OUP user menu

A hyperactive, Ca2+-dependent antifreeze protein in an Antarctic bacterium

Jack A. Gilbert, Peter L. Davies, Johanna Laybourn-Parry
DOI: http://dx.doi.org/10.1016/j.femsle.2005.02.022 67-72 First published online: 1 April 2005


In cold climates, some plants and bacteria that cannot avoid freezing use antifreeze proteins (AFPs) to lessen the destructive effects of ice recrystallization. These AFPs have weak freezing point depression activity, perhaps to avoid sudden, uncontrolled growth of ice. Here, we report on an uncharacteristically powerful bacterial AFP found in an Antarctic strain of the bacterium, Marinomonas primoryensis. It is Ca2+-dependent, shows evidence of cooperativity, and can produce over 2 °C of freezing point depression. Unlike most AFPs, it does not produce obvious crystal faceting during thermal hysteresis. This AFP might be capable of imparting freezing avoidance to M. primoryensis in ice-covered Antarctic lakes. A hyperactive bacterial AFP has not previously been reported.

  • Antifreeze protein
  • Thermal hysteresis
  • Calcium
  • Recrystallization inhibition
  • Ice
  • Antarctic bacterium

1 Introduction

Antifreeze proteins (AFPs) are a diverse group of ice-binding proteins that inhibit the growth of ice in two different situations. Prior to freezing they possess thermal hysteresis (TH) activity, which is the non-colligative depression of the freezing point of a solution containing ice below its melting point. In the frozen state they show ice recrystallization inhibition (RI), whereby the proteins inhibit the growth of large crystals at the expense of small crystals at high subzero temperatures. AFPs have been found in animals [1,2], plants [37], fungi [8,9] and eubacterial species [8,1012]. AFPs isolated from animals typically have substantial TH activity. This is consistent with these species having a freeze-avoidance strategy, whereby they depress the freezing point of their body fluids to prevent freezing of tissues [1,2,13,14]. However, AFP-producing plants and bacteria reported to date show substantially lower thermal hysteresis activity than do animals. This is consistent with a freeze-tolerance strategy whereby these organisms readily freeze but use the RI activity of AFPs to control the size of ice crystals [7,15]. The bacterial AFPs that have been assessed for TH, have demonstrated low activity [8,1012] compared to insect and fish AFPs, which have up to 5 and 2 °C of TH, respectively [1,1618].

In a previous study, Gilbert et al. [19] isolated 11 bacterial strains with RI activity from several Antarctic lakes in the Vestfold Hills, Eastern Antarctica (68°S, 78°E). A bacterium identified as M. protea in the original study [20] is now thought to be a strain of Marinomonas primoryensis [21] based on 16S rDNA comparison (results not shown). It was one of the most abundant AFP-active bacterial strains isolated in that study. The lakes from which M. primoryensis was cultured were isolated from the sea following the last glacial maximum around 10,000 years ago when the land rose up as the ice cap retreated [22]. They are approximately half as saline as seawater, and ice-covered, usually with a short ice-free period of 3–6 weeks in late summer. Their brackish salinity helps maintain the water temperature in the water column between −1 °C (at the ice–water interface) and +1 °C (at the base of the oxygenated water) [19], thereby constituting an ice-laden but thermally buffered environment. This study describes the preliminary characterisation of a novel, hyperactive AFP from M. primoryensis (MpAFP), revealing several unusual properties, including Ca2+-dependence.

2 Materials and methods

2.1 Protein preparation

M. primoryensis was cultured in 1 l of 50% sea water broth (SWB) growth medium (per L: yeast extract (Sigma) 1 g, peptone (Sigma) 1 g, sea salt (Coral Life, Aquatics Online) 19 g), with shaking at 4 °C for 4–5 days, or until the optical density at 600 nm reached 0.8–0.9. The cultures were centrifuged (4500×g at 4 °C for 15 min in a Beckman JA4.2 swinging bucket rotor). The cell pellet was resuspended in 2.5 ml of ice-cold lysis buffer (25 mM Tris/HCl (pH 8.0), 10 mM CaCl2). Following sonication on ice, the lysate was centrifuged (27,000 ×g (15,000 rpm) at 4 °C for 1 h in a Beckman JA20 rotor) and the supernatant (cell-free extract) was removed and stored at −20 °C.

