Life under Nutrient Limitation in Oligotrophic Marine ... - Springer Link

Microb Ecol (2003) 45:203–217 DOI: 10.1007/s00248-002-3008-6 2003 Springer-Verlag New York Inc.

Life under Nutrient Limitation in Oligotrophic Marine Environments: An Eco/Physiological Perspective of Sphingopyxis alaskensis (Formerly Sphingomonas alaskensis) R. Cavicchioli, M. Ostrowski, F. Fegatella, A. Goodchild, N. Guixa-Boixereu School of Biotechnology and Biomolecular Sciences, The University of New South Wales, Sydney, UNSW, 2052, Australia Received: 11 April 2002; Accepted: 26 November 2002; Online publication: 14 March 2003

A

B S T R A C T

The oceans of the world are nutrient-limited environments that support a dynamic diversity of microbial life. Heterotrophic prokaryotes proliferate in oligotrophic regions and affect nutrient transformation and remineralization thereby impacting directly on the all marine biota. An important challenge in studying the microbial ecology of oligotrophic environments has been the isolation of ecologically important species. This goal has been recognized not only for its relevance in defining the dynamics of community composition, but for enabling physiological studies of competitive species and inferring their impact on the microbial food web. This review describes the successful isolation attempts of the ultramicrobacterium, Sphingopyxis alaskensis (formerly described as Sphingomonas alaskensis) using extinction dilution culturing methods. It then provides a comprehensive perspective of the unique physiological and genetic properties that have been identified that distinguish it from typical copiotrophic species. These properties are described through studies of the growth phase and growth rate control of macromolecular synthesis, stress resistance and global gene expression (proteomics). We also discuss the importance of integrating ecological and physiological approaches for studying microorganisms in marine environments.

Introduction Oligotrophic environments are defined by a low nutrient flux of a fraction of a milligram of carbon per liter per day [71] and by low absolute concentrations of nutrients [57]. According to this definition, around one-third of the world’s oceans may be considered oligotrophic. OligoCorrespondence to: R. Cavicchioli; E-mail: [email protected]

trophic environments are generally found in the open sea, while eutrophic regions are typical for coastal areas. Beyond these general principles, upwelling eutrophic waters may also be encountered within oligotrophic zones. Despite the low-level of nutrients in oligotrophic waters, microbial numbers persist on the order of 0.5–5 · 105 cells mL)1 [94]. As a result, marine microorganisms contribute a large proportion of the worlds biosphere

204

in terms of carbon, nitrogen and phosphorus. Furthermore, of the three largest microbial habitats (seawater, soil, and sediment/soil subsurface), the rates of cellular activity and turnover are highest in the open ocean [94]. In this oligotrophic environment, prokaryotes play an essential role in regulating the accumulation, export, remineralization and transformation of the world’s largest pool of organic carbon [16]. Marine bacteria also dominate in terms of biomass. As a result, the open ocean is composed primarily of a microbial food web where prokaryotes represent the most important biological component.It is often assumed that all bacterial cells in a community have similar metabolic activities and therefore contribute equally to biomass and activity. However measurements from oligotrophic environments indicate that not all cells are active [22, 35, 48]. Moreover, while there is a lack of information regarding metabolic activities of specific microbial assemblages in planktonic systems, it would appear that specific prokaryotic taxa play key roles [19, 49, 66, 67].This review examines the advances which have been made in isolating abundant heterotrophic bacteria from oligotrophic environments. In particular it focuses on the ultramicrobacterium Sphingopyxis alaskensis and attempts to integrate an ecological perspective of this bacterium with a growing understanding of its physiology.

Oligotrophic Isolates and Community Composition The microbial community composition in oligotrophic marine waters has been studied using molecular and culturing methods. Molecular analyses have revealed a broad diversity of bacteria, particularly members of the a- and cProteobacteria, as well as members of the Cytophaga– Flavobacterium–Bacteroides group. Studies comparing culture-dependent with culture-independent methods have often produced conflicting results about community composition. A good example is the high abundance of members of the Photobacterium and Vibrionaceae groups, which are frequently isolated from oligotrophic waters, but have lower representation in molecular analyses [26]. This may reflect biases in clone libraries (overrepresentation of particular clones) or growth media which support how low-abundance species (often copiotrophs) but not predominant species. Very recently Bruns et al. [6] identified

