The McMurdo Dry Valleys of East Antarctica (Figure 1A) comprise a windswept polar desert that receives minimal annual precipitation (3–50 mm water equivalent per year; Fountain et al., 2009) and consists of nutrient-poor, rocky–sandy soils. The valleys contain several permanently ice-covered lakes that contain a broad diversity of microorganisms adapted to harsh physicochemical conditions, including constant freezing temperatures, nutrient limitation, and low incident light with long periods of almost total darkness during the austral winter. Lake Fryxell, located in Taylor Valley, has a water column of 18 m covered by 5–6 m of ice that prevents wind mixing and limits light penetration (Figure 1B). Beneath the ice cover, Lake Fryxell waters are fresh and supersaturated with trapped gasses, especially O₂ from phytoplankton photosynthesis (Vincent, 1981; Priscu et al., 1987; Figure 1C). Deeper strata are brackish and anoxic and experience the upward diffusion of dissolved sulfide produced by sulfate-reducing bacteria (Karr et al., 2005; Sattley and Madigan, 2010; Figure 1C). Dissolved O₂ and benthic sulfide coexist as opposing gradients to produce a biologically active chemocline from 9–10.5 m, wherein numbers of sulfur-oxidizing bacteria peak (Sattley and Madigan, 2006).

Several studies have characterized novel bacterial isolates from Lake Fryxell microbial mats (Van Trappen et al., 2004a,b) and sediments (Sattley and Madigan, 2007, 2010), as well as the water column itself (Jung et al., 2004; Sattley and Madigan, 2006; Sattley et al., 2008). Other studies have employed culture-independent methods to examine these habitats (Brambilla et al., 2001; Karr et al., 2003, 2005, 2006; Kwon et al., 2017). Taken together, these investigations have revealed that extensive microbial diversity exists in Lake Fryxell and that only a handful of species have been isolated and characterized. Of the Lake Fryxell water column isolates, bacteria that grow phototrophically or drive microbial sulfur cycling have been the best characterized, including anoxygenic phototrophs, sulfur-oxidizing bacteria, and sulfate-reducing bacteria (Karr et al., 2003, 2005; Sattley and Madigan, 2006, 2010). By contrast, very few enrichments targeting aerobic, heterotrophic bacteria from the Lake Fryxell water column have been performed.

Although Antarctic lakes exhibit a surprising degree of microbial diversity, we have relatively few isolates from these ecosystems compared to temperate aquatic environments and know little of their genetic or biochemical potential. We have thus employed both defined and complex culture media and specific incubation conditions to enrich for cold-adapted, chemoorganotrophic bacteria from different depths of the Lake Fryxell water column. These enrichments yielded pure cultures of eight diverse strains of bacteria, and here we describe the morphology, physiology, and phylogeny of these isolates with a focus on Pseudomonas strain LFY10, a potentially useful strain for teaching laboratories in microbiology.

Phylogenetic analyses showed the Lake Fryxell strains to lie within the families Sphingomonadaceae, Micrococcaceae, Microbacteriaceae, Caulobacteraceae, and Pseudomonadaceae. These families contain predominantly terrestrial and aquatic bacteria, along with outliers of clinical significance. Of note was the isolation of two strains of Arthrobacter (strains FY10 and FYS10), a genus primarily known for inhabiting soil. At least nine other recently described Antarctic species were assigned to this genus, including isolates from Wright Valley cyanobacterial mats (A. flavus and A. roseus), aquatic sediments (A. antarcticus and A. ardleyensis), soil (A. cryotolerans, A. gangotriensis, and A. livingstonensis), penguin guano (A. psychrochitiniphilus), and seawater (A. kerguelensis) (Reddy et al., 2000, 2002; Gupta et al., 2004; Chen et al., 2005; Wang et al., 2009; Pindi et al., 2010; Ganzert et al., 2011a). Arthrobacter strains FY10 and FYS10 showed close phylogenetic alignment (>99% identity) to Wright Valley isolate A. flavus and also had a similar phenotype. The Lake Fryxell Arthrobacter strains and A. flavus are all non-motile coccobacilli (Figure 2) with distinctive bright yellow pigmentation, and growth occurs within similar temperature and salinity ranges (Reddy et al., 2000). Interestingly, however, Arthrobacter strains FY10 and FYS10 showed considerably higher metabolic flexibility compared to A. flavus, a bacterium that utilizes only sorbitol as a carbon source (Table 2; Reddy et al., 2000).

