Growth kinetics of microorganisms isolated from Alaskan soil and permafrost in solid media frozen down to −35°C


  • Editor: Rosa Margesin

Correspondence: Nicolai S. Panikov, Department of Chemistry & Chemical Biology, Stevens Institute of Technology, Castle Point on Hudson, Hoboken, New Jersey 07030, USA. Tel.: +201 216 8193; fax: +201 216 8240; e-mail:


We developed a procedure to culture microorganisms below freezing point on solid media (cellulose powder or plastic film) with ethanol as the sole carbon source without using artificial antifreezes. Enrichment from soil and permafrost obtained on such frozen solid media contained mainly fungi, and further purification resulted in isolation of basidiomycetous yeasts of the genera Mrakia and Leucosporidium as well as ascomycetous fungi of the genus Geomyces. Contrary to solid frozen media, the enrichment of liquid nutrient solutions at 0°C or supercooled solutions stabilized by glycerol at −1 to −5°C led to the isolation of bacteria representing the genera Polaromonas, Pseudomonas and Arthrobacter. The growth of fungi on ethanol–microcrystalline cellulose media at −8°C was exponential with generation times of 4.6–34 days, while bacteria displayed a linear or progressively declining curvilinear dynamic. At −17 to −0°C the growth of isolates and entire soil community on 14C-ethanol was continuous and characterized by yields of 0.27–0.52 g cell C (g of C-substrate)−1, similar to growth above the freezing point. The ‘state of maintenance,’ implying measurable catabolic activity of non-growing cells, was not confirmed. Below −18 to −35°C, the isolated organisms were able to grow only transiently for 3 weeks after cooling with measurable respiratory and biosynthetic (14CO2 uptake) activity. Then metabolic activity declined to zero, and microorganisms entered a state of reversible dormancy.


Permafrost, which occupies >20% of the world's land surface, is extremely sensitive to global climate warming and its degradation could have a dramatic impact on the entire Earth system (Nelson et al., 2001; Elberling and Brandt, 2003; Callaghan et al., 2004). An accurate prediction of cryosphere dynamics critically depends on understanding the behavior of permafrost, including its biotic component. For many years permafrost was considered a depository of ancient microbial life but recently-discovered intensive winter gas fluxes from tundra to atmosphere (Whalen and Reeburgh, 1988; Zimov et al., 1993; Hobbie and Chapin, 1996; Oechel et al., 1997) stimulated a search for organisms able to metabolize below the freezing point.

Numerous reports have been published recently on microbial activity detected at various temperatures from 0 to −20°C in permafrost and similar frozen habitats, such as sea and lake ice, glaciers and polar snow. A summary of the available literature is given in Table 1, which contains the referenced technique used, natural habitat and temperature range in which the activity was recorded.

Table 1.   Techniques to measure microbial activity in permafrost and other naturally frozen habitats
Technique HabitatTemp (°C)Ref
Incorporation of labeled precursors intoDNA (3H-thymidine) and proteins (14C-Leucine)Glacial ice bacteria−15Christner (2002)
South Pole Snow−17 to −12Carpenter et al. (2000)
Proteins (15 14C-amino acids)Arctic sea ice−1.3 to −1Ritzrau (1997)
Lipids (14C-acetate)Siberian permafrost−20 to −0Rivkina et al. (2000)
Gas fluxesCO2Barrow permafrost, AK−40 to −0Panikov et al. (2006a)
Tussock tundra, AK−12 to −0Mikan et al. (2002)
CH4Siberian permafrost−16.5 to −0Rivkina et al. (2002)
N2OAlpine tundra, Colorado−5 to −0Brooks et al. (1997)
Gradient of entrapped gasesMountain glacier, Bolivia−40 to −0Campen et al. (2003)
O2Antarctic peat−1 to −+1Wynn-Williams (1982)
Photosynthetic 14CO2 uptakeAlpine, Tibet−10 to −+20Kato et al. (2005)
Endolithic lichen, Antarctica−24 to −+5Kappen & Friedmann (1983); [24]Kappen (1993)
Dark 14CO2 uptakePermafrost and tundra, AK−80 to −0Panikov & Sizova (2006)
Organic matter decompositionNet N mineralization and nitrificationTaiga & tundra soils, AK−5 to −+5Clein & Schimel (1995)
Tundra, AK−30 to −+5Schimel et al. (2004)
Plant litter weight lossTussock tundra, AK−30 to −+Hobbie & Chapin (1996)
Loss of K, Mg, P, phenolics and carbohydratesSubarctic woodland, CanadaNDMoore (1983)
14C-glucose oxidationBarrow permafrost, AK−40 to −0Panikov et al. (2006a)
UV MicroscopyStaining of DNA (DAPI) and respiration products (CTC)Arctic Sea Ice−20 to −2Junge et al. (2004)

A survey of the most recent literature highlighting molecular adaptation, biodiversity and microbial dynamics in the cold (Deming, 2002) led to the conclusion that our understanding of microbial life in frozen habitats should be based on comprehensive studies of complex interactions among various biological, chemical and physical factors at water–ice interfaces, e.g. solute concentration, pressure and the physical state of the water and ice, rather than the effect of temperature alone. In addition to ‘psychrophiles,’ the novel ecophysiological group of ‘eutectophiles’ was introduced to designate specialized organisms living at the critical interface inherent to the phase change of water to ice.

