OUP user menu

A comparison of the growth and starvation responses of Acanthamoeba castellanii and Hartmannella vermiformis in the presence of suspended and attached Escherichia coli K12

Zoë L. Pickup, Roger Pickup, Jacqueline D. Parry
DOI: http://dx.doi.org/10.1111/j.1574-6941.2006.00224.x 556-563 First published online: 1 March 2007


The growth and starvation responses of Acanthamoeba castellanii and Hartmannella vermiformis were investigated in the presence and absence of Escherichia coli on an agar surface or within shaken suspensions. The amoebae perceived all the suspended systems to be unfavourable for growth, despite being challenged with high levels of prey, and as a consequence they exhibited a starvation response. However, the response differed between species, with A. castellanii producing characteristic cysts and H. vermiformis producing round bodies. These amoebic forms were reactivated into feeding trophozoites in the presence of bacterial aggregates, which formed in the suspended systems after 68 h of incubation. In contrast, both species of amoebae grew well in the presence of attached E. coli at a concentration of 1 × 106 cells cm−2 of agar and yielded specific growth rates of c. 0.04 h−1. Starvation responses were induced at the end of the growth phase, and these were equivalent to those recorded in the suspended systems. We conclude that, when suspended, amoebae in the ‘floating form’ cannot feed effectively on suspended prey, and hence the starvation response is initiated. Thus the majority of amoebic feeding is via trophozoite grazing of attached bacterial prey.

  • amoeba
  • cyst
  • yield
  • growth rate
  • ingestion rate
  • biofilm


Free-living protozoa are the most abundant group of phagotrophic organisms in the biosphere (Finlay, 2001). Their predatory role in planktonic systems is undisputed; they feed primarily on bacteria, often consuming 30–100% of bacterial production per day (Sherr et al., 1983). In addition, protozoa play a crucial role in the cycling of nitrogen and phosphorus, particularly in oligotrophic environments (Goldman et al., 1985; Eccleston-Parry & Leadbeater, 1995), and constitute an important component of the ‘microbial loop’ (Azam et al., 1983). Less is known about the role of protozoa in surface-associated communities, such as biofilms, primarily due to the belief that bacteria are afforded refuge from grazing by being embedded within the exopolymer matrix (Costerton et al., 1981). This has since been disproved (Heaton et al., 2001), and subsequent work has shown that protozoa can effectively graze on attached bacteria, with their activity influencing the topography of the biofilm matrix (Ardnt et al., 2003; Matz et al., 2004; Parry, 2004; Matz et al., 2005; Weitere et al., 2005).

Amoebae constitute an important component of surface-associated protozoan communities (Parry, 2004), with published amoebic ingestion rates of attached bacteria ranging from 0.2 to 1465 bacterial cells per amoeba cell per hour (Rogerson et al., 1996; Butler & Rogerson, 1997; Mayes et al., 1997; Heaton et al., 2001; Huws et al., 2005). These predators are thought not to feed on suspended bacterial cells (Martin, 1985; Rogerson & Laybourn-Parry, 1992), even though there are reports of successful experiments being carried out in suspended systems (Weekers et al., 1993). This contradiction, and the assumption that amoebae can only effectively feed on attached prey, has never been convincingly demonstrated (Rogerson et al., 2003), and this is the reason why this study was initiated.

Naked amoebae are normally thought to exist in three morphological forms, i.e. the motile trophozoite, the ‘floating form’, and the much smaller cyst (Page, 1988), but a fourth form, known as a ‘round body’, has also been reported (Band, 1963; Griffiths, 1970; Weisman, 1976). The lack of pseudopodia on the round body results in a cessation of movement and feeding (McConnachie, 1969; Griffiths, 1970), and it is a transient form that, in naked amoebae, leads to cyst formation under conditions of starvation, osmotic stress, anaerobiasis, or treatment with chemicals (Band, 1963; Neff et al., 1964; Griffiths & Hughes, 1969; Weisman, 1976; Lasman, 1987; Turner et al., 1997; Boucard et al., 2004). The round bodies of the social amoeba Dictyostelium discoideum can revert back to active trophozoites if nutrients are replenished (Yarger et al., 1974), but no corresponding data exist for naked amoebae.

