OUP user menu

Phages of the marine cyanobacterial picophytoplankton

Nicholas H. Mann
DOI: http://dx.doi.org/10.1016/S0168-6445(03)00016-0 17-34 First published online: 1 April 2003


Cyanobacteria of the genera Synechococcus and Prochlorococcus dominate the prokaryotic component of the picophytoplankton in the oceans. It is still less than 10 years since the discovery of phages that infect marine Synechococcus and the beginning of the characterisation of these phages and assessment of their ecological impact. Estimations of the contribution of phages to Synechococcus mortality are highly variable, but there is clear evidence that phages exert a significant selection pressure on Synechococcus community structure. In turn, there are strong selection pressures on the phage community, in terms of both abundance and composition. This review focuses on the factors affecting the diversity of cyanophages in the marine environment, cyanophage interactions with their hosts, and the selective pressures in the marine environment that affect cyanophage evolutionary biology.

  • Bacteriophage
  • Phage
  • Cyanophage
  • Marine
  • Synechococcus
  • Prochlorococcus
  • Picoplankton

1 Introduction

1.1 Discovery of cyanophages in the oceans

Phages, viruses that infect bacteria, were discovered in the second decade of the 20th century. However, it is only very recently that their importance in marine ecological and biogeochemical processes has begun to be appreciated (for review see [1]), following the discovery that viruses are abundant in aquatic ecosystems [2]. Strains of unicellular cyanobacteria of the genera Synechococcus and Prochlorococcus are abundant in the world's oceans and constitute the prokaryotic component of the picophytoplankton. Together they contribute between 32% and 89% of primary production in oligotrophic regions of the oceans [36]. Viral infection of marine unicellular cyanobacteria was first reported in 1990 [7, 8]. It is these viruses, cyanophages, which infect this ecologically important group of cyanobacteria that are the subject of this review. The topics of viruses in aquatic systems in general and cyanophages in particular have been comprehensively reviewed by Wommack and Colwell [9] and Suttle [10] respectively. It is the intention here to examine the factors affecting the diversity of cyanophages in the marine environment, cyanophage interactions with their hosts, and the selective pressures in the marine environment that affect cyanophage evolutionary biology.

1.2 The hosts

Marine unicellular cyanobacteria possessing phycoerythrin as their primary accessory light-harvesting pigment were first identified and recognised as a major component of the phytoplankton in 1979 [11, 12]. They were classified as marine cluster A (MC-A) Synechococcus and distinguished from marine cluster B (MC-B), members of which have phycocyanin as their major light-harvesting pigment [13]. The term Synechococcus will from this point be used to refer to MC-A, except where explicitly stated. Furthermore, several isolates of Synechococcus in common laboratory use are known by multiple synonyms and to avoid confusion these are listed in Table 1. Members of the genus Prochlorococcus were first discovered in 1988 [14] and have a markedly different light-harvesting system that employs a chlorophyll a2/b2 complex, rather than a phycobilisome [15]. Despite this major difference in their photosynthetic apparatus, they are very closely related on the basis of the sequences of their rpoC1 and 16S rRNA genes [16, 17]. Members of both genera are considered to be cyanobacteria and constitute the prokaryotic component of the picophytoplankton. The comparative oceanic distribution of Synechococcus and Prochlorococcus has been reviewed by Partensky et al. [18]. Prochlorococcus is essentially ubiquitous between 40°N and 40°S, but is not found in waters where the temperature is less than 10°C. This contrasts with Synechococcus, which has a somewhat broader distribution in that it tolerates a wider range of temperatures and is found, albeit in low concentrations, in waters with temperatures as low as 2°C. The vertical distributions of Prochlorococcus and Synechococcus populations tend to exhibit certain general features. Typically Synechococcus populations are about 10-fold less abundant than Prochlorococcus populations, except in coastal waters. However, their abundance maxima are similar at around 106 cells ml−1 whilst their relative abundance may change on a seasonal basis. In the oligotrophic regions of the open oceans the most common pattern is for the Prochlorococcus population to extend from the surface to the bottom of the euphotic zone with a typical population density of about 105 cells ml−1 with the Synechococcus population approximately one order of magnitude lower. The population density has profound implications for interactions with the co-occurring viral community (see Sections 3.4 and 3.5).

View this table:
Table 1

Synonyms used for marine Synechococcus isolates

Original isolate nameWoods Hole Culture Collection designationNorth East Pacific Culture Collection designationProvasoli-Guillard National Centre for Culture of Marine Phytoplankton designation
L1604 (C-7)WH8018CCMP838
WH 1μ filtrateWH8101

Cultured isolates of Synechococcus and Prochlorococcus exhibit a variety of significant physiological differences in terms of their responses to light intensity, pigment composition, nutrient utilisation, motility etc. which enable them to adapt to particular sets of conditions in the marine environment. It is important to recognise this genetic diversity between different ecotypes of the two genera, as it is likely to be reflected in the ability of some phages to infect different strains, i.e. to be associated with host range. A considerable amount of work has been done on analysing genetic diversity amongst Synechococcus and Prochlorococcus strains (for review see [19]). Only the key findings are summarised here and are focused on Synechococcus as little is known at present about the phages of Prochlorococcus. The main approaches to studying genetic diversity of Synechococcus have involved 16S rDNA sequences [20] and, more extensively, the rpoC1 gene encoding a subunit of RNA polymerase [2126]. Use of the rpoC1 gene has resolved Synechococcus into at least seven clades [2325] that have been associated with particular phenotypes such as chromatic adaptation [26], and non-flagellar motility [25]. Synechococcus and Prochlorococcus ecotypes have been resolved by using the 16S–23S rDNA internal transcribed spacer sequences [27]. This approach separated marine Synechococcus into six clades, three of which were associated with particular phenotypes, namely motility, chromatic adaptation, and lack of the pigment phycourobilin. In the case of Prochlorococcus the analysis confirmed the existence of high-light and low-light ecotypes and resolved the low-light group into four distinct clades. Most recently, 16S rRNA sequencing of Synechococcus cultures, together with some environmental sequences, has revealed 10 phylogenetically discrete clades, including one clade that contains halotolerant strains (e.g. WH8101) that can be phylogenetically included within MC-A (N.J. Fuller, D. Marie, F. Partensky, D. Vaulot, A.F. Post and D.J. Scanlan, unpublished results).

1.3 Phage families and lifestyles

More than 5100 phages belonging to 13 families have been described since 1959 and 96% are tailed phages [28]. In keeping with this observation, tailed phages are the most abundant phages in the marine environment (for review see [9]). Indeed, it has been estimated that there are more than 1030 tailed phages in the biosphere [29]. For the purposes of this review only tailed phages, with dsDNA genomes, belonging to three phage families need concern us. The myoviruses, typified by the coliphage T4, have long contractile tails. The siphoviruses, represented by λ, have long non-contractile tails and the podoviruses, of which coliphage T7 is the archetype, have short, or apparently non-existent tails. It should be said though that this phenetic classification of phages is being called into question [30] and genome-based taxonomic approaches are being explored [31].

Bacteriophages may be distinguished as obligately lytic or temperate (lysogenic). Temperate phages when they infect a susceptible host may either cause lysis and release of progeny phage or enter a stable relationship with the host, termed lysogeny, which does not involve lysis. A temperate phage in the lysogenic state is referred to as a prophage and, in the case of the large majority of temperate phages, lysogeny involves integration of the phage genome into the host cell chromosome as a prophage. The prophage may be induced back into the lytic cycle in response to environmental factors, leading to production of progeny phage and cell lysis. Obligately lytic phages do not have the option to enter the lysogenic state and always cause lysis of the infected host. However, there is the phenomenon of pseudolysogeny, which is important when considering the relationship between phage and nutrient-starved Synechococcus host (see Section 3.10). A pseudolysogen has been defined as “a phage-infected cell that grows and divides even though its virus is pursuing a lytic infection” [32].

What are the challenges and selection pressures that determine which of the different phage lifestyles is going to be the fittest for a given environment? Stewart and Levin [33] proposed two major selective advantages for temperate phages. Firstly, lysogenised cells are immune to infection resulting in lysis by the same phage strain. Secondly, lysogeny is an effective strategy to maintain the phage population when host abundance is too low to support maintenance of the population by lytic infections, as prophages will be replicated along with the host cell chromosome. Both advantages of lysogeny may be of significance in the marine environment, but low host abundance is clearly of importance. The tendency of a temperate phage to exist as a lysogen is essentially determined by two parameters, the probability of entering the lysogenic state on infection and the probability of induction once the lysogenic state has been established. A high frequency of lysogenisation and a low rate of induction might be favoured when sensitive hosts are rare. Conversely, a low frequency of lysogenisation and a high rate of induction would be favoured when sensitive hosts are abundant. Resource-based modelling suggests that in an equable environment with constant inputs of nutrients, a temperate phage with a low probability of lysogenisation and a low frequency of induction will compete well [34]. The same approach shows that in a variable environment temperate phages with higher probabilities of lysogeny and higher induction rates are likely to be favoured. However, the probabilities of lysogenisation and induction are not constant for a given phage and can be significantly altered by environmental and host genetic factors.

