The uracil-auxotrophic strain S. acidocaldarius SK-1 (Suzuki and Kurosawa, 2016) and all derived mutant strains were cultivated in liquid in basic Brock medium (Brock et al., 1972) supplemented with 0.1% (w/v) N-Z-Amine, 0.2% (w/v) sucrose and 20 μg mL⁻¹ uracil, acidified to pH 3.0 with sulfuric acid. Liquid S. acidocaldarius cultures were incubated at 75°C while shaking. Growth was monitored by measuring optical density at 600 nm (OD₆₀₀). Cells were cultivated until reaching an OD₆₀₀ of between 0.4 and 0.5 (exponential growth phase) or OD₆₀₀ of 1.0 (stationary growth phase). Nutrient starvation was performed based on Haurat et al. (2017). In this case, cells were cultivated until reaching OD₆₀₀ 0.4 and collected by centrifugation at 4,000 g for 10 min at room temperature. The cell pellet was resuspended in fresh prewarmed Brock medium supplemented with 20 μg mL⁻¹ uracil lacking N-Z-Amine and sucrose and incubated at 75°C for 4 h while shaking.

For cultivation of S. acidocaldarius on solid medium, Brock medium was supplemented with 1.5 mM CaCl₂, 5 mM MgCl₂ and 0.6% Gelrite as a solidifying agent (Wagner et al., 2012). For selection of marker-containing integration mutants (“pop-in” step), the medium was solely supplemented with 0.1% (w/v) N-Z-Amine, 0.2% (w/v) sucrose, while for selection of markerless knock-in mutants (“pop-out” step), the medium was supplemented with 0.1% (w/v) N-Z-Amine, 0.2% (w/v) sucrose, 20 μg mL⁻¹ uracil and 200 μg mL⁻¹ 5-FOA. Solid medium was prewarmed to 75°C before performing plating (Wagner et al., 2012).

E. coli strain DH5α was used for cloning and for propagation of plasmid constructs and was cultivated in Lysogeny Broth (LB) medium supplemented with 50 μg mL⁻¹ ampicillin, if required. For selection of transformants from Golden Gate cloning, LB medium was used in which NaCl was omitted and 5% (w/v) sucrose was added. All strains used in this work are provided in Supplementary Table S1.

Gibson assembly (Gibson et al., 2009) was used to construct a Golden Gate destination vector named pYX2304 (Figure 1). To this end, the backbone of suicide plasmid vector pSVA431 (Wagner et al., 2012) was amplified using oligonucleotides YX123 and YX124 (Supplementary Table S2) and the sacB gene was amplified from the pRN1-carrier_2_paqci plasmid vector using primers YX125 and YX126. Polymerase Chain Reaction (PCR) amplification was performed with KAPA HiFi DNA polymerase (Roche) using 10 ng plasmid DNA and 20 pmol of each primer in a 50-μL reaction with following PCR conditions: 3 min at 95°C, 35 cycles of 20 s at 98°C, 15 s at 60°C and 1 min/kb at 72°C, followed by a final step of 5 min at 72°C. PCR amplification was verified by agarose gel electrophoresis and products were purified using the Wizard® SV Gel and PCR Clean-Up System (Promega). Next, Gibson assembly was performed by incubating 100 ng of amplified pSVA431 backbone with an equimolar amount of the sacB gene fragment in NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs) at 50°C for 1 h, followed by heat-shock transformation into chemically competent E. coli DH5α. Transformants were screened by colony PCR using primers YX127 and YX128 and the sequence of the obtained construct was verified by Sanger sequencing (Eurofins Genomics).