The feeding form of amoebae, the trophozoite, possesses numerous pseudopodia, and has been observed to feed on attached prey (Arndt, 1993; Preston et al., 2001). It is presently unclear whether ‘floating forms’ of amoebae feed on suspended prey. Their elongated pseudopodia comprise a more stable cytoskeletal structure that resists compression (Ueda & Ogihara, 1994), and although this is thought to aid in the drifting mechanism of these cells, the stiffened pseudopodia are not thought to provide a method of propulsion or feeding (Smirnov, 2002; Smirnov & Brown, 2004). However, growth rates of amoebae in shaken suspended systems have been reported (Weekers et al., 1993), so it remains unclear as to whether ‘floating forms’ can indeed feed planktonically or whether the experimental system affords a subhabitat for trophozoites to feed on attached bacteria. Attachment can occur in unshaken suspended systems, with participating organisms settling onto, or attaching to, the bottom of the container.

This study tested the hypothesis that ‘floating forms’ of amoebae do not feed effectively on suspended prey but that the majority of amoebic grazing occurs on surfaces, by trophozoites. In order to test this, the growth and starvation responses of two species of amoebae were compared in the presence and absence of attached and suspended Escherichia coli. Growth responses were evaluated by determining the specific growth rates, ingestion rates and yields of the amoebae, and starvation responses were evaluated from observations on the inducement and extent of either round body or cyst formation.

Materials and methods


Escherichia coli K12 was maintained as streak plates on Nutrient Agar incubated at 25°C for 24 h. Two spread plates were flooded with distilled water, the bacterial cells were dislodged with a sterile spreader, and the suspension was collected in a sterile McCartney bottle, which was then sonicated for 10 min using a Camlab Transonic T460 water bath (Highfield frequency = 35 Hz). The concentration of bacterial cells was determined in a subsample fixed in glutaraldehyde (0.4% v/v final concentration). The cells were 4′-6-Diamidino-2-phenylindole (DAPI) stained (Porter & Feig, 1980), filtered through a 0.2-μm-pore black track-etched polycarbonate filter (Osmonics), and enumerated with a Leitz Labourlux S epifluorescence microscope under UV excitation (BP340/LP430 filter block) at a final magnification of × 1250. At least 400 cells per filter were counted before the cell concentration (cells mL−1) was determined.

The amoebae Acanthamoeba castellanii [Culture Collection of Algae and Protozoa (CCAP) 1501/1A] (c. 25–40 μm) and Hartmannella vermiformis (CCAP 1524/7A) (c. 15–28 × 5 μm) were routinely maintained on Amoeba Saline (AS) agar plates (Page, 1988), supplemented with a streak of E. coli K12, at 20°C. Ten agar plates, which had been incubated for 6 days, were examined microscopically for the presence of cysts, and these areas, together with any remaining streak of E. coli, were removed prior to flooding the plates with AS broth, dislodging the attached amoebic cells with a sterile spreader, and collecting the suspension in a 40-mL tissue culture flask. The amoebae were rested overnight at 20°C to allow the cells to attach to the bottom of the flask, and then the overlying AS broth was carefully removed to eliminate suspended cysts, debris and bacteria. The adherent cells were washed three times with fresh AS medium, after which the flasks were vortexed gently for 20 s to dislodge the amoebae into suspension. The concentration of amoebic cells was determined on a subsample, fixed with glutaraldehyde (1% v/v final concentration), using a haemocytometer. This procedure resulted in monaxenic amoebic suspensions with an E. coli concentration of c. 1 × 104 cells mL−1.

Experimental design

A bacterial concentration representative of that found in the natural plankton (c. 1 × 106 cells mL−1) (Caron, 1991) was selected for the first suspended system (Suspended-Low). The amoeba species was added at a concentration of 5 cells mL−1 to yield a predator/prey ratio of 1 : 2 00 000 and give a low enough starting density to allow for prolonged exponential growth within the batch culture. These concentrations were reproduced as lawn densities on agar medium in the attached system, i.e. 5 amoeba cells cm−2 and 1 × 106 bacterial cells cm−2 (Attached). As the prey cells would be in closer proximity on the agar surface (10 μm apart – immobilized) compared to the suspended system (100 μm apart – if immobilized), which might be advantageous to the amoeba, a third system was introduced that attempted to reproduce the attached system in three dimensions (Suspended-High). An initial concentration of 1 × 109E. coli cells mL−1 would be required to provide distance of 10 μm between E. coli cells in suspension (if immobilized), but this bacterial concentration resulted in methodological problems, e.g. poorly suspended cells, and hindrance of amoebic counts. Hence, E. coli was inoculated into flasks at 2 × 108 cells mL−1 (17 μm apart – if immobilized), and the amoeba was inoculated at 1000 cells mL−1; thus, a predator/prey ratio of 1 : 2 00 000 was maintained in each experimental system.

