Membrane Lipid Composition of the Moderately Thermophilic Ammonia-Oxidizing Archaeon “Candidatus Nitrosotenuis uzonensis” at Different Growth Temperatures

The core membrane lipids detected comprised 12 GDGTs, detected in all four replicate “Ca. Nitrosotenuis uzonensis” cultures at all growth temperatures. These were GDGT-0; GDGT-1 to GDGT-3, each with one isomer; GDGT-4, with two isomers; and crenarchaeol (cren) and its isomer (crenʹ, in which one cyclopentane moiety has a different stereochemistry) (31) (see Fig. 2 for structures). Previously, a Bligh-Dyer extract of “Ca. Nitrosotenuis uzonensis” was subjected to ether cleavage in order to examine the biphytanes (BPs) released from the GDGTs (31). The range of BPs produced were in agreement with the results of this study: in addition to a BP with no cyclopentane moieties and one with one cyclopentane moiety (data not shown), three BPs with two cyclopentane rings were produced (x, y, and I in Fig. 2 of reference 31) alongside two BPs with two cyclopentane rings and one cyclohexane ring (II and III in Fig. 2 of reference 31). The specific structural configuration of the 12 GDGTs cannot be determined from these data and would require nuclear magnetic resonance (NMR) analysis (31) or a selective sn2 ether cleavage protocol (56). Indeed, if both parallel and antiparallel (56) GDGT configurations are taken into consideration, the BPs detected could lead to 2 possible isomers of GDGT-0, 3 of GDGT-1, 11 of GDGT-2, 9 of GDGT-3, 15 of GDGT-4, and 18 of cren (of which cren′ would be one).

At a growth temperature of 37°C, GDGT-4 was dominant (28% ± 6% of total core GDGTs), followed by cren (25% ± 2%) and GDGT-1 (11% ± 3%) (Table 1). All other core GDGTs represented between 1 and 7% of the total. At 46°C and 50°C, cren was the dominant GDGT (51% ± 7% and 59% ± 8%, respectively), followed by GDGT-4 (16% ± 3% and 11% ± 4%, respectively) and crenʹ (12% ± 2% and 11% ± 3%, respectively), while all the remaining GDGTs accounted for between 0.4 and 6%. Eight of the core GDGTs exhibited an overall decrease in fractional abundance as the growth temperature increased (Table 1): GDGT-0 to GDGT-3, the GDGT-1 to GDGT-3 isomers, and GDGT-4. Generally, they all saw the greatest decrease in fractional abundance between temperatures of 37°C and 46°C, with only a slight or no decrease between 46°C and 50°C. The two isomers of GDGT-4 saw no overall change in fractional abundance at temperatures between 37°C and 50°C. Both cren and crenʹ increased in fractional abundance with increasing growth temperature, again with the greatest change between temperatures of 37°C and 46°C.

Five different GDGT polar headgroups were detected in the “Ca. Nitrosotenuis uzonensis” cultures: monohexose (MH), dihexose (DH), two isomers of trihexose (TH1 and TH2), and hexose/phosphohexose (HPH) (see Fig. 2 for structures). Both of the TH isomers underwent the same fragmentation under high-performance liquid chromatography–ion trap mass spectrometry (HPLC-ITMS) (Table 2), which indicates that both sets of isomers have the same distribution of sugars around the GDGT core (as opposed to the constitutional isomers described in reference 57). Therefore, it can be assumed that they are different sugar stereoisomers. The DH-GDGT was assigned a structure with one sugar moiety on each end of the GDGT, based on comparison of the LC-ITMS² fragmentation (Table 2) with those reported previously (57). The LC-ITMS² fragmentation (Table 2) allowed us to further assign the two TH isomers a structure with two sugar moieties on one side and one on the other.

It should be noted that the intact polar lipids (IPLs) were examined in terms of their MS peak area response, and thus, the relative abundances reported here may not be their true relative abundances. However, this method allows for direct comparison between the “Ca. Nitrosotenuis uzonensis” cells cultured at different temperatures that were analyzed in this study. The headgroup distributions (Table 3) at 37°C and 46°C were quite similar, with the most dominant polar headgroup being HPH (fractional abundances of 36% ± 15% and 33% ± 3%, respectively), followed by DH (25% ± 4% and 29% ± 3%), TH1 (22% ± 3% and 23% ± 3%), and TH2 (12% ± 8% and 10% ± 5%), with the least abundant headgroup being MH (4% ± 1% and 5% ± 1%). With the increase in growth temperature from 46°C to 50°C, the headgroup distribution changed noticeably (Table 3); MH increased (from 5% ± 1% to 10% ± 5%), as did DH (from 29% ± 3% to 36% ± 8%). Indeed, DH had become the most abundant GDGT headgroup at 50°C. Considering the variability between replicate cultures, the fractional abundances of TH1 and TH2 did not exhibit an overall change with increasing temperature. The HPH GDGT headgroup decreased with the increase in growth temperature from 46°C to 50°C, from 33% ± 3% to 22% ± 9%.

