding for this work was provided by the following sources: Appalachian State University Department of Biology, the Office of Student Research grants to CKW, WPB, ADA, EAV, JPZ, ANM (https://osr.appstate.edu/) and by NIH-NIAID (https://www.niaid.nih.gov/) (Grant number AI096358 to E.K.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Biofilms are communities of microbes that exist attached to surfaces and/or each other. In most cases, biofilm microbes are encased in a self-produced hydrated matrix of an extracellular polymeric substance (EPS) . The major components of the EPS are polysaccharides, proteins, and DNA . Bacteria may form biofilms as a survival mechanism as such structures have been shown to aid in protection against pH extremes, osmotic stress, UV radiation, antimicrobials, and the host’s immune response [3–6]. Biofilm formation is regulated by a variety of cues which include nutrient availability, quorum sensing, surface composition, osmolarity, and small molecules such as polyamines [7,8].
Polyamines are flexible aliphatic chains containing two or more amine groups that are positively charged at physiological pH. These natural polycations are involved in cell growth and development of both prokaryotic and eukaryotic cells . While polyamines are believed to perform different functions in different organisms, they have been shown to play a role in biofilm formation in a variety of bacteria, including Vibrio cholerae, the causative agent of the diarrheal disease cholera [8,10]. Norspermidine is a relatively uncommon polyamine that is produced mainly by the Vibrionales [11,12]. It is also found in some extremophilic Archaea, in some aquatic and terrestrial plants, and lower aquatic eukaryotes including some diatoms, algae, arthropods, mollusks, and sea squirts [13–17].
Norspermidine is synthesized in V. cholerae through the decarboxylation of carboxynorspermidine by the enzyme NspC, encoded by the nspC gene (Fig 1A). Deletion of nspC was shown to reduce biofilm formation, which could be rescued by uptake of norspermidine from the growth medium . Norspermidine import into the cell is mediated by the ABC transporter binding protein, PotD1, presumably together with the PotA, PotB, and PotC proteins, which comprise the rest of the transporter (Fig 1B) . It has been suggested that adequate levels of cytoplasmic norspermidine are required for supporting robust biofilm formation by an unknown mechanism in V. cholerae . Norspermidine forms the backbone of the V. cholerae siderophore vibriobactin and iron deficiency has been shown to inhibit biofilm formation [19,20]. Therefore, one possible mechanism by which intracellular norspermidine affects biofilm formation may involve iron acquisition; however, the effect of vibriobactin synthesis on biofilm formation has not been studied.
(A) Synthesis of norspermidine and vibriobactin. Norspermidine can be synthesized by the enzyme NspC through decarboxylation of carboxynorspermidine. Norspermidine is utilized by VibF to form the backbone of the siderophore vibriobactin. Norspermidine backbone outlined by the grey box.(B) Norspermidine synthesis, utilization, transport, and signaling pathways. Norspermidine can be synthesized by NspC and can also be imported from the environment presumably through the PotABCD1 ABC-type transporter. Vibriobactin, synthesized from norspermidine, is secreted into the environment and binds to iron. This ferric-vibriobactin complex is recognized by the outer membrane protein ViuA and transported into the periplasm, which is then imported into the cell by an ABC-type transporter (not shown). Exogenous norspermidine is sensed by the NspS/MbaA signaling complex, which leads to increased VPS production and biofilm formation, presumably through increasing c-di-GMP levels in the cell. Norspermidine is represented by a zig-zag and carboxynorspermidine is represented by a branched zig-zag. VPS, Vibrio polysaccharide. The PotABCD1 ABC-type transporter is denoted as A, B, C, and D1.
Norspermidine also acts as an extracellular signal that enhances V. cholerae biofilm formation . The effect of norspermidine is mediated by the periplasmic binding protein NspS, which has been shown to bind norspermidine in vitro . NspS itself is also an activator of biofilm formation. NspS is believed to interact with the GGDEF-EAL family protein MbaA, a cyclic diguanylate monophosphate (c-di-GMP) phosphodiesterase, and down-regulate its enzymatic activity, leading to increased c-di-GMP levels. Increased c-di-GMP activates expression of the vps (Vibrio polysaccharide) genes encoding proteins responsible for the production of the biofilm polysaccharide in V. cholerae, which in turn increases biofilm formation. Norspermidine binding to NspS is believed to enhance the inhibitory effect of NspS on MbaA, leading to further increases in biofilm formation [8,18]; (Fig 1B).
The purpose of the current study was to delineate the interplay of norspermidine synthesis, transport, utilization, and signaling pathways on the regulation of biofilm formation in V. cholerae. We constructed various mutants in norspermidine-related pathways and assessed the resulting biofilm phenotype as well as cellular polyamine content in V. cholerae. We demonstrate that the NspS/MbaA signaling pathway is epistatic over norspermidine synthesis, import, and utilization pathways in regulating biofilms.
Materials and methods
Bacterial strains, plasmids, and media
The V. cholerae strain used was O139 MO10. More information on the bacterial strains and plasmids used in this study can be found in Table 1. Primers are listed in Table 2. All experiments were done in Luria-Bertani (LB) broth. To generate iron-deplete conditions, the ferric iron chelator ethylenediamine di(ortho-hydroxyphenylacetic acid) (EDDA) was added to LB at a concentration of 100 μg/ml after being deferrated according to the method of Rogers . Streptomycin, ampicillin, and kanamycin were used at 100 μg/ml unless otherwise specified; tetracycline was used at 2.5 μg/ml. Primer synthesis and DNA sequencing were performed by Eurofins MWG Operon.