The pYX2304 destination vector was subsequently used to construct all suicide knock-in plasmid vectors with Golden Gate assembly (Engler et al., 2008). For each of the targeted genomic locations, primers (Supplementary Table S2) were designed to enable amplification of fragments of approximately 650 bp up- and downstream of the target site, respectively. These PCR reactions were performed using KAPA HiFi DNA polymerase (Roche) with 10 ng genomic DNA (gDNA) of S. acidocaldarius SK-1, which was extracted from liquid culture employing a QuickPick SML gDNA kit with magnetic bead purification (BioNobile). The lacS reporter cassette was PCR-amplified using KAPA HiFi DNA polymerase (Roche) with primers YX133 and YX134 and pYX2301-UPDD plasmid DNA as a template. The latter was derived from pSVA431 by replacing the PmalE promoter driving lacS expression with a synthetic constitutive promoter named P_{Ec10_Sa1}. This was realized via Gibson assembly, with the 56-bp P_{Ec10_Sa1} promoter sequence being introduced by primer YX091 (Supplementary Table S2). Golden Gate assembly was performed in 20-μL reactions consisting of 2 μL T4 ligase buffer (ThermoFisher), 1 μL T4 ligase (ThermoFisher), 1 μL PaqCI (Bioké), 0.3 μL PaqCI activator (Bioké), 100 ng pYX2304 plasmid DNA and equimolar amounts of the fragments. Reaction mixtures were subjected to following conditions: 50 cycles of 3 min at 37°C, 4 min at 16°C, followed by a final step of 5 min at 37°C and 5 min at 60°C. Subsequently, reaction mixes were heat-shock transformed into chemically competent E. coli DH5α and transformants were screened by colony PCR using customized primers (Supplementary Table S2). Finally, the sequence of the obtained construct was verified by Sanger sequencing (Eurofins Genomics). An overview of all plasmids constructed and used in this work is provided in Supplementary Table S3.

Construction of S. acidocaldarius markerless knock-in mutant strains was performed using a classical “pop-in pop-out” strategy as described (Wagner et al., 2012; Suzuki and Kurosawa, 2016) (Supplementary Figure S1). Competent S. acidocaldarius SK-1 cells were prepared by harvesting cells from a 100-mL culture at an OD₆₀₀ of 0.2 through centrifugation at 6574 g for 20 min, followed by two washing steps in 20 mM sucrose and a final resuspension of the cells in this 20 mM sucrose solution, reaching a final concentration of 2×10¹⁰ cells mL⁻¹. Genetic transformation was performed with a Gene Pulser II electroporator (Bio-Rad) at 1.5 kV, 25 mF and 600 W with Gene Pulser 1-mm cuvettes (Bio-Rad) (Wagner et al., 2012). After 5 days of incubation at 75°C on solid Brock medium lacking uracil, transformant colonies were screened for β-galactosidase activity by spraying with 5 mg mL⁻¹ 5-bromo-4-chloro-3-indolyl-beta-D-galacto-pyranoside (X-Gal) (ThermoFisher), followed by an incubation at 75°C for 30 min. Positive “blue” integrants were further confirmed by PCR analysis and were cultivated in liquid culture without addition of uracil. 50 μL of liquid culture was subsequently plated on Gelrite plates with uracil (20 μg mL⁻¹) and 5-FOA (200 μg mL⁻¹) for the second selection of “pop-out” of the vector backbone by homologous recombination and incubated for 5 days at 75°C. Transformant colonies were screened for β-galactosidase activity as described previously and successful integrants were confirmed through PCR analysis.

Culture samples of 4 mL were centrifuged for 15 min at 4000 g and pellets were stabilized using an equal volume of RNAprotect (Qiagen) and subjected to RNA extraction using an SV Total RNA Isolation System kit (Promega), followed by removal of residual genomic DNA using a Turbo DNase kit (Ambion Life Technologies). Next, cDNA was prepared from 1 μg RNA using a Go-Script Reverse Transcriptase kit (Promega). Three biological replicates were included in this experiment, except for the stress condition, which was conducted without replication.