Another factor that must be considered for both temperate and lytic phages is lysis timing (see Sections 3.8 and 3.10). Abedon [35] has demonstrated theoretically that, other things being equal, in an environment with a high host density a phage that lyses its host quickly and releases fewer progeny will have an advantage over a phage with a longer latent period and larger burst size. It is important to note that the values for host density used for modelling studies tend to be much higher than those found for natural populations of marine cyanobacterial picoplankton.

Finally, laboratory studies show that bacteria rapidly evolve resistance to bacteriophage infection (for review see [36]). Resistance may be partial or complete and there are variations in the physiological cost of the mutation. Of course, it is also possible for the phage to mutate in such a way that the resistance mutation in the host is overcome. This leads to the idea of a continual arms race between phage and host, commonly known as the Red Queen hypothesis, in which there are continual co-evolutionary changes that lead to host and parasite being in the same position relative to each other. Again, the caveat regarding the high host densities used in such laboratory studies and those obtaining for natural marine Synechococcus and Prochlorococcus populations must be borne in mind.

2 Characterisation of marine cyanophage isolates

2.1 Isolation methods

Phage samples from marine waters are commonly filtered through 0.22-μm filters to remove cells and then may be used directly, or after concentration by ultracentrifugation or ultrafiltration. Samples may be stored in the dark (with or without chloroform) and phage can still be isolated after several years. Two approaches have been successfully adopted to the isolation of phages infecting MC-A Synechococcus strains. These are plaque formation on a host growing on a solid medium [37, 38] and liquid serial dilution [37, 39, 40]. The former method yields useful information regarding plaque morphology. The latter method allows for the fact that growth by MC-A Synechococcus strains on solid media is a very variable property [39]. Obviously, the choice of host strain is important in terms of both the diversity of phages isolated and the choice of isolation method. It could also affect the isolation of temperate phages that exhibit a high frequency of lysogenisation. Two distinct plaque morphologies are commonly observed when assaying natural samples. Apart from variations in plaque diameter, there are clear plaques with a roughly circular periphery and irregularly shaped plaques with an area of central host growth. In studies on phage abundance in the Gulf of Aqaba (Red Sea) the irregularly shaped, turbid plaques were more commonly obtained in samples from greater depths in the autumn, but were only observed with strain WH7803 as the host and not with WH8103 or a MC-B strain (A. Millard and N.H. Mann, unpublished results). The clear plaques are thought to be produced by obligately lytic phages or temperate phages with a very low frequency of lysogenisation. Conventionally, the interpretation of irregular plaques with turbid centres is that they are caused by temperate phages. In the initial stages of growth of such a plaque the multiplicity of infection (MOI) is low and the cells are growing with plenty of nutrients. As the plaque expands the local MOI increases and nutrients near the centre of the plaque are not utilised, because there are no cells growing there. The plaque ceases to expand as the cells in the lawn stop growing as nutrients become exhausted. Any cells that have become lysogenised near the centre of the plaque will be able to grow on the nutrients remaining there. This explanation might suffice for rapidly growing hosts like Escherichia coli where plaques are observed in hours and therefore nutrient diffusion would be limited, but this is unlikely to be the case for Synechococcus plaques, which take over a week to appear.

Temperate phages may be identified and isolated by another method. Recently, we have adopted in this laboratory a liquid co-culture technique in which two strains of Synechococcus are grown together. Any prophages spontaneously induced in one strain may infect and be propagated on the second strain. These phages are then detected by subjecting the culture supernatant to filtration through a 0.22-μm filter and plaque assaying the filtrate using each of the original co-culture strains individually.

2.2 Families

The isolation and characterisation of cyanophages infecting marine Synechococcus strains was first reported in 1993 [3739]. Waterbury and Valois obtained 75 Synechococcus phages from both inshore and open ocean samples [39]. These phages exhibited marked morphological diversity and there were representatives of all three families of tailed phages, though the large majority of isolates were myoviruses with contractile tails. About 80% of Synechococcus phage isolated in river estuaries were found to be myoviruses [40]. There were representatives of all three families of tailed phages among the seven Synechococcus phages isolated by Suttle and Chan [37] and the five Synechococcus phages isolated by Wilson et al. [38] were either myoviruses or siphoviruses. Thus, it appears that the large majority of Synechococcus phages in the marine environment are myoviruses, though of course it may be that Synechococcus myoviruses are just easier to isolate than podo- and siphoviruses.

2.3 Host range

The large majority of tailed dsDNA phages possess thin tail fibres attached to the baseplate or the tail shaft that constitute the structure involved in initial attachment to the host and individual tail fibre genes often appear as mosaics with parts derived from a common gene pool (for review see [41]). Duplication of a small domain in the tail fibre adhesin can extend the host range of T4 [42] and recombination between the tail fibre genes of T-even phages has been shown to alter their adhesin specificity [43]. Some phages carry more than one tail fibre gene, encoding proteins with different adhesin specificities, which allows them to infect a wider range of hosts, e.g. bacteriophage K1-5 [44].

Waterbury and Valois [39] found considerable variation in the host ranges of their Synechococcus phage isolates. Some phages would infect as many as 10 of the 13 strains tested, whereas others would infect only the strain used for isolation. One phage isolated on a MC-A Synechococcus strain would infect other MC-A strains, but also a MC-B strain (WH8101). None of the phages would infect the freshwater strain Synechococcus sp. PCC 6307. Suttle and Chan [37] isolated phages on both MC-A and MC-B hosts. Host range was clearly not tied to the geographical locations where the phages and hosts were isolated. Phages from all three tailed families were isolated that infected MC-B strains, but all of the phages capable of infecting MC-A strains were myoviruses. One of their isolates, a myovirus (S-PWM3), infected a green Synechococcus (presumably MC-B) as well as three MC-A strains. Again no infectivity against freshwater strains was observed. Thus, myoviruses, as well as appearing to be more abundant, tend to exhibit broader host ranges. Lu et al. [40] also isolated phages against phycoerythrin-containing MC-A and phycocyanin-lacking (presumably MC-B) hosts. Again they found that phages infecting MC-A Synechococcus strains had a broader host range and that there were phages infecting hosts from both MC-A and B strains.

Very recently, a new set of marine cyanophage has been isolated using fresh isolates of Synechococcus and multiple strains of the closely related Prochlorococcus. Cross-infection experiments designed to examine the host range of these marine cyanophage suggest that the cyanopodoviruses are very specific, almost exclusively infecting only their original hosts, while the cyanomyoviruses are capable not only of cross-infecting other strains of the same genus, but also of cross-infecting strains of the other cyanobacterial genus (M. Sullivan, J.B. Waterbury and S.W. Chisholm, personal communication). The fact that there are phages capable of infecting both Synechococcus and Prochlorococcus strains has very great implications for the interpretations of contact rates calculated for Synechococcus and associated phages as the Prochlorococcus population is usually of the order of 10-fold higher than that of Synechococcus.