R. Cavicchioli et al.

discrepancies between culturing (in this case MPN) and molecular methods for samples isolated from water and sediment samples in the central Baltic Sea. They concluded that a very high level of microbial diversity in the samples, in combination with a small sampling size of isolates, may produce significantly different outcomes between the methods. A number of excellent reviews cover this topic [36, 50]. However, irrespective of the differences between the culturing and molecular methods, it is clear that physiological studies necessitate isolates, which demands that culturing methods (e.g., [6]) continue to be developed. The difficulty in isolating oligotrophs is well documented (reviewed in [79]). A fundamental problem is that obligately oligotrophic bacteria are inherently sensitive to nutrients. As a result, only oligotrophs able to adapt to the nutrient composition of the solid growth media will be isolated. The ability to adapt therefore implies that the isolates are facultative oligotrophs. A number of aquatic facultative oligotrophs have been isolated including Caulobacter, Hyphomicrobium [55, 71], Cydoclastus oligotrophus [93], and S. alaskensis [76, 79]. Some of the factors which may restrict the ability to adapt and isolate oligotrophs include (1) intolerance to high concentrations of nutrients, (2) inappropriate growth substrates, (3) the absence of specific vitamins or growth factors, (4) inhibitory growth substrates or other additives, (5) inactivation by the close proximity to other cells (in colonies on agar plates), (6) susceptibility to the oxidative respiratory burst upon upshift and outgrowth in the presence of fresh nutrients, and (7) the deleterious effects of lytic phage. The most successful isolation technique is the extinction dilution method [7], which was used to obtain numerically abundant strains of S. alaskensis [7, 77] and C. oligotrophus [7, 93]. In the original description of this procedure seawater samples were diluted in filtered and autoclaved natural seawater without additional nutrients until only a few organisms remained in each dilution tube [7]. Long-term incubation of these cultures at stationary phase (6–12 months) in the dark and at 5C initiated an unknown mechanism that enabled the cells to grow on a rich nutrient medium [76]. The nature of cell transformation from an initially obligate oligotrophic to facultative oligotrophic state is still unclear [80]. However, despite an incomplete understanding of the mechanisms of adaptation, the method has proven to be reproducible with the isolation of a new strain of S. alaskensis from

Life under Nutrient Limitation

oligotrophic waters [25]. Other reports exist on the use of the extinction dilution method [4, 26]; however, they have not focused on the isolation of numerically abundant, oligotrophic isolates. An exception to this is the very recent report of the cultivation of a new a-proteobacterium [73]. Despite the inability to culture the isolate on solid media, the capacity to generate biomass provides a rare opportunity for beginning to characterize a member of the ‘‘SAR11’’ group of bacteria. Due in part to the inability to cultivate other suitable isolates, the majority of studies on the eco/physiology of oligotrophs have focused on S. alaskensis.

Predicting the Properties of Typical Oligotrophic Bacteria A description of characteristics for a model oligotroph was introduced in the Dahlem Conference in 1979 by Hirsch et al. [38] and was structured according to a role in nutrient uptake or nutrient utilization. Nutrient uptake characteristics included a high surface-to-volume ratio (cells were expected to be small); preferential usage of metabolic energy for nutrient uptake especially during nongrowth periods; a constant nutrient uptake ability; possession of high-affinity, low-specificity transport systems for simultaneous uptake of mixed substrates; and the establishment of accumulation reserves following nutrient uptake [38]. Poindexter [71] extended the qualification of a ‘‘high-affinity’’ nutrient uptake system to include either a low-energy activation system for substrate binding (a high number of binding sites per uptake site), or a system with efficient conformational changes of the binding site to minimize energy requirements for every uptake event. Nutrient utilization systems were expected to be efficient and inducible, carriers to be constitutive, and only a minimal level of catabolite repression to occur to ensure simultaneous utilization of mixed substrates. In addition, biosynthetic rates were expected to be regulated in accordance with the rate of nutrient uptake [71]. The final nominated characteristic was the ability to store diverse nutrients as reserve materials [38]. Since these descriptions for a model oligotroph were proposed, a range of physiological studies have been performed to assess them (reviewed in [79]). It is important to note, however, that many of these studies have not been conducted with oligotrophic microorganisms, thereby highlighting the need to do so with appropriate isolates such as S. alaskensis (see below).

