To elucidate the function of MinD2 in H. volcanii, we started by characterizing the growth of a mutant strain lacking MinD2 (∆minD2). Growth rate measurements showed that the ∆minD2 strain exhibited a growth pattern similar to that of the wild-type (WT) strain H26 (Supplementary Figure 1), suggesting that MinD2 does not directly influence growth in H. volcanii as previously shown by Nußbaum et al. (2020).

Then, we assessed the impact of MinD2 on morphology, using phase contrast microscopy to compare the cell shape and size of WT vs. ∆minD2 cells during different growth stages. As previously described, we found that WT H. volcanii (H26) undergoes growth-dependent alterations in its cell morphology. The time point of this shape transition has been observed to differ greatly based on medium components and is also affected by the presence of plasmids (Li et al., 2019; de Silva et al., 2021; Patro et al., 2023). During the early log phase, the liquid culture is predominantly composed of rod-shaped cells (R), which gradually transition into an intermediate state (I) and then into plate-shaped cells (P) as the optical density (OD₆₀₀) of the culture increases. The categorization of cell shape was conducted as previously described in Patro et al. (2023). Indeed, H26 (WT) cells predominantly exhibited rod-shaped morphology until an OD₆₀₀ of 0.03, with a minority showing I or P shapes (R = 41%; I = 26%; P = 33%) (Figure 1 and Supplementary Figure 2a). Starting from an OD₆₀₀ of 0.06, a noticeable transition towards P cells became evident, with plates representing the majority (P = 55%). This trend persisted as the culture progressed, with the frequency of P cells reaching 81% at an OD₆₀₀ of 0.2. By contrast, the majority of ∆minD2 cells exhibited a plate-shaped morphology even in the early log phase (OD₆₀₀ 0.01), with few I and R cells (R = 17%; I = 33%; P = 50%) (Figure 1 and Supplementary Figure 2a). The frequency of P cells continued to increase at higher culture densities, with more than 80% of the cell population adopting a plate-shaped morphology at OD₆₀₀ above 0.06 (Figure 1 and Supplementary Figure 1).

Additionally, a comparison of the cell area in both strains displayed a similar trend, with ∆minD2 cells having significantly smaller areas than WT cells at an OD₆₀₀ of 0.01 (Supplementary Figure 2b). Nevertheless, the areas of both WT and ∆minD2 cells were comparable at higher densities (OD_{600 ≥} 0.03).

Collectively, these analyses demonstrate that while MinD2 does not affect cell growth, it significantly influences cell shape, with the deletion of minD2 resulting in the loss of the ability to maintain a rod-shaped morphology, particularly in the early growth phase.

Previously, we found that the presence of a plasmid played a transient role in maintaining rod-shaped morphology in H. volcanii (Patro et al., 2023). To test this, we looked at the effects of deleting the MinD2 gene on cell shape and size, with or without the empty plasmid pTA1392. In the WT (H26) cells, when the plasmid was present, the cells changed from rods to plates between OD₆₀₀ 0.06 and 0.1. Without the plasmid, this switch happened earlier, between OD₆₀₀ 0.03 and 0.06 (Supplementary Figure 1b; Patro et al., 2023). Similarly, in MinD2 mutants, more cells were rod-shaped at OD₆₀₀ 0.01 when pTA1392 was present. Without the plasmid, these mutant cells were mostly plate-shaped at OD₆₀₀ 0.01. In MinD2 mutants with pTA1392, cells shifted to an intermediate shape (39%) at OD₆₀₀ 0.03 and mostly plate-shaped (40%) at OD₆₀₀ 0.06. This shape change happened one generation earlier in the MinD2 mutants with pTA1392 compared to WT cells with the plasmid. These results highlight MinD2’s role in modulating cell shape, especially in becoming plate-shaped. The plasmid increases the number of rod-shaped cells in the MinD2 mutant, promoting rod development. Furthermore, while cells that contained plasmids maintained rod-shaped morphology longer throughout the different growth phases, the absence of minD2 still reduced the ability to maintain a rod-shaped morphology throughout all phases (Supplementary Figures 1b, 2c,d).