2.4 Diversity

Apart from the fact that phages could be isolated infecting Synechococcus that belonged to all the families of tailed phages, the extensive morphological diversity was particularly remarked upon and taken to indicate that there are a number of cyanophage species [39]. Considerable structural variation was observed within the cyanomyovirus group, reflecting further family-specific diversity [37], which has been confirmed by molecular approaches (see below). Restriction analysis of cyanomyovirus DNA revealed clearly different restriction patterns [38, 40], but a few bands were found to cross-hybridise indicating a limited degree of genetic conservation [38]. This conserved region was analysed by Fuller et al. [45] who showed it to contain a gene that was homologous to g20 of coliphage T4. Gene g20 of phage T4 encodes the portal vertex protein involved in capsid assembly. Significant sequence diversity was observed between the g20 genes from three Synechococcus myoviruses and this diversity was also apparent at the amino acid level of the encoded proteins. However, there were regions of sufficient conservation to permit the design of polymerase chain reaction (PCR) primers specific for Synechococcus myoviruses [45], which has led to the analysis of diversity in natural assemblages. The g20 primers were used to amplify Synechococcus myovirus DNA from samples taken on a cruise between the Falkland Islands, in the south Atlantic ocean, to the UK [46, 47]. It was found that genetic diversity of g20 genes within a single water sample was as great as between the g20 genes of phages from different oceans. Denaturing gradient gel electrophoresis of the PCR products revealed changes in the phage population structure in the surface water along the transect and throughout depth profiles. These observations were extended by Zhong et al. [48] who designed an additional primer, which, in combination with one of the primers designed by Fuller et al. [45], yielded a larger, c. 592-bp product, which has been used to reveal the enormous diversity of Synechococcus myoviruses. Sequence analysis of the cloned PCR products from six natural virus concentrates from estuarine and oligotrophic offshore sites revealed that a total of 114 different gp20 homologues clustered into nine phylogenetic groups. There was no correlation of phylogenetic grouping with geographical separation. However, the population composition and structure of cyanomyovirus communities from the estuarine and open ocean samples differed and unique phylogenetic clusters were found for each environment. Changes in diversity were also noted from surface waters to the deep chlorophyll maximum. Only three of the phylogenetic groups defined by this study contained known cyanophage isolates. Given the recent isolation of cyanomyoviruses that infect both Prochlorococcus and Synechococcus (see Section 2.3) it will be important to establish whether the gp20 PCR primers will amplify the g20 product from these phages. This will establish whether these phylogenetic groups are composed of phages that infect different Synechococcus strains only, or also Prochlorococcus. It is important to note these various aspects of cyanophage diversity in the context of phage–host interactions (Section 3).

3 Phage–host interactions

3.1 Synechococcus phage abundance

Total viral abundance in marine samples can be estimated in a variety of ways including transmission electron microscopy (TEM), fluorescence microscopy and flow cytometry. However, the only commonly employed methods of measuring the abundance of phages infecting Synechococcus are to use either plaque assay on solid medium or the most probable number method in liquid culture. Obviously, it is appropriate to use a strain that is thought to be infected by a wide variety of Synechococcus phages and strain WH7803 (alias DC2) is commonly employed [37, 40]. Other strains used include WH8101 [40], WH8018, WH8017 and WH8012 [39]. Viral abundance in Woods Hole Harbor ranged from 1.9×101 ml−1 in June to 1.14×104 ml−1 in July [39]. Suttle and Chan [37] reported abundances ranging from undetectable to in excess of 105 ml−1. The highest reported abundance of Synechococcus phages is c. 106 ml−1 [49].

The direct approaches to Synechococcus phage enumeration employ a single strain of Synechococcus as the host and given that the phages so far characterised exhibit varying degrees of host specificity, must underestimate the true abundance. Even if two strains are employed it cannot be easily established whether the phage titres are additive or overlapping. The extent to which cyanophage abundance is underestimated has major implications for any analysis of contact rates between phage and host (see Section 3.4). One clue to the extent of the underestimation comes from the use of fluorescently stained phages [50]. The relatively broad host range Synechococcus phage S-PWM3 (a myovirus) was found to attach to only about 3% of the Synechococcus cells in two samples from the Gulf of Mexico suggesting that the degree of underestimation may be very considerable.

3.2 Enumeration of infected cells

The most direct method of detecting Synechococcus phage interactions is to use TEM to determine the proportion of cells that contain visible mature phages. Other, less direct, methods involve the calculation of virus–host contact rates and virus decay rates (see Section 3.4). The cycle of phage infection of a bacterial cell can be divided into stages. The time from the initial infection of the cell to lysis is known as the latent period. However, mature progeny phages arising from the initial infection appear in the cell well before lysis and the time from infection to the first appearance of mature phage in the cells is termed the eclipse period. By definition mature phage particles are only present in the infected cell from the end of the eclipse period and so a calculation can be made as to the total number of infected cells by estimating the ratio of the eclipse period to the latent period. Thus, there is a conversion factor to infer the frequency of infected cells from the frequency of visibly infected cells (FVIC). However, estimation of the value of the conversion factor and calculation of absolute viral mortality are fraught with problems.

This approach was first used with marine Synechococcus by Proctor and Fuhrman [8]. They found that, depending on the sampling station, between 0.8% and 2.8% of cyanobacterial cells contained mature phage. Working on the assumption that the eclipse period was 90% of the latent period, they estimated that percentage of infected cells was actually 10-fold greater than the observed frequency (a conversion factor of 10). Proctor and Fuhrman [8] went on to suggest that to estimate the absolute mortality rate, the relationship between the latent period and the host generation time needed to be known. Assuming that there is a steady state where cell division is balanced by mortality (i.e. half of all daughter cells survive to divide themselves), and if the latent period is roughly equal to the generation time, then the total mortality attributable to phages is twice the number of infected cells at any given time.

This conversion factor of 10 used by Proctor and Fuhrman [8] was questioned by Waterbury and Valois [39] who suggested that, on the basis of observations made by Padan and Shilo [51], the eclipse period was more like 50% of the latent period and that a conversion factor of 2 was appropriate, leading to values for the proportion of infected cells ranging from 1.6% to 5.6%. However, Padan and Shilo [51] used a filamentous cyanobacterium for their studies and it is questionable whether the results can be extrapolated to a unicellular species. Waterbury and Valois [39] using one of their podovirus isolates found that assembled phages were visible for 60% of the latent period implying a conversion factor of 1.7. Suttle [10] suggests that, on the basis of studies by MacKenzie and Haselkorn [52] on a freshwater Synechococcus strain, a conversion factor of 4 would be reasonable. Thus, depending on the conversion factor used, the lowest estimate of the proportion of infected cells is 1.4% and the highest is 28%, but these values are based on only one data set [8] and there exists the possibility of a greater range of values.

The idea that absolute mortality is twice the number of infected cells [8] has been questioned by Binder [53]. The ‘factor of two rule’ assumes that steady-state conditions apply, that the latent period equals the generation time, and that infected cells are not grazed. Binder proposes that absolute mortality rate or the fraction of mortality from viral lysis (FMVL) is approximated by the following equation:

Embedded Image 1

where γ is the ratio between the latent period and the generation time and ε is the fraction of the latent period during which viral particles are not visible. Calculation of FMVL is very sensitive to changes in ε and γ. Wilson et al. [54] observed a latent period of approximately 9 h for the myovirus S-PM2 infecting a culture of Synechococcus WH7803 which grew with a generation time of approximately 24 h, giving a value for γ=0.375. Assuming the suggestion of Suttle [10] regarding the length of the eclipse period then ε=0.75. When these values are substituted into Eq. 1 and the maximum values for FVIC (2.8%) from Proctor and Fuhrman [8] are used the absolute mortality rate is 48%. On the other hand, if we take a value of ε=0.5 as suggested by Waterbury and Valois [39] and γ=1 together with the lowest value FVIC of 0.8%[8], an absolute mortality rate of 2.3% is obtained.

Thus, it is clear that until there are reliable values of ε and γ for Synechococcus strains and their phages there is a very great problem in estimating, with any appreciable degree of accuracy, the absolute value for viral mortality of Synechococcus populations. This problem is exacerbated by the fact that the values of ε and γ are likely to vary with different host–phage combinations and with the same host in different physiological conditions. Furthermore, there is a strong selection pressure on the length of the latent period, which is determined by the concentration of sensitive host cells (see Section 3.8).

3.3 Burst size

An important factor in assessing the impact of phages on Synechococcus communities is the burst size, the average number of phages released when a single infected cell lyses. The extensive literature on the phages of heterotrophic bacteria indicates that burst size varies from phage to phage and host to host, as well as being affected by host physiology and environmental factors such as temperature. The archetypal myovirus T4 has a burst size of only 20 and 25 when infecting E. coli B growing on media affording doubling times of 190 and 210 min respectively ([55] as cited by [56]). By contrast, when infecting E. coli B growing with a doubling time of 20 min, the T4 burst size was increased by an order of magnitude to 200–300. Obviously, it would be more useful to have an idea of the burst size in natural assemblages and an average value derived from TEM-determined burst sizes of bacterioplankton in aquatic environments is 24 [9]. In the case of phages infecting Synechococcus a burst size of 250 was reported for the siphovirus S-BB1, using a one-step growth curve [37]. A similar analysis indicated a burst size of 22 for the cyanomyovirus S-PM2 [54]. Estimates of burst sizes based on balancing estimates of viral decay and Synechococcus contacts yielded values ranging from 93 to 324 [57]. A TEM-based approach indicated a burst size of 81 [58]. Thus, there is a considerable range of values spanning an order of magnitude that can be used in calculations relating to the maintenance of phage populations (see Section 3.4)

3.4 Adsorption, contact rates and viral decay

A critical factor in phage–host interaction is the frequency with which phages collide with their host cells and then adsorb to them leading to infection. Calculation of such contact rates provides another approach, in addition to the direct TEM method, to assessing the significance of phage infection for Synechococcus populations by estimating the theoretical maximum number of phage–host contacts, given the known abundances of phage and host. A different approach to assessing the contribution of phages to Synechococcus mortality stems from the idea that phages are subject to a variety of decay processes and that to maintain a standing stock of virus net viral removal rates must be balanced by viral production. Consequently estimation of viral decay rates can be used to infer the proportion of the Synechococcus population that must be infected and lysed to maintain the population of phage. Such calculations, however, are very sensitive to the value assumed for the burst size (see Section 3.3).