A marked variability in the distribution of core lipids of each intact polar lipid was observed (Table 3 and Fig. 3). The LC-ITMS method utilized did not resolve the isomers of the GDGTs, nor did it resolve cren from crenʹ; hence, the IPL-bound GDGT cores are defined as GDGT-0 to GDGT-4 and cren. For example, cren and GDGT-0 were predominantly contained in the HPH IPL, while GDGT-4 and, to a lesser extent, GDGT-3 were predominantly found in the DH, TH1, and TH2 IPLs. Consistent with what was seen for the hydrolysis-derived core lipids, the cultures grown at 46°C and 50°C exhibited similar IPL-bound core lipid distributions, while those grown at 37°C were more distinct (Fig. 3). Overall, in the cultures grown at 46°C and 50°C, cren became more dominant for the MH and HPH IPLs and increased in DH and TH2. For the TH1 IPL, there was no change in the core lipid distribution with increasing temperature.

The relative abundance of the core lipids of “Ca. Nitrosotenuis uzonensis” varied with increasing growth temperature, with more cren and crenʹ and less GDGT-0 to GDGT-4 (including isomers). It is well established that Thaumarchaeota increase their cren and crenʹ proportions at higher temperatures, which explains the fundamental role of crenʹ in the sea surface temperature (SST) proxy TEX₈₆ (tetraether index of tetraethers consisting of 86 carbons) (63), particularly at relatively high temperatures of >20°C. The TEX₈₆ SST proxy, and its low-temperature (<15°C) and high-temperature (>15°C) versions TEX^L₈₆ and TEX^H₈₆, respectively, have been applied to temperature reconstructions in a wide range of marine and lacustrine settings (see references 29 and 64 for reviews). We calculated the TEX₈₆ values for the replicate stationary “Ca. Nitrosotenuis uzonensis” cultures (Table 4) and applied both the core-top and mesocosm-based TEX^H₈₆ calibration described previously (65) to calculate estimated temperatures. Using the core-top TEX^H₈₆ calibration model (65), the calculated temperatures were 22°C ± 4.7°C, 32°C ± 1.7°C, and 30°C ± 2.4°C, while they were 29°C ± 3.6°C, 37°C ± 1.3°C, and 35°C ± 1.8°C, respectively, using the mesocosm TEX^H₈₆ calibration model (Table 4). The temperature calculation that gave the most similar results to the actual culture temperatures was the mesocosm TEX^H₈₆ calibration model, developed using enrichment cultures (65). However, the estimated temperatures were still on average 10°C ± 4°C lower than the actual growth temperatures, while the core-top TEX^H₈₆ calibration model gave results that were on average 16°C ± 3.4°C lower than the actual growth temperatures (Table 4). Previous studies have reported poor correlations between TEX₈₆ values and temperature, and, hence, inaccurate temperature estimates, in thermophilic Thaumarchaeota cultures and in samples from thermal environments (33, 51, 53, 61). In this context, it is important to note that neither TEX₈₆ nor TEX^H₈₆ was designed to be used in terrestrial thermal environments such as hot springs. In this context, it is also interesting to keep in mind that TEX₈₆ was found to correlate with the concentration of bicarbonate, not temperature, in a range of Nevada hot springs (51). Our results now further demonstrate that TEX₈₆ does not reflect well lipid membrane adaptation for thermophilic Thaumarchaeota. As culturing conditions other than temperature were kept constant in this study, we cannot examine the relationship that variables such as growth phase, bicarbonate concentration, ammonium oxidation rates, and pH would have on the GDGT distribution in the “Ca. Nitrosotenuis uzonensis” cultures.

It is also well established that temperature is a primary factor controlling the number of GDGT cyclopentane moieties with increasing temperature leading to an increasing number of cyclopentane moieties (63, 66–69). However, this effect was only minor for “Ca. Nitrosotenuis uzonensis,” as the average number of cyclopentane moieties increased from 3.1 ± 0.3 at 37°C to 3.7 ± 0.3 at 46°C and 50°C (Table 4).