In H. volcanii, cell shape is linked to swimming motility. During the early logarithmic phase, rod-shaped H. volcanii cells exhibit motility, powered by the assembly of polar bundles of archaella (Li et al., 2019). However, as cells transition into plate shape during stationary phase, the archaellum filaments are lost, resulting in motility cessation. Given the observed impact of MinD2 deletion on cell shape, we investigated whether the ∆minD2 also showed changes in archaella assembly and motility. For this, we used transmission electron microscopy (TEM) to characterize WT and ∆minD2 cells collected at an early log phase (OD₆₀₀ 0.03). However, H. volcanii cells express several pili throughout different growth phases (Esquivel et al., 2016), which are difficult to differentiate in diameter and size from archaella. Therefore, we further deleted pilB3, the pilus assembly ATPase, in both the H26 and ∆minD2 strain. These ∆pilB3 mutants lack pili, which facilitates visualization of archaella in these strains.

The results revealed that the rod-shaped H26ΔpilB3 cells, both with and without plasmid pTA1392, displayed archaellation in 80 and 83.7% of the cells, respectively (Figures 2a,c). By contrast, the ∆minD2ΔpilB3 strain, characterized by a discoid-shaped phenotype, only the few rod-shaped cells observed showed archaellation (~20% of the cells). In the presence of pTA1392, where the percentage of rod-shaped cells is slightly increased (24%; Supplementary Figures 2a,b), the percentage of archaellated cells is also observed to increase to ~32% (Figures 2b,c).

Given the observed effects of MinD2 deletion on archaellation levels, we investigated whether this impacted the motility of the ∆minD2 strain, by growing the cells on soft agar plates. First, we found that ∆minD2 strain grown in nutrient-rich YPC media showed reduced motility compared to H26 (data not shown). However, due to the composition of the medium, cell growth was rapid, making it difficult to differentiate growth from swimming halos in soft agar plates. Therefore, we switched to a different medium (CA), where strains grow slower. In order to support growth on CA medium, which lacks uracyl, strains were transformed with a plasmid (pTA1392) containing the pyrE2 locus (which enables uracyl biosynthesis). Using this strategy, we observed that ∆minD2 + pTA1392 cells exhibited a swimming defect, with their motility being only ~25% of what was observed for H26 + pTA1392 cells (Figure 2d). We were able to complement the ∆minD2 motility defect by expression of MinD2 on plasmid pTA1392 (pSVA6011, Figure 3). Additionally, the observed reduction in motility in ∆minD2 was consistent with the lower number of mutant cells assembling archaella.

The hallmark of ParA/MinD superfamily proteins is the Walker A (WA) and Walker B (WB) motif (Leipe et al., 2002; Hu and Lutkenhaus, 2003). Mutation of both the motif in MinD4 had previously displayed motility phenotype and disrupted MinD4’s localization at poles (Nußbaum et al., 2020). We wanted to investigate the role of the WA and WB motifs of MinD2 by generating plasmids with WA (K16A) mutant and WB (D117A). Expressing these in ∆the minD2 strain, we observed that mutation of WA did not have an effect on the swimming motility, and the strain had 100% activity as visualized in the wild type (H26). Whereas the WB mutant showed a defect in motility (~30%) similar to ∆the minD2 deletion mutant strain (Supplementary Figure 4).

Thus, these results confirm that MinD2 not only modulates cells shape but also impacts cell motility, and that Walker B motif of MinD2 plays a role in the motility phenotype.

The observed impact of MinD2 on motility is reminiscent of the role of MinD4, which we previously showed to be a MinD homologue that oscillates along the cell axis in H. volcanii and stimulates the formation of chemosensory arrays and archaella at the cell poles, thereby regulating motility (Nußbaum et al., 2020). Therefore, we proceeded to examine the localization pattern of MinD2, by generating fluorescent fusion proteins. For this, the MinD2 protein was tagged at the N-or C-terminus with various tags and linkers, which were then expressed in the ΔminD2 deletion mutant. To test whether the fluorescent tag had an impact on the function of MinD2, we performed complementation experiments and assessed their motility using motility assays (Figure 3b). Notably, when a tag-less MinD2 variant was used (pSVA6011), the swimming phenotype could be restored to ~91% of the activity observed in WT cells. Expression of an N-terminal His-tag MinD2 variant (pSVA6010) restored ~62% of the motility activity, whereas the strain expressing an N-terminally tagged GFP variant (pSVA3926) had a motility defect similar to the ΔminD2 strain (Figure 3b) and showed diffuse localization in the cells (Supplementary Figure 3). Expression of a variant in which MinD2 was fused to an N-terminal GFP tag with a semi-flexible linker (Ithurbide et al., 2024) (pSVA6059) could also not restore motility. Tagging MinD2 on the C-terminus was, in general, more effective at restoring motility. Using only an HA tag at the C-terminus complemented the swimming back to 100% (Supplementary Figure 4). And, the strain expressing a C-terminally GFP tagged MinD2 variant (pSVA3920) retained ~60% of the motility displayed by WT cells, and the introduction of a semi-flexible linker between MinD2 and a C-terminal mTurquoise tag (pSVA6307) led to a restored swimming efficiency up to ~80% of that observed in H26 cells (Figure 3 and Supplementary Table 4).