2.2 Thermal hysteresis measurement and image capture

TH measurements were made using a Clifton Nanolitre Osmometer (Clifton Technical Physics) as previously described [23]. To study the effect of [AFP] on TH activity, serial dilutions of the lysate supernatant were made in lysis buffer. Digital images of ice crystals formed during TH assay were taken using a Nikon COOLPIX 4500 digital camera mounted on a Leitz dialux 22 microscope with a Leitz Wetzlar 160/- EF 10/0.25 objective.

2.3 Determining Ca2+-dependency

To determine the dependency of MpAFP activity on divalent metal ions, samples containing MpAFP were subjected to exhaustive dialysis against 25 mM Tris/HCl (pH 8.0), from which metal ions (e.g., Ca2+) had been chelated by passage through a Chelex-100 chelating resin column (iminodiacetic acid immobilised on 1% cross-linked polystyrene 50–100 dry mesh), (Sigma) at 0.5 ml min−1. Approximately 10 ml of crude lysate was dialysed in 3500 MWCO (molecular weight cut-off) dialysis tubing (Fisher) at 4 °C against 1 l of Ca2+-free 25 mM Tris/HCl (pH 8.0) for ?15 h, with three changes of the dialysis buffer. Following dialysis, metal salts (CaCl2, MnCl2, MgCl2, ZnCl2, CuSO4, NiSO4) were individually added to the solution to a final concentration of 10 mM to determine the effect of each upon the TH activity of the dialysed lysate supernatant.

2.4 The effects of proteolysis and chelation

The effects of proteases on TH activity were determined as follows. Pronase and trypsin (Sigma) were used at 37 °C at a 1:10 enzyme to substrate ratio. The protein concentration of the crude lysate supernatant was determined by Bradford assay (Bio-Rad) and a concentration of ?6.5 mg ml−1 was used for these experiments. Aliquots of the digests were measured for TH activity every 30 min. Proteinase K (MBI Fermentas) was added to the crude lysate supernatant to a final concentration of 2 mg ml−1 and incubated at 37 °C. TH activity again was recorded every 30 min. All protease experiments were performed in the presence and absence of Ca2+, where Ca2+ was removed from crude lysate supernatant by dialysis as described above. The effects of EDTA, EGTA and O-phenanthroline on MpAFP TH activity were recorded following incubation on ice for 1 h.

3 Results

3.1 Thermal hysteresis activity of crude lysate and growth medium

A typical preparation of crude lysate supernatant contained 11 mg ml−1 of protein and had an activity of 0.8 °C (±0.07 °C). There was little variation in concentration or activity between preparations. No activity was found in the growth medium recovered after pelleting the cells, even after 100-fold concentration.

3.2 Characterisation of MpAFP

An initial attempt to purify the MpAFP by size-exclusion chromatography of lysate supernatant resulted in the complete loss of TH activity. From this, we reasoned a low molecular weight factor was required for activity (not shown). This result was confirmed when extensive dialysis of the lysate supernatant in 3500 MWCO membranes also caused the complete loss of activity (Fig. 1). The addition of 10 mM CaCl2 to the dialysate restored full activity. The addition of other divalent metals ions (Mn2+, Co2+, Ni2+, Zn2+, Cu2+, Mg2+) to the Ca2+-free dialysate had no effect on activity.

Figure 1

Effect of various conditions on MpAFP activity in crude lysate supernatant (error bars show standard deviation): (a) control: ?6.5 mg ml−1 crude lysate supernatant; (b) exhaustive dialysis; (c) dialysate 10 mM Ca2+; (d) 10 mM EDTA; (e) 100 mM EDTA; (f) 10 mM EGTA; (g) 10 mM EDTA + 20 mM Ca2+; (h) 10 mM o-phenanthroline; (i) proteinase K (+ or −Ca2+); (j) pronase (−Ca2+); (k) trypsin (+ Ca2+); (l) trypsin (−Ca2+).