To make the new concept credible, we need more careful studies of isolated eutectophiles in frozen media that have ice–water interfaces similar to permafrost or glaciers. So far the growth of microorganisms below 0°C was explicitly demonstrated only for supercooled liquid media with added antifreeze compounds (Breezee et al., 2004).

Recently we have traced the oxidation of organic compounds and the dark 14CO2 fixation in Alaskan permafrost down to −40°C (Panikov et al., 2006); we have also found that gases and volatiles (H2, CH4, ethanol) are able to penetrate the intact permafrost body and be used as growth substrates by permafrost microorganisms (Panikov & Sizova, 2006). In this paper, we describe a novel technique for isolation and cultivation of bacteria and fungi in solid icy media frozen down to −35°C without artificial antifreezes. We followed growth dynamics and measured kinetic and stoichiometric parameters to assess growth efficiency and characterize the physiological state of metabolically active cells in frozen media.

Materials and methods


Sampling was carried out in May, 2004 by Dr V. Romanovsky in Smith Lake, a forest wetland close to the gas flux site of the University of Alaska, Fairbanks AK (64°52′N 147°51′W). The soil was coarse-loamy, mixed, superactive gelic ruptic histoturbel, with the second permanently frozen, buried organic layer at 50–65 cm depth; the active layer typically 35–40 cm. Permafrost cores (diameter ∼10 cm) were extracted by column rotation drilling from the surface to ∼1 m, packed into sterile disposable plastic containers in portable cooler with ‘blue ice’ and delivered to the New Jersey laboratory within 40 h. The temperature was recorded continuously with the micrologger STOW AWAY (Onset Computer Corporation, USA). In the lab, frozen cores were subsectioned into 5-cm long fragments and then crushed under aseptic and cold (−20°C) conditions into aggregates of diameter 3–8 mm. The obtained ground material was used immediately for isolation and activity measurements or packed into airtight wide-mouth sterile plastic containers and stored at −40°C before further analysis.

Isolation of microorganisms using liquid media

Frozen soil (layer depth 10–20 cm) or permafrost (layer 50–55 cm) were slowly defrosted on ice and homogenized with a spatula. One gram was mixed in 30-mL glass vials with 10 mL of precooled liquid medium (see composition below); the vials were closed with rubber septa and aluminum seals and incubated at ∼1°C (in an ice bath inside a refrigerator) for a duration of 1–3 months. The mineral base nutrient solution (MB) contained (g L−1): K2HPO4, 2.0; KH2PO4, 1.5; (NH4)2SO4, 1.0; MgSO4, 0.4; CaCl2·2H2O, 0.1. Trace elements (mg L−1): FeCl3·6H2O, 1; KI, 0.2; CoCl2·6H2O, 0.2; MnCl2·4H2O, 0.8; ZnSO4, 0.8; H3BO3, 0.1; Na2MoO4·2H2O, 0.1; CuCl2, 0.1; NiCl2·6H2O, 0.1, and Na-EDTA, 10 as a chelating agent. Ethanol–mineral media (EMM) was made by adding ethanol to the MB at concentration 1.0 g L−1, if not specified otherwise. Other C sources (succinate, citrate) were tested at the same concentration. The chemolithotrophic H2-consuming bacteria were grown on MB under a gas mixture of the following composition (vol%): CO2 : H2 : N2 : O2=5 : 20 : 60 : 15. The growth of the enrichment cultures was monitored by periodic measurements of turbidity. The positive bottles were serially diluted with fresh media and grown for another 1–3 months. Then the last positive dilution was plated on Nutrient Agar (DIFCO) to isolate individual strains. Enrichments and pure cultures were also grown below the freezing point in supercooled liquid media that contained 1–20% of glycerol as antifreeze (Breezee et al., 2004).

Solid-state enrichment without antifreeze

We tested several materials as solid supports for the cultivation of psychroactive microorganisms below the freezing point: microcrystalline cellulose (MCC), polypropylene film, paper strips and membrane filters. The best results were obtained with MCC, which has been selected for further studies and is described below.

One gram of sterile MCC was mixed with 0.5 mL of ice-cold EMM and 1.0 g of crushed permafrost or frozen soil, and the mixture was placed in Hungate tubes and incubated at −1 to −20°C. Growth was recorded as the rate of CO2 production: every week 0.5 mL of headspace air sampled from each tube was analyzed for CO2; gas loss was compensated with equivalent amount of CO2-free air. Positive tubes were identified as those with steady CO2 increase. To make a ‘dry dilution,’ the powder in a positive tube was homogenized with a spatula in a sterile mortar placed on an ice bath and about 200 mg of the powder was transferred to the tube containing 1.0 g of fresh MCC-EMM medium, mixed and incubated at −5 to −7°C until the end of intensive CO2 production (typically 1–3 months). The transfers were repeated four to five times, and then the powder was suspended in ice-cold water and plated on Nutrient Agar (DIFCO) or Malt Extract Agar (DIFCO) at 1°C to isolate pure cultures.

