Microorganisms in the Accreted Ice of Lake Vostok, Antarctica

See allHide authors and affiliations

Science  10 Dec 1999:
Vol. 286, Issue 5447, pp. 2144-2147
DOI: 10.1126/science.286.5447.2144


Analysis of a portion of Vostok ice core number 5G, which is thought to contain frozen water derived from Lake Vostok, Antarctica (a body of liquid water located beneath about 4 kilometers of glacial ice), revealed between 2 × 102 and 3 × 102 bacterial cells per milliliter and low concentrations of potential growth nutrients. Lipopolysaccharide (a Gram-negative bacterial cell biomarker) was also detected at concentrations consistent with the cell enumeration data, which suggests a predominance of Gram-negative bacteria. At least a portion of the microbial assemblage was viable, as determined by the respiration of carbon-14–labeled acetate and glucose substrates during incubations at 3°C and 1 atmosphere. These accreted ice data suggest that Lake Vostok may contain viable microorganisms.

The existence of subglacial lakes in East Antarctica has been known for nearly three decades, but only recently have their large numbers and dimensions been revealed (1). Lake Vostok, one of nearly 80 subglacial lakes that have been discovered and mapped by means of airborne 60-MHz radio-echo sounding (2), is the largest (∼14,000 km2surface area and ∼1800 km3 volume) and deepest (up to 670 m) of these unusual subglacial environments. The fresh water in Lake Vostok is kept liquid by the pressure of the ice overburden (equivalent to ∼350 atm) and, perhaps, by geothermal heating. This lake and others like it may contain previously undescribed relic populations of microorganisms that are adapted for life in these presumably oligotrophic (low-nutrient, low-biomass, and low–energy flux) habitats.

In 1998, a team of Russian, U.S., and French scientists completed the drilling of Vostok hole number 5G (72°28′S, 106°48′E). At a termination depth of 3623 m, this is the deepest ice core ever obtained. The bottom of the core is ∼120 m from the ice–Lake Vostok water interface. The upper 3300 m of Vostok ice core 5G provides a continuous record of Earth's paleoclimate over the past 400,000 years, including four complete glacial-interglacial periods (3). Ice samples extracted from core depths of 1500 to 2750 m (with corresponding ages ranging from 110,000 to 240,000 years) have shown (i) the presence of a diverse assemblage of prokaryotic and eukaryotic microorganisms (0.8 × 103to 11 × 103 cells per milliliter of ice melt), (ii) a positive correlation between the presence of dust and the number of microorganisms, and (iii) the presence of viable mesophilic microorganisms as revealed by the consumption of14C-labeled organic substrates (4).

At greater depths in Vostok ice core 5G, between 3311 and 3538 m, the layers are disturbed by ice sheet dynamics; and beneath 3538 m, changes in the crystal structure, electrical conductivity, and stable isotope and gas composition of the ice suggest that the basal ice at this location (3538 to 3743 m) is refrozen Lake Vostok water (3, 5). Because this lake is so remote and is largely inaccessible, the accreted ice provides the most reliable surrogate sample of the Lake Vostok ecosystem before the actual penetration of the ice-lake boundary and the collection of water samples.

A sample of the accreted Lake Vostok ice was analyzed for (i) microbial cell enumeration by epifluorescence microscopy, scanning electron microscopy (SEM), and dual laser flow cytometry (Figs. 1 and2); (ii) microbial biomass estimation with two independent biomarker compounds (Table 1): adenosine-5′-triphosphate (ATP) and lipopolysaccharide (LPS); (iii) microbial cell viability and potential metabolic activity by analysis of rates of 14C-CO2 production and14C-incorporation into macromolecules after timed incubations with exogenous 14C-labeled organic substrates (Tables 1 and 2); and (iv) the presence of potential carbon and nitrogen growth substrates (6–9). Our measurements from ice collected at 3603 m complement the independent ice core analyses of a sample from 3590 m (10). A major difference is that our core contained no sediment inclusions. Therefore, the results presented here may not be directly comparable to those of Priscu et al.(10), despite the fact that both samples were obtained from the accreted ice of Lake Vostok.