We then proceeded to characterize the localization of the most functional C-term fluorescently tagged MinD2 variants. For this, samples were collected at similar OD₆₀₀s as that of the cell shape experiments. Visualization of both the C-terminally GFP tagged and C-terminally mTurquoise tagged MinD2 variants showed a diffused localization throughout the cells (Supplementary Figure 3b) and from early log phase through mid-log phase. Based on our observations, MinD2-linker-mturquoise did not show a change in localization during the transition from rod to plate phases and does not show a different localization in rods versus in plates. Collectively, these data show that MinD2 does not localize to the cell poles, suggesting that its impact on the polar organization of the motility machinery likely requires the interaction of MinD2 with other proteins. Furthermore, the inability of N-terminally tagged MinD2 to complement the swimming phenotype suggests that the N-terminus of MinD2 is important for its activity and interaction with protein partners in vivo.

Given that the ΔminD2 mutant exhibited a motility defect and displayed reduced assembly of archaella, we hypothesized that MinD2 may have a function similar to that of MinD4 with respect to the cellular positioning of the motility and chemotactic machineries. Therefore, we investigated the localization of the motility machinery in ΔminD2. For this, we used strains expressing ArlD-GFP, a fluorescently tagged version of a well-established marker protein, which is part of the cytoplasmic archaellum motor complex, used as an indicator to identify cells with archaella (Li et al., 2019). Upon expression of ArlD-GFP in the ΔarlD strain at OD₆₀₀ 0.01, we observed that most cells (77%) had fluorescent foci at the cell poles, with only a few cells displaying diffused fluorescence (23%) (Figures 6a,b), similar to results previously observed in H. volcanii (Li et al., 2019). However, when ArlD-GFP was expressed in ΔminD2ΔarlD cells, fluorescent foci at the poles were only detected in 25% of the cells (Figures 6a,b). This result agrees with the low abundance of archaella observed in TEM, where these structures were present in only ~32% of ΔminD2ΔpilB3 (+ pTA1392) cells (Figure 2).

As described above, in previous studies we established a correlation between MinD4 and the positioning of the archaellum machinery. Therefore, we decided to explore the links between MinD2 and MinD4 in the regulation of this process, by comparing the localization of the motility machinery in cells lacking either one of the MinD homologues (ΔminD2ΔarlD or ΔminD4ΔarlD) or both homologues (ΔminD2ΔminD4ΔarlD). In the ΔminD4ΔarlD mutant, we observed that the number of cells with ArlD polar foci was ~20% at an OD₆₀₀ of 0.01 (Nußbaum et al., 2020). In the ΔminD2ΔminD4 mutant, the formation of polar foci decreased, being present in only ~7% of cells at an OD₆₀₀ of 0.01, and continuing to decrease with increasing OD₆₀₀. Additionally, it was previously observed that all cells in the stationary phase form foci corresponding to the archaellum motor complex. We observed comparable results in the ΔarlD and ΔminD2ΔarlD mutants, with all cells having foci upon reaching the stationary phase at OD₆₀₀ 0.6 (Figure 6b). However, in the ΔminD2ΔminD4ΔarlD mutants, we observed a consistent reduction in foci formation even at high OD₆₀₀, which was not observed in the individual mutants (Figure 6b). Which suggests a synergistic effect or partially redundant role of MinD2 and MinD4 in the archaellum polar assembly.

In H. volcanii, chemotaxis involves the assembly of chemosensory arrays, which are preferentially localized at the cell poles during the early log phase and become diffused as the cells enter stationary phase (Li et al., 2019). Therefore, we investigated whether MinD2 also influences the localization of chemosensory arrays, using the chemotaxis protein CheW as a marker for these clusters (Li et al., 2019). For this, we expressed GFP-CheW in cells lacking CheW (ΔcheW) and cells lacking both CheW and MinD2 (ΔminD2ΔcheW). In ΔcheW cells, expression of GFP-CheW led to polar foci in 78% of cells and diffused localization in 22% of cells (Li et al., 2019 and Figure 6c). By contrast, in ΔminD2ΔcheW cells, expression of GFP-CheW led to only 5% of the cells displaying polar chemosensory foci, while the remaining 95% of ΔminD2ΔcheW cells showed diffused GFP-CheW localization, suggesting the absence of chemosensory arrays localization (Figure 6c). Further analysis showed that the number of cells with CheW foci formation decreased substantially in ΔcheW as the OD increased, while the localization of CheW reduces 5–0% in ΔminD2ΔcheW cells (Figure 6d).