Cultivation of isolates in solid frozen media

The cultivation of isolates (pure cultures and consortia) on MCC-EMM was done in a similar way with the only difference being that the amount of MCC was lowered from 1 g to 200 mg per tube. Occasionally we used plastic films as solid supports for growing cells instead of MCC: 50 μL of microbial cell suspension in EMM was added to a sterile transparent polypropylene bag 5.1 × 15.2 cm2 (2 × 6″, Research Products International Corp.) and evacuated (−27 psi) for uniform distribution of suspension inside the bag into a thin film ∼4 μm deep. The bag was sealed, rolled up and frozen inside the standard Hungate tube (16 × 125 mm). The plastic bags are transparent and autoclavable; they allow gas exchange but retained cells, ice and water. Growth was recorded by CO2 analysis as described above. As a control we used tubes containing sterile medium and inoculum killed by autoclaving (121°C, 30 min).

Identification of isolates

Phenotypic tests to determine the key taxonomical characteristics of bacteria, including Gram's stain, the presence of oxidase and catalase, and spectrum of C-substrate assimilation, were carried out following standard methods (Smibert & Krieg, 1994). The membrane fatty acids were extracted and analyzed by GC using the Microbial Identification System at Microbial ID, Inc (Newark, DE). The morphology of bacteria, yeasts and filamentous fungi was observed with a LEO 982 FEG scanning electron and Leica DMLB UV microscope. A K3 filter was used to visualize DTAF-stained microbial cells. For scanning electron microscopy, microbial cells were fixed with 4% paraformaldehyde and dehydrated by successive passages through 50%, 80% and 100% ethanol and then air-dried.

The prokaryotic 16S rRNA gene was analyzed as described earlier (Sizova et al., 2003) and compared with available 16S rRNA gene sequences using the blast and RDP-II release 9.0 online resources (Cole et al., 2003).

DNA from fungi was extracted by a Qiagen robot and two ribosomal DNA regions were sequenced: the D1/D2 variable domains of the large subunit rRNA gene and the internal transcribed spacers (ITS) 1 and 2 (Scorzetti et al., 2002). DNA extraction, PCR amplification and sequencing were performed at Laragen Inc., CA.

Experiments with 14C-Ethanol

14C-Ethanol (CH314CH2OH, 10 mCi mmol−1, Sigma) was mixed with MB and unlabeled ethanol to bring its concentration to 1.0 g L−1; 200 μL of obtained mixture were added to 2 g of crushed permafrost or 0.2 g of MCC-powder inoculated with Pseudomonas sp. 3–2005 or Arthrobacter sp. 9–2 in standard Hungate tubes. The bacteria were grown first at 0±1°C on EMM for 2 months; cells were collected by centrifugation (6000 g), washed with MB and mixed with MCC aseptically on ice. The control tubes were inoculated with heated cells (autoclave, 121°C, 30 min). All tubes were incubated in an aluminum block with linear temperature gradients from −20 to 0°C. The 14CO2 produced was trapped by filter paper strip soaked in 0.5 M NaOH, which were replaced weekly. To remove 14C-ethanol vapor, the strips were heated at 90°C for 1 h before adding them to the scintillation cocktail. At the end of the experiment, residual 14C-ethanol was recovered by repetitive washing of MCC or soil placed on Millipore membrane filters (0.45 μm) with unlabeled ethanol. The washing was stopped when zero radioactivity was found in the filtrate; all washings were then combined and counted by scintillation counter. The incorporation of 14C into the cells was measured in the washed solids after combustion at 900°C. The exhaust was passed through an alkali trap, and radioactivity in the alkali was measured as 14CO2 (Solid Sample Module of TOC-VE, Shimadzu).

Temperature shift-down experiment

The consortium containing two eukaryotic components (filamentous fungi Geomyces spp. and yeasts Leucosporidium spp.) was grown in 24 replicate (Hungate) tubes at −8°C on MCC-EMM with weekly recording of respiration rates until the start of intensive CO2 production. Each of the 24 tubes was then flushed with CO2-free air and transferred to a different position in the aluminum block with temperature gradients from −8 to −35°C. Three or four replicate tubes were used at each temperature and incubation was continued with weekly measurement of respiration and dark CO2 fixation (DF). Control tubes containing sterile MCC-EMM inoculated with killed cells (121°C, 30 min) were incubated at the same temperatures.

Measurement of cell respiration rate (RR) and dark 14CO2 fixation (DF)