Quantitative reverse transcriptase PCR (qRT-PCR) was performed to determine relative transcriptional gene expression levels of lacS employing tbp (saci_1336) as a reference gene. qRT-PCR primers were designed with Primer3 software (Untergasser et al., 2012) (Supplementary Table S2) and tested for efficiency using S. acidocaldarius gDNA as a template. 20-μL qRT-PCR reactions were performed in an iCycler qPCR device (Bio-Rad) with each reaction mixture containing 1 μL 10-fold diluted cDNA, 1.6 pmol of each primer and GoTaq qPCR master mix (Promega). The following PCR conditions were used: 3 min at 95°C and 40 cycles of 10 s at 95°C and 30 s at 55°C. Quantification cycle (C_T) values were determined with iQ5 software (Bio-Rad). For each gene and biological replicate, three technical replicates were performed, for which the average C_T value was determined. Relative expression levels were normalized with respect to the reference gene and ratios were calculated using the ΔΔC_T method relative to the strain displaying the lowest expression level. Graphpad Prism 9 was employed to perform a two tailed, one sample T test for statistical analysis.

Western blotting was performed based on the His-tagged LacS. To this end, 4 mL of S. acidocaldarius cells, cultivated until an OD₆₀₀ of 0.4, were pelleted by centrifugation at 4000 g for 15 min at 4°C. The pellet was resuspended in 200 μL extraction buffer (50 mM Tris–HCl, 50 mM NaCl, 15 mM MgCl₂, 1 mM DTT) to which 0.1% Triton X-100 was added. Cells were incubated at 4°C for 30 min while rotating for lysis. Total protein concentration was determined using Bradford Assay solution (TCI Chemicals), using BSA for the generation of the standard curve.

Normalized protein extracts (30 μg) were mixed with LDS (ThermoFisher) and subjected to sodium dodecyl sulfate poly-acrylamide gel electrophoresis (SDS-PAGE) using 4–12% NuPAGE™ Bis-Tris Mini Protein gels (ThermoFisher) and NuPAGE™ MES SDS Running Buffer (ThermoFisher). After SDS-PAGE, proteins were electroblotted onto a polyvinylidene difluoride (PVDF) membrane (Bio-Rad) using a Trans-Blot Turbo transfer system (Bio-Rad) operated at 1.3 A, up to 25 V for 7 min. The membrane was then incubated in Phosphate Buffer Saline (PBS) buffer with 5% (v/v) milk and 0.1% (v/v) Tween20 for 1 h at room temperature followed by an overnight incubation at 4°C with anti-His-tag mouse antibody (Proteintech, Cat No. 66005-1-Ig) diluted at 1:5000 in the same buffer. Excess primary antibody was washed away with PBS buffer containing 0.1% Tween20, followed by incubation with a secondary goat anti-mouse HRP-conjugated antibody (Proteintech, Cat No. SA00001-1) diluted at 1:5000 in the same buffer. After 1 h incubation at room temperature, unbound antibodies were washed away with PBS buffer containing 0.1% Tween20, twice, and PBS buffer for final wash. Visualization was performed with a Pierce™ ECL Western Blotting Substrate kit (ThermoFisher) and an ImageQuant800 imager (Cytiva). Band intensities were quantified using ImageJ (Schneider et al., 2012) as integrated density values (area × mean gray value). Values are reported in arbitrary units.

β-galactosidase assays were performed as described (Van der Kolk et al., 2020) with some modifications. The σ-nitrophenyl-β-D-galactopyranoside (ONPG) conversion rate was measured spectrophotometrically at 410 nm in a microplate reader (Infinite M Nano+, TECAN) every 5 min, while being incubated at 42°C for 4 h. The slope of the conversion curves was taken as the value for β-galactosidase activity, which is a proxy for LacS expression. For each sample, the assay was performed in triplicate. The slopes were determined by fitting a Monod-type model with smoothing to the conversion data:

where sl = slope, m = maximum value, t = time, t_off = offset and d = period around the offset that the transition takes place. The slope was determined by fitting the model to ranges of datapoints from t₀-t_max to t₀-t₅. Local minima of the fit error of these fits were then searched and from those fits, the slope of the fit resulting in the slope with the lowest standard error was selected. When the resulting fit was not deemed adequate, the average value of the fitted d parameter of the other technical and/or biological replicates was used as fixed value for a subsequent fitting. The geometric mean of the slopes was calculated, together with the geometric standard error. If the fit error or error from the technical replicates was larger than the geometric standard error, that one was used instead.