Phages adsorb onto the host cell through highly specific binding to receptors on the cell surface and potentially any surface structure may act as a phage receptor. However, there is no information yet as to the nature of the receptors for Synechococcus phages. There are certain general principles which govern this initial interaction between the phage and host that are important for predicting events occurring in natural assemblages, particularly for the prediction of phage contribution to host cell mortality. Phage adsorption to the surface of a potential host cell follows first-order kinetics and may be described by Eq. 2[59], where P is the number of phage particles remaining unadsorbed after a time t, N is the concentration of host cells and k is the adsorption rate constant:

Embedded Image 2

Integration of this equation gives the expression:

Embedded Image 3

indicating that the logarithm of the fraction of initial phages, P0, remaining unadsorbed decreases linearly with time at a rate determined by the phage adsorption constant and the number of host cells.

The phage adsorption rate constant can be theoretically predicted from the following equation ([60] as cited by [59]):

Embedded Image 4

where R is the radius of a sphere whose surface area is equal to that of the bacterial cell, C is the diffusion constant of the virus and f is the fraction of collisions between phage and host leading to adsorption. The diffusion constant of T-even myoviruses has been experimentally estimated at 2.4×10−6 cm2 min−1 ([61] as cited by [59]). A theoretical value of the adsorption rate constant k from Eq. 4 (0.24×10−8 cm3 min−1) is in very good agreement with an experimentally determined value of 0.25×10−8 cm3 min−1 [59]. Similar absorption rate constants have been experimentally determined for Synechococcus phages in the range of 0.61×10−8 ml min−1 for the myovirus S-PM2 [54] and 0.39×10−8 ml min−1 for the siphovirus S-BBS1 [37].

Waterbury and Valois [39] used a form (Eq. 5) of Eq. 3 to calculate the time necessary for a phage to contact a Synechococcus cell

Embedded Image 5

where N is the number of Synechococcus cells, P is the phage titre and t is the time in minutes for a phage to contact a host cell. The authors used Eq. 3 to calculate the adsorption constant applying Putnam's value ([61] as cited by [59]) of the diffusion constant and using a value of 0.65 μm for the spherical radius R. It was assumed all collisions lead to adsorption (i.e. the value of f was 1), that adsorption led to cell lysis and that all cells in the Synechococcus population were sensitive to infection. These values and assumptions led to Eq. 6[39], which predicts that:

Embedded Image 6

Application of this equation to Synechococcus populations in Woods Hole Harbor in 1991 indicated that between 0.005% (at the end of the spring bloom) and 3.2% (during a Synechococcus peak in July) of the Synechococcus population was contacted, and assumed to be infected, on a daily basis. It was thought though that these calculations were an overestimate as the majority of the population was resistant to the co-occurring phages and therefore not all contacts would lead to infection. Thus, it was concluded that lytic phages have a negligible effect in regulating the densities of marine Synechococcus populations. However, if the true phage abundance was underestimated (see Section 3.1) the contact rates might be significantly increased.

Murray and Jackson [62] developed a model for the interaction of viruses to particles (hosts) based on diffusive transport that allows inclusion of the motion of water and host cell in estimation of the contact rate and has been used for this purpose with Synechococcus-cyanophage assemblages [57, 58]. Underlying the model is the concept that small particles like viruses are subject to the random wandering of Brownian motion. Consequently, viruses obey the laws of diffusion in their approach to larger particles such as hosts. This model gives rise to Eq. 7 that predicts the total rate (R) at which viruses attach to particles:

Embedded Image 7

where Sh is the Sherwood number, d is the diameter of the host cell, Dv is diffusivity of the virus, V is the virus concentration and P is the particle (host) concentration. The Sherwood number is the factor by which diffusion transport is increased due to local fluid motion. Eq. 7 has been used [57, 58] to calculate contact rates between phage and Synechococcus cells and with the parameter values provided by Murray and Jackson [62]Eq. 7 simplifies to:

Embedded Image 8

In fact the predictions made by Eqs. 6 and 8 are extremely close as they differ only in terms of the two constants f (the fraction of collisions leading to adsorption) and Sh (the Sherwood number), which both have values of approximately unity for an adsorbent host cell that is small or immobile.

Solar irradiation is a major determinant of phage decay in the open ocean, but absorption to particles can be a major component of the phage removal in coastal waters, where the transparency is lower [63]. As described above the contribution of phages to Synechococcus mortality can potentially be inferred from the phage decay rate since this must be balanced by production of new phages to maintain the standing stock. The first attempt to infer infection rates from phage decay rates by Suttle and Chan [57] led to the calculation of an average decay rate in the surface mixed layer of 2 day−1. Assuming a very large value for the burst size of 250, c. 5–7% of the Synechococcus population would have to be infected and lysed per day on average to replenish the phage population. However, if a burst size of 50 was applied as many as 33% of the Synechococcus population would have to have been lysed daily at one of the sampling stations. Contact rates were also calculated using the diffusive transport approach [62]. Comparison of contact rates with predicted infection rates required that for the farthest offshore station most Synechococcus cells be susceptible to infection, that most contacts result in infection and that the burst size was 324. However, in nearshore waters it was calculated that 80% of the Synechococcus population would be contacted, but only 1% of contacts would result in lytic infection [57]. A subsequent study [58] indicated that values of 0.53–0.75 day−1 might be more appropriate for the phage decay rates and a much lower value of the burst size (81) was applied. The changes in the values of the decay rates and burst sizes largely compensated for each other and figures for the proportion of the Synechococcus community infected ranged from 1 to 8% for offshore waters. In nearshore waters only 0.01–0.02% of the Synechococcus were lysed on a daily basis. In all cases the efficiency of infection was very low, with only 1.01–3.18% of contacts leading to infection. The observations for nearshore waters are in good agreement with those of Waterbury and Valois [39] and the low values of contacts leading to infection are consistent with a significant fraction of the Synechococcus population not being sensitive to infection. The discrepancy between the values for offshore waters between Waterbury and Valois [39] and Suttle and co-workers [57, 58] is attributed to the higher phage abundances reported in the Gulf of Mexico [9, 10].

The reliance that we can put on these approaches is determined by the accuracy with which we can measure infectable host and infective phage abundances. Total Synechococcus abundance can be measured accurately, but it would be extremely difficult and time-consuming to analyse the population composition in genetic terms. Consequently, it is commonly assumed that the population is homogeneous with respect to infection and lysis, but this is unlikely to be the case in many situations (see Sections 3.5 and 3.6). Measurement of virus abundance is equally problematic because of the tendency of the methods used to lead to underestimates (see Section 3.1).

There are other caveats that should be borne in mind when interpreting contact rate data. Firstly, the situation is complicated by the fact that some phages that infect the highly abundant Prochlorococcus can also infect MC-A Synechococcus strains and some phages that infect MC-B Synechococcus can also infect MC-A Synechococcus (see Section 2.3). Secondly, it has been reported, albeit in the case of a coastal strain of Synechococcus, that phage would only bind to about 10% of the cells, suggesting that only a small proportion of the population expressed the particular phage receptor at any particular time [10]. Thirdly, a significant fraction of the infected cells might be grazed by heterotrophic nanoflagellates. Finally, a variable proportion of the phage population, depending on prevailing conditions, could arise from their induction in lysogenic Synechococcus hosts.