Respiration at all temperatures was measured as the CO2 production rate by the manual withdrawal of small subsamples of headspace air from Hungate tubes (usually 0.5 mL). CO2 was quantified as a spike after the gas sample injection into the IR gas analyzer LiCor 840 was continuously flushed with the gas carrier N2, 60 mL min−1. The flow rate was controlled with an electronic flowmeter (Alikate Technology, USA). The spike height and area were converted to gas concentration vs. standard mixtures (Matheson Gas Products, Inc.) using homemade software profiler-2000. The respiration rate (RR) was calculated by the regression of CO2 headspace content vs. time; correction factors were used to account for CO2 removal and headspace dilution with CO2-free air. Dark CO2 fixation (DF) was measured in cultures grown on EMM in Hungate tubes with MCC. Labeled 14CO2 was added to tubes by syringe to bring air radioactivity to the level 10 000 DPM cc−1, and tubes were incubated 1 week. After that, the actual isotope dilution in each tube was determined by sampling the headspace for total CO2 (see above) and 14CO2 (0.5 mL of air was trapped by alkaline scintillation cocktail and radioactivity counted). Then the excess 14CO2 was removed from the tube headspace by suction and passing through 1 N NaOH while the tubes were kept frozen at their respective below-zero temperature. After that, the content of tubes (MCC+cells) was oven dried for 3 h at 95°C. The 14C incorporated into microbial cells was released as 14CO2 by ignition of the MCC powder at 900°C (Solid Sample Module of TOC-VE, Shimadzu), trapped into 1 N NaOH and counted. An alternative way to assess microbial 14C was the extraction of dried MCC with 5.0 mL of 0.002 M HCl under continuous shaking (150 rpm) for 60 min. One milliliter of the extract was mixed with the scintillation cocktail and the radioactivity was counted. The conversion factor from extracted to total 14C was found to be 0.60±0.05, n=12, i.e. extracted 14C accounted for 60% of the total 14C released by ignition. To calculate DF, the amount of label incorporation, DPM, was multiplied by the isotope dilution factor (mg C DPM−1).

All experiments and preparations were carried out with strict temperature control. The tubes were incubated in either an alcohol bath inside a freezer or in the temperature-gradient aluminum block. The temperature gradient was supported by two FP50-HL (Julabo, USA) thermostats continuously circulating the bath liquid (ethanol or polydimethylsiloxane) through ports at opposite sides of the aluminum block. One tube in the alcohol bath or three tubes per each block (at the coldest, warmest and middle position) contained sensors and logged the temperature results to STOW AWAY micrologger (Onset Computer Corporation, USA). The described setting provided the required temperature stability, ±0.1°C, for all incubations even with frequent access of tubes for sampling.

Results and discussion

Below-zero growth in supercooled liquid and solid media

In preliminary experiments we compared supercooled liquid media containing glycerol as antifreeze and solid media containing MCC for the purpose of microbial cultivation below the freezing point. Enrichments were obtained at ∼0°C on three substrates: ethanol, succinate and citrate. After three transfers, these enrichments were used to inoculate both liquid and solid media with subsequent incubation at several degrees below zero. Microbial growth was monitored as RR. It is based on the widely used assumption that a substrate-sufficient RR is proportional to the total biomass of respiring microorganisms (Panikov & Sizova, 1996; Anderson & Domsch, 1978). Results are shown in Fig. 1. With succinate and citrate, bacterial growth was approximately the same in liquid and MCC media at moderate cooling (temperatures −0.5 and −1.5°C), while at −5°C the supercooled liquid was clearly better (Fig. 1, top). With ethanol (Fig. 1, bottom), the results were reversed: MCC media supported intensive linear growth during the entire 30-d experiment while glycerol suppressed growth already at −0.5°C. Further cooling to −3 and −5°C slowed down growth on ethanol at least twice as much compared with MCC. Thus, although each cultivation method had its own substrate preference, the solid-state technique turned out to be less restrictive.

Figure 1.

 Comparison of supercooled liquid medium with glycerol (a, c) and solid MCC powder (b, d) for microbial cultivation below the freezing point on succinate (a, b) and ethanol (c, d). Note the difference in scale for (c). Each data point is the average amount of produced CO2 for two to three replicates; the vertical error bars indicate standard deviations. The control is CO2 formation in tubes containing killed cells.

Another significant advantage of solid MCC media was its independence from the temperature range while glycerol remains efficient at best above −9°C; further cooling produced frequent failure as some tubes from the same batch stayed liquid while others underwent freezing. Thus, solid-state cultivation is more stable, predictable and reproducible. Besides, frozen cellulose more closely imitates the natural microenvironment in permafrost or top soils. Therefore, in the remaining part of the experimental work we relied completely on solid-state enrichment and cultivation.

Figure 1 also demonstrates that control samples containing full sterile media+killed cells produced a negligible amount of CO2. Similar results were obtained in other experiments; for clarity we skip plotting the control data on other graphs.

Another solid-state technique, with plastic bags, was found to be acceptable only for cultivation but not for primary enrichment (Fig. 2). The reproducibility of this technique turned out to be worse than EMM-MCC, as indicated by greater data scattering when growth rates were plotted vs. temperature (Fig. 2c). On the other hand, transparent plastic gives a potential opportunity for noninvasive microscopic examinations and nondestructive microbial growth detection, e.g. using fiber-optic sensors.

Figure 2.

 Dynamics of bacterial growth in frozen solid media. A suspension of Arthrobacter sp. 9-2 (0.05 mL) was frozen inside a 2 × 6″ plastic bag with EMM, rolled up and placed in a Hungate tube. (a) and (b): Growth dynamics recorded as RR; numbers indicate incubation temperature. Each data point is average for two to three replicates; the vertical error bars indicate standard deviations. (c) Plot of the initial RR rate vs. incubation temperature; each data point was calculated using LINEST Excel function for the initial linear part of the RR dynamics, the error bar is standard error.