Figure 1

Microscopic analyses of melt samples from the accreted Vostok ice. (A) Epifluorescence. SYBR Green I– stained image of a small coccoid-shaped bacterium (right, bright yellow-green cell) and associated red-fluorescing nonliving particulate matter. (B) Epifluorescence. SYBR Green I–stained image of a rod-shaped bacterium. (C) Field emission SEM image of a coccoid bacterium on a 0.02-μm Anopore filter viewed at a magnification of ×150,000 in the decontaminated ice melt. (D) SEM image of a rod-shaped bacterium as in (B).

Figure 2

Flow cytometric particulate matter analysis of a Hoechst 33342–stained top melt sample from the accreted Vostok ice. (A) A plot of log red fluorescence (relative units) versus log blue fluorescence (relative units) along with reference beads of known size and fluorescence intensity. Green indicates the population of particles with low blue (low DNA) and variable size [see (B)]; they were not counted as bacteria. (B) A plot of log blue fluorescence, same scale as in (A), versus log right angle side scatter (relative units) along with the same reference beads. We identified at least two particle categories: “heterotrophic bacteria” and high blue– high red–fluorescing particles. The heterotrophic bacteria designation is based on the known properties of marine bacteria with cell dimensions similar to those of the small coccoid-shaped bacteria in the ice core melts (Fig. 1). On the basis of these criteria, we estimate the bacterial abundances in the top melt and bottom melt samples to be 498 and 712 cells per milliliter, respectively. These values conform to those determined with direct microscopy and to cell carbon estimates derived from LPS concentrations (Tables 1 and 3).

Table 1

Selected chemical, biochemical, and microbiological measurements of decontaminated meltwater samples from the Vostok accreted ice at 3603 m.

View this table:
Table 2

Respiration and net incorporation of14C-labeled acetate and glucose by microorganisms in decontaminated meltwater samples from the Vostok accreted ice at 3603 m. 14C activity is expressed as DPM of the specific14C-labeled organic substrate that was respired (CO2) or incorporated into cellular biomolecules [nucleic acid (NA) and protein] per 2-ml sample of ice melt per incubation time, as indicated. All samples were corrected for time zero14C activities in the respective fractions: 190 (±16) DPM for 14CO2 in glucose and 11,611 (±1556) DPM for 14CO2 in acetate. No significant (above instrument background) amount of 14C was detected in either the NA or protein fractions, so time zero procedural blanks were used for error propagation estimation: 25.9 (±1.9; n = 8) DPM of14C for NA and 25.8 (±3.8; n = 8) DPM of14C for protein. Values shown are mean net (experimental minus controls, or blanks) determinations ±1 SD (n = 3) for the 2- and 11-day incubations; all others are single net measurements. All incubations were conducted at 3°C in the dark, except as noted.*

View this table:

Epifluorescence microscopic examination of decontaminated, melted ice samples revealed numerous inorganic particles, many of which fluoresced under ultraviolet (UV) illumination (Fig. 1). The presence of these particles complicates precise enumeration of putative microbial cells; however, microbial cells (presumably bacteria) were readily and unequivocally detected (Fig. 1, A and B). There was a spectrum of cell sizes and morphologies, ranging from the abundant small (0.1 to 0.4 μm) coccoid cells that represented about half (43 ± 6%) of the community to a diverse mixture of thin rods and vibrios (0.5 to 3 μm) that made up the remainder (Fig. 1). Enumeration revealed a relatively low abundance of 2 × 102 to 3 × 102 cells per milliliter of melted ice, which extrapolates to ∼3 ng of C per liter (Table 1). These biomass estimates are at least an order of magnitude lower than estimates of total prokaryotic cells present in low-nutrient, deep ocean environments (Table 3).

Table 3

Cross-ecosystem comparisons of microbial biomass for a variety of natural aquatic habitats. Habitat data are from a variety of sources (22). DMC, direct microscopic count.

View this table:

Dual laser–based flow cytometric analyses of the ice melts confirmed the two main conclusions of the direct microscopic observations: (i) the presence of a broad size spectrum of particles with variable blue fluorescence, including DNA-containing microorganisms and (ii) the presence of fluorescent (especially red) inorganic particles (Fig. 2, A and B). Prokaryotic cell enumeration by the Hoechst 33342 staining method yielded estimates of 5 × 102 to 7 × 102 cells per milliliter of melt, a value that is about two to three times greater than the estimates based on microscopic observations. We believe that these higher estimates from the flow cytometric analyses are due to the inability of the method to discriminate between stained cells and stained nonliving particles of similar size (Fig. 2). In any case, even these estimates from flow cytometric analyses are low compared to biomass estimates from other oligotrophic aquatic ecosystems (Table 3).