Construction of mutants
Mutants were constructed by double homologous recombination with sucrose selection as described by Metcalf et al. . The viuA::tetR mutant was constructed by mating E. coli SM10λpir containing pPAC20 with wild-type V. cholerae to generate the V. cholerae viuA::tetR mutant. E. coli SM10λpir containing pWCW3 was mated with wild-type V. cholerae or with the V. cholerae ΔnspS mutant to generate the V. cholerae ΔvibF and ΔnspSΔvibF strains, respectively. E. coli SM10λpir containing pAR17  was mated with the V. cholerae ΔvibF mutant to generate the V. cholerae nspC::kanRΔvibF mutant. E. coli SM10λpir containing pAR17  was mated with the V. cholerae ΔnspS mutant or the V. cholerae ΔmbaA mutant to generate V. cholerae nspC::kanRΔnspS and nspC::kanRΔmbaA mutants, respectively. E. coli SM10λpir containing pMM9  was mated with the V. cholerae nspC::kanRΔnspS mutant to generate nspC::kanRΔnspSΔpotD1. After mating, strains were streaked on selection plates containing 100 μg/ml streptomycin and 50 μg/ml ampicillin. For generation of the viuA::tetR and nspC::kanR strains, selection plates also contained tetracycline (2.5μg/ml) or kanamycin (50 μg/ml). Single colonies were restreaked on selection plates to confirm single cross-over events followed by streaking on LB plates without antibiotics in order to facilitate a second recombination event. After incubation overnight, multiple isolated colonies were streaked on sucrose plates without antibiotics or containing tetracycline (2.5 μg/ml) or kanamycin (50 μg/ml) if relevant and incubated for 48 hours. Isolated colonies were patched on LB plates containing either streptomycin (100 μg/ml) or ampicillin (50 μg/ml) in order to screen for streptomycin resistance and ampicillin sensitivity and confirming successful homologous recombination. Colonies that grew on streptomycin but not ampicillin plates were confirmed for the recombination event by colony PCR. Primers used to verify construction of the ΔvibF and viuA::tetR mutants are listed in Table 2. Primers used to verify ΔpotD1 and nspC::kanR have been published in  and , respectively.
Cloning of the potD1 gene
Primers PA138 and PA144 were used to amplify the potD1 gene from V. cholerae chromosomal DNA. This PCR product was first TOPO-cloned into the pCR2.1 vector (Invitrogen) and correct sequence was verified. It was then excised with EcoRI and NcoI and subcloned into pACYC184 digested with the same restriction enzymes generating pAR2.
Extraction, benzoylation, and detection of polyamines
Bacteria were grown at 27°C to mid-exponential phase, pelleted, washed twice with 1X PBS, and resuspended in 10 μl water per milligram wet cell weight. Then, 250 μl of the cell suspension corresponding to 25 mg of cells was lysed using sonication and the cell debris was removed by centrifugation. Cellular proteins were precipitated with 50% (w/v) trichloroacetic acid and centrifuged. The supernatant containing the polyamines was removed and benzoylated as described previously . For measurement of polyamines in the spent medium, cultures were centrifuged, 1 ml of the media was removed and passed through a 0.22 μm syringe filter to remove any remaining cells, proteins were precipitated with trichloroacetic acid, and the supernatant was for polyamine extraction. Briefly, samples were extracted twice with chloroform, evaporated to dryness, and dissolved in 100 μl of 60% methanol in water. A standard mix containing 0.1 mM of each polyamine was also prepared and benzoylated each time. The set of benzoylated polyamine samples were separated using a Phenomenex Spherclone ODS column (5 μm, 250 x 4.6 mm) that was fitted with a 4.0 x 3.0 mm guard cartridge with the system described above. The runs were performed using a gradient of 45–60% methanol in water for 30 minutes with a 10-minute isocratic equilibration of 45% methanol in water.
Bacteria were diluted in 0.3 ml LB at an OD655 of 0.02 taken using a Bio-Rad iMark MicroPlate Reader and incubated in borosilicate test tubes for either 18, 24, or 34 hours at 27°C or 37°C without shaking. After 18, 24, or 34 hours, planktonic cells were discarded, and the biofilm was washed once with 0.3 ml of 1x PBS. The biofilm was then mechanically disrupted in 0.3 ml of 1x PBS by vortexing with glass beads, and the cell density was measured at OD655 or OD595 depending on the availability of the filter. In assays using the pEVS143 vector system, we found that there was a lot of day to day variation in biofilm formation; therefore, the output was reported as percent wildtype to illustrate the trends in the data better. Three independent colonies were used for each experiment. All experiments were performed in triplicate and repeated multiple times to ensure reproducibility. Student’s t-tests were used to detect significant differences between various strains and treatments.
All the data used to generate the figures and supplemental figures in this study can be found in Supplemental file (S1 File).
Biofilm deficiency of nspC mutants is not due to a defect in vibriobactin production
In V. cholerae, norspermidine forms the backbone of the siderophore vibriobactin. Vibriobactin chelates ferric iron and ultimately transports the iron into the cell [20,35]. Vibriobactin is synthesized by three proteins: VibF, VibB, and VibH. VibB and VibH act as courier proteins by bringing the vibriobactin precursors to VibF, which is responsible for attaching oxazoline rings to a norspermidine backbone (Fig 1B); . Like most other bacteria, V. cholerae requires iron for various cellular processes that are necessary for cell growth and survival . In addition, low iron levels inhibit biofilm formation . The enzyme carboxynorspermidine decarboxylase, which catalyzes the last step of norspermidine synthesis in V. cholerae is encoded by the nspC gene . Deletion of nspC has been shown to lead to a reduction in biofilm formation; however, the mechanism behind this effect has not been elucidated . We hypothesized that the decreased ability of the nspC mutant to form biofilms may be due to the inability of these strains to synthesize vibriobactin, resulting in reduced iron levels in the cell. To test this hypothesis, we first constructed two mutants defective in vibriobactin-mediated iron uptake into cells, ΔvibF and viuA::tetR. The ΔvibF mutant is not capable of synthesizing vibriobactin whereas the viuA::tetR mutant has a disruption in the gene encoding the outer membrane receptor required for uptake of the vibriobactin-Fe3+ complex into the periplasm. As seen previously with the wild type, iron depletion resulted in a decrease in accumulation in biofilms for all of the strains . However, neither of the mutants showed a reduction in biofilm levels in iron-replete or iron-deplete conditions as compared to the wild type (Fig 2A). To our surprise, deletion of vibF generated a 56 and 64% increase in biofilm formation in iron-deplete and iron-replete conditions, respectively, whereas biofilm forming ability of the viuA mutant was similar to the wild-type strain in both conditions (Fig 2A). These results indicate that the reduction in biofilms in nspC mutants is not a result of defective vibriobactin production. Furthermore, inability to synthesize but not to import vibriobactin enhances biofilms.