In this study, we used a knock-in suicide plasmid vector that is based on the pSVA431 plasmid vector harboring a S. solfataricus pyrEF gene cassette and a pGEM-T Easy cloning vector containing an ampicillin resistance cassette and f1 origin of replication for E. coli (Wagner et al., 2012). To facilitate the construction of knock-in suicide vectors, we have chosen to employ a one-step Golden Gate assembly approach. To this end, pSVA431 was converted into a Golden Gate destination vector named pYX2304 (Figure 1a). This vector combines the pSVA431 backbone with a sacB gene cassette flanked by type IIS restriction sites. The sacB gene encodes levansucrase, an enzyme that converts sucrose into levans, which accumulates in the E. coli periplasm causing toxicity (Gay et al., 1985). During cloning, it can thus be used as a negative selection marker in presence of sucrose, enabling the selection of transformants harboring a vector in which the sacB cassette is excised and replaced by the fused lacS cassette and target up- and downstream regions (Supplementary Figure S2).

The designed knock-in suicide plasmid vector (Figure 1b) can be used to create markerless knock-in mutant strains with a single-crossover recombination approach, employing the pyrEF cassette as a selection marker (Wagner et al., 2012) and the uracil auxotrophic S. acidocaldarius SK-1 (Suzuki and Kurosawa, 2016) as a host strain (Supplementary Figure S1). S. acidocaldarius SK-1 is derived from S. acidocaldarius M31, which harbors a 31-bp deletion in the pyrEF region (Reilly and Grogan, 2001), and was further engineered by deleting the suaI endonuclease-encoding gene, thereby inactivating its restriction-modification system (Suzuki and Kurosawa, 2016). As compared to S. acidocaldarius MW001, SK-1 offers the advantage that methylation of the plasmid DNA is not required, thereby simplifying the genetic transformation procedure. Otherwise, given its uracil auxotrophy, the use of SK-1 as a host strain enables a classical “pop-in pop-out” scheme in which, in a first selection step directly following transformation, cultivation on solid medium lacking uracil results in selection of integrant mutant strains, with the entire plasmid vector being integrated into the genome (“pop-in” step) (Wagner et al., 2012). Successfully obtained integrant mutant strains are then subjected to a second selection step, which entails a counterselection because of the presence of uracil and 5-FOA, resulting in the selection of strains that have undergone a second single-crossover recombination event (“pop-out” step), leading either to the wild-type genomic sequence or to the desired knock-in mutant sequence (Supplementary Figure S1).

The lacS reporter cassette consisted of a lacS gene from S. solfataricus, previously validated as a reporter system in S. acidocaldarius (Berkner et al., 2007), fused to a C-terminal 6xHis-tag and under control of a synthetic Sulfolobus promoter named P_{Ec10_Sa1}, which is based on the native Saci_2137 promoter, driving expression of a putative aminotransferase (Liu et al., 2014). This in-house developed promoter has a small size of 56 bp (Supplementary Table S1), thereby contributing to a minimal vector size, as well as a constitutive expression profile of moderate strength and a non-native sequence, which helps to avoid unwanted homologous recombination with the genome.

We selected 11 target positions for insertion within the S. acidocaldarius SK-1 genome, based on (i) variations in the relative distance of the site to the nearest origin of replication, (ii) transcriptional expression levels of nearby or overlapping genes under physiological conditions (Baes et al., 2023) and (iii) predicted essentiality assessed based on homology with S. islandicus genes (Zhang et al., 2018) (Figure 2; Table 1). Additionally, the target sites were chosen to be evenly distributed between the two chromosomal compartments A and B and for this reason, target sites were selected in two chromosomal segments: between oriC1 and oriC2 and between oriC1 and oriC3 (Figure 2; Table 1). In all cases with one exception (cmp), the reporter gene cassette was specifically targeted to intergenic regions to minimize unintended effects on the expression of adjacent genes (Figure 3; Supplementary Figure S3). Intergenic regions include those located between genes transcribed in the same direction, whether part of an operon (e.g., ccc1) or not (e.g., sdhC), as well as between convergently (e.g., cmp) or divergently transcribed genes (e.g., acad). The insertion positions and corresponding knock-in strains were named based on the nearest gene (Table 1).