Brevundimonas strains FL10 and FS17 had 16S rRNA gene sequences identical to three strains of Brevundimonas characterized in our previous study of Lake Vanda (Figure 3), an ice-covered lake located in Wright Valley, which lies adjacent to Lake Fryxell (Vander Schaaf et al., 2015). All five of these Antarctic strains are closely related to the temperate aquatic species Brevundimonas intermedia and B. mediterranea (Poindexter, 1964; Abraham et al., 1999; Fritz et al., 2005). Despite having identical 16S rRNA sequences, strains FL10 and FS17 exhibited differences in fructose utilization and motility. Moreover, these strains were isolated from depths in Lake Fryxell having different physicochemical conditions, with strain FL10 originating from much lower dissolved sodium and sulfide concentrations at 10 m than the brackish, sulfide-rich waters from the strain FS17 isolation depth of 17 m. Species of Brevundimonas have a widespread and diverse aquatic distribution, especially in marine habitats, and the isolation of strains FL10 and FS17 from a coastal and brackish Antarctic lake is thus consistent with this generalization (Fritz et al., 2005; Ryan and Pembroke, 2018).

Two strains from the family Microbacteriaceae were isolated from Lake Fryxell. Phylogenetic analyses of one of these isolates, designated Microbacteriaceae sp. FT17, revealed closely related species from three different genera: Leifsonia rubra; Plantibacter sp. ARCTIC-P5 (from an unpublished study of Arctic bacteria); and Salinibacterium amurskyense, a halotolerant marine bacterium from the Sea of Japan (Han et al., 2003). A marine origin seems likely for strain FT17, but it is unclear whether it represents a novel species. Another isolate of the family Microbacteriaceae, which we designated Leifsonia sp. FO17, was morphologically and phylogenetically similar to Leifsonia rubra; both bacteria produce bright red-to-orange colonies, have a curved-rod morphology, stain Gram positively, and are non-motile. Interestingly, of 17 reported Leifsonia species, six (including L. rubra) are either psychrotolerant or psychrophilic and originate from Antarctica, with Leifsonia pindariensis, isolated from a Himalayan glacier in India, as the sole exception (Reddy et al., 2003, 2008; Pindi et al., 2009; Ganzert et al., 2011b).

As previously described in our study of Lake Vanda water column isolates (Vander Schaaf et al., 2015), Sphingobium sp. FO10 was phylogenetically identical to Sphingobium sp. VY55 and closely related to Sphingomonas spp. Ant17 and BP-7 (Sakai et al., 2007). Although Sphingobium spp. VY55 and FO10 share the same optimum temperature and bright yellow pigmentation, strain FO10 showed lesser halotolerance and a slightly wider temperature range than its Lake Vanda counterpart (Vander Schaaf et al., 2015).

Phylogenetic analyses clustered Pseudomonas sp. LFY10 within the P. fluorescens group of the genus Pseudomonas (Figure 5). Under our culture conditions, colonies of strain LFY10 were opaque and did not produce the signature fluorescent yellow pigmentation exhibited by many species in the P. fluorescens group. The fluorescent pigments, called pyoverdines, specifically chelate ferric (Fe³⁺) iron and play a key role in Fe³⁺ scavenging during iron-limiting conditions (Ringel and Brüser, 2018). Although it may be that genes for pigment biosynthesis simply were not expressed under the culture conditions employed, another explanation is that strain LFY10 may not have the genetic capacity to produce these pigments. Levels of dissolved iron (Fe²⁺) in Lake Fryxell are approximately 0.2 μM L⁻¹ at 10 m (Lee et al., 2004). Therefore, strain LFY10 could presumably obtain iron more efficiently by importing Fe²⁺ through a cation transporter than by sequestering Fe³⁺ through the production of fluorescent siderophores, potentially rendering genes for pyoverdine biosynthesis expendable.