Isolation of psychroactive microorganisms

Table 2 shows the full list of isolates obtained in this study, both with conventional liquid and new solid-state enrichment techniques. Liquid enrichment resulted in the isolation of mainly bacteria with the lowest growth temperature varying from −1°C (Polaromonas sp. strain hydrogenovorans, bacterium utilizing molecular hydrogen) to −17°C (Pseudomonas sp. 3–2005 and Arthrobacter sp. 9–2). They were closely related to bacteria isolated from sea ice and various Antarctic habitats (Irgens et al., 1996; Junge et al., 1998; Staley & Gosink, 1999; Christner, 2002; Van Trappen et al., 2002; Christner et al., 2003a b; Geoffroy & Meyer, 2004; Shivaji et al., 2004). Application of solid-state enrichment led to the preferential isolation of eukaryotic organisms: basidiomycetous dimorphic yeasts Mrakia sp. MS-2 and Leucosporidium spp. MS-1, MS-3 and ascomycetous mycelial fungi Geomyces spp. FMCC-1, FMCC-2, FMCC-3, and FMCC-4. The low-temperature limit for their growth varied from −12°C (Mrakia sp.) to below −35°C for other isolates (see results below). Contrary to prokaryotes, yeasts and filamentous fungi grew in frozen media more quickly and always displayed exponential growth rather than linear or decelerating patterns typical for bacterial cultures. The upper temperature limit for growth was 18–20°C for fungi and 25°C for bacteria and therefore these two groups of isolates should be formally classified, respectively, as ‘psychrophiles’ and ‘psychrotolerants’ (Morita, 1975). A significant deviation from conventional definition was that low-temperature limit for our ‘psychrotolerant’ isolates went down to −17°C, which was normally attributed to specialized, strictly psychrophilic species.

Table 2.   Psychroactive microorganisms isolated in the present study
Organism, strainEnrichment conditionsGrowth temp
Closely related phylotypes (blast)Genbank Accession
Pseudomonas sp. 3-2005Liquid medium with ethanol−1725Antarctic bacterium R-9113 isolated from lake mat (96%)DQ 094182
Arthrobacter sp. 9-2Liquid medium with ethanol−1725Arthrobacter sp. An 16 isolated from deep sea sediment (98%)DQ 094184
Polaromonas sp. strain “hydrogenovorans”Liquid mineral medium under H2:CO2, 0°C−125Polaromonas naphtalenivorans (99%)DQ 094183
Leucosporidium spp. MS-1, MS-3Ethanol-MCC, −5°C to −8°C−3520Cryptococcus sp. Ytty94 Y24 (99%), Leucosporidium scottii isolate (97%)DQ 295018, DQ499475
Mrakia sp. MS-2Ethanol-MCC, −5°C−1218Mrakia sp. and M. frigida, isolated from various Antarctic habitats (100 %)DQ 295019
Geomyces spp. FMCC-1, FMCC-2, FMCC-3, FMCC-4Ethanol-MCC, −8°C−3518Geomyces sp. GS4N1a (99%); endophytic fungus in white pine (99%); Geomyces pannorum VKM FW-857, VKM FW-969, VKM FW-2236, from cryopegs (98%); Aleurodiscusfarlowii Burt, decomposing fungi in boreal zone of North America (100%); uncultured fungal clones SPR99B09, SPR99C09, WIN99B11 retrieved from winter tundra (99%)DQ499471-74 (ITS region) DQ520619-22 (LSU rRNA)

Growth dynamics of fungal consortium in frozen media at −8°C

Figure 3 shows the long-term growth dynamics of microorganisms in 29 individual tubes with EMM-MCC. As inoculum we used consortium that contained filamentous ascomycetous fungi Geomyces spp. and dimorphic basidiomycetous yeasts Leucosporidium spp. The consortium developed as a result of four consecutive transfers of enrichment primarily obtained from permafrost on EMM-MCC at −5 to −7°C. Plating on Nutrient Agar and direct microscopy of consortium with DTAF revealed only two fungi, mentioned above, and a full absence of prokaryotic species.

Figure 3.

 Growth dynamics of eukaryotic consortium in frozen ethanol-MCC powder. Eukaryotic consortium (Geomyces spp.–Leucosporidium spp.) was used as inoculum of 29 tubes containing EMM-MCC. Growth was followed at −8°C as CO2 production rate in each tube calculated with LINEST (the vertical error bar shows average standard error returned by LINEST function). Note that six out of 29 tubes displayed higher growth rates (1) then other slow growers (2). Heavy solid curves are the best-fitting equation (1) for respectively fast and slow growing populations. The inset 3 shows growth rate frequency distribution constructed with FREQUENCY Excel function.

The microbial growth recorded as RR (Fig. 3) varied in a wide range. It was much faster in six tubes where RR peaked at the end of the first month and then declined due to exhaustion of C-source (declined parts were omitted to keep the graph simpler). In the remaining tubes RR gradually declined over the first 20 days and then displayed exponential growth superimposed on regular RR oscillations with progressively increasing periods of from 22 to 34 days. Note that the oscillations were not induced by changes in growth conditions; the temperature remained perfectly stable starting from the second day of this incubation experiment.