Total LPS concentration was used as an independent estimate of the presence of bacterial cells. This cell wall biomarker is known to correlate with Gram-negative bacterial cell mass (11–13). From these LPS determinations, we estimate the Gram-negative bacterial biomass to be 0.5 to 1.6 ng of C per liter of ice melt (Table 1). This is about an order of magnitude lower than LPS concentrations reported from deep ocean environments (Table 3) and is consistent with the relatively low microbial biomass estimated from direct microscopy (Table 1). The lower LPS-extrapolated biomass, relative to direct microscopy and flow cytometry, might be expected if the microbial assemblages were not exclusively Gram-negative bacteria.

Attempts to measure ATP, an independent biomarker for microbial biomass (13, 14), were negative. This was in part due to the relatively high detection limit (≥0.5 pg of ATP ml−1 of melt, using a 50-ml sample) resulting from limitations in total sample volume. If total microbial biomass was ∼1 to 3 ng of C per liter, as the direct count and LPS data suggest, then ATP in these samples would have been ∼0.01 pg ml−1, or undetectable by our protocols. Larger sample volumes, however, should provide an unequivocal quantification of viable cells by ATP detection.

Incubation of the ice melts with 14C-labeled acetate and glucose documented the production of14C-CO2, indicating the presence of metabolically active cells (Table 2). Acetate was respired 800 times more rapidly than was glucose (Tables 1 and 2). These respiration rates correspond to turnover times of approximately 2 and 2000 years, respectively, for the added organic substrates. Compared to14C-CO2 production, rates of 14C incorporation into macromolecules (such as nucleic acid and protein) were even lower (Table 2), if detectable at all. For14C-labeled acetate, the 14C incorporated into macromolecules (the sum of that incorporated into nucleic acid and protein) was only 0.1 to 0.2% of the 14C respired; for14C-glucose, the relative incorporation was higher (10 to 40%), even though the total metabolism was lower (Table 1). Transfer of subsamples to 23°C after an initial 11-day incubation at 3°C stimulated 14C-CO2 production and14C-incorporation into macromolecules (Table 2). It should be emphasized, however, that these rates of organic matter mineralization are potential rates; in situ rates under ambient conditions (350 atm and subzero temperatures) may be much lower. Furthermore, the presence of liquid water would be required for cellular metabolism.

Inorganic and organic nutrient supply is the key to survival in all Earth habitats. The detection of relatively low total organic carbon (∼7.5 μM; Table 1) in the accreted ice, a value that is about five times lower than in deep ocean habitats (15), suggests that Lake Vostok is oligotrophic. In addition, we detected both oxidized and reduced nitrogen in the melt samples (Table 1). A reduction of nitrate plus nitrite in the accreted ice relative to concentrations in meteoric ice (16) might indicate net denitrification. The redox state of the lake is unknown, and some geochemical models predict that most free oxygen might be sequestered in gas hydrates (17).

To survive in a liquid habitat for extended periods of time, microorganisms must have an exploitable energy source. If the energy for contemporary microbial populations in Lake Vostok is supplied from above, this lake may be one of the most oligotrophic habitats on Earth. Basal melting of the ice core and gravitational transport would deliver deposited materials and ice-rafted debris, including microorganisms and potential growth substrates. If basin overflow connects these subglacial habitats to the sea (Lake Vostok is presently below sea level), this transport pathway could represent another allochthonous source of reduced carbon and energy. Alternatively, metabolic processes in the lake may be fueled by geothermal energy, analogous to microbial life discovered at deep sea hydrothermal vents (18). Lake Vostok may be associated with an intracontinental rift zone similar to that of East Africa, so this latter process remains a possibility (19). Finally, the subglacial lakes of East Antarctica may be among the most isolated ecosystems on Earth and could serve as terrestrial analogs to guide the design of samplers and experiments to be used in life probe missions to the ice-covered ocean of the jovian moon Europa (20).


View Abstract

Stay Connected to Science

Navigate This Article