205

Ecology and Isolation of S. alaskensis S. alaskensis ultramicrobacteria have been isolated from three geographically separated locations. S. alaskensis RB2256 and a number of other isolates were obtained from Resurrection Bay in Alaska and morphologically and physiologically similar strains were isolated from the North Sea. Strain RB2256 was isolated from a 106-fold dilution of sea water where the standing bacterial population was 0.2 · 106 cells mL)1. The presence of this species in Resurrection Bay and the North Sea was also demonstrated by the hybridization of species-specific probes FP1 and FP4 to eight isolates from Resurrection Bay and one isolate (strain NS1619) from the North Sea [77]. The persistence of this species in Resurrection Bay was shown by the detection of cells using Southern hybridization of extracted community DNA, more than 2 years after the initial isolation. The newest isolate, strain AFO1 from North Pacific waters, was characterized using DNA–DNA hybridization, 16S rRNA sequencing, lipid profiling, genome size, morphology and stress resistance, and was demonstrated to be a new strain of the same species (e.g., 80% DNA–DNA hybridization). Strain AFO1 was isolated from seawater with a bacterial count of 3.1 · 105 cells mL)1 using a dilution of 105 [25]. Strains AF01 and RB2256 were isolated from depths of 350 m and 10 m, respectively. The circulation of ocean currents in the North Pacific provides a means for distributing these strains between these two sites, which are about 10,000 km apart. These studies indicate that S. alaskensis was one of the most numerically abundant microorganisms at geographically remote sites and over a period of time spanning 10 years. This indicates that it has the capacity to be a major contributor to microbial biomass in oligotrophic marine waters and may have a significant impact on nutrient cycling in these environments. To date, the S. alaskensis strains which have been formally described include the type strain, RB2256, strains RB255, RB2515, RB2510 [89], and strain AFO1 [25]. The strains from the North Sea are no longer available [J. Gottschal, personal communication]. A range of other isolates of the Sphingomonas genus have been isolated from oligotrophic and eutrophic marine environments, ranging from polar to temperate waters and including coral pathogens and hydrocarbon degraders [12, 26, 68, 69]. It is noteworthy that prior to 1990 the genus Sphingomonas was described as Pseudomonas [97]. Re-

206


Fig. 1. Distance-matrix tree highlighting the phylogenetic position of Sphingopyxis 16S rDNA sequences. DNA sequences corresponding to the E. coli 16S rRNA gene positions 27 to 1433 were aligned using the programs PILEUP and ClustalX [87]. Genetic distances were calculated using the method of Jukes and Cantor, and the phylogenetic tree was reconstructed using the neighbor-joining algorithm of Saitou and Nei [75] as implemented within ClustalX. The phylogenetic tree was plotted using the program nj-plot. Numbers on branches represent bootstrap values for 1000 repeats [30]. The bar indicates 2 nucleotide changes per 100. Accession numbers: (1) AF145754, (2) AF378795, (3) AF378796, (4) AF145753, (5)AB022601, (6) AB015049, (7) AF367204, (8) AF181572, (9) D13723, (10) AF327069, (11) AY081981, (12) D17322, (13) D13727, (14) U20756, (15) X94102, (16) D16144, (17) X87161 (18) AF125194, (19) M96746, (20) AF510191, and (21) AF327028.

cently the genus Sphingomonas was reclassified into four genera [86]. S. alaskensis was not described in this study; however, it is clear from our phylogenetic analysis that it clusters with Sphingopyxis, including the type strain Sphingopyxis terrae (Fig. 1). We therefore propose that the species now be refered to as Sphingopyxis alaskensis.

Description and Physiology of Sphingopyxis alaskensis Ultramicro Size and Genetic Capacity S. alaskensis is rod shaped (0.3 · 0.9 lm) with a typical volume of 0.05 lm3 [24, 25, 77], which places it in the class ultramicrobacteria (less than 0.1 lm3) [14, 79]. This characteristic is shared by all members of the species [24, 25]. When grown in defined or complex marine media, cells exhibit only minor changes in size (approximately two-fold) irrespective of growth phase or media richness. The retention of an ultramicro size distinguishes S. alaskensis from other bacteria from oligotrophic waters which increase cell volume during cell growth (e.g., Vibrio angustum S14).

A consistent finding in field studies is the observation of cells with an ultramicro size [79]. The isolation and demonstration that S. alaskensis retains a permanent ultramicro size demonstrates that in the oligotrophic marine environment ultramicrobacteria have the potential to be metabolically active and contributing to mineralization. The fact that S. alaskensis is an ultramicrobacterium has an impact on discussions concerning the minimum size of a structure that may support life [13, 14]. Reports of the presence of nanobes and other structures in the Martian meteorite ALH84001 [53], subterranean rocks [88], human kidney stones [42], and other environments [32] are extremely controversial [1, 15, 44, 51, 99]. In this regard, the dimensions of S. alaskensis provide an upper limit for the minimum size of a self-replicating cell and demonstrate that cells as small as 0.05 lm3 are not only viable, but capable of proliferating in natural environments [13, 14]. Despite the small size of S. alaskensis, a moderate genome size is maintained. Estimates of the genome size have been derived from flow cytometry [8, 9, 77], and more recently it was determined by pulse-field gel electrophoresis to be 3.1–3.2 Mb [25]. Interestingly, Prochlo-