The initial decline in RR could be part of the oscillation cycle or a consequence of the short-term wasteful respiration (so called ‘overflow metabolism’) normally observed in any batch culture started using starving cells as inoculum (Tempest & Neijssel, 1984; Panikov, 1995). RR oscillations are often observed in the yeasts (e.g. Saccharomyces cerevisiae, Schizosaccharomyces pombe) chemostat cultures (Richard, 2003), especially if grown on ethanol (Keulers et al., 1996). It was speculated that collective synchronization is orchestrated by acetaldehyde, which is a first product of ethanol oxidation, H2S, an intermediate of intracellular S-metabolism (Lloyd & Murray, 2005), or the release of NH3 (Palkova et al., 2002); all three metabolites are gases and could play a signal role in porous frozen media. The period of oscillations observed under optimal growth conditions (30°C) could be independent of reproduction frequency or match exactly the duration of the full cell cycle (Tu et al., 2005). It is tempting to think that RR oscillations with periods of 22–34 days observed in our experiment at −8°C reflect the reproduction frequency of our consortium, doubling their cell mass every 22–34 days.

To test this assumption, we derived a simplified kinetic equation which relates cell growth to the observed RR dynamics including its wasteful and productive components [see the full model in Panikov (1995)]:


where p is CO2 concentration, t is time, μ is specific growth rate, and Yx/p and A are stoichiometric constants. Total cell mass (xtot) is represented by an ‘active’ part (x) contributing to growth and the rest (xtot−x) which is represented by cell constituents expressed only under starvation conditions including alternative oxidase participating in wasteful respiration. The terms x0 and x0tot are initial values of active and total cell mass, respectively.

The integration of (1) gives an expression explicitly describing RR dynamics and contains desired value of specific growth rate μ:


The best fit of equation (2) to the experimental RR data produces a set of μ-values, the average specific growth rates in each of the 29 tubes. The frequency distribution (Fig. 3, inset) was strongly skewed: the highest number of consortia were slow growers with μ-values ranging from 0 to 0.05 day−1 with the mode 0.03 day−1, a few isolates grew with the specific rates 0.08–0.15 day−1. Respective doubling time periods (td), found as td=(ln 2)/μ, varied between 4.6–8.8 days for the fast-growing isolates and 14–35 days for the slow-growing consortia with median 22 days. The observed oscillations in RR could therefore be attributed to synchronized cell division events in organisms growing in the frozen medium.

Growth kinetics of various organisms: yeasts vs. filamentous fungi and fungi vs. bacteria

What organisms were responsible for differences between fast- and slow-growing consortia? Direct microscopy of MCC powder stained with DTAF revealed significantly higher yeast populations in fast consortia. For example, two randomly selected ‘fast’ and ‘slow’ tubes contained, respectively, (7.0±0.15) × 107 and (1.9±0.68) × 107 yeast cells per g MCC overlaid by sparse and highly irregular fungal mycelium. After the consortium is split into six axenic cultures, two strains representing the genus Leucosporidium and four isolates of the genus Geomyces, we have found that yeast display a growth rate that is several times higher not only in frozen EMM-MCC media but also in liquid EMM and malt agar at +1 and +15°C.

A comparison of Figs 2 and 3 gives us a chance to discuss the difference in growth kinetics between fungi (yeasts and mycelial) and bacteria. None of the tested bacteria displayed exponential RR dynamics; instead it was either linear or progressively declined curvilinear dynamics at all tested temperatures. Some isolates displayed oscillations in RR, but no true synchrony characteristics for fungi and yeasts was observed.

What kinetic mechanisms are responsible for linear and exponential growth patterns? The linearity of growth is actually evidence of its continuous decline: after each cell division two daughter cells do the same biosynthetic job as did a single mother cell of the previous generation. In exponentially growing populations, biosynthetic activity per cell remains constant and the job done by two daughter cells is doubled as compared with the single mother cell. From this reasoning, it follows that linear growth of bacteria in frozen MCC has to be restricted by some factors other than availability of C-substrate because ethanol is volatile at −8°C and therefore spread uniformly by diffusion throughout the entire frozen MCC-space. It is probable that bacteria suffer from a deficiency of liquid water or mineral nutrients packed into frozen liquid phase.

Contrary to bacteria, mycelial fungi and even unicellular dimorphic yeasts forming pseudomycelium are better colonizers of the solid media. The linear extension rate of hyphae is much higher than linear growth rate of bacterial cells; instead of cell division they branch and form a mycelial network the density of which depends on the availability of nutrient resources: in poor microloci branch initiation is rare while in favorable microhabitats frequent branching produces dense networks resulting in the efficient utilization of deficient resources. There should be a specific advantage of a mycelial growth strategy in frozen environment. Probably, mycelial network contains continuous intracellular channels of unfrozen liquid along hyphae facilitating delivery of metabolites to extension zones. Fungi are also known as stronger cellulolytic organisms, however in our experiments we observed zero RR in the control tubes with MCC without added ethanol. The immunity of MCC against cellulolytic organisms in the primary enrichment and in consortium could be explained by repression of cellulase biosynthesis by ethanol. Besides, any extracellular enzymatic activity is expected to be inhibited below the freezing point due to the low mobility of enzyme molecules.