(A) Biofilm formation of ΔvibF and viuA::tetR mutants. (B) Biofilm formation of ΔvibF, nspC::kanR, and nspC::kanRΔvibF mutants. Biofilms were formed in borosilicate tubes in LB broth for 24 h at 27°C and quantified as described in Materials and Methods. EDDA was added to chelate iron to generate iron-deplete conditions. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between wild type and the mutants. A double star indicates a statistically significant difference between growth media conditions. A p-value <0.05 was considered significant. WT, wild type.
In a vibF mutant, norspermidine is not converted to vibriobactin and may therefore accumulate in the cell. However, in a viuA mutant norspermidine is utilized to synthesize vibriobactin, which is then transported out of the cell. We hypothesized the increase in biofilm formation of the ΔvibF mutant could be due to the accumulation of norspermidine in the cell in the absence of the VibF protein. To determine whether this increase required the synthesis of norspermidine, we constructed a mutant defective in both norspermidine and vibriobactin synthesis, nspC::kanRΔvibF, and assessed its biofilm forming ability. Both the nspC::kanRΔvibF and nspC::kanR mutants exhibited a 40% decrease in biofilm formation in iron-replete conditions compared to wild type. Biofilm formation was reduced even further in iron-deplete conditions (Fig 2B). To further confirm that the altered biofilms seen in the nspC::kanR and ΔvibF mutants are due to the deletions in these genes rather than unintended alterations elsewhere in the genome, we attempted to complement these mutants by introducing the genes on a plasmid. The nspC::kanR mutant containing the empty vector was reduced in its ability to form biofilms compared to the wild-type bacteria (S1 Fig). While this difference was not statistically significant, the trend was the same in each biological replicate. Reintroducing nspC from a plasmid led to a very large increase in biofilm formation, which was higher than that formed by the wild type. We have previously reported that introducing this plasmid into wild-type cells also has the same effect on biofilm formation; the reason for this is not known . Nevertheless, these experiments show that reintroducing the nspC gene corrects the inhibition of biofilm formation. We were not successful in cloning the vibF gene despite multiple attempts; hence, we were not able to perform complementation experiments. Therefore, while it is highly unlikely that the altered biofilm phenotype is a result of an unintended alteration in the genome, this still remains a possibility.
Next, we extracted and quantified cellular polyamines from the wild type and the ΔvibF mutant. Under the conditions of our experiment, V. cholerae can synthesize the polyamines putrescine, diaminopropane, cadaverine, and norspermidine. It does not synthesize spermidine, but can import it from the media which contains spermidine. There was no difference in norspermidine levels between wild type and ΔvibF, indicating increased levels of cellular norspermidine is not the cause of increased biofilms in the ΔvibF mutant (S2 Fig). It remained possible that cellular norspermidine levels did indeed increase initially; however, excess norspermidine was exported out of the cell to maintain homeostasis. To determine whether the increase in biofilm formation seen in the ΔvibF mutant is due to extracellular norspermidine, we quantified the polyamines in the media (S3 Fig). We did not detect any norspermidine in the spent media of either of the strains. This result is consistent with our previous report that showed V. cholerae does not secrete norspermidine . In addition, other extracellular polyamine levels were similar in both of these strains; therefore, differences in levels of other extracellular polyamines cannot explain the difference in the biofilm phenotypes. Thus, the increased capacity of the ΔvibF mutant to form biofilms requires the presence of the nspC gene, but is not a result of increased norspermidine levels in the cell or the presence of norspermidine in the media.
The NspS/MbaA signaling pathway is epistatic over NspC in regulating biofilm formation
Previous studies have found that biofilm levels can be restored in a nspC mutant through the addition of norspermidine to the growth medium . The authors of this study concluded that adequate norspermidine in the cell was required for normal biofilm formation in V. cholerae. Additionally, we have shown that deletion of the nspS gene significantly reduces biofilms; however, exogenous norspermidine is not able to restore biofilm levels in this mutant, suggesting that norspermidine synthesized and/or transported into the cell may not play a role in the ability of V. cholerae to form biofilms . To ensure the drastic biofilm defect in the ΔnspS mutant could be complemented, we reintroduced this gene from a plasmid and compared biofilm formation to that of the wild type and ΔnspS mutant cells carrying the empty plasmid. Complementation with the pnspS plasmid restored the biofilm forming ability to the ΔnspS mutant (S4 Fig). To determine whether nspC or nspS is the dominant mediator of the low biofilm phenotype, we constructed a double nspC::kanRΔnspS mutant. This mutant is incapable of synthesizing norspermidine; therefore, as expected the cells did not contain norspermidine (Fig 3, gray line). Putrescine, diaminopropane, and cadaverine could still be synthesized and spermidine was imported from the media. V. cholerae does not synthesize spermidine under the conditions of our experiment, but is able to import it in a process mediated by PotD1 (21). Inability to synthesize norspermidine also led to increased levels of diaminopropane, a diamine that is used to synthesize norspermidine . Addition of norspermidine to the culture medium resulted in accumulation of norspermidine in the cell (Fig 3, dashed line). As we have previously shown, presence of norspermidine in the culture medium also inhibited spermidine uptake as these two polyamines are presumably imported by the same transporter, PotABCD1 .
Polyamines were extracted from cells, derivatized by benzoylation and analyzed by HPLC as described in Materials and Methods. Labeled peaks on the chromatogram correspond to putrescine (put), diaminopropane (dap), cadaverine (cad), norspermidine (nspd), and spermidine (spd). AU254, absorbance units at 254 nm. Only 4–14 minutes of a 40-minute run are plotted for clarity.