Since previous studies have shown that MinD4 regulates the positioning of chemosensory arrays, we further investigated the links between MinD2 and MinD4 in modulating CheW localization. For this, we created the triple deletion mutant ΔminD2ΔminD4ΔcheW, which has then been used to study the localization of GFP-CheW. Our results showed a further reduction in CheW foci formation in ΔminD2ΔminD4ΔcheW cells compared to ΔminD2ΔcheW cells, with only 2% of cells showing polar GFP-CheW foci at OD₆₀₀ of 0.01. In the transition of OD₆₀₀ from 0.01 to 0.03, we observed GFP-CheW foci formation to be completely absent, with no cells displaying CheW foci (Figure 6d).

Collectively, these data show that MinD2 impacts the localization of both the motility and chemosensory machineries, in addition to the effect that MinD4 has on the localization of these cellular components.

Given the pronounced effects of minD2 deletion on cell shape and motility, we explored whether these phenotypic changes can be explained by transcriptional alterations in the relevant genes. Therefore, RNA-seq analysis was performed on WT and ΔminD2 strains at early-log (OD₆₀₀: 0.02), mid-log (OD₆₀₀: 0.3) and stationary phase (OD₆₀₀: 2) to identify any differentially expressed genes (log₂-fold changes: Supplementary Table 6, transcript per million values: Supplementary Table 7). Despite the clear morphological and motility defects, no significant transcriptional changes in the core genes related to archaellum assembly or chemotaxis were consistent over all three growth stages (Figure 7a). Instead, some of the genes showed stage-dependent differences, including downregulation of CetZ2 (log₂-fold change: −2.3) and upregulation of PssA (log₂-fold change: 1.1) during early growth phase, and upregulation of CheW (log₂-fold change: 0.9) during stationary phase. In general, many significant transcriptional changes were observed in the early (down: 21.9%; up: 20.9%) and stationary phase (down: 12.1%; up: 19.8%), but not at mid-exponential (down: 0.4%; up: 0.4%) (Figure 7b). From the 33 differentially regulated genes at OD₆₀₀ of 0.2 (mid), the loss of polar localization of the motility machinery could not be explained (Figure 7c). To detect any general trends, a functional enrichment analysis using clusters of orthologous groups was performed (Figure 7d and Supplementary Table 8). This analysis revealed the absence of consistently regulated and functionally related genes upon minD2 deletion. Instead, a counter-regulation of metabolism groups C (energy production and conversion), G (carbohydrate transport and metabolism), E (amino acid transport and metabolism), and I (lipid transport and metabolism) could be observed, with genes that are downregulated during early growth phase but are later upregulated.

Together, these RNA-seq findings complement the previous localization studies and suggest that MinD2 regulates these processes at the level of protein localization rather than gene expression. This underscores the importance of MinD2 in maintaining cell morphology and motility through mechanisms beyond transcriptional control.

Escherichia coli strains were cultured in LB (Luria Broth)-medium or grown on LB agar plates, with the necessary antibiotics (100 μg/mL ampicillin, 30 μg/mL chloramphenicol, 25 μg/mL kanamycin) and grown at 37° C. Liquid cultures were constantly shaken at 150 rpm.

Haloferax volcanii H26 cells were grown in YPC medium [0.5% (w/v) yeast extract (Difco), 0.1% (w/v) peptone (Oxoid), and 0.1%(w/v) casamino acids (Difco)] dissolved in 18% buffered Salt Water (SW) (144 g/L NaCl, 18 g/L MgCl2 * 6 H2O, 21 g/L MgSO4 * 7 H2O, 4.2 g/L KCl, 12 mM Tris/HCl, pH 7.5), supplemented with 3 mM CaCl_2, adjusted to a pH of 7.2 with KOH for transformations. For experiments, CAB medium was used [i.e. CA medium (0.5% (w/v) casamino acid) dissolved in 18% SW, supplemented with 3 mM CaCl₂, and 0.8 μg/mL of thiamine, and 0.1 μg/mL of biotin, adjusted to a pH of 7.2 with KOH and supplemented with trace elements solutions (Duggin et al., 2015)] was used. For each experiment, a single colony was inoculated into 5 mL medium and diluted to a larger volume on the subsequent day. This dilution was crucial to ensure an appropriate cell density for subsequent experiments. By adjusting the OD₆₀₀ to the desired value, it was possible to obtain a consistent starting point on the day of the experiment.