Stoichiometry of 14C-ethanol utilization in the temperature range from 0 to −20°C

This experiment was designed to answer the question: what fraction of consumed substrate is converted into cell material? The weekly rates of 14CO2 production were summed up for an entire 60-day period and at the end of the experiment residual 14C-ethanol and total 14C incorporation into microbial cells were measured. The recorded data close the 14C-growth budget according to the equation:


The growth yield, Y, is defined as the amount of cell mass (Δx) produced per unit of substrate consumed (Δs). In our study, label recovery at the end of experiment (14C in residual ethanol+CO2+cells) was only 70–90% due to 10–30% evaporation loss of ethanol from the alkaline paper strips used to capture 14CO2 (see Materials and methods); therefore in the Y calculation Δs was substituted by Δx+Δp from equation (3):


Results are shown in Fig. 4 as a plot of oxidation and incorporation vs. temperature. With temperature decrease, the amounts of oxidized and incorporated 14C declined to roughly the same degree, while the yield factor Y remained relatively constant in bacteria or slightly declined in the soil community. The average yield calculated from equation (4) was 0.28±0.12 g cell C (g C ethanol)−1 for Arthrobacter sp., 0.27±0.08 g C (g C)−1 for Pseudomonas sp. and 0.52±0.11 g C (g C)−1 for the community. This growth efficiency remains close to what has been reported for unfrozen laboratory media or growth stoichiometry of soil microbial communities in situ (Mayberry et al., 1967; Payne, 1970). For instance, Pseudomonas fluorescens and Arthrobacter globiformis had yields of 0.46 and 0.58 g C (g C)−1, respectively, under optimal growth conditions, which dropped to 0.16 and 0.33 g C (g C)−1 under severe substrate limitation in dialysis culture (Panikov, 1995). The two-fold higher yield for soil community indicates that majority indigenous populations are better adapted to functioning below the freezing point than isolated bacteria.

Figure 4.

 Growth of Arthrobacter sp. 9-2 (a), Pseudomonas sp. 3-2005 (b) and soil microbial community (c), soil 5–10 cm) on 14C-ethanol. Legend: ΔR stands for respiration, the cumulative amount of 14CO2 resulted from oxidation of 14C-ethanol, ΔG is cell growth measured as 14C incorporation into solids. Growth yield is calculated from 14C mass balance, equation (4). Each ΔR and ΔG data point was calculated as average of two replicates, the vertical error bar is standard deviation. The standard error for Y-value was calculated from respective individual SD values for ΔR and ΔG using quotient differentiation rule.

Published data on growth yield of bacteria grown below zero are scarce. In supercooled liquid media (Bakermans & Nealson, 2004), the growth yield of Psychrobacter cryopegella [renamed to P. cryohalolentis (Bakermans et al., 2006)] at −4 and −10°C was 0.26 and 0.08 g C (g C)−1 which is close to our results. The main discrepancy was that the yield of P. cryohalolentis was found to decline with cooling while in our studies it was constant even at wider temperature range. The indicated discrepancy could originate from differences in the cultivation approach (liquid vs. solid frozen media) and methodology used to measure the yield (label incorporation vs. dry weighting).

Evidence against postulated ‘maintenance state’

It was hypothesized (Bakermans et al., 2003; Price & Sowers, 2004) that in natural frozen habitats such as permafrost and glaciers, nutrient flux is so low that it could support only the ‘maintenance state’ of microbial cells, which stands for slow catabolic activity without growth. Earlier the ‘maintenance state’ was proposed in a purely deductive way for any terrestrial or oceanic oligotrophic habitat (Morita, 1997). In microbial physiology the maintenance concept has been introduced mathematically (Pirt, 1975) through the balance of energy source utilization:


If we divide (5) by xΔt and switch to infinitesimally small increments, then:


where q is specific rate of energy source consumption [(1/x)ds/dt], μ is specific growth rate, m is specific maintenance rate (i.e. the nonrelated to growth consumption of energy source per unit time and per unit cell mass) and Ym is the yield under hypothetical conditions of zero maintenance requirements.

In practice, a maintenance coefficient is found in chemostat culture run at several dilution rates, which gives several μ values and corresponding rates of energy source consumption (q) which could be rates of O2 uptake, CO2 production, heat generation, etc. Then a plot of q vs. μ is fitted to linear equation (6) containing m as intercept. What is important is that the maintenance coefficient m is a regression parameter which remains meaningful only within a given set of q (μ) data. Chemostat culture usually covers the μ range 0.1–0.9μm, so the slower growth could not obey equation (6), and we cannot claim that non-growing cells (μ=0) consume energy at the rate m. Several attempts to stop microbial growth by delivering the energy source at the rate exactly matched to m failed: some microorganisms started to lyse (μ<0) while others continued slow growth (Panikov et al., 1982; Dorofeev et al., 1984). The reason is that under chronic starvation caused by a low supply of the energy source, the assumption that maintenance is constant is no longer valid. The adaptive decline in maintenance requirements takes place due to changes in macromolecular composition of starving cells, loss of motility, reduction in turnover and osmoregulation work (Panikov, 1995).