Biofilm formation by the nspC::kanRΔnspS double mutant was comparable to that of the ΔnspS single mutant. In contrast to what was seen with the nspC::kanR mutant, exogenous norspermidine was not able to enhance biofilm formation in nspC::kanRΔnspS, despite the mutant’s ability to import norspermidine (Fig 4A). Furthermore, providing nspS from a multicopy plasmid to the nspC::kanRΔnspS mutant was able to restore the responsiveness of this mutant to norspermidine; whereas providing nspC was not (S5 Fig). These results indicate that nspS is epistatic over nspC in regulating biofilm formation.
(A) Biofilm assay of ΔnspS, nspC::kanR, and nspC::kanRΔnspS mutations, with and without exogenous norspermidine. (B) Biofilm assay of nspC::kanR, ΔmbaA, and nspC::kanRΔmbaA mutations, with and without exogenous norspermidine. Biofilms were formed in borosilicate tubes in LB broth for 18 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between wild type and the mutants. A double star indicates a statistically significant difference between growth media conditions. A p-value <0.05 was considered significant. WT, wild type.
In order to corroborate these findings, we constructed an additional double mutant, nspC::kanRΔmbaA. This mutant is incapable of synthesizing norspermidine and is also void of the phosphodiesterase MbaA, which is hypothesized to interact with NspS in a signal transduction pathway. MbaA is a repressor of biofilm formation and considering that reduced biofilm formation is correlated with degradation of c-di-GMP, it has been suggested that MbaA functions as a phosphodiesterase [8,37]. Indeed, we have previously shown that MbaA is able to break down c-di-GMP to pGpG, confirming its phosphodiesterase activity . The nspC::kanRΔmbaA double mutant formed significantly higher biofilms than wild type, with levels similar to that of the ΔmbaA single mutant (Fig 4B). Furthermore, complementing mbaA from a plasmid was able to restore the nspC::kanR phenotype (S6 Fig). However, norspermidine was not able to increase biofilm formation in this mutant most likely because of the high levels of the MbaA enzyme expressed from the multicopy plasmid (S6 Fig). Taken together, these findings indicate that the NspS/MbaA signaling pathway is the dominant mediator of biofilm formation over synthesis of norspermidine by NspC.
We also investigated whether the increase in biofilms seen in a vibF mutant would still be present in the absence of NspS. Biofilm formation was significantly inhibited in a ΔnspSΔvibF double mutant indicating that NspS is required for the high-biofilm phenotype of the vibF mutant (S7 Fig).
Intracellular norspermidine is not a determinant of biofilm formation
To further delineate the relative contributions of the norspermidine synthesis, transport, and signaling pathways to regulating biofilm formation, we utilized a double mutant incapable of norspermidine synthesis and transport, nspC::kanRΔpotD1. We have previously shown that this mutant is incapable of synthesizing or taking up norspermidine (S8 Fig); . The ΔpotD1 single mutant forms robust biofilms (Fig 5 and ). The mechanism behind this effect is not well understood; however, we have hypothesized that it may be due to the lack of spermidine in the cell. We were able to complement the phenotype of the ΔpotD1 mutant by introducing potD1 in trans on a plasmid (S9 Fig). The nspC::kanRΔpotD1 double mutant exhibited low biofilm levels, similar to the nspC::kanR mutant rather than the ΔpotD1 mutant. Norspermidine addition did not further enhance biofilm formation by the ΔpotD1 mutant, most likely because this mutant already has maximal amount of biofilm that can be supported by these media conditions. However, both the nspC::kanRΔpotD1 and the nspC::kanR mutants responded to norspermidine in the environment in a manner comparable to wild type (Fig 5). Furthermore, we confirmed that NspS was responsible for this norspermidine-dependent increase in biofilms as a nspC::kanRΔnspSΔpotD1 triple mutant had low biofilm levels and was unresponsive to exogenous norspermidine. These results indicate that inability to make or transport norspermidine does not lessen responsiveness to this polyamine. Additionally, these data suggest that intracellular norspermidine is not a determinant of biofilm formation, but rather that norspermidine in the environment acts as a signal through the NspS/MbaA pathway to regulate biofilm levels.
Biofilms were formed in borosilicate tubes in LB broth for 18 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between wild type and the mutants. A double star indicates a statistically significant difference between growth media conditions. A p-value <0.05 was considered significant. WT, wild type. The values for WT and nspC::kanR in Fig 4 are the same as these experiments were performed simultaneously.
Increase in cellular c-di-GMP levels reverses the biofilm defect of ΔnspS mutants
NspS is hypothesized to regulate biofilm formation through its effect on c-di-GMP signaling. More specifically, NspS is thought to inhibit the EAL domain of MbaA, which is responsible for the phosphodiesterase activity that degrades c-di-GMP. Thus, inhibition of MbaA is thought to lead to increased levels of c-di-GMP, which results in increased vps gene expression and biofilm formation . We recently quantified cellular c-di-GMP levels in the ΔnspS and ΔmbaA mutants and found that they were not different than that in the wild-type bacteria despite the drastic effects of this mutation on the biofilm phenotype. Therefore, this pathway is likely to affect local rather than global c-di-GMP pools to transmit signals . We hypothesized that artificially increasing c-di-GMP levels should circumvent the negative effect of a ΔnspS mutation and increase biofilms. To test this hypothesis, we introduced the qrgB gene, encoding a diguanylate cyclase from Vibrio harveyi, from a plasmid into a ΔnspS mutant and assayed biofilm formation . Overexpression of qrgB has previously been shown to enhance V. cholerae biofilm formation. . Indeed, presence of qrgB in the ΔnspS mutant increased biofilm formation by approximately 165% (Fig 6A). To confirm that this effect on biofilm formation was due to the increase in c-di-GMP levels and not a secondary effect, we also introduced a mutated qrgB gene, which is incapable of synthesizing c-di-GMP, into the ΔnspS mutant. Indeed, the ΔnspS with qrgBAADEF mutant , where the conserved glycines in the GGDEF motif have been replaced with alanines, exhibited biofilm levels similar to that of the ΔnspS mutant (Fig 6A). To further confirm the effect of a qrgB addition on biofilm formation, we introduced this gene into the triple mutant, nspC::kanRΔnspSΔpotD1. This mutant cannot synthesize, transport, or detect norspermidine. The nspC::kanRΔnspSΔpotD1 with qrgB also had increased biofilm levels by approximately 108%. However, addition of qrgBAADEF did not rescue biofilm formation; rather, this strain exhibited even lower biofilm levels than the nspC::kanRΔnspSΔpotD1 mutant (Fig 6B). The reason for this effect is not known.
