Resolving a piece of the archaeal lipid puzzle

Lipid membranes are common to all cells, despite occurring in many different forms across Earth’s great biotic diversity. Among the most distinctive membranes are those formed by the archaea, whose lipids are characterized by sn-2,3-glycerol stereochemistry (in contrast to sn-1,2-glycerol in bacteria and eukarya), isoprenoid rather than acetyl hydrophobic chains, and frequent occurrence as membrane-spanning macrocycle structures (1). The membrane-spanning lipids consist of mixed assemblages of structural isomers containing up to 8 internal cyclopentane rings (GDGT-0 through GDGT-8 [glycerol dibiphytanyl glycerol tetraethers with zero to 8 rings]) (Fig. 1). Many aspects of the biosynthesis of these unusual structures remain unknown, but, in PNAS, Zeng et al. (2) take an important step forward by revealing genes encoding for 2 enzymes involved in synthesis of the cyclopentane moieties. Pinpointing these genes is critical not only for understanding archaeal biosynthetic pathways but also for resolving questions about the primary sources of the GDGTs that are widely detected in the environment.

Fig. 1.

(A) GDGTs containing cyclopentane rings are widely distributed among the archaea. (B) GDGTs are membrane-spanning tetraether lipids with 2 C₄₀ isoprenoid chains (R, R’), each containing zero to 4 cyclopentane rings (and sometimes a cyclohexane ring, not shown). Tetraethers are formed from the diether precursor, DGGGP, in which the phosphate is replaced by an alternative polar group (X) before tetraether formation. (C) The sequence of biosynthetic steps leading from 2×[DGGGP] → GDGT remains unknown, but a plausible order would be tetraether synthesis using an MSS, a step that requires the Δ^14-15 double bond (22); GrsA and GrsB in sequence, as shown by Zeng et al. (2); and final saturation by GGR, which targets double bonds in the order Δ², Δ⁶, Δ¹⁰ (23).

Some history is as follows: In 1992, observations from independent disciplines yielded the remarkable conclusion that archaea—often considered “extremophiles”—must be widespread in the world’s oceans. A suite of C₄₀ isoprenoid hydrocarbons of archaeal origin was found in sediments apparently not associated with hydrothermal or methanogenic activity (3), and gene sequences of archaea were discovered in some of the earliest universal amplicon libraries (4, 5). This confluence launched a new era of research that identified the central role of ammonia-oxidizing Thaumarchaeota in the marine nitrogen cycle. Carbon isotopic analyses (6), detection of archaeal ribosomal RNA (7) and ammonia monooxygenase (amoA) genes (8), and isolation of the first pure culture (9) collectively established their physiology and global importance, while the high preservation potential of GDGTs in marine sediments led to development of a new lipid-based paleothermometer called TEX₈₆ (10). TEX₈₆ is a proportionality index of GDGTs formulated by analogy to the lipid composition of cultured thermophilic archaea, in which it is observed that a higher fractional abundance of cyclopentane rings is associated with higher growth temperature (1, 11). Refined calibrations of the TEX₈₆ index and its response to upper ocean temperatures have made it a key tool for the paleoclimate community (12).

However, archaeal groups in addition to Thaumarchaeota also make cyclopentane-containing GDGTs, including the Crenarchaeota and many divisions of Euryarchaeota (Fig. 1A). In particular, the surface-dwelling Marine Group II (MG-II) Euryarchaeota have been suggested to be GDGT sources (13). This would affect TEX₈₆ signals if their ring distributions have different physiological controls compared to Thaumarchaeota. Such differences in lipid response might be expected, because, to date, the known ammonia-oxidizing Thaumarchaeota are obligate autotrophs residing near the base of the photic zone (9, 14), while the uncultured MG-II is suggested to be heterotrophic, occupying a different niche space at shallower depths (15). Lack of knowledge about GDGT synthesis has inhibited resolution of this problem and contributes to ambiguity about the taxonomic sources of these lipids in marine systems.