Experimental systems

The amoeba species and E. coli were inoculated into triplicate 250-mL flasks containing 100 mL of AS broth to yield starting concentrations of 1 × 106E. coli cells mL−1 and 5 amoeba cells mL−1 (Suspended-Low test), or 2 × 108E. coli cells mL−1 and 1000 amoeba cells mL−1 (Suspended-High test). Triplicate control flasks containing the amoeba species alone (Suspended-Low and Suspended-High controls) and E. coli alone (bacterial controls) were also prepared. All flasks were incubated with rotary shaking at 100 r.p.m. at 20°C, and subsamples were removed twice daily for 4 days. The concentrations of bacteria and amoeba within the systems were determined with DAPI-staining/epifluorescence microscopy and light microscopy, respectively, as previously described. For the attached system (Attached test), the amoeba and E. coli were coinoculated into 2 mL of AS broth to give final concentrations of 160 and 3.2 × 107 cells mL−1, respectively. The suspension was poured onto the surface of an AS agar plate (90 mm) that had been prepared with 45 mL of molten medium to yield an agar thickness of 7 mm. The inoculated plate was allowed to dry in a laminar flow cabinet, after which the lawns were examined microscopically for 1-cm2 areas showing confluent bacterial lawns and the presence of about five amoeba cells. Three 1-cm2 areas of inoculated agar were aseptically excised, and each was placed onto 10 μL of sterile distilled water in a Sedgwick Rafter Counting Chamber. Controls of amoeba alone (Attached controls) and E. coli alone were also prepared, in triplicate. The counting chambers were incubated at 20°C, on top of moistened foam in Perspex boxes, to prevent desiccation. The agar sections were examined twice daily with a light microscope (× 400 to × 1200 final magnification, depending on species). The number of amoebae on the whole 1-cm2 surface was deduced. At the beginning and end of the experiment, additional agar surfaces of tests and bacterial controls were fixed in glutaraldehyde (0.4% v/v final concentration), and the surfaces were scraped to remove the adherent cells into suspension. The bacterial concentration (cells cm−2) was determined after DAPI staining and epifluorescence microscopy as previously described.


All analyses were carried out using spss statistics software and the microsoft excel statistics package. The specific growth rates (h−1) of the amoebae were determined by regression analysis of the linear portion of the graph of Ln amoeba mL−1 or cm−2 against time (h) after correcting for any amoebic growth in the control systems. Regression lines were statistically compared using ancova. The number of amoeba cells produced per ingested E. coli cell (Yield) (amoeba cells per E. coli cell) was determined by dividing the concentration of new amoebae produced in the system by the concentration of bacteria removed from the system. Ingestion rates (E. coli cells per amoeba cell per hour) were determined by dividing half the loss in bacterial concentration over the whole experimental period by the average concentration of amoebae following Sherr et al. (1983), a method that accounts for the growth of the amoeba population during the sampling intervals. Ingestion rates and yields were compared using a Student's t-test. The onset and extent of any amoebic round body or cyst forms were noted throughout the experiment.

Results and discussion

The growth and starvation responses of two species of amoebae were evaluated in the presence and absence of attached and suspended E. coli prey. The study aimed to test whether a suspended system afforded a positive growth response in the amoeba population or whether it induced one of starvation, as amoebae are considered to be surface feeders.

