The ability to trigger ice formation is a property shared by certain particles both organic and inorganic, including bacteria, viruses, phytoplankton, pollen, soot, dust. These little particles help water molecules come together and form ice at warmer temperatures than freezing point, in a process called ice nucleation. In nature these ice nucleators play important roles in cloud condensation, precipitation, climate, and Earth's radiative equilibrium.
Common ice nucleators include several bacterial species. One such species is Pseudomonas syringae, whose special outer-membrane protein engages in hydrogen bonding with water to line the molecules up in an orderly lattice, thereby catalyzing the formation of ice crystals. It is this unique ability that gives the bacterium a double-edged relationship with humans: P. syringae is a notorious crop pathogen that attacks plants by inducing frost damage on leaves, but it is also this ability that makes P. syringae useful for ski resort snow machines preparing winter-ready slopes. Other bacteria with similar special ice-inducing proteins include members of the genera Pantoea, Erwinia, and Xanthomonas. Of course, that outer membrane protein is exclusive to Gram-negatives. The basis for ice nucleation by Gram-positive bacteria lacking the outer membrane, however, is an area that has been a bit quieter.
A recent icebreaker was a 2021 paper from Failor et al. To peek into a Gram-positive mechanism for ice nucleation, the authors first identified involved genes in the Firmicute (aka Bacillota) Lysinibacillus parviboronicapiens (Lp) by comparing the genome of wild-type Lp with that of mutant non-ice-nucleating Lp. These candidate genes were winnowed further by a UV-mutagenesis screen to two genes that resulted in significant loss of ice-nucleating ability. Here came a surprise. One of these genes was predicted to be a type I PKS (polyketide synthase), and the other, an NRPS (non-ribosomal peptide synthetase), both of which were in the same operon—marking the first time that a PKS-NRPS has been associated with ice nucleation.
The authors wondered if the ice-nucleating particle from Lp could be secreted via vesicles, like fellow polyketide linearmycin. It is a possibility; supernatant from Lp cultures was filtered and subjected to gradient ultracentrifugation, and the ice-nucleating particles were found in the same fractions as linearmycin. Transmission electron microscopy of Lp then revealed that these particles form long extracellular filaments composed of many little "pearls" (Figure 2) that are on average 488 nanometers long and 5 nanometers wide. These "pearl chains" were not found on mutants incapable of ice-nucleation. Although how exactly these chains cause ice nucleation remains unknown, this study demonstrates that polyketides, like the proteins of Gram-negative bacteria, are capable of triggering it.
Although this is the first PKS-NRPS to be linked to the nucleation of ice, it is not the first to be linked to nucleation as a broader phenomenon. For example, there is a PKS known to be involved in nucleation of calcium carbonate in the process of biomineralization. The authors suggest that similar polyketide-based molecules may underlie the ice-nucleating ability of pollen and of fungi like Fusarium and Mortierella, whose mechanisms also remain enigmatic.
Beyond examples of smaller-scale ice crystal formation, one might also wonder how much of a role microbes have played on a grander scale, in global freeze-thaw cycles throughout the history of Earth… Earth has had five major ice ages—periods in which colder temperatures cause ice sheets to spread across the globe. The most recent ice age 20,000 years ago was replete with woolly mammoths and saber-toothed tigers, popularized by the animated film and franchise. The climate of this last ice age has been reconstructed using well-established methods that include simulations of particles acting as ice nucleators. Of these nucleating particles important to the inception of ice ages, one might wonder what proportion was bacterial.