Cells were grown in 16×25 mm glass tubes with 3 mL of Hv-Cab, a semi-defined medium based on casamino acids that improve growth and cell shape maintenance (de Silva et al., 2021). Cultures were incubated at 42°C under constant agitation with the inducers D-xylose (Thermo Scientific Chemicals), L-arabinose (Thermo Scientific Chemicals), IPTG (Fisher Bioreagents), D-maltose (Fisher Bioreagents), or L-tryptophan (Thermo Scientific Chemicals) in the concentrations indicated. Plasmid and strain list can be found in Tables 1, 2.

All oligos used to make our constructs can be found in Table 3. The eTR8 vector was constructed by Gibson assembly (Gibson et al., 2009) from two PCR fragments using the oligos oTR26 and oTR27 (msfGFP) and a linearized pTA962 digested with NdeI.

The eAD08 vector was constructed by Gibson assembly from two PCR fragments using the oligos oBL340 and oJM96 (to amplify the msfGFP fragment) and the linearized pTA962 digested with NcoI.

The pAL250::msfGFP and pAL750::msfGFP vectors were constructed by Gibson assembly from three PCR fragments using the oligos oBL169 and oBL345 (msfGFP, eTR8 as a template), oBL343 and oBL344 (the first 250 bp upstream to the HVO_B0027 start codon using the H. volcanii strain DS2 gDNA as a template) or oBL343 and oBL345 (the first 750 bp upstream to the HVO_B0027 start codon using the H. volcanii strain DS2 gDNA as a template), and a linearized pTA962 previously digested with KpnI and BamHI.

The pAL750 vector was created by Gibson assembly from three PCR fragments using the pAL::msfGFP vector as a template and the oligos oZC23 and oTR150 (pHV2 ori), oTR151 and oSB33 (pyrE2-Pxyl₇₅₀ region), and oBL97 and oBL105 (E. coli oriC and AmpR cassette).

The pAL750::ftsZ1 and pAL750::ftsZ2 vectors were cloned by Gibson assembly of two fragments using the oligos oTR151 and oTR152 (ftsZ1) or oTR153 and oTR154 (ftsZ2) using the H. volcanii strain DS2 gDNA as a template, and a linear fragment of the pAL vector digested with NdeI.

All Gibson reactions were transformed into competent E. coli DH5ɑ cells and clones confirmed by whole-plasmid sequencing (Plasmidsaurus). Plasmid preps were then transformed into H. volcanii using the method previously described (Dyall-Smith, 2009), with 0.5 M EDTA (Thermo Scientific Catalog #J15694-AE) and PEG600 (Sigma, catalog # 87333-250G-F).

Cells were grown to an OD₆₀₀ of ~0.5 and diluted to an OD₆₀₀ of 0.05. Then, 200 μL of culture was placed in a 96-well flat-bottom plate (Corning Inc.). All wells surrounding the plate’s edge (A and H rows, 1 and 12 columns) were filled with 200 μL of water to prevent media evaporation. Growth curves of three biological triplicates were performed using an EPOCH 2 microplate spectrophotometer (Agilent) with constant orbital shaking at 42°C. Data points were collected every 30 min.

Cells were grown in Hv-Cab (de Silva et al., 2021) or YPC medium (Allers et al., 2004) to mid-exponential or stationary OD₆₀₀ without or with inducers as indicated. Cultures were then concentrated 10-fold by centrifuging (3,000xg for 2–5 min), and a 3 μL droplet of culture was placed on a 60×24 mm coverslip and gently covered with a 1.5×1.5 cm 0.25% Hv-Cab agarose pad (SeaKem LE Agarose, Lonza). Staining with Ethidium Bromide (Invitrogen, catalog #15585–011) was performed by adding 3 μg/mL of the dye into the cell culture and incubating at 42°C for 5 min. Cells were then imaged at 42°C using a Nikon TI-2 Nikon Inverted Microscope within an Okolab H201 enclosure. Phase-contrast and GFP-fluorescence images were acquired with a Hamamatsu ORCA Flash 4.0 v3 sCMOS Camera (6.5 μm/pixel), a CFI PlanApo Lambda 100x DM Ph3 Objective, and a Lumencor Sola II Fluorescent LED (380–760 nm). Image analysis was performed using FIJI (Schindelin et al., 2012). Cells were segmented using the rolling ball background subtraction (value = 20) on phase-contrast images, followed by thresholding (default) and using the analyze particles function with a minimum particle size of 0.2 μm². The mask was then used to acquire shape descriptor data and applied to the GFP channel (after background subtraction) for fluorescence quantification.