rococcus marinus has larger cellular dimensions than S. alaskensis and yet possesses a smaller, 1.8-Mb genome [85]. This indicates that genome size alone does not directly correlate with cell size. Moreover, it does not substantiate the suggestion that an oligotrophic bacterium would need to have a minimal genome [8]. A significant genetic complement may be required to effectively compete in an oligotrophic environment, which may include coping with stress and undergoing rapid responses to nutrient fluxes, including outgrowth from starvation [25]. General Growth Characteristics S. alaskensis is an obligately aerobic bacterium that forms opaque yellow, low convex, entire colonies on solid medium [25, 77, 89]. It has a specific growth rate (0.13–0.16 h)1 at 23C) which remains largely unchanged in VNSS with carbon concentrations between 0.8 and 800 mg L)1 [24]. Growth rate is similar in defined ASW with glucose as carbon source or complex VNSS medium. Cells grow marginally faster with trehalose (0.25 h)1 at 30C) as a carbon source in ASW compared with glucose (0.19 h)1) as the carbon source [Goodchild and Cavicchioli, unpublished results]. As trehalose is a disaccharide of glucose (thereby utilizing equivalent catabolic pathways), a higher growth rate with trehalose may indicate that the transport system(s) are more efficient for trehalose than for glucose. A number of other substrates can be used as sole sources of carbon including malate, acetate, alanine, and mixed amino acids, resulting in specific growth rates of 0.14, 0.02, 0.04, and 0.10 h)1, respectively [[24, 78]; Goodchild and Cavicchioli, unpublished results]. Amino acid transport is facilitated by high-affinity, broad-specificity uptake system(s). The alanine uptake system has an affinity for alanine that may exceed that of any previously reported transport system, is constitutive, and is capable of transporting nine other amino acids [78]. In contrast, the glucose uptake system is inducible and has narrow substrate specificity. Interestingly, alanine and glucose are simultaneously utilized and enable cells to grow at maximum specific growth rates that exceed growth with individual substrates. Both uptake systems use periplasmic binding proteins which provide unidirectional transport of substrates against a 105-fold concentration gradient. The uptake systems would enable S. alaskensis to efficiently scavenge and utilize substrates from the environment. Based on the uptake kinetics of these systems,

207

strain RB2256 could grow by using DFAA at an in situ doubling time of 12 h to 3 d [78, 80], which compares favorably with measured doubling times for bacteria in oligotrophic waters of 5 to 15 d [34]. It is also noteworthy that the ability to scavenge nutrients efficiently is a trait that was theoretically considered to be essential for an oligotrophic bacterium [38, 71]. Ribosomes Ribosome, protein and RNA content, and rates of macromolecular synthesis have been examined in S. alaskensis RB2256 [27, 29]. Ribosome content peaks during logarithmic growth at 2000 ribosomes per cell and decreases rapidly to 200 ribosomes per cell during late-logarithmic growth, and remains at this level for up to 7 d of starvation. The ribosome content is low compared to faster growing copiotrophs such as Escherichia coli, which have 6800 ribosomes per cell when grown at 0.6 h)1 and 72,000 at 2.5 h)1 [5]. The low number of ribosomes per cell is consistent with S. alaskensis containing a single copy of the rRNA operon, in contrast to 8–11 copies in marine Vibrio species [27]. The low ribosome content is also consistent with the inability to detect S. alaskensis using FISH, despite the ability to detect it using PCR [76]. It will be valuable to determine the ribosome content in other abundant isolates from oligotrophic environments to establish whether this is a common occurrence. Clearly a low ribosome content limits the utility of studies that rely on FISH for examining community composition. While the number of ribosomes per cell and rRNA operon copy number provide an explanation for the relatively low growth rate [46], a number of important observations indicate that ribosome content does not, in fact, limit growth rate. First, if the ribosome content is calculated on a per volume basis (rather than number per cell), the concentration of ribosomes (up to 40,000 ribosomes um)3) is equivalent to E. coli growing at a 10-times faster rate [27]. Secondly, when starved cells are subjected to nutrient excess they are able to immediately resume growth at maximum specific growth rates, despite the ribosome content being 10% of maximum cellular levels [27]. Thirdly, the number of ribosomes per cell peaks and subsequently falls to minimum levels before rates of protein synthesis and total protein content reach maximum levels, indicating that protein synthesis is, at times, uncoupled from ribosome synthesis [29]. These data collectively indicate that ribosomes do not limit growth rate and