As the temperature increased, specific GDGTs changed in their abundance relative to the abundance of their isomers (Table 5). With the increase in temperature from 37°C to 50°C, GDGT-2 and GDGT-2ʹ went from a distribution of 60:40 to 52:48, GDGT-3 and GDGT-3ʹ went from 71:29 to 54:46, and GDGT-4, GDGT-4ʹ, and GDGT-4″ went from 83:16:1 to 66:30:5 (Table 5). Interestingly, while the overall percentage of crenʹ increased with temperature, it remained constant relative to cren: the ratios of cren to crenʹ were 84:16 at 37°C and 85:15 at 50°C (Table 5). It has recently been revealed that crenʹ has a stereochemically different cyclopentane ring than that of cren, a difference in stereochemistry that has been postulated to have an effect on membrane fluidity, therefore playing a role in maintaining membrane homeostasis (31). In the temperature range examined for “Ca. Nitrosotenuis uzonensis,” crenʹ was not upregulated relative to cren as a membrane adaption to increasing temperature. However, GDGT-2ʹ was upregulated relative to GDGT-2, GDGT-3ʹ was upregulated relative to GDGT-3, and GDGT-4ʹ and GDGT-4″ were upregulated relative to GDGT-4. As explained in Results, we are not able to determine the stereochemistry of the different GDGT isomers detected in this study; however, the biphytanes (BPs) released from the GDGTs of “Ca. Nitrosotenuis uzonensis” grown at 46°C (31) included three different BPs with two cyclopentane rings and two BPs with two cyclopentane rings and one cyclohexane ring. Combinations of these BPs can give rise to a wide range of isomers. Here we hypothesize that the change in the composition of GDGT-2, GDGT-3, and GDGT-4 with increasing temperature represents, as per cren, changes in their cyclopentane ring stereochemistry, in order to maintain membrane homeostasis.

Whereas the core lipid compositions of the “Ca. Nitrosotenuis uzonensis” cultures grown at 46°C and 50°C were most similar to each other, the cultures grown at 37°C and 46°C were the most similar in terms of polar headgroup composition (Table 3 and Fig. 3). A similar observation was made previously by others (33), who noted that the core lipid and polar headgroup distributions in Thaumarchaeota are affected by different factors. With increasing growth temperature, the main change in the “Ca. Nitrosotenuis uzonensis” polar headgroup composition was that two of the smaller headgroups (MH and DH) increased in relative abundance, while one of the largest headgroups (TH2) decreased. However, when we calculated the average number of sugars per GDGT, we found no significant difference between the different growth temperatures (Table 3). To examine polar headgroup adaptions further, we calculated for each growth temperature the ratio of glycolipids to phospholipids, which was higher at 50°C (4.2 ± 2) than at 37°C and 46°C (2.2 ± 2 and 2.0 ± 0.3, respectively). Studies that describe the effect of temperature on archaeal polar headgroup composition are limited (for a review, see reference 70). The temperature-driven polar headgroup adaption reported for three strains of the Thaumarchaeota species Nitrosopumilus maritimus (71) (also from the NP order but a marine mesophile) was different from that seen in this study: all three N. maritimus strains generally decreased the relative percentage of MH lipids as the temperature increased, while the percentages of DH and HPH generally increased, and hence, in contrast to our findings, the ratio of glycolipids to phosphoglycolipids decreased as the temperature increased. However, it should be noted that the growth temperature range (18°C to 35°C) in the N. maritimus study was much lower. The results for “Ca. Nitrosotenuis uzonensis” are also different from those described previously for the Euryarchaeota species Thermoplasma acidophilum (thermophilic and acidophilic) (72), which was found to adapt to higher temperatures (and to lower pHs) by increasing the number of sugars in the polar headgroups. However, similar to our findings for “Ca. Nitrosotenuis uzonensis,” the ratio of glycolipids to phosphoglycolipids in T. acidophilum increased at higher temperatures. The decrease in phosphoglycolipids relative to glycolipids may relate to adaptions such as decreased proton permeability of the membrane (72–74) or could relate to stress adaption during which P-containing lipids are replaced with non-P-containing lipids in order to utilize the P for other essential cell processes. Replacement of phospholipids with nonphospholipids as a response to nutrient limitation or other stresses has been previously observed in bacteria (75), algae (76–78), and archaea (50, 74).