Considerable significance has been attached to the theoretical approaches to estimating phage-associated Synechococcus mortality, but are there any more pragmatic observations that can help us? Presumably there is a threshold density below which a phage population cannot be maintained by lytic infection. Wiggins and Alexander [64] tested this idea using obligately lytic phages of Staphylococcus aureus, Bacillus subtilis and E. coli and found that there was a threshold host population density of c. 104 below which the phage numbers failed to increase. They were working with host populations that could be assumed to be genetically homogeneous. Suttle and Chan [57], studying natural populations of Synechococcus, found that there was a marked increase in phage numbers above a host threshold of c. 1×103 cells ml−1, though a more extensive data set indicated a threshold of 1×104 cells ml−1 ([65] as cited by [10]), which agrees well with data of Wiggins and Alexander [64]. The results of Suttle and Chan [57] were obtained by plotting the number of infective cyanophages against Synechococcus abundance. They used two hosts, WH7803 (alias DC2) and SYN48 (alias WH6501), to assay the number of infective phage and therefore could have significantly underestimated the true abundance of total Synechococcus phage. The observations were made on natural populations of Synechococcus, which suggests, on the basis of the laboratory studies of Wiggins and Alexander [64], that the majority of the cells were sensitive to phage infection. Waterbury and Valois [39] argue, though, that their results indicate that there is sufficient time for phage adsorption to occur at host densities well below the 104 cells ml−1 threshold. They suggest that the explanation may be the rapid growth of the heterotrophic bacteria used by Wiggins and Alexander [64] in contrast to a longer doubling time for Synechococcus in the Woods Hole Harbor spring bloom. An alternative explanation would be that the true abundance of the phage population was underestimated. Kokjohn et al. [66] found that productive infection could occur at host (Pseudomonas aeruginosa) densities as low as 102 cells ml−1, which is in keeping with the observations of Waterbury and Valois [39] and theoretical calculations (see Section 3.5).

3.5 Host genetic diversity and resistance

In laboratory studies of communities of bacteria and bacteriophage there is a rapid evolution of resistance to infection (for review see [36]). Lenski and Levin [67] showed for chemostat cultures of E. coli infected with phage T4 that a virus-resistant, resource-limited bacterial population becomes established and can persist indefinitely. There was also a persistent phage population supported by the maintenance of a small population of sensitive cells that in turn persisted because of a growth rate advantage over the T4-resistant cells. At equilibrium the T4-resistant cell density was c. 5×108 cells ml−1 and the T4 density was c. 5×105 pfu ml−1. Thus, a trade-off between resistance and competitiveness allows the co-existence of sensitive and resistant cells. To what extent though can laboratory studies predict the behaviour of complex natural assemblages in which there are multiple Synechococcus hosts and multiple phages? In fact phage resistance does not appear to be a dominant factor in marine systems in general [1]. Mutation to phage resistance usually involves a physiological cost and mutation to resistance to one phage does not necessarily lead to resistance to other unrelated phages [36]. There is evidence that one particular Synechococcus strain may be susceptible to infection by several distinct phages (see Section 3.6). Consequently, one would expect that mutation to resistance to several co-occurring phages would impose a large trade-off in fitness.

Using the contact rate predictions of transport theory combined with the measurement of decay rates for natural populations of Synechococcus phages [57, 58] (see Section 3.4) it becomes possible to calculate a theoretical threshold at which the population of a lytic phage would increase rather than just be maintained. This threshold, which would presumably represent the point at which the phages begin to exert a significant selection pressure on the host, is the point at which the contact rate (Eq. 8), assumed to be equal to the infection rate, multiplied by the burst size equals the virus decay rate multiplied by the virus population density (Eq. 9):

Embedded Image 9

where B=burst size, P=Synechococcus abundance, V=virus abundance and D=the virus decay rate.

The threshold host population P is given by Eq. 10:

Embedded Image 10

Obviously, calculation of the threshold is sensitive to values of the decay rate and burst size. Decay rates during the winter months have been estimated to be of the order of 0.03 day−1 rising to 0.5 day−1 for the mixed layer during the summer [58]. A realistic range for burst sizes might be from 25 to 100 (see Section 3.3). These would lead to minimum and maximum values for the threshold of 91 cells ml−1 (B=100, D=0.03) to 6079 cells ml−1 (B=25, D=0.5). The upper value is similar to the threshold of 1×104Synechococcus cells ml−1 observed for natural populations above which phage numbers increased ([65] as cited by [10]). The lower value would be exceeded by almost all the Synechococcus abundances observed by Waterbury and Valois in Woods Hole Harbor [39].

Resistance to phage infection in natural Synechococcus populations could potentially occur in three ways. Firstly, there may be genetic diversity within the population with certain strains being intrinsically uninfectable by certain phages. This has been shown to be the case in laboratory studies on host range (see Section 2.3). This intrinsic resistance may not be absolute, but could rather be a difference in the efficiency of infection that could arise by a number of routes, e.g. restriction-modification. Secondly, there may be a selection pressure for the success of phage-resistant mutants of previously sensitive hosts, though these mutants may pay a physiological cost. Thirdly, normally sensitive cells may be uninfectable under certain physiological conditions or in particular phases of the cell cycle. It has even been suggested [1] that there may be ecological advantages in an oligotrophic marine environment to express decoy phage receptors leading to phage adsorption and DNA entry, but resistance to subsequent viral replication and lysis, thus permitting the incoming DNA to be used as a valuable source of carbon, nitrogen and phosphorus.

Waterbury and Valois [39] tested Synechococcus for resistance to their co-occurring phages. Clones of Synechococcus cells and phages were isolated from a single water sample. Enrichment, following dilution in 10-fold increments and isolation, yielded 10 Synechococcus clones that were then challenged with seven phage isolates. The Synechococcus clones isolated from the 100- and 1000-fold dilutions, which were taken to have arisen from the most abundant strains in the sample, were resistant to most of the phage isolates, whereas isolates from lower dilutions that were presumed to have out-competed the resistant strains during enrichment exhibited sensitivity to more of the phage isolates. This clear trend was interpreted to mean that a majority of the Synechococcus population was resistant to the co-occurring phages. However, this interpretation makes three assumptions: (i) that the enrichment process did not select against abundant strains which grew slowly (if at all) under the laboratory enrichment, (ii) that the sensitive strains had a growth advantage under laboratory conditions and (iii) that the phages isolated were truly representative of the abundant phages at the time of sampling rather than those that efficiently infected the hosts used for isolation. Estimation of the efficiency of encounters between phage and Synechococcus (i.e. contacts leading to infection) in the Gulf of Mexico and nearshore waters indicated that only c. 1–3% of contacts actually led to infections [58]. This is consistent with the results of Waterbury and Valois [39], but neither approach distinguishes between a genetically diverse population containing intrinsically resistant and sensitive hosts and one consisting of phage-resistant mutants (with a reduced growth rate) co-existing with their phage-sensitive progenitors. In this context an important question to ask is at what point the selection pressure for phage resistance becomes significant, leading either to the succession of intrinsically resistant strains, or of resistant mutants. Data from natural Synechococcus populations (see Section 3.4) would suggest that a genetically homogeneous population would start to experience significant selection pressure when it reached a density greater than 104 cells ml−1.

One common reason for differential efficiency of infection amongst related hosts is the possession by the different host strains of different host restriction-modification systems. There are three strategies by which a bacterium can modify its DNA, such that it is protected from its own restriction system [68], and these strategies may also be employed by phages to evade host restriction systems. Particular bases in the recognition sequence of the restriction endonuclease can be modified by methylation by a cognate methylase. The endonuclease may require a specific methylation pattern that is absent in the host DNA. Thirdly, an unusual base (e.g. hydroxymethyl cytosine) may be substituted throughout the genome for one of the normal bases by a modification of the biosynthetic pathway. This latter approach is adopted by phage T4 and provides almost unlimited protection against nucleases that recognise unmodified sequences and protection against nucleases recognising modified sequences is conferred by glucosylation of the hydroxymethyl cytosine residues [68]. There is some evidence for the occurrence of base modification systems in Synechococcus strains. Many restriction endonucleases do not digest Synechococcus phage DNA effectively, suggesting the presence of modified bases [38, 40]. They also observed that EcoRI digested the DNA of four phages propagated on Synechococcus sp. WH7803, but failed to digest the DNA of a fifth, the myovirus S-WHM1, suggesting that it encoded its own restriction-modification system. It has been observed that two Red Sea phages propagated on Synechococcus sp. WH8103 would infect both WH7803 and a MC-B Synechococcus, but when the same phages were propagated on WH7803 they could not infect the MC-B Synechococcus (A. Millard, M. Clokie and N.H. Mann, unpublished observations). Thus strains WH8103 and WH7803 are likely to have different restriction-modification systems.