The effect of chelating agents on MpAFP activity in crude lysate supernatant was investigated (Fig. 1). Addition of 10 mM EDTA or EGTA on ice caused a 50% reduction in TH activity. Addition of 100 mM EDTA also caused a 50% reduction in activity. The saturation of 10 mM EDTA with 20 mM CaCl2 fully restored the activity, confirming that loss of activity with EDTA is due to chelation of Ca2+ ions. Addition of the iron chelator O-phenanthroline to 10 mM had no effect on activity.

The effect of proteolysis on MpAFP activity of the crude lysate supernatant, was tested with and without Ca2+ present (Fig. 1). Pronase, in the presence of Ca2+, caused a decrease in TH activity of 0.09 °C, for every 30 min of incubation at 37 °C (not shown). In the absence of Ca2+, activity was extinguished within 30 min. Proteinase K was more detrimental, eliminating activity within 30 min in the presence or absence of Ca2+. Trypsin had no effect on activity in the presence of Ca2+. There was no loss of activity following incubation at 37 °C with trypsin for 6 days; thereby indicating the thermal stability of MpAFP at 37 °C when Ca2+ is present. However, in the absence of Ca2+, activity was completely extinguished by 30 min.

3.3 Effect of dilution on crude lysate supernatant TH activity

Two-fold serial dilutions of this cell-free extract were prepared and assayed for TH. The plot of activity against protein concentration (Fig. 2) showed a sigmoidal relationship at low protein concentrations. TH activity was undetectable below 0.18 mg ml−1 but began to increase rapidly at 0.36 mg ml−1 and followed a more hyperbolic relationship with protein concentration from there on. The crude bacterial lysate supernatant was quite active in comparison to pure type III AFP. Concentrating the cell-free extract approximately 20-fold using a 30,000 MWCO Amicon ultra centrifugal concentrator (Millipore) increased the activity to >2 °C, higher than the maximal activity of most fish AFPs.

Figure 2

Thermal hysteresis activity versus protein concentration curves for pure fish type III AFP (♦) and crude lysate supernatant of M. primoryensis (▲). The insert shows the sigmoidal nature of the M. protea curve at low protein concentrations.

3.4 Ice crystal morphology

Ice crystals formed in the presence of MpAFP do not have distinct facets (Fig. 3(a) and (b)). They are typically rounded in shape and do not change their morphology during the course of the TH measurement. The ice crystal “burst” occurring at the end-point of TH is dendritic with hexagonal symmetry, suggesting growth from primary or secondary prism faces or edges (Fig. 3 (c)). This more closely resembles the “bursts” seen with insect AFPs than “bursts” that occur along the c-axis from crystals inhibited by fish AFPs [24].

Figure 3

Ice crystal morphologies obtained with MpAFP. (a) and (b) two different ice crystals obtained in the presence of M. primoryensis crude lysate supernatant at 0.3 °C of under-cooling. (c) a dendritic crystal growing extremely rapidly below the non-equilibrium freezing point. Size marker = 50 μm and relates to (a) and (b) only.

4 Discussion

In this study, we have partially characterised a Ca2+-dependent, bacterial AFP that demonstrates hyperactive thermal hysteresis. Previously, bacterial AFPs have demonstrated RI activity but only low levels of TH activity (<0.1 °C), e.g., Micrococcus cryophilus & Rhodococcus erythropolis [8], Pseudomonas putida [10], Moraxella sp. [11] and Pseudomonas fluorescens [12]. The presence of an AFP with low TH activity would suggest that these bacterial strains employ a freeze-tolerant strategy similar to that of plants, e.g., rye grass [6], carrot [4], winter rye [25] and bittersweet nightshade [3]. The presence of a highly active bacterial antifreeze in M. primoryensis suggests that some bacteria, like insects [26,27] and fish [2] have also evolved a freeze-avoidance strategy, which has not yet been reported in microorganisms.