We compared the core lipid composition of “Ca. Nitrosotenuis uzonensis” with the core lipid compositions of other thaumarchaeotal species reported in the literature (data used are listed in Table S1 in the supplemental material). To produce this simplified data set, the isomers of the GDGTs were grouped together, with the exception of cren and crenʹ, which were treated separately, while hydroxy-GDGTs were not included. First, we used principal-component analysis (PCA) (Fig. 4) to examine this data set. The first two principal components accounted for 32 and 31%, respectively, of the variability in the core lipid composition. GDGT-0 to GDGT-3 were negatively loaded on the first principal component, while GDGT-4 and crenʹ were positively loaded. cren was negatively loaded on the second principal component. The majority of the NP order members were grouped across the two negative quadrants of the first principal component, while the majority of the Nitrososphaerales (NS) order members were in the two positive quadrants of the first principal component, in the direction of crenʹ. The single Nitrosotaleales (NT) and Nitrosocaldales (NC) order members included in the data set were placed between the NP and NS orders (Fig. 4). There were three exceptions to this otherwise clear NP/NS separation. First, our three “Ca. Nitrosotenuis uzonensis” (NP order) cultures grown at different temperatures clustered between the NP and NS members. Second, a “Ca. Nitrososphaera gargensis” culture (moderately thermophilic; NS order), which was grown at 35°C (33), was placed within the NP order cluster. Third, “Ca. Nitrosocosmicus oleophilus” MY3 (NS order) (22) was placed within the NP order cluster. It should be noted that all members of the NS order that group closely together in the PCA are phylogenetically very closely related (Fig. S1). Without lipid analysis of further members of the genus Nitrosocosmicus, it is not possible to say whether “Ca. Nitrosocosmicus oleophilus” is an outlier or whether all Nitrosocosmicus members would group with the NP. Overall, the NP/NS cluster separation was driven by the fractional abundance of crenʹ (Fig. 4). Many of the NS members examined (e.g., “Ca. Nitrososphaera gargensis,” Nitrososphaera viennensis, and “Ca. Nitrososphaera sp.” strain JG1) contained a high percentage (14 to 29%) of crenʹ (33, 36, 40, 54, 63), while the majority of the NP members contained a lower fractional abundance (0 to 3%) of crenʹ (16, 32, 33, 37–39). This explains why the “Ca. Nitrosotenuis uzonensis” cultures did not cluster with the other NP order members, as all cells grown at the three different temperatures contained a relatively high fractional abundance of crenʹ (i.e., 5, 11, and 12%). It is commonly observed that the proportion of crenʹ is higher in thermophilic Thaumarchaeota than in mesophilic Thaumarchaeota (35, 36, 40), and our results suggest that this phenomenon is independent of order affiliation. The results of the PCA illustrate that while phylogeny seems to have a strong influence on GDGT distribution, environmental parameters like growth temperature can lead to inconsistencies between phylogenetic affiliation and GDGT composition, as exemplified by the moderate thermophile “Ca. Nitrosotenuis uzonensis” via its elevated amounts of crenʹ compared to other NP members. In this context, it should be kept in mind that factors not examined in this study have also been shown to have an effect on GDGT distribution (cf. lines linking points in Fig. 4), including O₂ concentration, pH, and salinity (52, 71, 79).

To further examine the relationship between crenʹ and temperature across the thaumarchaeotal orders, we calculated the cren′-to-cren ratio for all known thaumarchaeotal core lipid compositions reported in the literature (Table S1) and found a significant correlation with growth temperature (Spearman r = 0.72; n = 56; P = <0.001). However, there appears to be a “tipping point” in the crenʹ-to-cren ratio at 35°C (Fig. 5). In the temperature range of 4°C to 35°C, the crenʹ-to-cren ratio was on average 0.02 ± 0.03, while in the range of 36°C and above, the ratio was 0.3 ± 0.2. However, it should be noted that taxon sampling is still relatively skewed in this analysis, with many members of the genus Nitrosopumilus being well represented and many members of the genera Nitrosocosmicus, Nitrosotalea, and Nitrosocaldus still awaiting lipid composition analysis. As discussed above, crenʹ has a stereochemically different cyclopentane ring than that of cren (31), which could lead to the two isomers having different effects on the fluidity of a cell membrane. It is possible that this apparent tipping point represents a Thaumarchaeota-wide temperature above which the different stereochemistry of the crenʹ cyclopentane ring provides a beneficial effect to the membrane. The fact that all “Ca. Nitrosotenuis uzonensis” cultures were grown at temperatures above this 35°C tipping point would then explain why the ratio of crenʹ to cren did not change between these growth temperatures.