Another reason for differential efficiency of infection between host strains is immunity arising from lysogeny. Lysogenic cells are resistant to superinfection with the same temperate phage or secondary infection with homoimmune phages. As yet though there is no evidence for the extensive occurrence of lysogeny in Synechococcus (see Section 3.9). Some lytic phages are also able to prevent secondary infection by other lytic or temperate phages by a process known as superinfection exclusion.

3.6 Interactions and diversity

Theoretical analysis of host-selective lysis of bacterioplankton in aquatic systems predicts a reciprocal relationship between bacterial diversity and phages, in which the co-existence of competing bacterial species is ensured by the presence of phages that ‘kill the winner’, whereas the differences in substrate affinity between the co-existing bacterial species, leading to different growth rates, determine viral abundance [69]. In the context of this model, different Synechococcus strains sensitive to infection by different phages and with different growth rates in a particular set of environmental conditions could be treated as different species. As the Synechococcus density increases, phage selection pressure will influence the population structure and composition. This will in turn impose a selection pressure on the phage population structure and composition. Thus, there are likely to be co-evolutionary pressures on the phage and Synechococcus populations. This idea is supported by the observation that maximum Synechococcus myovirus diversity in a stratified water column correlated with maximum Synechococcus population density [46]. Further support comes from the fact that there are distinct cyanomyovirus population structures in estuarine water versus open ocean as well as surface water versus the deep chlorophyll maximum which suggests the presence of different host ecotypes in each environment and a dynamic interaction between cyanophage and host [48]. Thus, phages in one particular phylogenetic cluster based on g20 sequences specifically infected hosts that were adapted to oligotrophic environments [48]. In a study of such an environment (Gulf of Aqaba, Red Sea) over an annual cycle the Synechococcus diversity was monitored using the rpoC1 gene (M. Mühling, N. Fuller, A. Millard, D.J. Scanlan, A. Post, W.H. Wilson, D. Marie and N.H. Mann, unpublished results). There was considerable diversity in spring, with as many as 12 genotypes present, and this was followed by a strong decline in diversity towards the summer and autumn, where only one and two genotypes dominated respectively. In the following winter months there was an even greater Synechococcus diversity than there was in spring with as many as 24 genotypes present. The genetic diversity in the co-occurring cyanophage population was monitored using the cyanomyovirus g20. The seasonal changes in cyanomyovirus genetic diversity paralleled that of Synechococcus with greatest diversity in spring (28 genotypes) and winter (22 genotypes). However, cyanophage diversity was not reduced as much as Synechococcus diversity during the summer (17 genotypes) and autumn (15 genotypes). Given the short half-life of phages in the surface layers, the presence of cyanophage of multiple genotypes in the water column at times when the host population was dominated by only one or two genotypes indicates that cyanophage of more than one genotype were capable of infecting the dominant Synechococcus strains. However, the most abundant and second-most abundant cyanophages (based on g20 clones) showed respectively a parallel pattern of abundance to that of the most abundant and second-most abundant Synechococcus clones, which dominated the summer and autumn maxima, suggesting a specific cyanophage–host relationship.

3.7 Endolysins and holins

Double-stranded DNA phages achieve lysis of the host by degrading the peptidoglycan with a muralytic enzyme, known as an endolysin, and an associated holin protein (for review see [70]). The endolysins lack a secretory signal and it is the holin that permeabilises the cytoplasmic membrane and allows the endolysin to reach its target. The timing of endolysin release by the holin and the consequent lysis of the cell appears to be determined by the reduction of the energised state of the membrane to a critical threshold level in the case of phage λ[71]. Other auxiliary, or inhibitor, proteins may interact with the holin to affect lysis timing. It is presumably the genes encoding the holin and its ancillary factors that are the targets for selection in the evolution of phage latent periods. They also constitute the mechanism by which the latent period can be altered in response to variations in the host cell's physiology and environmental conditions. It is surprising, therefore, that no endolysin or holin genes were identified in the genome of the cyanopodovirus P60 [72]. However, a putative endolysin gene has been identified in the genome of the cyanomyovirus S-PM2 and a contiguous open reading frame (ORF) encodes a protein that has several of the predicted characteristics of a holin (this laboratory, unpublished results).

3.8 Evolutionary pressures on the latent period

A key determinant of the fitness of bacteriophages in a given environment is the length of the latent period, i.e. the time from infection of a cell to subsequent lysis. At any particular density of host cells there will be an optimal latent period that affords the most rapid rate of increase in the phage population. Thus, as alluded to previously, the genes encoding holins and auxiliary factors are under strong selection pressure. Mathematical modelling approaches have demonstrated that both host quantity and quality are involved in determining the optimal latent period [35, 73, 74]. A comparison of approaches for modelling phage absorption to host cells revealed that a complex exponential decay approach based on Eq. 3[59] gave accurate results at both high and low host concentrations [73]. In contrast, an approach that treated adsorption as a single variable, namely, host cell concentration [74], was accurate only at high host concentrations [73]. Application of the exponential decay approach to calculate latent period optima indicates that in an environment where the host cell density was of the order of 104 to 105 cells ml−1 latent periods of the order of 250 to 100 min respectively would be optimal [73]. Extrapolation of these data suggests a latent period optimum of approximately 17 h at a host cell density of 103 cells ml−1. These host concentrations are typically the range of population densities found for Synechococcus in situ. The cyanopodovirus P60 is capable of lysing a Synechococcus sp. WH7803 culture within 10 h [72]. Similarly a one-step growth curve for the cyanomyovirus S-PM2 on the same host indicated a latent period of approximately 9 h [54]. These values would be optimal for phages infecting a host with a density of between 103 and 104 cells ml−1, which would seem realistic for natural Synechococcus populations. It should be borne in mind that the simulations of phage growth assume constant host quantity and quality over time, which are conditions that manifestly do not occur in the marine environment. However, it is apparent that phages with different latent periods will be selected for as the Synechococcus abundance changes.

3.9 Lysogeny

Theory predicts that when the host is present in low abundance and is slow-growing, temperate phages will have a great selective advantage over obligately lytic phages (see Section 1.3). However, there are other factors relating to the interactions of temperate phages and their hosts that must be considered. In simple terms one might expect that the carriage of prophage DNA would impose a genetic burden on the host cell. Studies with phages of E. coli in fact show that prophage carriage can actually improve the fitness of the host in certain environments, e.g. the bor and lom genes of phage λ are involved in serum resistance of the host cell and adhesion to human epithelial cells respectively [75, 76]. A lysogenic phage may significantly alter the phenotype of the host, a phenomenon known as phage conversion, e.g. the gene for cholera toxin is carried by a prophage [77]. Furthermore, the lysogenised host will acquire immunity to superinfection by homoimmune phages. In the natural environment the prophage will be induced in a proportion of the lysogenic hosts leading to the release of infectious phages, which will select against sensitive non-lysogenic hosts.

There have been a variety of studies on lysogeny in marine heterotrophic bacterial communities (for review see [9]), but relatively little work has been done on lysogeny in marine Synechococcus. Waterbury and Valois [39] found that their Synechococcus phage isolates were virulent and that attempts to induce any putative prophages in Synechococcus strains using temperature shock, light shifts, UV and X-ray irradiation and mitomycin C were unsuccessful, suggesting that lysogeny was not widespread. Sode et al. [78] isolated a temperate phage infecting a marine Synechococcus strain, though the host did not belong to MC-A. There have been two recent reports dealing with lysogeny in natural populations of Synechococcus. In one study samples were taken during a Synechococcus bloom in a pristine fjord in British Columbia, Canada [79]. The samples were incubated with and without mitomycin C and the abundances of Synechococcus and infectious cyanophage (using strain WH7803 as a host) were measured. The Synechococcus population was initially c. 1.4×105 cells ml−1. Lysogenic phage production was estimated from the difference in cyanophage abundance between the mitomycin C-treated samples and the controls after 18 h. On this basis induction of prophages occurred in 0.6% of the Synechococcus population. This represents a minimum value as only phage that infect WH7803 would be counted and mitomycin C may not be an effective inducer of prophages in Synechococcus spp. In this context Sode et al. [80] have reported the induction of a temperate phage of a non-MC-A Synechococcus in response to copper at concentrations as low as 31 μM. In the second study McDaniel et al. [81] analysed lysogeny using mitomycin C during an annual cycle in Tampa Bay, Florida, USA. The frequency of lysogens was inversely correlated with Synechococcus abundance. Lysogens were primarily detected during the late winter months, though lysogens were also detected in August, preceding a secondary autumnal Synechococcus bloom.