Growth response of attached trophozoites

Each amoeba species was fed with E. coli at a concentration of 1 × 106 cells cm−2 on an agar surface (Attached test). The trophozoites of both species were observed to feed and divide immediately under these conditions, resulting in no apparent lag phase in their growth curves (Fig. 1; open symbols). As the cells continued to divide, the Ln concentration of both species of amoebae increased linearly (Fig. 1) but there was no significant difference between the calculated specific growth rates of the two species (P>0.1; Table 1). Similarly, there was no significant difference between the calculated ingestion rates and yields of the two species (P>0.1 for both parameters) (Table 1). After 68 h, the population densities remained relatively stable at around 200 and 500 cells cm−2 for A. castellanii and H. vermiformis, respectively (Fig. 1).


The change in the cell concentration (±SEM) of Hartmannella vermiformis (circles) and Acanthamoeba castellanii (squares) in the absence (controls, closed symbols) and presence (tests, open symbols) of Escherichia coli on an agar surface at 1 × 106 cells cm−2, at 20°C. *Time at which cyst formation or round bodies were initiated in the A. castellanii and H. vermiformis populations, respectively.

View this table:

Net specific growth rates (h−1), ingestion rates (Escherichia coli cells per amoeba cell per hour) and yield (amoeba cells per E. coli cell) (±SEM) for Hartmannella vermiformis and Acanthamoeba castellanii feeding on E. coli at 1 × 106 cells mL−1 (Suspended-Low), 2 × 108 cells mL−1 (Suspended-High) and 1 × 106 cells cm−2 (Attached), at 20°C

Specific growth rate (h−1)Ingestion rate (E. coli cells per amoeba cell per hour)Yield (10−3 amoeba cells per E. coli cell)
A. castellanii
Attached0.039 (0.004)964 (78)0.029 (0.002)
H. vermiformis
Suspended-HighND783 (69)0.024 (0.004)
Attached0.040 (0.003)139 (29)0.042 (0.003)
  • ND, not determined.

Published specific growth rates and ingestion rates of amoebae determined with agar-based experimental systems are rare, and data are highly variable. Ingestion rates range from 0.2 to 1465 bacteria per amoeba per hour at 20°C (Rogerson et al., 1996; Butler & Rogerson, 1997; Mayes et al., 1997; Heaton et al., 2001), whereas those for specific growth rate range from 0.01 to 0.15 h−1 at 20°C (Baldock & Baker, 1980; Baldock et al., 1980; Laybourn & Whymant, 1980). Calculated growth rates and ingestion rates in the present study (Table 1) fall within the published range, but a direct comparison of parameters between experiments is difficult, because other authors employed different amoeba species, heat-killed fluorescently labelled prey, and a direct method for determining ingestion rate (fluorescent prey per amoeba over time). This contrasts with the live E. coli employed in this study, coupled with an indirect method for determining ingestion rate (loss of external prey over time). In addition, the initial concentration of attached prey as cells cm−2 is often not reported in agar-based studies, but this is also a problem with nonagar-based systems such as a settled suspended system, where prey concentration is expressed as cells cm−3. The only published values of amoebic yield, originating from settled suspended systems, range from 0.25 to 60 × 10−3 amoeba cells per prey cell for Acanthamoeba sp., Hartmannella hyaline and Vexillifera sp. feeding on a yeast, bacterium and diatom, respectively (Cutler & Crump, 1927; Heal, 1967; Bunt, 1970). These are higher than the yields deduced in this experiment (Table 1), but once again are not directly comparable, for the reasons stated above.