A direct comparison of the intact polar lipid (IPL) composition of “Ca. Nitrosotenuis uzonensis” with those of other thaumarchaeotal species reported in the literature is less straightforward than for core lipids due to variability in analytical methods used between studies and the nonquantitative nature in which IPL data have often been reported. In Table 6, we summarize, in a qualitative manner, thaumarchaeotal IPL distributions reported in the literature from studies that utilized a normal-phase liquid chromatography-mass spectrometry (LCMS) method comparable to the one used in this study. The “Ca. Nitrosotenuis uzonensis” cultures grown at 37°C and 46°C contained HPH as the dominant polar headgroup, as has been previously reported for a range of NP species (Table 6) and also for a moderately thermophilic terrestrial NS member, “Ca. Nitrososphaera gargensis.” Conversely, the “Ca. Nitrosotenuis uzonensis” cultures grown at 50°C were dominated by DH, a characteristic IPL reported in high abundance for the NS members “Ca. Nitrososphaera sp.” JG1 and N. viennensis (40). Neither form of TH detected in “Ca. Nitrosotenuis uzonensis” was reported in other members of the NP (16, 37, 38), but they have been reported in the NS member N. viennensis (40). Previously, the lipid compositions of a range of cultured representatives of the four thaumarchaeotal orders were examined (33), and it was suggested that the core lipid composition reflects phylogenetic orders, while the polar headgroup composition reflects habitat (either terrestrial thermophiles, marine mesophiles, or soil mesophiles). Knowledge of the lipid composition of “Ca. Nitrosotenuis uzonensis,” which, unlike the other mesophilic members of the NP order, is moderately thermophilic (55), further confirms that environmental niche or habitat is a driver of headgroup composition. The “Ca. Nitrosotenuis uzonensis” cultures are more similar in headgroup composition to the three terrestrial NS members, all of which were cultured at temperatures above 35°C, than the five other NP members, all of which were cultured at temperatures below 35°C (Table 6).

The “Ca. Nitrosotenuis uzonensis” cultures were not found to contain IPLs with a core hydroxy-GDGT (MH-OH or DH-OH), whereas these have been detected in all other members of the NP order examined to date (16, 33, 37, 38). It should be noted that hydroxy-GDGTs were not included in the PCA (Table S1). Their absence in “Ca. Nitrosotenuis uzonensis” may mean that hydroxy-GDGTs are found only within specific clades of the NP order. However, hydroxy-GDGTs have also been associated with growth temperature: a decrease in temperature has been observed to lead to an increase in hydroxy-GDGTs in both thaumarchaeotal cultures (80) and environmental samples (80–83). Hence, it is possible that their absence in “Ca. Nitrosotenuis uzonensis” relates to the high cultivation temperatures, reflecting their thermophilic nature.

Highly enriched “Ca. Nitrosotenuis uzonensis” cultures that contained no other archaea (55) were grown in medium containing (per liter) 54.4 mg KH₂PO₄, 74.4 mg KCl, 49.3 mg MgSO₄·7H₂O, 584 mg NaCl, 33.8 μg MnSO₄, 49.4 μg H₃BO₃·7H₂O, 43.1 μg ZnSO₄·7H₂O, 37.1 μg (NH₄)₆Mo₇O₂₄·4H₂O, 97.3 mg FeSO₄·5H₂O, 25.0 μg CuSO₄·2H₂O, and 4.0 g CaCO₃. Fresh medium batches were allowed to equilibrate for 2 weeks to ensure reaching low levels of hydrogen peroxide forming during the preparation process. Four biological replicates were grown at 37°C, 46°C, and 50°C, in 250-ml Schott flasks, in the dark. All replicates went through two 10% transfers at the respective temperatures to dilute out lipids from the inoculum culture. The cultures were fed with 1 mM NH₄Cl (final concentration), and depletion was monitored with Nessler’s reagent. The final batches were refed multiple times, consumed 4 mM ammonium in total, and depleted ammonium before the biomass was harvested. This final biomass production step took about 8 weeks, ensuring that at the end, the vast majority of the cells produced were in stationary phase.