(A) Biofilm assay of ΔnspS with qrgB and ΔnspS with qrgBAADEF mutants. (B) Biofilm assay of nspC::kanRΔpotD1ΔnspS with qrgB and nspC::kanRΔpotD1ΔnspS with qrgBAADEF mutants. Biofilms were formed in borosilicate tubes in LB broth for 18 h at 37°C and quantified as described in Materials and Methods. Relative biomass was calculated using the following equation OD655 mutant/OD655 wild type (Y-axis). Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between wild type and the mutants. A p-value <0.05 was considered significant. WT, wild type.
Previous studies on the effects of polyamines on V. cholerae biofilms have suggested that biofilm formation is enhanced by both intracellular and extracellular norspermidine [8,18,34]. In this study, we sought to determine whether norspermidine present in the cells or the environment is the main determinant that regulates biofilm formation in this bacterium. Considering norspermidine also forms the backbone of the siderophore vibriobactin, we also investigated a potential link between vibriobactin synthesis, iron acquisition, and biofilm formation.
Because V. cholerae requires iron for cellular processes, we considered that perhaps the low biofilms formed by the nspC::kanR mutant were due to the bacterium’s inability to synthesize the siderophore vibriobactin, which contains a norspermidine backbone. Under our conditions, however, the ΔvibF mutant that is incapable of synthesizing vibriobactin exhibited not a decrease but rather a significant increase in biofilm formation. The nspC::kanRΔvibF double mutant formed low levels of biofilm, comparable to the nspC::kanR single mutant. Thus, inhibition of vibriobactin synthesis increases biofilm formation; however, this increase is nullified through inhibition of norspermidine synthesis. Therefore, nspC is epistatic over vibF in regulation of biofilm formation and the inability to synthesize vibriobactin does not diminish biofilm capability under our experimental conditions. We also found that the biofilm forming ability of a viuA mutant, which is capable of synthesizing but not utilizing vibriobactin, was unaffected. Although iron limitation has been shown to inhibit biofilm formation in V. cholerae, a finding we confirmed in this study, the media utilized for these assays contains sufficient iron. In addition, V. cholerae has alternative iron uptake pathways that are presumably used under these conditions ; therefore, it is not surprising that inability to utilize the vibriobactin pathway to acquire iron did not inhibit biofilm formation. Furthermore, as both vibF and viuA mutants are unable to utilize iron through the vibriobactin pathway, they are likely to have similar levels of iron. Therefore, the difference in the biofilm phenotype between these two mutants cannot be explained by iron availability. The most likely explanation is that the cellular consequence of the inability to utilize the norspermidine to produce vibriobactin in the vibF mutant, but not in the viuA mutant, somehow leads to an increase in biofilm formation. We were unable to demonstrate an accumulation of norspermidine in the cell or in spent media that would explain this result. However, polyamine levels are tightly regulated in cells to avoid polyamine toxicity; therefore, it is also not surprising to see unchanged norspermidine levels in the vibF mutant .
Our results indicate that the NspS/MbaA signaling pathway is the main determinant of biofilm formation by V. cholerae in response to norspermidine. In the absence of the NspS periplasmic binding protein, biofilm formation is almost completely abolished and cannot be restored by norspermidine synthesis or import of exogenous norspermidine. Additionally, in the absence of the c-di-GMP phosphodiesterase MbaA, biofilm formation is very high and blocking norspermidine synthesis in this mutant does not have any negative effect on biofilm levels. Although the levels of biofilm formed by both nspC::kanR and ΔnspS mutants are low, our results suggest that the presence of nspS is responsible for restoring biofilm formation in response to exogenous norspermidine; therefore, we conclude that intracellular levels of norspermidine are not a significant determinant of biofilm formation.
Although our data show that intracellular levels of norspermidine do not seem to affect biofilm levels under our conditions tested, it is clear that NspC plays a significant role in biofilm formation of V. cholerae. One explanation could be that intracellular norspermidine synthesized by NspC is exported to the periplasm, where it is then sensed by NspS to enhance biofilm formation. We have previously found that intracellular levels of norspermidine do not increase in a mutant overexpressing the nspC gene. Furthermore, norspermidine is not detected in the spent medium of this mutant, indicating that norspermidine neither accumulates in the cytoplasm, nor is it exported into the extracellular environment . Recently, a novel ABC-type exporter, SapBCDF, was identified in E. coli, which mediates export of putrescine from the cytosol to the extracellular environment to regulate intracellular concentrations of this polyamine . ABC-type transporters are located in the cell membrane; therefore, presumably, this transporter exports putrescine into the periplasm which then diffuses out of porins to the extracellular environment. V. cholerae may utilize a similar polyamine exporter to transport norspermidine from the cytosol to the periplasm in order to regulate intracellular levels. It is possible that norspermidine exported into the periplasm may either be trapped in this compartment by being sequestered by NspS, or accumulate in the periplasm because it is unable to diffuse out of the porins or both. Norspermidine in the periplasm may then act as a signal to regulate biofilm levels through NspS/MbaA (Fig 7). Therefore, periplasmic norspermidine may be a determinant of biofilm formation; however, further experiments must be conducted to test this hypothesis. This model may also explain the increased biofilms we see in the ΔvibF mutant, as excess norspermidine may be exported to the periplasm. The requirement for NspS for the biofilm phenotype of the vibF mutant is also consistent with this model.