All steps were performed at room temperature unless otherwise indicated. Cells were grown to an OD₆₀₀ of ~0.5 in the presence of xylose (10 mM), arabinose (10 mM), IPTG (1 mM), and maltose (1 mM) and harvested by centrifugation (4,500xg for 10 min). RNA extraction was performed using 1 mL of TriZol (Invitrogen) per sample, followed by vigorous pipetting to lyse all cells in the sample. 200 μL of chloroform was added to the sample mixture, and samples were vortexed for 90 s. Samples were then centrifuged at 12,000xg at 4°C for 15 min, and the supernatant was collected and mixed with 2 μL of glycogen (Sigma Aldrich) and 400 μL of isopropanol (Thermo Scientific Chemicals). Samples were incubated overnight at −20°C. Samples were centrifuged at 21,000xg, 4°C for 30 min and washed twice with 75% ethanol (Zaffagni et al., 2022). Samples were treated with DNAse I (NEB) for 12 min at 37°C, followed by a second ethanol precipitation. Purified RNA was sent to SeqCenter LLC (Pittsburgh, United States, seqcenter.com) for ribosome depletion using Haloferax-specific probes (Supplementary Table S1) and sequencing. Results were mapped to the H. volcanii DS2 genome and analyzed using Geneious 2002.2. Transcripts per million separated by ORF and inducer can be found in Supplementary Table S2. The complete raw RNA-seq datasets presented in this study can be found in NCBI GEO online repositories: PRJNA953041 (no induction), PRJNA953037 (xylose), PRJNA953035 (IPTG), PRJNA953033 (Arabinose), PRJNA953034 (Maltose).

To find native inducible promoters, we investigated four different sugars frequently used as inducers in other microbial models: arabinose (Guzman et al., 1995), xylose (Kim et al., 1996), maltose (Ming-Ming et al., 2006), and the sugar analog IPTG (Dubendorff and Studier, 1991). To determine the highest sugar concentrations we could use in H. volcanii cultures without compromising growth rates, cell size, and morphology, we titrated each sugar from 1 mM to 100 mM. Based on concentrations used to induce promoters in bacteria, we expected that concentrations higher than 1 mM could yield slow-growing, smaller cells. Surprisingly, we could only observe this outcome from cultures above 10 mM of xylose (Figures 1A,B). Therefore, we focused on concentrations at 10 mM.

Next, we employed RNA-seq from mid-exponential cultures with or without each of the four inducers to map candidates for new inducible promoters. Using a set of specific probes for H. volcanii (Supplementary Table S1) (Pastor et al., 2022), the total ribosomal RNA detected in our samples was between 0.21 to 0.37% of the total number of reads from each sample. By comparing the relative fold change of transcripts per million, we obtained 58 genes from which mRNA levels were significantly upregulated (log₂ fold change ≥1 and value of p ≤0.05) for xylose, 42 genes upregulated by maltose, 0 for IPTG, and 45 genes upregulated for arabinose. We identified 15 downregulated genes (log₂ fold change ≤ −1 and value of p ≤0.05) for xylose, 44 genes for maltose, 3 for IPTG, and 10 for arabinose (Supplementary Table S2).