Starvation response of attached trophozoites

The degree of encystment in amoebic populations can be highly variable (1–100%) and has been shown to depend on the culturing conditions prior to starvation (Band, 1963; Neff et al., 1964; McConnachie, 1969; Griffiths, 1970; Lasman, 1987), with more optimal preculturing conditions leading to increased cyst formation (Neff et al., 1964). In this study, the onset and extent of cyst formation in each species were compared with and without prefeeding of the amoeba population with E. coli. In the absence of feeding (Attached control), both amoebae underwent some degree of cellular division on the agar surfaces (Fig. 1; closed symbols). This is not uncommon in protozoan studies, as starved cells are known to divide at least once before producing resting stages (Fenchel, 1982). The resting stages of amoebae are normally cysts, but these cysts are thought to be preceded by round bodies. Even so, round bodies were not observed within the A. castellanii population, with or without feeding, but considering that this cell form only exists for 7–12 h in this species (Weisman & Moore, 1969; Weisman, 1976), it may have been missed due to infrequent sampling. However, the cells of A. castellanii did show encystment in both systems. In the absence of feeding (Attached control), cyst formation was evident by 32 h (Fig. 1), and by 92 h, 70–100% of the population had encysted. In the presence of feeding (Attached test), cyst formation was evident by 86 h, which was 18 h into the stationary phase (Fig. 1) and earlier than the 32 h recorded in the complete absence of feeding (Attached control). By 92 h, 20% of the population had encysted (Fig. 1). Visual observation of these latter agar systems showed that cyst formation in a given area was initiated by one amoebic cell, and then the surrounding cells encysted in concentric ‘waves’, suggesting the involvement of some form of cell–cell signalling. Synchronous encystment has been observed by other authors (Band, 1963; Neff et al., 1964; Griffiths & Hughes, 1969; Weisman, 1976; Turner et al., 1997), and the supernatants of encysting or stationary cells have been shown to induce encystment in trophozoites of A. palestinensis (Lasman, 1987) and D. discoideum (Yarger et al., 1974). No significant encystment was detected in the population of H. vermiformis. Instead, the majority of the population existed as round bodies from 6 h in the absence of E. coli (Attached control) and from 86 h in the presence of E. coli (Attached test).

Growth and starvation responses of suspended amoebae

Visual observation of the amoeba cells within all the suspended flasks indicated that the transformation from suspended trophozoites to ‘floating forms’ occurred by 6 h into the experiment, affording the trophozoites only a short period of time to ingest suspended prey. However, suspended trophozoites of A. castellanii have been shown to ingest an average of 0.66 latex beads per cell over a period of 30 min (Avery et al., 1995), so some level of ingestion, although low, would have been expected in our suspended systems. However, both species of amoebae exhibited the same starvation response in the presence of two concentrations of E. coli (Suspended-Low and Suspended-High tests) and in its absence (Suspended-Low and Suspended-High controls), and neither species showed any significant increase in population size (Fig. 2), indicating that they perceived all suspended conditions to be unfavourable for growth. This response could not have been due to oxygen depletion, as the suspensions were well aerated by agitation. In addition, this level of agitation should have increased the probability of prey capture and subsequent ingestion of prey (Avery et al., 1995). Thus, the onset of unfavourable conditions would have been due to the presence of either the incorrect form of the prey (suspended) or the incorrect form of the amoeba (a nonfeeding form). The starvation responses of both suspended amoebae were independent of the E. coli concentration (equivalent in Suspended-Low and Suspended-High), which suggests that the transition from trophozoite to ‘floating form’ initiated the starvation response, indicating that these latter cells do not feed on suspended prey, as predicted from their morphological features (Smirnov, 2002; Smirnov & Brown, 2004).


The change in the cell concentration (±SEM) of (a) Hartmannella vermiformis and (b) Acanthamoeba castellanii in the presence of Escherichia coli at 1 × 106 cells mL−1 (open circles) and 2 × 108 cells mL−1 (open squares), and the corresponding controls in the absence of E. coli (closed symbols) in suspension at 20°C. *Time at which cyst formation or round bodies were initiated in the A. castellanii and H. vermiformis populations, respectively.

The starvation responses exhibited in all the suspended systems (Suspended-Low and Suspended-High tests and controls) were compared to those observed in the absence of attached E. coli on the agar systems (Fig. 1; closed symbols). The round body form of H. vermiformis began to appear from 6 h in the absence and presence of suspended E. coli, which was similar to what was observed in the absence of the prey on the agar systems. Once again, minimal cyst formation was evident in the suspended population. With regard to A. castellanii, no round bodies were observed but encystment was evident from 32 h in all suspended systems, the same response as that recorded on agar devoid of prey. In this case, almost complete encystment of the population occurred earlier in the suspended systems (62 h) than in the agar system devoid of prey (92 h).