208

appear to be present in excess during some stages of the growth phase. It is possible that ribosomes have a role in the cell in addition to protein synthesis. This may include a role as a protein storage reserve, or by fulfilling alternative functional role(s). Ribosome-associated initiation and elongation factors in E. coli have been proposed to function as chaperones and assist in protein (re)folding [10], and similar roles may exist for ribosomes. In E. coli, during starvation two 70S ribosomes dimerize to form a 100S ribosome [91]. They are essential for cell survival during starvation [98], and while they do not participate in translation, they are thought to represent a storage form that may be used as a source of nutrients during outgrowth [90, 91]. In comparison to E. coli, during starvation in V. angustum S14 the majority of ribosomes remain 70S and capable of supporting protein synthesis, albeit at a 200-fold lower rate than during exponential growth [31]. The form of the ribosomes in S. alaskensis have not been examined. However, the rapid and immediate outgrowth response [27, 29] indicates that 70S ribosomes must be present, and it indicates that during starvation the ribosomes perform a function more in common with V. angustum than with E. coli. This is further supported by the fact that after 4 to 7 d starvation in E. coli, 100S ribosomes dissociate and cells rapidly lose viability [92], whereas viability is unaffected in S. alaskensis and V. angustum over the same period. This indicates that while ribosomes appear to have roles in addition to protein synthesis (e.g., when they are in excess), their maintenance during 7 d starvation is likely to be for protein synthesis and the outgrowth response of the cell. Starvation and Stress Resistance Irrespective of the whether a microorganism is copiotrophic or oligotrophic, it will encounter periods of starvation in the marine environment. The potential to starve can be rationalized on the basis of the total amount of carbon available in oligotrophic regions, relative to the amount of carbon required for the bacterial standing stock to perform one round of cell division [29]. It may also be considered as a result of the heterogeneous distribution of nutrients such as micro-patches of nutrients [2] which may form nutrient gradients around a nutrient source [29]. Changes in the type of nutrients that limit growth may also change the limiting nutrient and hence the cause of starvation. This may occur as a result of diel cycling [47]. All


of these are good reasons to consider that oligotrophic bacteria may be exposed to periods of starvation. The carbon starvation response in S. alaskensis RB2256 differs in many ways from the response in copiotrophs [24, 27, 29, 76, 79]. In addition to the absence of reductive cell division which is typical of copiotrophs such as Vibrio ANT 300, V. angustum S14, Aeromonas, Pseudomonas, and Alcaligenes [39, 45, 61], in S. alaskensis, starvation does not induce cross protection to the stress-inducing agents hydrogen peroxide, ethanol, heat [24], and UV-B [41]. Instead, S. alaskensis is inherently resistant to these stresses. For example, the viability of growing cultures decreases less than two-fold after 30 min exposure to 25 mM hydrogen peroxide [24, 25], whereas the viability of V. angustum S14 decreases 105-fold after exposure to only 2 mM hydrogen peroxide for 30 min [64]. Interestingly, starvation does confer a measurable level of resistance to pH stress and sonication [Eguchi et al., unpublished results]. This indicates that while the starvation response in S. alaskensis differs markedly from that in other bacteria, some parallels exist. This physiological manifestation of the starvation response is consistent with the large changes in gene expression which are observed during carbon starvation (see below). The inherently high levels of stress resistance in S. alaskensis extends to antibiotics (e.g., ampicillin, gentamycin, streptomycin, tetracycline) [Cavicchioli et al., unpublished results]. This may be a general characteristic of Sphingomonas, Sphingopyxis, and related genera as a high level of resistance to streptomycin has previously been reported [72]. Using an electrophoretic mobility assay, S. alaskensis cells were found to have a low surface charge [Morisaki and Eguchi, unpublished results]. This may facilitate resistance to charged antibiotics by minimizing their interaction with the cell [3]. One of the few agents to which S. alaskensis appears to be sensitive is the detergent SDS [Cavicchioli et al., unpublished results]. Like most gram-negatives, S. alaskensis contains phosphotidylethanolamine as the major phospholipid component [25]. Its presence in the membranes relies on interactions with other phospholipids, proteins, and stabilization by cationic interaction with the charged head group, and all of these interactions may be destabilized by SDS [D. Nichols, personal communication]. The resistance of S. alaskensis to a broad range of stressing agents, and the sensitivity to SDS, indicates that the membrane plays an important role. However, it is unlikely that the membrane acts simply as a physical


barrier preventing all stress-inducing agents from entering the cell. For example, hydrogen peroxide and ethanol would freely diffuse into the cell and yet the cells are very resistant. The resistance mechanism may involve nonspecific, high-efficiency export and/or membrane detoxification systems. It is also possible that detoxifying compounds are actively targeted to the membrane and/or secreted. Evidence for this comes from experiments on ethanol resistance. When logarithmic-phase cultures are starved by resuspending cell pellets in nutrient-free medium, the cells show significantly higher levels of sensitivity compared with actively growing or 24-h starved cells [24]. This indicates that a component in the medium which was produced by the cells (during growth or starvation) augments their resistance to ethanol. Consistent with this, the cells produce a compound which absorbs strongly in the UV spectral range [Ostrowski and Cavicchioli, unpublished data]. This component is unrelated to the carotenoid components (probably nontoxanthin] which are also produced by the cells [[40]; Ostrowski et al., unpublished results] and which may be involved in stabilizing the membrane and resistance to UV and oxidative stress [33, 65].