To create a shortlist of promoter candidates, we could further validate experimentally, we arbitrarily selected genes above 5-fold or higher ratio change upon the addition of the inducer (Figure 1C). Above this threshold, we assigned 1 promoter for maltose (HVO_0562–64); 4 for xylose (HVO_B0027-29, HVO_B0030-32, HVO_B0035, and HVO_B0036-38); 3 for arabinose (HVO_B0027-29, HVO_B0030-32, and HVO_B0035; and none for IPTG (Figures 1C,D). In the first two cases, candidate gene expression was upregulated upon adding xylose or arabinose (regions 1 and 2, Figure 1D). In contrast, the genomic region comprising gene HVO_B106 did not pass our requirements; it called our attention for being arabinose-specific and not being significantly upregulated under xylose addition (Figure 1C, orange arrowhead). Likewise, genes HVO_0562-HVO_0566 are specifically upregulated under maltose induction (Figure 1C, green arrowhead), and three regions (HVO_A0173, HVO_B0032, and HVO_B0303) respond to IPTG (Figure 1C, pink arrowheads).

Despite observing an approximately 10-fold increase in mRNA levels upon inducer addition, we wanted to ensure these genes were tightly controlled by an inducible promoter with precise linear titration power and not already constitutively expressed at a high basal expression level. To understand the basal expression levels of genes, including regions 1 and 2, we plotted the raw transcription profile (Figure 1E, left panel). We compared it to the transcriptional levels of all other genes (Figure 1E, right panel). As expected, after ribosomal RNA depletion, the transcripts with higher relative values in our datasets were the ones mapping to the S-layer glycoprotein (csg, over 26,000 TPM - Transcripts Per Million - across samples) and the translational factor EF1a transcripts (above 10,000 TPM across samples). On the other hand, both promoter regions 1 and 2 transcriptional levels are placed in the low quartile range of our dataset in the uninduced sample (1.44 TPM and 23.65 TPM, respectively).

Altogether, genomic regions 1 and 2 are promising candidates for new xylose- and arabinose-inducible promoters for H. volcanii. For the context of this work, and based on the dynamic range suggested by our RNA-seq dataset above, we focused on the characterization of the promoter regulating the gene xacE (HVO_B0027), or genomic region 1, under the induction of xylose.

To test the xacE promoter region, we sub-cloned the fluorescent protein msfGFP under the control of two different putative promoter regions: 250 bp (P250) and 750 bp (P750) upstream to xacE (Figure 2A). Plasmids were then transformed into H. volcanii H26 (ΔpyrE2), and selected on plates without uracil. As a negative control, we used the H26 strain transformed with the empty pAL750 vector (labeled from now on as wild type). As a comparison, we analyzed the expression of msfGFP under the control of the popular inducible promoter PtnaA, which activates upon tryptophan addition (Allers et al., 2010). As a benchmark, we also inserted msfGFP under the control of PfdX, a strong and constitutive promoter used to express the pyrE2 gene in our vectors as a selective transformation marker (PfdX-msfGFP-pyrE2 operon).

Imaging of live cells by phase-contrast and epifluorescence microscopy showed cytoplasmic signal emitted by the msfGFP fluorescent protein from single cells (Figure 2B). Cells carrying empty plasmids (Figure 2B, first column) showed low auto-fluorescence at 488 nm excitation compared to cells carrying vectors inducing msfGFP under the control of PfdX and Pxyl promoters. For a quantitative picture of the induction power of each promoter, we segmented individual cells and measured the mean fluorescence per cell from three biological triplicates and graphed using SuperPlots (Lord et al., 2020; Goedhart, 2021). Surprisingly, the P250 promoter was insensitive to xylose and constitutively expressed 2-fold above the PfdX control (10,048 ± 2,256 and 4,967 ± 929 AU, respectively) (Figure 2C).