A megaBLASTn search revealed that two 16S rRNA gene sequences identical to that from Pseudomonas strain LFY10 had been previously identified in two dissimilar studies, one analyzing the root microbiota of Vitis vinifera from an Argentinian vineyard (Salomon et al., 2013) and the other an uncultured clone from an unpublished study in Arunachal Pradesh, India (Figure 5). Strain LFY10 also aligned closely with P. azotoformans, a mesophilic pathogen of rice and other cereals (Fang et al., 2016), as well as the aforementioned aquatic and cold-adapted Antarctic species P. antarctica, P. meridiana, and P. proteolytica (Reddy et al., 2004). With growth temperatures ranging from −2 to 36°C and an optimum growth temperature of 24°C (Table 1 and Figure 6), the psychrotolerant Arctic isolate P. psychrophila (Yumoto et al., 2001) most closely matches the growth temperature response of strain LFY10.

Considering the physiological and phylogenetic characteristics of the bacteria isolated and described in this study, as well as the fact that Lake Fryxell is situated fewer than 10 km inland from the sea, it seems probable that at least some of these strains have a marine origin. With the possible exception of strain FO10, which shows little halotolerance, these bacteria may be derived from marine microorganisms that were deposited into Lake Fryxell, either by past flooding of Taylor Valley or aerosolization, and subsequently adapted to a planktonic lifestyle in an ice-sealed, stratified water column.

In addition to its physiological properties described above, an unexpected morphological feature of strain LFY10 was the bleb-like structures extending from the outer membrane of some but not all cells (Figure 4). Although the extensions had a phage-like morphology, closer inspection revealed that their dimensions and overall shape varied considerably. The structures had diameters ranging from 30 to 100 nm, and therefore, with such variation, it is unlikely that they were bacteriophages or particles of some other gene transfer agent, which have a defined and constant size (Solioz and Marrs, 1977; Lang et al., 2013, 2017). In addition to this, cultures of strain LFY10 grew robustly upon repeated transfer in both liquid and solid culture media—showing no universal virion-coating of cells in electron micrographs or other signs of phage infection (such as cell lysis)—and produced no detectable free viral particles.

It is likely that the observed surface structures on cells of strain LFY10 were membrane vesicles (MVs). The production of MVs is widespread among microorganisms, discovered first in Gram-negative bacteria and since identified in species from all three domains of life (Deatherage and Cookson, 2012; Anand and Chaudhuri, 2016). Pseudomonas fluorescens, a close relative to strain LFY10, produces MVs that are formed when the outer membrane of the cell wall bulges outward, fills with periplasm, and then pinches off to form a spherical vesicle (Guyard-Nicodème et al., 2007; Anand and Chaudhuri, 2016), a process that is consistent with our observations of strain LFY10. In addition, MVs vary in size from 20 to 250 nm (Beveridge, 1999; Nevot et al., 2006; Anand and Chaudhuri, 2016; Pathirana and Kaparakis-Liaskos, 2016), dimensions that correspond with the diameters of the structures extending from the cells of strain LFY10 (Figure 4).

Many Gram-negative bacteria shed MVs during the course of normal growth (Beveridge, 1999; Hagemann et al., 2014; Pathirana and Kaparakis-Liaskos, 2016). In the pathogen Pseudomonas aeruginosa, MVs play key roles in biofilm formation and pathogenesis (Park et al., 2014; Metruccio et al., 2016). However, in several cold-adapted Antarctic bacteria, including strains of Pseudomonas syringae, Shewanella vesiculosa, Pseudoalteromonas sp. M4.2, Psychrobacter fozii, and Marinobacter guineae (Frias et al., 2010; Pérez-Cruz et al., 2013, 2015; Kulkarni et al., 2015), vesicle shedding is a reaction to stress through SOS response activation and can be triggered by conditions such as extreme temperature or hyperosmolarity (Guyard-Nicodème et al., 2007; Frias et al., 2010; Metruccio et al., 2016).