Physiological Responses to Starvation The outgrowth response of S. alaskensis appears to be different to copiotrophs. Following 24-h carbon starvation in defined medium, cells resume maximum rates of growth without a detectable lag [24, 27]. Cells starved for longer periods (up to 7 d) also resume growth without a lag, albeit at a lower growth rate. If a typical stringent response [11] occurred, a growth lag would be expected while the cell biosynthesizes a sufficient amino acid pool to accommodate growth demands. The lack of stringent response is consistent with the pattern of rates of protein synthesis which occurs as cells enter starvation. In contrast with a spike in rate of protein synthesis which accompanies the initial starvation phase, rates of synthesis remain low in S. alaskensis [29]. Throughout the growth phase, the concentration profile of cAMP in the cell matches the pattern of protein synthesis (Fig. 2). The main peak in intracellular cAMP levels during late-logarithmic growth coincides with a rapid decrease in glucose concentration in the medium (Fig. 2). Increases in cAMP concentration and cAMP-dependent

209

Fig. 2. cAMP levels throughout the growth phase in S. alaskensis RB2256. Growth-phase dependent synthesis of intracellular cAMP (d), rates of protein synthesis (e), and glucose concentration of the growth medium (j) plotted against the optical density (433 nm) of S. alaskensis. Cultures were grown at 30C in ASW medium with 3 mM glucose as described previously [24]. For cAMP determinations, 5 mL of cells was collected by centrifugation, snap frozen in liquid nitrogen, and stored at )20C, prior to assaying with a BioTRAK enzyme-linked immunoassay kit (Amersham) according to manufacture’s instructions (Protocol 1: Non-acetylation EIA procedure). cAMP data are from the average of duplicate measurements with a maximum coefficient of variation of 29%. The remaining data are from Fegatella et al. [27].

gene expression are strongly correlated with changes in nutrient concentration [59, 60]. This may indicate that the cells respond to falling levels of glucose by increasing synthesis of cAMP. cAMP may be involved in inducing gene expression during late-logarithmic growth, which leads to the downregulation of macromolecular synthesis and the preparation of the cell for the onset of starvation. In E. coli, a similar peak in cAMP synthesis occurs prior to starvation; however, while intracellular levels in E. coli remain constant for 4 d starvation [21, 60], they drop approximately 10-fold in S. alaskensis over the same time period (data not shown). In addition to cAMP, other small intracellular molecules, including (p)ppGpp [11] and acetyl phosphate [62] are known to play a role in regulating starvation gene expression. To date, these have not been examined in S. alaskensis. Extracellular, quorum-sensing compounds may also contribute to regulation [20, 83, 84]. The presence of extracellular acetylated-homoserine lactones from S. alaskensis was examined using Chromobacterium violace-

210


Fig. 3. Western-blot analysis of RpoS in S. alaskensis and E. coli. Samples for Western blots (C, D) were taken throughout the growth phase of S. alaskensis (B) and E. coli (A). A total of 20 lg protein from whole cell lysates of E. coli K12 and S. alaskensis RB2256 was treated with SDS sample buffer [0.05 M Tris, pH 6.8; 0.025% (w/v) bromphenol blue; 2% SDS; 10% (v/v) glycerol] and separated on 12% polyacrylamide gels. Proteins on gels were transferred to polyvinylidene difluoride membranes (Amersham), blocked overnight in 5% skim milk in TTBS [10 mM, Tris-HCl; 0.1 M, NaCl; 0.1% (v/v) Tween-20], probed with E. coli polyclonal anti RpoS antibody (KT-2, supplied by Prof. Ishihama, National Institute of Genetics, Department of Molecular Genetics,

Japan), and incubated with goat-anti-rabbit IgG alkaline-phosphate conjugate (Immunopure, Pierce). The blots were developed with a chromogenic substrate (BCIP/NBT). Boxed protein bands in panel C indicate RpoS in E. coli. Arrows denote bands with nonspecific binding in the negative control lane (N), which was obtained from an E. coli sample not treated with RpoS antibody. The size standard (M) was the Bio-Rad Unstained Precision Protein Marker stained with Coomassie Brilliant Blue R250 and is identified with the molecular weight (kDa) of the proteins. Sample numbers from the growth curves (A, B) correspond to lane numbers (C, D).

um and Agrobacterium tumefaciens [52, 81]; however, no activity was detected in the bioassays [Fegatella et al., unpublished results]. Clearly a variety of metabolites could be important to the cell. Future studies will have to address this, not only for the purposes of understanding the physiology of S. alaskensis, but also for the effects the metabolites may have on other marine microorganisms (e.g., community composition). The potential effects are illustrated by the broad-range of antimicrobial activities displayed by the surface-colonizing bacterium Pseudoalteramonas tunicata [23]. Despite the considerable potential ecological impact, few studies of this kind have been performed using natural free-living marine bacteria [6].