In contrast, the P750 promoter harboring the whole upstream intergenic sequence showed approximately a 5.5-fold repression (1,827 ± 848) without induction compared to P250. However, msfGFP levels of non-induced P750 cells were still relatively high, 5.7-fold higher than uninduced PtnaA cells (320 ± 14) (Figures 2B,C). To minimize the transcriptional leakage from P750, we tested whether H. volcanii cells would present catabolite repression upon adding glucose. This strategy has been successful in various bacterial and yeast systems (Gancedo, 1998; Deutscher, 2008). Adding 20 mM glucose to cultures decreased leakage of the Pxyl promoter, but msfGFP intensity is still 2.7-fold higher than PtnaA cells (Figure 2D).

However, different from the P250 promoter, adding 10 mM xylose to P750 cells resulted in a 10.3-fold increase (18,792 ± 4,129) in fluorescence intensity with a wider heterogeneity across the population compared to P250. Nevertheless, the unusual decrease in msfGFP expression observed between P250 and P750 suggests that xacE’s upstream region between 250 bp and 750 bp might have important regulatory elements.

We also inspected the dynamic range of our P750 promoter compared to PtnaA. By titrating xylose (0 to 25 μM) and tryptophan (0 to 2 mM), we observed a significant improvement from an 11.2-fold linear range for the PtnaA promoter to a 23-fold for the P750 promoter (Figure 2E). Providing the relatively higher leakage of the P750 promoter levels compared to PtnaA, we concluded that this new construct is a valuable tool for titration experiments and overexpression at high protein levels for functional studies and protein purification directly from H. volcanii. Notably, P250 could be used simultaneously with the above inducible promoters as another constitutive promoter in the Haloferax community.

Furthermore, we tested the ability to induce the P750 promoter in cells growing in Hv-YPC, a rich medium based on yeast extract instead of casamino acids (Allers et al., 2004). In contrast with Hv-Cab, cells at mid-exponential growth phase showed lower levels of promoter leakage (222 ± 188), 3.6-fold higher than in Hv-Cab (Figure 3). However, together with the tighter expression of P750, induced cells in Hv-YPC expressed msfGFP at 7.6 times lower than in Hv-Cab (1,636 ± 1,258), possibly due to catabolic repression in response to a component in yeast extract.

In addition to Hv-YPC, we also measured the expression profile of P750 at stationary growth phase and upon the addition of arabinose instead of xylose (Figure 3). While we observed a similar induction profile from cells in stationary phase and under arabinose induction, both cases showed a wider signal distribution (9,851 ± 617 and 11,699 ± 797, respectively) compared to xylose-based induction during exponential growth (12,395 ± 800).

Recently, Liao and colleagues reported the role of two tubulin paralogs (FtsZ1 and FtsZ2) in H. volcanii’s cell division (Liao et al., 2021). The authors used PtnaA-controlled overexpression of FtsZ1 and FtsZ2 independently and observed distinct, specific morphological phenotypes related to the overexpression of each paralog. Cells under PtnaA-FtsZ1 overexpression were slightly larger but significantly misshaped compared to the control, whereas PtnaA-FtsZ2 cells looked significantly smaller but showed a more consistent morphology.

To confirm if those phenotypes are reproducible or even enhanced in our new Pxyl system, we cloned ftsZ1 and ftsZ2 in the pAL vector and analyzed the cell size and circularity compared to PtnaA-induced cells. Cells overexpressing ftsZ2 under the Pxyl promoter are slightly smaller (1.2-fold decrease in average cell area) than cells overexpressing ftsZ2 under the tryptophan-inducible PtnaA (Figure 4A). Meanwhile, cells overexpressing ftsZ1 under the Pxyl promoter seemed to have more drastic phenotypes than those overexpressing ftsZ1 using PtnaA, with deformed and enlarged cells (Figures 4A,C). Curiously, a fraction of the population exhibits narrow “cell bridges” connecting two enlarged cells (Figure 4B), similar to those previously described (Rosenshine et al., 1989; Sivabalasarma et al., 2021; Von Kügelgen et al., 2021). This indicates that, in our samples, the cell-to-cell bridges observed in phase contrast could be formed by a product of incomplete cell division.