By identifying 2 unique S-adenosylmethionine (SAM) proteins required for the formation of cyclopentane rings—which they call GDGT ring synthases GrsA and GrsB—Zeng et al. (2) open a window into resolving the taxonomic question, while also providing an opportunity to understand more about both the physiological functions and biosynthetic mechanisms of such rings. Notably, they report that open ocean metagenomes and MG-II metagenome-assembled genomes (MAGs) yield a low diversity of Grs sequences, all of which are affiliated with Thaumarchaeota (i.e., no instances in MG-II MAGs). If this result holds, it is good news for the TEX₈₆ proxy, which assumes marine sedimentary GDGTs mainly derive from planktonic Thaumarchaeota (11, 12). However, although DNA sequence identification provides taxonomic assignment, it raises questions about temporal or spatial adequacy of sampling, and potential taxonomic differences with respect to cellular growth rate or activity. Copy numbers do not equal production rates, and, now that the GDGT ring synthases have been identified, more work will be needed to demonstrate in situ activity. More information may also be available through data mining of existing metagenomes and metatranscriptomes from other types of environments.

The identification by Zeng et al. (2) of GrsA and GrsB in the genetically tractable thermophile Sulfolobus acidocaldarius also may result in better understanding of the physiological controls on biosynthesis of ring-containing GDGTs. Formation of a variable number of rings reflects the balance between 2 categories of reactions: saturation by the enzyme geranylgeranyl reductase (GGR) vs. ring formation by GrsA/GrsB (Fig. 1). Double-bond reductions by GGR require the organisms to dedicate net reducing power to the process, whereas internal cyclization does not change the oxidation state. Because the extremophile nature of archaea can be described as adaptation to chronic energy stress (16), this balance is particularly relevant—both the biosynthetic pressure for ring synthesis (less demand for electron donor) and the outcome (different physicochemical membrane properties) may be under evolutionary and environmental selection. The preference between these 2 steps may be regulated by many variables affecting cellular homeostasis, including not only temperature (1, 11) but also other factors that affect transmembrane potential, including environmental pH, Eh, substrate availability, and growth rate (17–19). The resulting assemblage of GDGT-0 through GDGT-8 affects membrane stiffness and diffusive properties, including the rate at which transmembrane potential is dissipated (20). Sulfolobus provides a model experimental system to test how GDGT production responds to environmental pressures.

Finally, many questions remain about the complete biosynthesis of GDGT core structures. Zeng et al. (2) demonstrate that GrsA acts prior to GrsB but decline to speculate on whether cyclization occurs before or after saturation by GGR. However, given the presence of geranylgeranyl chains in the intermediate structure digeranylgeranylglycerol phosphate (DGGGP) (Fig. 1B), I suggest a reasonable hypothesis is that GrsA/GrsB would act before GGR to exploit the existing double bonds: A typical mechanism of SAM enzymes is free radical formation at an sp³ carbon, followed by internal attack on an sp² carbon (21). If true, this order provides a framework in which to unite additional, independent observations. Zeng et al. (2) note that the substrate-specific behavior of GrsA and GrsB appears to require prior formation of a membrane-spanning tetraether. Here, I call this step “MSS” to indicate a hypothetical membrane-spanning synthase. The terminal (Δ^14-15) double bond apparently is required for the MSS reaction, and the intermediate may again be a carbon radical (22) (Fig. 1C). Together, these observations imply that diether condensation and ring formation begin at the hydrophobic ends of DGGGP and proceed in the direction of the glycerol moiety. In contrast, saturation by GGR begins nearest the glycerol and proceeds sequentially in the other direction (23). By pinpointing both the type of enzyme and the preferred substrates for GrsA and GrsB, Zeng et al. (2) provide evidence for a testable, albeit still quite speculative, GDGT biosynthesis scheme having the order MSS, GrsA, GrsB, GGR (Fig. 1).

Of course, I have conspicuously neglected any mention of the unusual cyclohexane ring (24) that occurs in the GDGTs of Thaumarchaeota—it is a mystery that reminds us how many more secrets the archaea have yet to reveal and a reminder that there is much more work still ahead.