Regulatory molecules directly or indirectly affect proteins involved in regulating gene expression (e.g., cAMPCRP). The most well characterized starvation-responsive global regulator is RpoS (rs). Effector molecules which directly interact with RpoS have not been identified; however, these will be conveyed through cognate partner proteins. Interestingly, while RpoS plays a pivotal role in the starvation response of c-Proteobacteria such as E. coli and V. angustum, the rpoS gene does not appear to be present in many a-Proteobacteria [74], including the available DNA sequence databases (data not shown). Polyclonal antibodies raised against E. coli RpoS were used to examine the presence of RpoS in S. alaskensis (Fig. 3).


211

Despite the specific detection of a band of approximately the correct molecular weight, band intensity remained constant (Fig. 3B, D). This was in contrast to growthphase-regulated expression of the protein in E. coli (Fig. 3A, C). It remains to be determined whether the reactive band is an RpoS homolog or not. Recently, it was found that rpoS was not present in the genome of the c-ProteobacteriumCampylobacter jejuni [43]. Furthermore, the aProteobacterium C. crescentus was shown to express a starvation-specific, transcriptional regulatory protein SkgA, which was important for hydrogen peroxide resistance [74]. The presence of bacteria specific regulators, and possibly multiple regulators, highlights the complexity of control pathways which is likely to exist in S. alaskensis. A useful approach to unraveling these networks and identifying key regulators is to use proteomics (see section below).

Physiological Responses to Nutrient-Limited Growth In addition to being exposed to periods of starvation in the ocean, the predominant nutritional state is likely to be one of nutrient limitation. Perhaps more than any other physiological ability, an oligotrophic bacterium must possess the capacity to proliferate (rather than cease growth) when nutrients are limiting. Chemostats provide the practical means of studying cells during nutrientlimited growth [37]. Using mini-chemostats, glucose-limited growth has been shown to have a profound effect on the physiological state of S. alaskensis [64]. At a specific growth rate of 0.14 h)1 or higher, chemostat grown cells are as inherently resistant to stress (e.g., hydrogen peroxide) as logarithmic or starved batch grown cells. However, cells grown at 0.13 h)1 or less are 104-fold more resistant. The sudden change in stress resistance is also observed for cells that are limited for carbon using mixed amino acids [64]. In contrast, nitrogen-limited cultures are less resistant and do not display growth-rate-dependent resistance to hydrogen peroxide, thereby indicating that the stress resistance is mediated by carbon- and/or energylimited growth. We recently extended this work by examining other sources of nutrients for carbon limitation. Malate-limited cultures exhibited a similar degree of hydrogen peroxide resistance to glucose-limited cultures; cells grown at 0.13 h)1 were 103-fold less resistant than cells grown at 0.06 h)1 or 0.1 h)1. In contrast, trehalose-limited cultures were

Fig. 4. Effect of growth rate during glucose-limited growth on stress resistance in S. alaskensis RB2256. Cells were grown at low (0.02 h)1 to 0.07 h)1; solid bars), medium (0.08 h)1 to 0.13 h)1; horizontal hatched bars) or high (0.16 h)1 to 0.20 h)1; open bars) growth rates and exposed to 25 mM hydrogen peroxide for 60 min, 20% ethanol for 15 min, heat stress at 56C for 15 min, ultraviolet (UV) light at 2000 Jm)2, or 500 mM paraquat for 15 min. Handling and treatment of chemostat cultures was described previously [64]. Procedures for heat and ethanol stress are described in Eguchi et al. [24], and for UV treatment in Joux et al. [41] using an Amersham UVR crosslinker, model RPN 2500. Paraquat (methyl viologen) resistance was determined in a similar way to hydrogen peroxide resistance [64]. Colony-forming units were determined on Marine Broth 2216 (Difco) agar. The data are the average from two independent experiments, with the exception of the hydrogen peroxide data which were from Ostrowski et al. [64].

more sensitive to hydrogen peroxide and largely unaffected by growth rate. Survival after 15 min in 25 mM hydrogen peroxide was 0.06, 0.01, and 0.01% for cells grown at 0.06, 0.1, and 0.16 h)1, respectively. This was in contrast to 85% survival for glucose-limited cultures at 0.08 h)1, after 60 min in 25 mM hydrogen peroxide [64]. As trehalose is a disaccharide of glucose, the results were surprising. The data may indicate that substrate transport, but not carbon metabolism is linked to the growth-rate control of hydrogen peroxide resistance.