All genetic constructs were transformed into S. acidocaldarius SK-1 followed by a selection in uracil-free medium, yielding high numbers of colony-forming units (CFUs). Upon X-gal treatment, enabling colorimetric detection of lacS⁺ colonies, it appeared that only a small fraction of CFUs represented “pop-in” lacS⁺ integrants (Figure 4a). The nature of the “background” lacS⁻ CFUs is unknown. On average, the fraction of “pop-in” lacS⁺ CFUs was higher for locations in compartment A as compared to compartment B (an average of 0.56% versus 0.22%) (Figure 4a). Transformations that yielded a higher “pop-in” lacS⁺ integrant fraction, such as for slaA and ccc1 knock-in constructs, were possibly characterized by more a favorable genomic recombination as compared to transformations with knock-in constructs like acad or arlJ.

Purified “pop-in” integrant strains were subsequently cultivated on 5-FOA- and uracil-containing plates to select for “pop-out” lacS⁺ integrant strains, which still contained the lacS cassette, but with the remainder of the vector removed by a crossover event (Figure 1b). For all knock-in constructions, with the exception of sdhC and cmp, “pop-out” lacS⁺ integrant strains were easily obtained and the correct integration of the lacS cassette was verified by colony PCR (Figure 4b). In contrast, for the construction of the acs and cmp knock-in strains, colony PCR revealed partial or full presence of the wildtype genotype and despite multiple selection attempts, we failed in obtaining these knock-in strains. For the cmp knock-in strain this could be possibly attributed to disruption of expression of the cmp gene itself, which is a highly essential gene (Table 1).

A qRT-PCR approach was used to evaluate transcriptional expression levels of the lacS reporter gene in the different knock-in strains (Figure 5). Although all strains were cultivated in identical conditions and despite lacS being under control of the same constitutive promoter, large variations in transcriptional levels were observed (Figure 5a). Relative transcription is calculated with respect to the knock-in strain displaying the lowest lacS transcription, arlJ.

Transcriptional levels were observed to be highest in the slaA knock-in strain, 55-fold higher than in the arlJ knock-in strain. This could be attributed to the target site position, immediately downstream of the slaA stop codon, making it highly probable that slaA transcription, which is known to be constitutively expressed at high level (Baes et al., 2023), runs through to the lacS cassette. The same applies for other mutant strains that display significantly higher transcription levels with respect to the arlJ knock-in strain (Figure 5a). For the vapC knock-in strain, in which lacS is transcribed at a 6.2-fold higher level, the lacS cassette was integrated downstream of the vapC gene preceding a putative terminator site (Figure 5a; Supplementary Figure S3). For the ccc1 knock-in strain, in which lacS is transcribed 5.2-fold higher than in the arlJ knock-in strain, the target site is located in between two operonic genes (Figure 5a; Supplementary Figure S3).

Only in the clsN and acad knock-in strains, the target sites are located in regions where no read-through transcription from adjacent genomically encoded transcription units is expected, namely in an intergenic region in between the two divergently oriented putative promoters (Figure 3; Supplementary Figure S3). For these strains, transcriptional expression of lacS is expected to be driven solely by the P_{Ec10_Sa1} promoter and indeed, both strains have very similar transcriptional levels with respect to the arlJ knock-in strain, namely 1.88-fold and 1.86-fold for clsN and acad, respectively (Figure 5a).

Although a trend was observed of compartment A-targeted knock-in strains having higher transcriptional expression levels than compartment B-targeted knock-in strains, this cannot be linked to a global effect, but rather to the expression levels of adjacent genes, which are not insulated from the lacS cassette. To a certain extent, a correlation was observed between previously detected expression levels of the adjacent gene (Baes et al., 2023) expected to cause read-through transcription and the relative lacS transcriptional level (Supplementary Table S4). Given that for the two “insulated” knock-in strains clsN and acad, which both have similar transcriptional expression levels and of which the clsN target site is located in compartment A and the acad target site in compartment B, a global gene silencing effect was not apparent.