Thus, the ‘state of maintenance’ is unlikely for any natural habitats, and permafrost should be no exception. Whatsoever small energy flux is available for active microbial cells, it will be distributed between the maintenance and growth functions in such a way that growth will not stop completely. Undoubtedly, this statement is applicable only to true microbial growth, which is often hidden by any elimination process, e.g. grazing, lysis or migration:


What about frozen habitats? Maybe there are some specific factors that prevent microbial replication without stopping their catabolic activity? For instance, active microbial cells could be encaged in a rigid icy ‘corset,’ which would restrict any cell volume enlargement. However, our experiments with 14C-ethanol (Fig. 4) clearly demonstrate that it is not the case. Cooling decreases the respiration and growth rate at roughly the same degree, their ratio remaining more or less constant.

Steady-state and transient growth dynamics after temperature shift-down from −8 to −35°C

The purpose of this experiment was to record the transient dynamics of activity and growth of psychroactive microorganisms after sudden decrease of incubation temperature. Initially, the eukaryotic consortium was grown at −8°C and then subjected to cooling of various degrees from −10 to −35°C. The growth dynamic recorded as RR was biphasic (Fig. 5): during the first 2 weeks (the early phase of the transient process) the consortium sustained measurable RR even at the lowest tested temperature −35°C, and then (late phase) the process smoothly declined to zero in the temperature range −35 to −20°C or to some nonzero constant rates under warmer conditions. The rates of 14CO2 uptake (DF) and respiration were plotted separately for early and late phases (Fig. 6). The late-phase data form an almost perfect exponential curve (linear graph in semi-log scale) with apparent Q10=270, while the early phase data points deviated from this exponential curve starting from −10°C, reached a minimal level at −20°C and then remained constant down to the lowest tested temperature −35°C.

Figure 5.

 Transient reaction of eukaryotic consortia grown on EMM-MCC to the temperature shift-down. The CO2 accumulation dynamics is shown in individual tubes shifted at time zero from −8°C to various temperatures indicated by numbers. Each data point was calculated as average of two to four replicates. The vertical error bar is the standard deviation.

Figure 6.

 Plot of DF and RR vs. temperature separately for the early and late phases of the transient process shown in Fig. 5. Each data point was calculated as slope of linear parts of CO2 dynamics (Fig. 5) using LINEST. The vertical bars stands for standard error. Exponential equation describes effects of temperature on the stabilized (late) metabolic rate.

The temperature threshold located between −18 and −20°C separates two domains in physiological state of studied organisms. Above this threshold, i.e. at moderate cooling, the psychroactive microorganisms preserve the ‘normal’ physiological state: they grow exponentially, consume ethanol, oxidize about half of it to CO2 and incorporate the rest into expanding cellular material, eventually leading to cell division/budding or hyphae extension and branching, the only difference from above-zero conditions being severe rate reduction. As shown above, the doubling time of the slow growers at −8°C was 14–35 days. Using an exponential rate-temperature relationship for late response (Fig. 6), we calculate that doubling time increases with further cooling, reaching 20 years at −18°C and 62 years (!) at −20°C. Note that our calculation refers to laboratory experiments with frozen solid media with a plentiful supply of ethanol, one of the optimal growth substrates in frozen media. The in situ growth rate could be even lower because of quite probable substrate limitation; however some yet unknown indigenous organisms may be better adapted to life in permafrost than our isolates and thus grow faster.

Below a threshold of −20°C, the cold stress is much deeper and the physiological state of microbial cells is different: cells are not able to sustain metabolic vigor continuously and after 2 weeks enter the dormant state with zero metabolic activity. One observation we found surprising and somewhat counterintuitive: the early reaction to cooling below −20°C was the uniform activation of cells independent of ambient temperature; the same metabolic activity was detected in the range from −20 to −35°C. Another remarkable feature of the early phase below −20°C was that the transient activation affected both catabolic reactions (RR) and anabolic biosynthetic processes (DF). Just as a reminder, DF refers to the uptake of CO2 by chemoorganotrophic organisms in so-called anaplerotic reactions which should, in theory, be close to total cell synthesis (=growth) for organisms synthesizing all their proteins from inorganic N and simple C-compounds such as sugar or ethanol. The empirically found ratio between DF and cell-C growth varies from 1% to 10% with an average of about 6% (Hesselsoe et al., 2005; Santruchkova et al., 2005).

We can hypothesize that a probable metabolic strategy of the mentioned short-term activation is the synthesis of intracellular constituents responsible for the long-term survival of a frozen population under deep freezing when a steady metabolic function is prohibited. Contrary to the hypothetical ‘maintenance state,’ which assumes zero growth and nonzero wasteful respiration, the actual physiological state of microorganisms in deeply frozen media is characterized by tight coupling between respiration and biosynthesis; their rates are similar during the early phase and synchronously decline to zero later, when cells enter a state of dormancy.


This research was supported by the NSF grant MCB-0348681. We thank Dr V. Romanovsky for permafrost sampling. Drs J. Fell, J.P. Sampaio, N. Ivanushkina and S.M. Ozerskaya provided valuable assistance in preliminary identification of isolated fungi and yeasts; the ongoing collaborative studies with J.P.S. and S.M.O. are aimed at full identification and description of these organisms. We thank Thomas Cattabiani for reading and correcting the manuscript.