Visualization of Loki- and Heimdallarchaeia (Asgardarchaeota) by fluorescence in situ hybridization and catalyzed reporter deposition (CARD-FISH)

Metagenome-assembled genomes (MAGs) of Asgardarchaeota are starting to be recovered from a variety of habitats, broadening their environmental distribution and providing access to the genetic makeup of this archaeal lineage. Despite their singular phylogenetic position at the base of the eukaryotic tree of life, the morphology of these bewildering organisms remains a mystery. In order to visualize this elusive group, we applied a combination of CARD-FISH and epifluorescence microscopy on coastal hypersaline sediment samples, using specifically designed CARD-FISH probes for Heimdallarchaeia and Lokiarchaeia lineages and provide the first visual evidence for both these groups. Here, we show that while Heimdallarchaeia are characterized by a uniform cellular morphology typified by central DNA localization, Lokiarchaeia display a plethora of shapes and sizes that likely reflect their broad phylogenetic diversity and ecological distribution.

During microscopic inspections, we carefully checked for potential non-specific or 92 autofluorescent signals at wavelengths not interfering with the probe signal and found no 93 overlap for any of the inspected cells. A set of negative controls were conducted to rule out 94 false-positive signals due to unspecific binding of dye or nucleic acid components of probes 95 by using a nonspecific probe (NON338, Supplementary Figure 7)(16). To avoid false-positive 96 signals from cellular peroxidases, we performed additional control experiments including the 97 CARD reaction only (without probes). All these control treatments resulted in low, unspecific 98 background signals (comparable to those obtained in samples with probes for 99 Asgardarchaeota), but no obvious staining of cells (Supplementary Figures 4, 7). Further 100 evidence of specificity was seen in all cells hybridized with the Heimdallarchaeia probe, both 101 the shapes and staining patterns coupled to DAPI were remarkably consistent. 102 conclude whether the condensed DNA, particularly in Heimdallarchaeia cells is indicative of a 104 proto-nucleus. Microscopic images of bacterial cells with apparently eukaryotic-features have 105 been misinterpreted before, e.g. in the case of the phylum Planctomycetes (17). 106 Examining these first images of Asgardarchaeota, we could not avoid to note the unusual 107 happenstance of their naming with shapes and ecology. Their initial baptism after 108 mythological characters from Norse mythology (Odin, Thor, Loki and Heimdall (3,4)) has 109 been unusually prescient (at least for Loki-and Heimdallarchaeia). The many-faced Loki

Competing Interests 127
The authors declare no competing financial interests in relation to the work described.

Supplementary methods
During in silico testing of the designed probes for specificity and group coverage, we identified two sequences that behaved aberrantly, i.e. could theoretically be regarded as outgroup hits for probe loki1-1184 (GU363076 and EU731577, Supplementary Table 2). However, these were discarded after closer examination because of pintail values of 20 and 0, respectively, indicating a chimeric origin (1). Pintail values are measures of chimeric nature for rRNA sequences in the databases, a value closer or equal to 100 indicating a nonchimeric sequence (1). Additional outgroup candidates for probe loki1-1184 (KU351219, AY133348, JQ817340, Supplementary Table 2) were initially located within Heimdallarchaeia in the basic tree provided by SILVA, but turned out to belong to Lokiarchaeia after alignment optimizations and RAxML tree reconstruction.
To be absolutely certain to avoid false-positive signals, i.e., to target no other organisms, we designed a set of competitor oligonucleotides that bind specifically to those rRNA sequences that have a single mismatch with our probes (2). We designed three distinct competitor probes for Heimdallarchaeota and two for Lokiarchaeota (Supplementary Table 1). Each competitor was used in the same concentrations as the CARD-FISH probes in order to prevent non-specific binding. The usage of specific probes together with competitors has been previously shown to work very well for visualizing cell-morphology and enumeration and was applied numerous times (3)(4)(5)(6)(7). Probes were tested in silico (8) and in the laboratory with different formamide concentrations in the hybridization buffer until stringent conditions were achieved (Supplementary Table 1).
We tested these probes in sediment samples from two sites from where recently several Asgardarchaeota genomes were recovered by metagenomics (Lakes Tekirghiol and Amara, Romania) (9). Sediment sampling was performed using a custom mud corer on 22 April 2018 at 12:00 in Tekirghiol Lake, Romania, (44°03.19017 N, 28°36.19083 E) and Amara Lake (44°36'N, 27°20'E, 23 April 2018, 14:00). Seven sediment layers (0-70 cm, in 10-cm ranges) were sampled in Lake Tekirghiol, and the top 10 cm was sampled in Lake Amara. Samples were fixed with formaldehyde for 1 h and washed three times with 1 x PBS, with centrifugation at 16000 g for 5 min between washes, and a final resuspension in a 1:1 mixture of PBS and ethanol. A treatment of sonication (20 sec, minimum power) on ice, vortexing and centrifugation to detach cells from sediment particles was applied (10) and aliquots diluted with PBS were filtered onto white polycarbonate filters (0.2 µm pore size, Millipore). CARD-FISH was conducted as previously described with fluorescein labelled tyramides and included permeabilization steps with lysozyme and achromopeptidase and an inactivation step for cellular peroxidases with methanol (10). Filters were counterstained with DAPI and inspected by epifluorescence microscopy (Zeiss Imager.M1) with filter sets for DAPI (filter set 01: BP 365/12, FT 395, LP 397), fluorescein (filter set 10: BP 450-490, FT 510, BP 515-565), and autofluorescence (filter set 15: BP 546/12, FT 580, LP 590). Micrographs of CARD-FISH stained cells were recorded with a highly sensitive chargecoupled device (CCD) camera (Vosskühler) at fixed exposure times (70 and 100 ms for DAPI, 100 and 200 ms for CARD-FISH and 100 and 400 ms for autofluorescence for magnifications of 400 x and 1000 x respectively) and cell sizes were estimated with the software LUCIA (Laboratory Imaging Prague, Czech Republic).