Transcriptional expression of lacS was also monitored in response to stress conditions, either nutritional starvation (Figure 5b) or stationary phase growth (Figure 5c). Although the P_{Ec10_Sa1} promoter is expected to drive transcription of lacS in a constitutive manner, differential transcription was observed in several knock-in strains, either positively or negatively. Trends are similar for both nutritional starvation and stationary growth phase. In the slaA and vapC knock-in strains, a transcriptional downregulation of lacS is observed, while in the sdhC and arlB knock-in strain, an upregulation is observed (Figures 5b,c). Upon comparing the regulatory effects on lacS expression with the differential transcriptional patterns of nearby genes observed in previous transcriptomic datasets under nutritional starvation and stationary phase conditions (Supplementary Tables S5, S6) (Bischof et al., 2018; Takemata et al., 2019), indications of read-through effects were found in some but not all “non-insulated” knock-in strains. For example, in the arlB knock-in strain, the cassette is inserted directly downstream of the arlB ORF (saci_1178) (Figure 3; Supplementary Figure S3) and an upregulation of lacS is observed (Figures 5b,c), consistent with the known upregulation of arlB under nutritional starvation and stationary phase conditions (Supplementary Tables S5, S6). A similarly clear correlation was found for the ccc1 and sdhC knock-in strains. In contrast, for the slaA, vapC, l2p, and arlJ knock-in strains differential expression of lacS was not entirely correlated with differential expression of adjacent genes expected to cause read-through expression in both transcriptomic datasets (Figures 5b,c, Supplementary Tables S5, S6). In almost all cases, there was a lack of correlation in differential expression observed for the gene of interest either under nutritional starvation or in stationary phase growth (Supplementary Tables S5, S6). Moreover, in the arlJ knock-in strain, in which the lacS cassette is integrated into the intergenic region between the converging 3′ ends of the arlJ (saci_1172) and arnR1 (saci_1171) genes (Lassak et al., 2013) (Figure 3), it is unclear which of the two genes, if any, lacS expression would be expected to correlate with, especially since no obvious transcriptional terminator sequences were identified within this region (Supplementary Figure S3).

With exception of lacS expression in the acad knock-in strain under nutritional starvation, for which an upregulation was observed, little or no transcriptional regulation was observed for the “insulated” knock-in strains clsN and acad (Figures 5b,c). This insulation is further strengthened by the observation that several adjacent genes (saci_0047 for clsN and saci_1123 and saci_1124 for acad) display transcriptional regulation under nutritional starvation and stationary phase conditions (Supplementary Tables S5, S6).

To assess the abundance and activity of LacS, which is expressed as a His-tagged protein, on the protein level, we performed anti-6xHis western blotting and ONPG-based LacS activity assays (Figure 6). Both approaches demonstrated a variability among the knock-in strains with compartment A strains showing a relatively higher average activity as compared to compartment B strains (an average of 2.412 × 10⁻⁴ versus 1.419 × 10⁻⁴) (Figure 6b), with the clsN and slaA knock-in strains seemingly exhibiting the highest activity, together with the vapC knock-in strain.

An apparent higher translational activity in compartment A versus compartment B knock-in strains might be, at least partially, reflecting differences in transcriptional activities, which follow the same trend (Figure 5a). The observed differences in translational activity for the different strains were less pronounced as those observed at the transcriptional level. For instance, while compartment A strains, such as the clsN and slaA knock-in strains, exhibited slightly higher protein activities as compared to compartment B strains, such as the arlJ knock-in strain, the variation in protein activity was relatively modest. Moreover, the correlation between transcriptional and translational levels was very limited with a R² of 0.1296 and a lack of statistical significance (p value = 0.3413) (Figure 7a). Interestingly, a somewhat larger correlation can be found for protein activity and distance to the closest origin of replication (R² = 0.432) than for transcriptional expression (R² = 0.3471), although with low statistical significance (p values of 0.0543 and 0.0951, respectively) (Figures 7b,c).