As this strain of M. primoryensis was isolated from a saline (1.9%), permanently cold (−1 to +1 °C) Antarctic lake with 1–2 m deep ice cover [19], we hypothesise that this bacterium has evolved a highly active AFP to resist freezing in the water column. An AFP with TH activity comparable to that found in marine fish should be sufficient to inhibit the growth of ice crystals that might otherwise be propagated into the bacterium, especially near the ice–water interface of the lake. The other AFP-active bacterial species were isolated from harsher, frozen environments, i.e., high Arctic plant rhizosphere (P. putida [10]), mid-gut of frozen beetle larvae (R. erythropolis), frozen/chilled pork sausages (M. cryophilus [8]), and Antarctic soil (Moraxella sp [11] and P. fluorescens [12]). Bacteria in these environments would require an RI-active protein to prevent lethal recrystallization. Even the most potent AFPs are unlikely to prevent a bacterium from freezing in these harsh environments, and high TH activity will tend to cause rapid, uncontrollable ice crystal growth when freezing inevitably occurs.

Most previously reported bacterial AFPs (see above) were shown to be exported from the cell [8,1012]. However, in M. primoryensis AFP activity was only found in the cellular lysate supernatant. While we have not yet determined the location of MpAFP within the cell, we suggest that it would be most effective if localised to the periplasmic space, where it would be likely to encounter, bind and inhibit embryonic ice crystals from the extracellular environment before they damage the cell.

Many AFPs impart a characteristic morphology to ice crystals as a result of their crystal plane-specific binding. For example, fish AFP type I from winter flounder binds to the {20–21} pyramidal planes of the ice crystal, forming hexagonal bipyramids [28]. This morphology is assumed on initial cooling at the start of the TH-gap (temperatures between non-equilibrium melting and freezing points). Below the non-equilibrium freezing point, fish AFPs tend to exhibit uncontrollable growth along the c-axis [24]. In comparison, the insect AFP from Choristonerua fumiferana binds to both the primary prism planes and basal planes, to produce a hexagonal plate morphology that is stable throughout the TH-gap. The structurally similar but unrelated AFP from Tenebrio molitor produces an even more complex lemon-shaped crystal morphology [16] suggesting possible binding to more than just the basal and primary prism planes. For both insect AFPs, uncontrollable ice growth below the non-equilibrium freezing point occurs primarily along the a axes, producing a dendritic crystal pattern with hexagonal symmetry [29]. This “burst” pattern is also seen with MpAFP. Other bacterial AFPs have been shown to produce hexagonal crystal shaping [1012], but as yet their binding plane specificity has not been reported.

Unlike most AFPs, which show a hyperbolic relationship between TH activity and AFP concentration [30], the sigmoidal shape of the dilution curve at low protein concentrations of MpAFP is suggestive of cooperativity in binding to ice or inhibition of its growth. A similar result has been reported for one of the smaller AFGPs [17].

To date, Ca2+-dependent AFP activity has only been shown in fish AFP type II from rainbow smelt (Osmerus mordax) and Atlantic herring (Clupea harengus) [31]. The herring AFP, like MpAFP, demonstrates a conformational change from a protease-sensitive, inactive form to a protease-resistant, active form with the addition of Ca2+ [31]. Type II AFPs are homologous to the carbohydrate-recognition domain of the Ca2+-dependent (C-type) animal lectins [32]. Although there are lectins in bacteria that bind Ca2+ [33], there is no evidence at this time that MpAFP is a homolog. Stressman et al. [34] noted that during freeze/thaw cycling, the AFP from winter rye (Secale cereale) bound Ca2+ to the detriment of its activity. However, if the AFP was maintained at 4 °C, Ca2+ had no effect on its activity. We note that the bacterial AFP from P. putida, AfpA, contains a series of putative Ca2+-binding repeats that could be involved in a possible hemolysin-like, Ca2+-binding secretion domain [35]. However, there have been no reports as yet of Ca2+-dependent bacterial AFPs. As such the Ca2+-dependency in MpAFP is novel for AFPs from this kingdom.

In summary, we report the discovery of a Ca2+-dependent, hyperactive bacterial antifreeze protein from an Antarctic lake strain of the bacterium M. primoryensis.


This work was funded by a grant from the Canadian Institutes of Health Research (CIHR). P.L.D holds a Canada Research Chair in Protein Engineering. We are grateful to Chris Marshall and Dr. Andrew Scotter for comments and suggestions.


  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].
View Abstract