The observation of turbid plaques during phage isolation (see Section 2.1) supports the notion that some Synechococcus phages are temperate. Use of the co-culture technique (see Section 2.1) has suggested that temperate phages, characterised as myoviruses, capable of infecting MC-A Synechococcus may be carried by some MC-A strains and also by green MC-B strains isolated from the Gulf of Aqaba (A. Millard and N.H. Mann, unpublished results).

3.10 Effect of host physiology on lysis time and the concept of pseudolysogeny

In laboratory studies on phage–host systems infection is commonly carried out under optimal conditions for host growth, but in natural environments bacteria are subject to an alternating feast and famine existence. The physiological status of the host can have profound effects on the course and outcome of phage–host interactions, beginning with adsorption and ending with lysis, and has been well reviewed in the context of phages in aquatic environments by Wommack and Colwell [9]. It was shown as early as 1940 that the adsorption rate constant under optimal growth conditions was more than 60 times greater than under poor conditions [82]. There is clear evidence from studies on P. aeruginosa phages that starvation conditions cause a marked increase in the latent period and reduction in burst size [66, 83]. Phage T4 adsorption rate and burst size were reduced when the growth rate of the host was reduced, whereas eclipse period and latent period increased [84]. However, certain phage receptors might be envelope proteins associated with high-affinity nutrient transport systems [85] and susceptibility to infection might only occur when the cells were deprived for a particular nutrient. There is very little work done in this area on Synechococcus phage except one laboratory study that showed that the phage adsorption rate was unaffected when the cells were phosphate-starved, but that there was a much longer latent period [54]. However, it would seem sensible to assume that Synechococcus–phage interactions are sensitive to the host physiological status in the natural environment.

The term pseudolysogeny has been used to describe a phage–host relationship in which a phage-infected cell grows and divides even though its virus is pursuing a lytic infection, e.g. T3 infection of F+E. coli under starvation conditions [32]. Potentially, both obligately lytic and temperate phages could enter the pseudolysogenic state. It has been suggested that one way in which pseudolysogeny may occur is when the host cell is starved and may explain the persistence of phage in natural ecosystems [86]. Continuous culture studies employing both temperate and obligately lytic phages of P. aeruginosa showed that the frequency of pseudolysogen occurrence increased with increasing nutrient deprivation, but that the dependence on overall nutrient concentration was more marked than the dependence on growth rate [87]. One study suggests that obligately lytic Synechococcus phages can enter the pseudolysogenic state. When the obligately lytic phage S-PM2 (a myovirus) was used to infect Synechococcus sp. WH7803 cells grown in phosphate-replete or phosphate-depleted media there was an apparent 80% reduction in the burst size under phosphate-depleted conditions [54]. However, a more detailed analysis showed that 100% of the phosphate-replete cells lysed, compared to only 9% of the phosphate-depleted cells, suggesting that the majority of phosphate-depleted cells were pseudolysogens. Similar observations were made with two other obligately lytic Synechococcus myoviruses, S-WHM1 and S-BM1.

3.11 Effect of infection on host cell physiology

Photosynthesis in unicellular freshwater cyanobacteria is typically unaffected by phage infection until the cell lyses and phage production is sensitive to the photosynthesis inhibitor DCMU (for review see [10]). The effect of phage infection on photosynthesis has only been studied in the case of Synechococcus sp. BBC1, which lacks phycoerythrin, infected with the phage S-BBS1 (a siphovirus), where photosynthesis was not affected until near the onset of lysis [37]. The archetypal myovirus T4 causes a cessation of the transcription of host genes and translation of host mRNA, coupled ultimately with degradation of the host genome [88]. Do Synechococcus myoviruses (and other families) operate in the same fashion? This question has significance for the continued functionality of photosystem II (PSII). The D1 protein and, to a lesser extent, the D2 protein of PSII in oxygenic phototrophs are subject to rapid turnover as a result of photodamage (for review see [89]). Are the host D1 and D2 genes continuously expressed during the latent period to allow turnover essential for the maintenance of PSII functional integrity and continued non-cyclic photosynthesis or is there some phage-encoded mechanism? Sequencing of the cyanophage S-PM2 genome has revealed phage-encoded D1 and D2 genes (this laboratory, unpublished results), but it is not yet known under what conditions they are expressed.

3.12 The persistence problem

Given the observed rates of cyanophage destruction (see Section 3.4) coupled with the fact that the host Synechococcus population density can fall below the threshold at which the population of obligately lytic phages can be maintained, there is a question of how such phages persist. Obviously, this problem does not apply when considering temperate phages that can persist in the lysogenic state. Pseudolysogeny (see Section 3.10) may represent one strategy by which obligately lytic Synechococcus phages persist when population density is low due to nutrient starvation. An alternative explanation may arise from the observation that virus particles are abundant in marine sediments probably as a result of adsorption to suspended material in the water column that settles out and contributes to the benthic viral population [90]. Indeed, infective cyanophage can be recovered from considerable depths in marine sediments [10]. Thus, sediment may act as a phage reservoir, particularly in coastal marine systems where the whole water column may be mixed at times. Finally, decay rates are considerably reduced in the winter when host abundance is also low [58].

4 Genomics and evolutionary biology

4.1 Genome size and genome size distributions in the sea

Pulsed-field gel electrophoresis (PFGE) has become a popular technique for profiling the genome size distribution and abundance of marine virioplankton (e.g. [9093]). Wommack et al. [90] found viral genome sizes ranging from less than 23 kb up to c. 300 kb in water samples from Chesapeake Bay. 75% of the viruses had genome sizes of less than 97 kb and viruses with genomes in the range 23–48.5 kb accounted for c. 30–60% of the total population. Changes in the structure of the virioplankton community were correlated with time, geographical location, and the extent of water column stratification. Steward et al. [92] studied the virioplankton genome size distributions in a variety of extremely diverse marine environments. The genome size distribution was multimodal with major peaks occurring at 31–36 kb and 58–63 kb. On average in surface seawater more than 90% of the viral genomes were in the 26–69-kb range. The largest viral genomes were greater than 200 kb in size. Changes in the structure of the virioplankton population were again found to alter on a temporal and geographical basis. Fuhrman et al. [93] studied virioplankton genome size distribution in the San Pedro Channel, California, USA. They found relatively few differences with depth in the euphotic zone (45 m), although the water column was stratified below about 5–10 m. They consistently observed 10–15 distinct bands by PFGE, but the most abundant genome sizes were in the ranges 30–40 and 50–75 kb. The largest genome sizes were greater than 242 kb. The PFGE patterns changed further down the water column below the euphotic zone. These sets of observations agree well, with the most abundant genome sizes being in the ranges 30–40 kb and 50–75 kb, but genome sizes of greater than 242 kb were consistently detected. There is some question, though, as to whether the size distribution may be biased towards the smaller phages [92]. Wichels et al. [94] studied the genome sizes of 22 marine phages and found siphoviruses to have genome sizes in the range 25–35 kb, podoviruses 38–48 kb and myoviruses 32–100 kb. At the time of writing the NCBI viral reference genomes site (http://www.ncbi.nlm.nih.gov/PMGifs/Genomes/28883.html) indicates that of the 69 dsDNA tailed bacteriophage genomes sequenced 45 are siphoviruses with genome sizes ranging from 14.5 to 134.4 kb, 13 are podoviruses (11.7–47.9 kb) and 11 are myoviruses (11.6–168.9 kb).

How do these observations on total viral genome size distributions correlate with what is known about Synechococcus phage genomes? The total viral community assayed by TEM, at around 109 virus-like particles ml−1 (for review see [1]), is usually two to six orders of magnitude greater than the cyanophage population assayed by infectivity and consequently little can be inferred from the overall pattern of viral genome size distributions. The one fully sequenced Synechococcus podoviral genome, P60, is 47.872 kb [72]. Analysis by PFGE of 25 Synechococcus myoviruses, isolated from Red sea water samples, has revealed a range of genome sizes of 107–172 kb (A. Millard and N.H. Mann, unpublished results). Similar large genome sizes were found by PFGE for the Synechococcus myoviruses S-PM2 (c. 194 kb) and S-WHM1 (c. 170 kb), together with a smaller size for the podovirus S-BP3 (43 kb) [95]. These estimates of genome size are likely to be much more reliable than those based on the addition of the size of restriction fragments [38], which are likely to be serious underestimates, particularly for the larger genomes. The S-PM2 genome was estimated at 90–100 by restriction analysis, but at 194 kb by PFGE. The majority of Synechococcus phage isolates are myoviruses (see Section 2.2) and the trend is for myoviral genomes to be larger than siphoviral and podoviral genomes. Consequently, it seems likely that the majority of Synechococcus phages in nature are myoviruses with genomes in the range 100–200 kb assuming that the frequency of phage isolation reflects their frequency in nature.