Freeze-dried biomass lipids were extracted using a modified Bligh-Dyer procedure (84). Briefly, the biomass was treated ultrasonically three times for 10 min with a solvent mixture of methanol (MeOH), dichloromethane (DCM), and phosphate buffer (2:1:0.8, vol/vol/vol). After sonication, the combined supernatants were phase separated by adding additional DCM and buffer to a final solvent ratio of 1:1:0.9 (vol/vol/vol). The organic phase containing the intact polar lipids (IPLs) was collected, and the aqueous phase was reextracted three times with DCM. Finally, the combined extract was dried under a stream of N₂ gas.

In order to remove the headgroups from the IPLs and to obtain the remaining core lipids, the Bligh-Dyer extract was hydrolyzed with 5% (vol/vol) HCl–MeOH by refluxing (3 h). The hydrolysate was neutralized with KOH to pH 7/8, extracted with DCM, and dried over Na₂SO₄.

The hydrolyzed Bligh-Dyer extracts were analyzed using high-performance liquid chromatography/atmospheric-pressure chemical ionization mass spectrometry (HPLC/APCI-MS) on an Agilent 1100/Hewlett Packard 1100 MSD instrument equipped with automatic injector and HP-Chemstation software according to methods described previously (85), with the following modifications. Separation was achieved in normal phase with two Prevail Cyano columns in series (150 mm by 2.1 mm; 3 μm) with a starting eluent of hexane-propanol (99.5:0.5, vol/vol) and a flow rate of 0.2 ml min⁻¹. This remained isocratic for 5 min, and thereafter, there was a linear gradient to 1.8% propanol at 45 min. The injection volume was 10 μl.

The ratios and calculations that were carried out on the core lipid data are as follows:

The Bligh-Dyer extracts were directly analyzed for IPLs. Extracts were redissolved in a mixture of hexane–2-propanol–water (72:27:1, vol/vol/vol) at a concentration of 10 mg ml⁻¹. IPL extracts were analyzed by HPLC-ion trap mass spectrometry (ITMS) according to methods described previously (86), with modifications as described previously (87). The analysis was performed on an Agilent 1200 series LC instrument (Agilent, San Jose, CA), equipped with a thermostated autoinjector and column oven, coupled to an LTQ XL linear ion trap with an Ion Max source and an electrospray ionization (ESI) probe (Thermo Scientific, Waltham, MA). Separation was achieved on a LiChrospher diol column (250 by 2.1 mm, 5-μm particles; Alltech) maintained at 30°C. The following elution program was used with a flow rate of 0.2 ml min⁻¹: 100% eluent A for 1 min, followed by a linear gradient to 66% eluent A–34% eluent B in 17 min, maintained for 12 min, followed by a linear gradient to 35% eluent A–65% eluent B in 15 min (where eluent A is hexane–2-propanol–formic acid–14.8 M NH₃(aq) [79:20:0.12:0.04 {vol/vol/vol/vol}] and eluent B is 2-propanol–water–formic acid–14.8 M NH₃(aq) [88:10:0.12:0.04 {vol/vol/vol/vol}]). The lipid extract was analyzed by an MS routine where a positive-ion scan (m/z 1,000 to 2,000) was followed by a data-dependent MS² experiment where the base peak of the mass spectrum was fragmented (normalized collision energy [NCE] of 25, isolation width of 5.0, and activation Q of 0.175). IPLs were examined in terms of their MS peak area response. Thus, the relative abundance of the peak area does not necessarily reflect the actual relative abundance of the different IPLs; however, this method allows for comparison between the strains analyzed in this study. The peak areas were determined from extracted ion chromatograms of the [M + NH₄]⁺ ion for each individual IPL species.

The ratios and calculations that were carried out on the intact polar lipid data are as follows:

Principal-component analysis (PCA) was done using XLSTAT 2018 (Addinsoft, New York, NY). The data used for the PCA were fractional abundances (percent) that totaled 100 for each species and were not transformed further before analysis. Spearman rank correlation was done using SigmaPlot (SigmaPlot for Windows version 14; Systat Software Inc., Germany).

An amino acid alignment of 34 universal marker genes was extracted from previously reported ammonia-oxidizing archaeon (AOA) genomes using CheckM (88). A maximum likelihood tree was constructed using IQTREE multicore version 1.6.2 (89) with 1,000 ultrafast bootstraps (90) under the best-fit model LG_F_R4, determined using ModelFinder (91) (where LG = general amino acid exchange matrix [92], F = empirical amino acid frequencies from the data, and R4 = rate heterogeneity calculated under the FreeRate model [93, 94] with four categories).