Environmental norspermidine may primarily derive from endogenously-produced norspermidine that is released during cell lysis or exported to the periplasm by an unknown transporter. It may also be provided by nearby eukaryotic organisms. Norspermidine may also act as a quorum sensing molecule, allowing V. cholerae to detect this signal, recognize that it is in the presence of other Vibrios, and respond appropriately by forming the Vibrio polysaccharide. In this way, norspermidine may allow V. cholerae to persist in its biofilm form in its natural environment.
Considering that NspS is thought to form a signaling complex with the c-di-GMP phosphodiesterase MbaA, we hypothesized that NspS mediates biofilm formation by modulating levels of c-di-GMP. Therefore, we sought to determine the relationship between NspS and c-di-GMP synthesis by providing a diguanylate cyclase in trans to ΔnspS and nspC::kanRΔnspSΔpotD1 mutants. Indeed, overexpression of the qrgB diguanylate cyclase corrected the low biofilm phenotype of both of these strains. Additionally, overexpressing a mutated form of the diguanylate cyclase that is unable to synthesize c-di-GMP did not rescue the low biofilm phenotype. Thus, while norspermidine acting through the NspS/MbaA signaling complex is epistatic over both norspermidine synthesis and transport, modulating c-di-GMP levels can overcome the effects of mutations in both norspermidine signaling and synthesis pathways. This is consistent with the prediction that the NspS/MbaA signal transduction system affects biofilm formation through regulating intracellular c-di-GMP levels.
Our results indicate that exogenous norspermidine is an important signal for V. cholerae to accumulate in biofilms. This suggests that norspermidine produced by prokaryotic or eukaryotic neighbors of V. cholerae signals an environment favorable for biofilm formation. Although norspermidine is a relatively rare polyamine, it has been found in a number of aquatic plants and lower eukaryotes. In aquatic plants, norspermidine is synthesized by polyamine oxidases as a byproduct of thermospermine metabolism at very low levels [13,14,16,40]. Therefore, it is unlikely that aquatic plants could produce enough norspermidine to support an association with V. cholerae. There are, however, a number of lower eukaryotes that contain higher levels of norspermidine. Norspermidine produced by some of these organisms may be released and could potentially mediate an association with V. cholerae. For example, V. cholerae has been shown to associate with Volvox species isolated from river water  and one species of Volvox has been reported to contain norspermidine . This association presents an intriguing possibility that warrants further investigation.
Another exciting possibility is that norspermidine present in the environment of V. cholerae is primarily derived from endogenously-produced norspermidine that is released during cell lysis by members of the Vibrionales . Norspermidine may then act as a quorum sensing molecule to signal that either more V. cholerae or other Vibrio species are present. This may be recognized as a favorable environment, leading V. cholerae to respond by forming biofilms. Norspermidine may allow V. cholerae to persist longer in its biofilm form in its natural environment, potentially in association with various eukaryotic aquatic organisms or other surfaces, as association with surfaces has been shown to improve survival of bacteria in the aquatic environments [43,44]; (Fig 7).
S1 Fig. Complementation of the nspC::KanR mutant.
Biofilms were formed in borosilicate tubes in LB broth for 24 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference from wild type. A p-value <0.05 was considered significant. pnspC, pACYC184::nspC.
S2 Fig. Polyamine composition of V. cholerae ΔvibF cells.
Polyamines were extracted from cells, derivatized by benzoylation and analyzed by HPLC as described in Materials and Methods. Labeled peaks on the chromatogram correspond to putrescine (put), diaminopropane (dap), cadaverine (cad), norspermidine (nspd), and spermidine (spd). AU254, absorbance units at 254 nm. Only 10–28 minutes of a 40-minute run are plotted for clarity. WT, wild type.
S3 Fig. Quantification of polyamines in spent media. (A) Polyamine composition of V. cholerae ΔvibF spent medium. (B) Polyamine composition of wild-type V. cholerae spent medium.
Polyamines were extracted from media, derivatized by benzoylation and analyzed by HPLC as described in Materials and Methods. Labeled peaks on the chromatogram correspond to putrescine (put), diaminopropane (dap), cadaverine (cad), norspermidine (nspd), and spermidine (spd). AU254, absorbance units at 254 nm. Only 10–28 minutes of a 40-minute run are plotted for clarity. Polyamine std, polyamine standard.
S4 Fig. Complementation of the nspS mutant.
Biofilms were formed in borosilicate tubes in LB broth for 24 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of five biological replicates. A star indicates a statistically significant difference from wild type. A p-value <0.05 was considered significant. pnspS, pACYC184::nspS.
S5 Fig. Role of NspS and NspC on biofilm formation of V. cholerae. Biofilm assay of nspC::kanRΔnspS with pnspS or pnspC, with and without exogenous norspermidine.
Biofilms were formed in borosilicate tubes in LB broth for 24 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between growth media conditions. A p-value <0.05 was considered significant. pnspC, pACYC184::nspC; pnspS, pACYC184::nspS.
S6 Fig. Role of MbaA and NspC on biofilm formation of V. cholerae.
Biofilm assay of nspC::kanRΔmbaA with pmbaA, with and without exogenous norspermidine. Biofilms were formed in borosilicate tubes in LB broth for 18 h at 37°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between wild type and the mutants. A double star indicates a statistically significant difference between growth media conditions. A p-value <0.05 was considered significant. WT, wild type; pmbaA, pVC0703.
S7 Fig. Biofilm assay of ΔvibF, ΔnspS, and ΔnspSΔvibF.
Biofilms were formed in borosilicate tubes in LB broth for 24 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference between wild type and the mutants. A p-value <0.05 was considered significant. WT, wild type.
S8 Fig. Polyamine composition of V. cholerae nspC::kanRΔpotD1 cells, with and without exogenous norspermidine.