4.2 Information from sequenced genomes

There are currently more than 100 fully sequenced publicly available phage genomes, of which only two, roseophage S101 [96] and the Synechococcus phage P60 [72], are marine phages. The genome of the Synechococcus myovirus S-PM2 is currently being sequenced and is more than 95% complete (this laboratory, unpublished results). P60 is a lytic podovirus which has a comparatively restricted host range within marine Synechococcus strains [72]. The genome is 47.872 kb with 80 potential ORFs of which 19 were assigned putative replication and morphogenesis functions. Proteins encoded by two of these ORFs exhibited striking similarity to ORFs in marine Synechococcus and Prochlorococcus strains encoding ribonucleoside triphosphate reductase A and B subunits, though the similarity does not extend to the nucleotide level. Another ORF exhibited considerable similarity to putative thymidylate synthases from marine Prochlorococcus strains, Synechocystis PCC 6803 and the other fully sequenced marine phage, roseophage S101. These two enzymes of nucleotide metabolism are also encoded by the marine cyanophages S-PM2 (unpublished results, this laboratory) and by a variety of other phages. Host chromosomal DNA can be degraded and the nucleotides incorporated into progeny phage DNA. However, it is thought that in the case of coliphage T4, which also encodes these enzymes, the bulk of dNTPs arise by de novo synthesis [97]. The genomes of one marine Synechococcus and three Prochlorococcus strains are completed or under way and so far there is no evidence for the presence of prophages in these genomes.

4.3 Evolution

Marine cyanophages exist in a constantly changing environment both in physico-chemical terms, and more importantly in terms of host abundance, genetic diversity and physiological condition. The selection pressure of host density on the latent period was discussed in Section 3.8 and the comparative benefits of the lysogenic versus the lytic lifestyle were considered in Section 1.3. However, Synechococcus phage can potentially recombine with each other during co-infection of a single host cell or indeed may recombine with phages that have a broader host specificity.

The study of phages in general has led to ‘the modular theory’, which proposes that the evolution of phages is based on a series of exchangeable and interchangeable, multigenic, genetic elements (modules) that are in themselves functional units and are exchanged between phages by recombination (for review see [98]). Recently, the modular theory has been extended by the suggestion that all dsDNA phages are mosaics and have access by horizontal exchange to a common gene pool, though not all phages have equal access to this pool [99]. Analysis of the genomes of ‘lambdoid’ phages reveals that genomes that share similar gene organisation (synteny) are mosaic with respect to each other [100]. Juhala et al. [101] have proposed that mosaic genomes arise through recombinational processes in which most of the surviving boundaries are located at gene boundaries or, in some cases, at protein domain boundaries within genes. In the case of modules, recombination might occur through short regions of homology between modules, or recombination might occur randomly across the genome, followed by rigorous selection for functional phages (for review see [100]). Such models account for the exchange of modules, but do not account for the entry of novel genes into the genome. In the case of temperate phages novel genes are commonly located near the prophage attachment site and may have been acquired by imprecise prophage excision (for review see [100]). There is also evidence for an incremental addition of novel genes that may have been selected for either by increasing the fitness of the host in the case of temperate phages or by enhancing lytic growth of obligately lytic and temperate phages. One class of novel genes found in phage genomes, located between two genes that are adjacent in phages, and with their own promoter and terminator, has been termed ‘morons’ [101]. The accretions of morons into phage genomes may be a major evolutionary pathway [102].

The first evidence that Synechococcus phages might both exhibit the mosaicism seen in other far more extensively characterised groups of phages and also have access to the dsDNA phage gene pool came from the discovery that phage S-PM2 carried a gene encoding a homologue of the T4 portal vertex protein (gp20) [45]. This observation was extended to show that the g20 gene of phage S-PM2 was merely one within a head–tail morphogenesis gene cluster [95]. There was a set of genes corresponding to the T4 genes, g18 (contractile tail sheath protein), g19 (tail tube protein), g20 (portal vertex protein), g21 (prohead protease), g22 (prohead core) and g23 (major capsid protein). The genes were contiguous except that g21 and g22 were separated by a small ORF encoding a polypeptide with no significant similarity to any known proteins in the databases. The gene order was the same as that found in T4. The current S-PM2 genome sequencing project has extended this morphogenesis gene cluster and shown it to be bounded by regions of DNA containing genes unrelated to those found in T4 (this laboratory, unpublished results).

Sequence analysis of the Synechococcus podovirus P60 showed it to have a very similar gene arrangement to the coliphage T7 and the sequence of the DNA polymerase gene confirmed its phylogenetic relatedness to T7-like phages [72]. There are other genes that exhibited considerable similarity to genes from other phages including Mycobacterium phage D29 and bacteriophage phi-YeO3-12. There is evidence for genetic exchange between P60 and marine cyanobacteria in the genes which encode thymidylate synthase, and the A and B subunits of ribonucleoside triphosphate reductase [72]. These genes are also found in S-PM2 (this laboratory, unpublished results) and the marine roseophage S101 [96]. Thus, it appears for the two Synechococcus phages so far characterised at the nucleotide sequence level that they have mosaic and modular genomes and that they may have acquired individual genes. So the genomic evolutionary processes of Synechococcus phages are probably no different from those of other groups of phages despite enormous differences in the phylogenetic affiliations and environmental niches of their hosts.

Of course, phages may also influence the evolution of their host through horizontal gene transfer in the form of generalised or specialised transduction. Transduction has been shown to occur with isolates of marine heterotrophic bacteria and concentrated natural communities [103]. There is no evidence, as yet, for transduction by Synechococcus phages though host chromosomal DNA has been detected in rigorously purified phage particles by real-time PCR (M. Clokie and N.H. Mann, unpublished results).

5 Conclusions

All the Synechococcus phages so far isolated and characterised are dsDNA tailed phages, with the large majority being contractile-tailed myoviruses. Whilst estimates of the contribution of phages to the mortality of Synechococcus populations are variable, there is agreement that phages exert a significant selection pressure on Synechococcus community structure. Both morphological and molecular analysis of the phages has revealed considerable diversity, and in the marine environment diversity varies down the water column, and on seasonal and geographical bases. Sequence analysis of Synechococcus phage genomes reveals an evolutionary mosaicism and indicates that these phages have access to a common gene pool with phages from other distinct environmental niches and phylogenetically distant hosts. The selection pressures operating on the Synechococcus phage community stem largely from host abundance, genetic diversity and host physiological condition, which in turn are determined by the prevailing physico-chemical conditions of the aquatic environment. One of the major gaps in our knowledge concerning these phage is the nature of the receptors that they interact with on the host cell and the ways in which these receptors might vary with host genotype and physiology.


I would like to thank Noel Carr, Nick Fuller, Martin Mühling, Sandy Murray, Dave Scanlan and John Waterbury for their constructive criticism of the manuscript and Martha Clokie, Andrew Millard, Martin Mühling, Matt Sullivan, Penny Chisholm and John Waterbury for their permission to quote unpublished results. Work in my laboratory on Synechococcus phages has been supported by grants from the Natural Environment Research Council.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
  34. [34].
  35. [35].
  36. [36].
  37. [37].
  38. [38].
  39. [39].
  40. [40].
  41. [41].
  42. [42].
  43. [43].
  44. [44].
  45. [45].
  46. [46].
  47. [47].
  48. [48].
  49. [49].
  50. [50].
  51. [51].
  52. [52].
  53. [53].
  54. [54].
  55. [55].
  56. [56].
  57. [57].
  58. [58].
  59. [59].
  60. [60].
  61. [61].
  62. [62].
  63. [63].
  64. [64].
  65. [65].
  66. [66].
  67. [67].
  68. [68].
  69. [69].
  70. [70].
  71. [71].
  72. [72].
  73. [73].
  74. [74].
  75. [75].
  76. [76].
  77. [77].
  78. [78].
  79. [79].
  80. [80].
  81. [81].
  82. [82].
  83. [83].
  84. [84].
  85. [85].
  86. [86].
  87. [87].
  88. [88].
  89. [89].
  90. [90].
  91. [91].
  92. [92].
  93. [93].
  94. [94].
  95. [95].
  96. [96].
  97. [97].
  98. [98].
  99. [99].
  100. [100].
  101. [101].
  102. [102].
  103. [103].
View Abstract