Given that amoebae were not feeding in suspension, no true values for specific growth rate, ingestion rate or yield could be deduced from the suspended systems; however, experimental ‘artefacts’ did lead to apparent parameter values in the Suspended-High test system supplemented with 2 × 108E. coli cells mL−1 (Table 1). The ‘artefact’ in question was the formation of flocs (bacterial aggregates), which probably originated from the sloughing of attached growth on the inner surface of the flask (Butler & Rogerson, 1996). Flocs were present after a period of 68 h, and active trophozoites were clearly associated with them. This suggests that reactivation of dormant amoebae had occurred in the suspended systems due to the presence of surfaces for attachment, i.e. on the flask surface or on a floc. However, this reactivation takes some time to achieve (68 h here), which might explain why extensive lag periods are recorded in amoebic growth curves carried out in agitated suspended systems. For example, Weekers et al. (1993) recorded a lag phase of 60 h before a net positive growth rate was recorded for A. castellanii, A. polyphaga and H. vermiformis in a continuously agitated system with a bacterial concentration in excess of 1 × 109 cells mL−1. It may have been that this was the time necessary for adequate floc formation in this system, but this was not reported. In the current experiment, neither species of amoeba yielded a continuous increase in population density in the presence of 2 × 108E. coli cells mL−1 (Fig. 2); instead, H. vermiformis showed a slight increase in abundance after floc formation (Fig. 2a). This increase possibly indicates that the reactivation of H. vermiformis from round body to trophozoite is more rapid than the reactivation of cysts in the A. castellanii population, but this requires further study. With only two data points, an accurate specific growth rate could not be determined; however, calculated ingestion rate and yield values (over the 92-h period) gave values of the same order as those calculated from the attached system (Table 1).

Methodological considerations

Agar-based experimental systems have not been used extensively to monitor amoeba–bacteria interactions, and many researchers have employed a settled suspended system (Cutler & Crump, 1927; Heal, 1967; Bunt, 1970; Martin, 1985; Butler & Rogerson, 1996; Zubklov & Sleigh, 1999; Hauer & Rogerson, 2005; Weitere et al., 2005). However, the agar system is particularly useful for determining the functional response of amoebae (Baldock et al., 1980), as a known concentration of attached prey cm−2 is present on the agar surface. Because there is no overlying water, or shaking of the agar, losses of bacteria are unequivocally due to ingestion by the amoebae, and not to sloughing of bacteria from the surface. In addition, the agar-based system allows investigators to observe the behaviour of participating organisms, which is often missed if destructive sampling and fixation is employed. In this study, visual observation of the trophozoites on the agar not only showed synchronous encystment, but yielded information regarding their movement, via the examination of ‘amoeba trails’ in the bacterial lawn. For example, A. castellanii cells were observed to move rapidly on a meandering path, producing erratic trails, whereas H. vermiformis cells moved far less, often remaining in close proximity to their initial position on the agar. Upon division, the daughter cells of both species moved in opposite directions to each other, creating a T-junction in the amoeba trail, and daughter cells never ventured closer than 30–100 μm to each other. Thus, the development of an amoeba population from a single initial mother cell can be tracked and evaluated using this technique.


Acanthamoeba castellanii and H. vermiformis perceived all the suspended systems to be unfavourable for growth, despite being challenged with high levels of suspended prey. The results suggest that this is due to the inability of the amoebic ‘floating form’ to feed on suspended prey, causing a starvation response to be initiated in suspension. The amoebae exhibited different starvation responses, with A. castellanii producing characteristic cysts but H. vermiformis producing round bodies. However, these amoebic forms could be reactivated into trophozoites in the presence of a surface, i.e. bacterial aggregates or the surface of the experimental flask. Trophozoites were shown to feed effectively on attached prey at a relatively low concentration of 1 × 106 cells cm−2 of agar, and hence we conclude that the majority of amoebic feeding is via trophozoite grazing of attached bacterial prey. The long-held view that biofilm-associated bacteria are not grazed upon by protozoa cannot be true for amoebae, as otherwise their extinction would have been evident. In addition, many interactions between amoebae and pathogens are recognized to occur within biofilms (Abu-Kwaik et al., 1998; Brown & Barker, 1999; Abd et al., 2003), with the amoeba often ending up as the prey rather than the predator. Thus, complex interactions occur within these systems, and the use of the agar system should allow a detailed examination of not only basic growth and survival parameters, but also the behaviour of the participants while these parameters are achieved.


The authors express thanks to Dr Terry Preston, Dr Alexey Smirnov and Dr Sutherland Maciver for useful discussions during the preparation of this manuscript and to the anonymous reviewers for their useful comments.


  • Editor: Riks Laanbroek


View Abstract