Polyamines were extracted from cells, derivatized by benzoylation and analyzed by HPLC as described in Materials and Methods. Labeled peaks on the chromatogram correspond to putrescine (put), diaminopropane (dap), and cadaverine (cad). AU254, absorbance units at 254 nm. Only 10–28 minutes of a 40-minute run are plotted for clarity.
S9 Fig. Complementation of the potD1 mutant.
Biofilms were formed in borosilicate tubes in LB broth for 34 h at 27°C and quantified as described in Materials and Methods. Error bars show standard deviations of three biological replicates. A star indicates a statistically significant difference from wild type. A p-value <0.05 was considered significant. ppotD1, pACYC184::potD1.
We would like to thank Dr. Stephen Calderwood for the pCP20 and pWCW3 plasmids used in the construction of the ΔvibF and viu::tet A mutants and Dr. Chris Waters for the pCMW75, pCMW98, and pVC0703 plasmids. We are grateful to Dr. Marty Roop for helpful discussions and Dr. Tony Michael on helpful discussions and critical reading of the manuscript. We thank Dr. Sam Pendergraft for constructing the AK164 strain.
- 1. Karatan E, Watnick P (2009) Signals, regulatory networks, and materials that build and break bacterial biofilms. Microbiology and Molecular Biology Reviews 73: 310–347. pmid:19487730
- 2. Sutherland IW (2001) The biofilm matrix–an immobilized but dynamic microbial environment. Trends in microbiology 9: 222–227. pmid:11336839
- 3. Donlan RM, Costerton JW (2002) Biofilms: survival mechanisms of clinically relevant microorganisms. Clinical microbiology reviews 15: 167–193. pmid:11932229
- 4. Elasri MO, Miller RV (1999) Study of the response of a biofilm bacterial community to UV radiation. Applied and Environmental Microbiology 65: 2025–2031. pmid:10223995
- 5. Prigent-Combaret C, Vidal O, Dorel C, Lejeune P (1999) Abiotic surface sensing and biofilm-dependent regulation of gene expression in Escherichia coli. Journal of bacteriology 181: 5993–6002. pmid:10498711
- 6. McNeill K, Hamilton I (2003) Acid tolerance response of biofilm cells of Streptococcus mutans. FEMS microbiology letters 221: 25–30. pmid:12694906
- 7. Stanley NR, Lazazzera BA (2004) Environmental signals and regulatory pathways that influence biofilm formation. Molecular microbiology 52: 917–924. pmid:15130114
- 8. Karatan E, Duncan TR, Watnick PI (2005) NspS, a predicted polyamine sensor, mediates activation of Vibrio cholerae biofilm formation by norspermidine. Journal of bacteriology 187: 7434–7443. pmid:16237027
- 9. Tabor CW, Tabor H (1984) Polyamines. Annual review of biochemistry 53: 749–790. pmid:6206782
- 10. Lee J, Sperandio V, Frantz DE, Longgood J, Camilli A, et al. (2009) An alternative polyamine biosynthetic pathway is widespread in bacteria and essential for biofilm formation in Vibrio cholerae. Journal of Biological Chemistry 284: 9899–9907. pmid:19196710
- 11. Hamana K (1997) Polyamine distribution patterns within the families Aeromonadaceae, Vibrionaceae, Pasteurellaceae, and Halomonadaceae, and related genera of the gamma subclass of the Proteobacteria. The Journal of general and applied microbiology 43: 49–59. pmid:12501353
- 12. Hamana K, Sakamoto A, Tachiyanagi S, Terauchi E (2003) Polyamine profiles of some members of the gamma subclass of the class Proteobacteria: polyamine analysis of twelve recently described genera. Microbiol Cult Collect 19: 3–11.
- 13. Hamana K, Matsuzaki S (1982) Widespread occurrence of norspermidine and norspermine in eukaryotic algae. Journal of biochemistry 91: 1321–1328. pmid:7096289
- 14. Hamana K, Niitsu M, Samejima K (1998) Unusual polyamines in aquatic plants: The occurrence of homospermidine, norspermidine, thermospermine, norspermine, aminopropylhomospermidine, bis (aminopropyl) ethanediamine, and methylspermidine. Canadian journal of botany 76: 130–133.
- 15. Zappia V, Porta R, Carteni-Farina M, De Rosa M, Gambacorta A (1978) Polyamine distribution in eukaryotes: Occurrence of sym‐nor‐spermidine and sym‐nor‐spermine in arthropods. FEBS letters 94: 161–165. pmid:700128
- 16. Hamana K, Furuchi T, Nakamura T, Hayashi H, Niitsu M (2016) Occurrence of penta-amines, hexa-amines and N-methylated polyamines in unicellular eukaryotic organisms belonging to the phyla Heterokontophyta and Labyrinthulomycota of the subdomain Stramenopiles. The Journal of General and Applied Microbiology: 2016.2005. 2001.