212

Fig. 5. Growth yield (CFU mL)1) of S. alaskensis RB2256 (j) and V. angustum S14 (r) grown in glucose-limited chemo-

stats at different specific rates of growth. Cultures were grown aerobically in ASW medium supplemented with 3 mM D-glucose at 30C as described previously [64]. The effect of growth rate was further examined by studying the resistance to stresses other than hydrogen peroxide. Glucose-limited growth had the greatest effect on hydrogen peroxide resistance with less impact on resistance to heat, ethanol, UV, or paraquat (Fig. 4). Cells at the highest growth rates were always less resistant; however, the magnitude of the effect was dependent on the stress. The effect of growth rate was also examined on growth yield. A comparison was performed between S. alaskensis and V. angustum to determine whether S. alaskensis had a superior ability to convert substrates into biomass at low growth rates. In glucose-limited chemostats, the growth yield of S. alaskensis was consistent between growth rates from 0.02 h)1 to maximum growth rates (0.20 h)1) (Fig. 5); in fact, growth yields tended to increase slightly as growth rates were lowered. In contrast to S. alaskensis, the growth yield of V. angustum decreased to approximately 10% of maximum levels at the lowest growth rates tested (0.02 h)1). Growth yields started to decrease when growth rates fell below approximately 25% of maximum growth rates (Fig. 5). Above a growth rate of 0.20 h)1, growth yields for V. angustum were constant and equivalent to the growth yield obtained in batch cultures. The data for CFU were confirmed by optical density (433 nm) and direct counts of DAPI-stained cells for both S. alaskensis and V. angustum (data not shown).


These results may be explained in terms of maintenance energy demand. It has been postulated that below a threshold rate of growth, significant amounts of available energy are diverted to maintenance and survival functions rather than biomass production [70]. Our data indicate that the maintenance energy demand remains the same in S. alaskensis, even at low growth rates, whereas in V. angustum it increases below a growth rate of 0.20 h)1. It will be important to repeat these experiments using lower concentrations of substrate and different sources of nutrients (e.g., amino acids) to determine the responses of the two bacteria under conditions which more closely resemble natural conditions. However, the present data support the concept that oligotrophic bacteria are efficient at utilizing available nutrients, thereby enabling them to produce more cells from the same substrate pool as compared to other bacteria.

Global Changes in Gene Expression during Starvation and Nutrient-Limited Growth Two-dimensional gel electrophoresis (2DGE) was developed to examine global changes in gene expression in S. alaskensis RB2256 [28]. During starvation for 24 to 36 h the number of protein spots with increased or decreased intensity was 72 and 177, respectively, out of a total of 1500 spots [29]. A comparative study in V. angustum S14 revealed 157 and 144 changes in spot intensities [63]. It is possible that the larger proportion of starvation-induced proteins in V. angustum S14 relates to reductive cell division processes and acquisition of stress resistance [29]. It is clear, however, that a considerable number of changes in gene expression take place when S. alaskensis transits from growth to starvation. Proteomics has also been used to examine gene expression during nutrient-limited growth. In contrast to the high proportion of differentially expressed spots observed between starved and logarithmic-phase batch cultures, protein profiles of cells from glucose-limited chemostats grown at 0.026, 0.076, and 0.18 h)1 revealed comparatively few changes [64]. A total of 12 spots differed by more than two-fold between gels from any two growth conditions, with 7 specific to low rates of growth (0.026 and 0.076 h)1) and 3 specific to a high rate of growth (0.18 h)1). It was striking that even though the total numbers of spots on gels was equivalent to the number observed for starvation regimes, the number of differentially expressed proteins was much lower. Some of this difference may relate to the


use of radioactive imaging for the batch experiments [28]. However, it would appear that comparatively few changes in cellular proteins are required to dramatically change the stress resistance capacity of the cell [64]. Recent advances in mass spectrometry have provided new approaches for determining the identities of proteins from 2DGE [18, 54, 58, 95]. When coupled with genome sequence data, identification is relatively straightforward. However, even cross-species identifications are possible [17, 56, 82, 95, 96]. Using combinations of data from MALDI-TOF peptide-mass fingerprinting, MCLC MS/MS and amino acid analysis produced by mass spectrometry, and N-terminal sequence from Edman degradation, several proteins were identified from 2DGE of S. alaskensis [Ostrowski et al., manuscript in preparation]. The ability to identify proteins provides new scope for understanding the molecular basis of adaptation in this species.

Future Perspectives and Concluding Remarks The physiological and molecular data on S. alaskensis highlight the capacity of this oligotrophic ultramicrobacterium to alter gene expression, macromolecular synthesis, and stress resistance. These observations may be contrary to expectation if it was anticipated that gene expression in an oligotrophic bacterium would be largely constitutive and that the genetic capacity to respond, including genome size, would be small [9]. As this is clearly not the case in S. alaskensis, it will be valuable to assess the physiological and genetic properties of other oligotrophic bacteria in order to establish whether a typical oligotrophic physiotype exists, or whether a range of strategies have evolved. Despite these surprises, it is noteworthy that many traits predicted for an oligotroph are present in S. alaskensis, including a constant ultramicro size (