For strains with an auxotrophic mutation grown in CA/CAB medium, the medium was supplemented with 50 μg/mL uracil for ΔpyrE2. Alternatively, the strains were transformed with a plasmid carrying the respective gene for viable growth. For growth curve, cells were grown in 15 mL culture volume and measured with cell growth quantifier (CGQ) (Aquila biolabs GmbH) at 45°C and shaking at 120 rpm.

Transformation in H26 based on uracil selection via the Polyethylene glycol 600 (PEG600) along with gene deletion and expression studies were conducted as described previously (Allers et al., 2004). For transformation into H. volcanii, non-methylated plasmids were extracted from E. coli dam⁻/dcm⁻ (C2925I, NEB). Mutant strains generated and used are described in Supplementary Table 1. Plasmids created for knockout mutants are described in Supplementary Table 2. Primers to create knockout plasmids were based on pTA131, are described in Supplementary Table 3.

Glycerol stocks were streaked on solid agar medium substituted with uracil and incubated at 45°C for 5 days. A single colony from plates was used to inoculate 5 mL of media on day 1. Strains without plasmid were grown in CAB with 50 μg/mL uracil and strains with plasmid pTA1392 (containing pfdx-pyrE2) were grown in CAB medium without supplements. To generate a growth curve, the obtained culture was inoculated on day 2 at a starting OD₆₀₀ of 0.05. The cell density was measured with cell growth quantifier (CGQ) (Aquila biolabs GmbH) at 45°C and shaking at 120 rpm and measurement were taken every 300 s.

To assess the growth and viability of the H26 and ΔminD2 (transformed with pTA1392) strains, cells were cultured in 5 mL CAB medium at 45°C to an OD₆₀₀ of 0.2. On the next day, cell cultures were set to a theoretical OD₆₀₀ of 0.2. A serial dilution was prepared until a dilution of 10⁵ and 5 μL of each dilution per strain was spotted on CA-plates. The plates were incubated for 2 days at 45°C in sealed plastic bags.

The cell shape was analyzed by imaging the cells with an inverted phase contrast light microscope (Zeiss Axio Observer Z.1). The cells were grown in 5 mL of the respective medium and diluted in 20 mL media volume the next day in order to achieve an OD₆₀₀ of 0.01 the day after. For each culture, 5 μL sample was collected from different growth phases and dropped at the center of an agarose pad (0.3% (w/v) agarose dissolved in 18% SW). On drying, the pad was covered with a cover slip and imaged. The images were acquired at 100x magnification using the oil immersion phase contrast (PH3) channel. All sampled were analyzed in triplicated. Fluorescence microscopy images were acquired on Zeiss Axio Observer Z.1 (ex: 450–490 nm em: 500–550 nm filter from Chroma®), equipped with a heated XL-52000 Incubator running VisiVIEW℗ software for MinD4 and CheW.

The phase contrast images from the microscopy were analyzed using Fiji (Schindelin et al., 2012) combined with MicrobeJ plugin (Ducret et al., 2016). For the analysis, cells that formed aggregates or were fragmented were discarded from the calculation. The circularity of the cells was automatically calculated. The diameter of each analyzed cell was thereon calculated and grouped into 6 bins in the range interval of 0.1 to 1. The parameters used for circularity were as previously defined in Patro et al. (2023).

Cells were harvested at 2000 g for 15 min. The resulting pellet was resuspended to a theoretical O.D of 10. Five μL of cells of the cell suspension was applied to a glow discharged carbon coated copper grid (Plano GmbH, Wetzlar Germany) and incubated for 10 s. The excess liquid was blotted away. The cells were then stained with 2% uranyl acetate (w/v). Cells were imaged using Zeiss Leo 912 Omega (tungsten) operated at 80 kV and images were taken using Dual speed 2K on Axis charge-coupled device (COD) camera (TRS, Sharp-Eye).

Semi-solid agar plates were prepared using 0.33% agar in CA medium supplemented with 1 mM tryptophan. Cultures grown at OD₆₀₀ 0.3 were inoculated into the plates using stab techniques, and the plates were then incubated at 45°C for 4 days. To compare the motility of different strains, all strains were spotted on the same plate. For each strain, a minimum of 3 technical replicates and 3 biological replicates were conducted. After 4 days, the diameter of the motility ring was assessed.

The Alphafold models for MinD2 and HVO_0596 were generated with Alphafold 3 Google colab (Abramson et al., 2024) and further analyzed by ChimeraX (Meng et al., 2023).