- 17. Michael AJ (2016) Polyamines in Eukaryotes, Bacteria and Archaea. Journal of Biological Chemistry 291: 14896–14903. pmid:27268252
- 18. Cockerell SR, Rutkovsky AC, Zayner JP, Cooper RE, Porter LR, et al. (2014) Vibrio cholerae NspS, a homologue of ABC-type periplasmic solute binding proteins, facilitates transduction of polyamine signals independent of their transport. Microbiology 160: 832–843. pmid:24530989
- 19. Mey AR, Craig SA, Payne SM (2005) Characterization of Vibrio cholerae RyhB: the RyhB regulon and role of ryhB in biofilm formation. Infection and immunity 73: 5706–5719. pmid:16113288
- 20. Griffiths G, Sigel S, Payne S, Neilands J (1984) Vibriobactin, a siderophore from Vibrio cholerae. Journal of Biological Chemistry 259: 383–385. pmid:6706943
- 21. McGinnis MW, Parker ZM, Walter NE, Rutkovsky AC, Cartaya-Marin C, et al. (2009) Spermidine regulates Vibrio cholerae biofilm formation via transport and signaling pathways. FEMS microbiology letters 299: 166–174. pmid:19694812
- 22. Rogers HJ (1973) Iron-binding catechols and virulence in Escherichia coli. Infection and immunity 7: 445–456. pmid:16558077
- 23. Hanahan D (1983) Studies on transformation of Escherichia coli with plasmids. Journal of molecular biology 166: 557–580. pmid:6345791
- 24. Miller VL, Mekalanos JJ (1988) A novel suicide vector and its use in construction of insertion mutations: osmoregulation of outer membrane proteins and virulence determinants in Vibrio cholerae requires toxR. Journal of bacteriology 170: 2575–2583. pmid:2836362
- 25. Waldor MK, Mekalanos JJ (1994) Emergence of a new cholera pandemic: molecular analysis of virulence determinants in Vibrio cholerae 0139 and development of a live vaccine prototype. Journal of Infectious Diseases 170: 278–283. pmid:8035010
- 26. Haugo AJ, Watnick PI (2002) Vibrio cholerae CytR is a repressor of biofilm development. Molecular microbiology 45: 471–483. pmid:12123457
- 27. Metcalf WW, Jiang W, Daniels LL, Kim S-K, Haldimann A, et al. (1996) Conditionally Replicative and Conjugative Plasmids CarryinglacZα for Cloning, Mutagenesis, and Allele Replacement in Bacteria. Plasmid 35: 1–13. pmid:8693022
- 28. Waters CM, Lu W, Rabinowitz JD, Bassler BL (2008) Quorum sensing controls biofilm formation in Vibrio cholerae through modulation of cyclic di-GMP levels and repression of vpsT. Journal of bacteriology 190: 2527–2536. pmid:18223081
- 29. Dunn AK, Millikan DS, Adin DM, Bose JL, Stabb EV (2006) New rfp-and pES213-derived tools for analyzing symbiotic Vibrio fischeri reveal patterns of infection and lux expression in situ. Applied and environmental microbiology 72: 802–810. pmid:16391121
- 30. Massie JP, Reynolds EL, Koestler BJ, Cong J-P, Agostoni M, et al. (2012) Quantification of high-specificity cyclic diguanylate signaling. Proceedings of the National Academy of Sciences 109: 12746–12751.
- 31. Butterton JR, Choi MH, Watnick PI, Carroll PA, Calderwood SB (2000) Vibrio cholerae VibF is required for vibriobactin synthesis and is a member of the family of nonribosomal peptide synthetases. Journal of bacteriology 182: 1731–1738. pmid:10692380
- 32. Tashima KT, Carroll PA, Rogers MB, Calderwood SB (1996) Relative importance of three iron-regulated outer membrane proteins for in vivo growth of Vibrio cholerae. Infection and immunity 64: 1756–1761. pmid:8613388
- 33. Sobe RC, Bond WG, Wotanis CK, Zayner JP, Burriss MA, et al. (2017) Spermine inhibits Vibrio cholerae biofilm formation through the NspS-MbaA polyamine signaling system. Journal of Biological Chemistry: jbc. M117. 801068.
- 34. Parker ZM, Pendergraft SS, Sobieraj J, McGinnis MM, Karatan E (2012) Elevated levels of the norspermidine synthesis enzyme NspC enhance Vibrio cholerae biofilm formation without affecting intracellular norspermidine concentrations. FEMS microbiology letters 329: 18–27. pmid:22239666
- 35. Butterton JR, Stoebner JA, Payne SM, Calderwood SB (1992) Cloning, sequencing, and transcriptional regulation of viuA, the gene encoding the ferric vibriobactin receptor of Vibrio cholerae. Journal of bacteriology 174: 3729–3738. pmid:1317381
- 36. Mey AR, Wyckoff EE, Kanukurthy V, Fisher CR, Payne SM (2005) Iron and fur regulation in Vibrio cholerae and the role of fur in virulence. Infection and immunity 73: 8167–8178. pmid:16299312
- 37. Bomchil N, Watnick P, Kolter R (2003) Identification and characterization of a Vibrio cholerae gene, mbaA, involved in maintenance of biofilm architecture. Journal of bacteriology 185: 1384–1390. pmid:12562809
- 38. Igarashi K, Kashiwagi K (2010) Modulation of cellular function by polyamines. The international journal of biochemistry & cell biology 42: 39–51.
- 39. Sugiyama Y, Nakamura A, Matsumoto M, Kanbe A, Sakanaka M, et al. (2016) A novel putrescine exporter SapBCDF of Escherichia coli. Journal of Biological Chemistry 291: 26343–26351. pmid:27803167
- 40. Sagor G, Inoue M, Kim DW, Kojima S, Niitsu M, et al. (2015) The polyamine oxidase from lycophyteSelaginella lepidophylla (SelPAO5), unlike that of angiosperms, back‐converts thermospermine to norspermidine. FEBS letters 589: 3071–3078. pmid:26348400
- 41. Tamplin ML, Gauzens AL, Huq A, Sack DA, Colwell RR (1990) Attachment of Vibrio cholerae serogroup O1 to zooplankton and phytoplankton of Bangladesh waters. Applied and Environmental Microbiology 56: 1977–1980. pmid:2383016
- 42. Hamana K, Aizaki T, Arai E, Saito A, Uchikata K, et al. (2004) Distribution of norspermidine as a cellular polyamine within micro green algae including non-photosynthetic achlorophyllous Polytoma, Polytomella, Prototheca and Helicosporidium. The Journal of general and applied microbiology 50: 289–295. pmid:15747227
- 43. Kirchman D, Mitchell R (1982) Contribution of particle-bound bacteria to total microheterotrophic activity in five ponds and two marshes. Applied and Environmental Microbiology 43: 200–209. pmid:16345921
- 44. Pedros-Alio C, Brock T (1983) The importance of attachment to particles for planktonic bacteria. Archiv für Hydrobiologie 98: 354–379.