Some species of cold-adapted bacteria also produce extracellular polymeric substances (EPS), which form a thick capsular layer that yields mucoid colonies (Frias et al., 2010), a feature displayed by strain LFY10. In accordance with observations of other Antarctic bacteria made by Frias et al. (2010), electron micrographs of cells of strain LFY10 showed putative MVs in various stages of formation, including several that were detached and spherical (Figure 4D). Membrane vesicles, which embed themselves in EPS, have been shown to be an integral part of this capsular layer in cold-adapted microorganisms, such as Pseudoalteromonas antarctica (Nevot et al., 2006). While the EPS in the capsular layer have been implicated in functions such as cryoprotection (Arctic marine bacterium Colwellia psychrerythraea 34H) and trace metal binding (Antarctic marine Pseudoalteromonas spp.), the role of MVs in the capsular layers is less understood (Marx et al., 2009; Ewert and Deming, 2011; Carillo et al., 2015; Casillo et al., 2018; Tribelli and López, 2018). Proposed functions for MVs in capsular layers include intracellular protein exchange, cell-cell signaling, or survival functions similar to those of the exopolymeric capsular components, but remain unclear without further study (Frias et al., 2010).

A final notable feature of Pseudomonas strain LFY10 is that it exhibits a number of ideal characteristics for use as a model cold-active bacterium in microbiology teaching labs. First, it is an excellent substitute for more commonly used psychrotolerant bacteria. For example, species of Listeria are often used to demonstrate the growth of microbes at low temperature. However, these species often take up to a week to show visible growth on tryptic soy agar (TSA) plates incubated at refrigerator temperatures (4–5°C). By contrast, strain LFY10 produces visible colonies within 24–48 h at 4°C on the same medium. In addition to growth temperature characteristics, strain LFY10 may be used to demonstrate halotolerance, as it is capable of growth in TSA containing up to 8.5% (w/v) NaCl (a demonstration commonly performed in teaching labs; data not shown) and has been shown to grow at even higher salinities in defined media (Heinzelmann et al., 2015a,b). Moreover, growth of strain LFY10 on blood agar plates may also be used to demonstrate beta-hemolysis, as colonies of this bacterium produce a characteristic clearing of red blood cells when grown on this medium (data not shown). Finally, strain LFY10 does not grow at 37°C (approximate human body temperature), and close relatives of this strain are of environmental or marine origin and are not reported to cause pathogenesis in mammals, minimizing the biosafety concerns that arise when using certain other Pseudomonas spp. (e.g., Pseudomonas aeruginosa) in a teaching lab setting.

In conclusion, with its permanent ice cover, seasonal swings of constant darkness or daylight, steady and near-freezing temperatures, and water column that shifts from hyperoxic to anoxic and from freshwater to brackish, the physicochemical conditions of Lake Fryxell have selected for a diversity of hardy microorganisms. Harboring exclusively microbial species, Lake Fryxell has become a favored habitat for studying microbial nutrient cycling in an ice-capped, polar environment. The collective ability of these strains to catabolize a wide range of carbon sources, including sugars, alcohols, and fatty and organic acids (Table 2), suggests that such nutritional versatility provides a fitness advantage and may catalyze dissolved organic carbon (DOC; Figure 1) turnover in the ice-sealed Lake Fryxell water column. The wide growth temperature range of the bacteria described here, including an ability to grow at 0°C, and their impressive metabolic versatility could signal their importance as key contributors to carbon turnover in this challenging aquatic ecosystem.

The datasets generated for this study can be found in the GenBank nucleotide repository¹ using either the strain names or accession numbers highlighted in Figure 3.

NVS, AC, AH, CJR, CLR, EH, and WMS performed laboratory experiments. JB wrote the initial manuscript draft. WMS designed the experiments and supervised the research. MM wrote the grant proposal that funded sample collection, and WMS wrote grant proposals to support laboratory investigations. WMS and MM edited the final version of the submitted manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.