A histidine pseudokinase modulates polar growth and cell shape in Streptomyces venezuelae

Abstract

Polar growth and cell shape determination in mycelium-forming Streptomyces bacteria depends on the function of a polarly localised multiprotein complex that directs cell wall synthesis – the polarisome. This complex assembles around the essential cell polarity determinant DivIVA, alongside other largely unknown components. We report here the discovery of a conserved hybrid histidine kinase-like protein, PsmA, that interacts and co-localises with DivIVA at the hyphal tips. Deletion of psmA affects the shape and dynamics of the polarisome, leading to aberrant cell shape and hyphal hyperbranching. PsmA is a pseudokinase that lacks the critical histidine residue in its catalytic core. Our results suggest that PsmA tunes the dynamics and properties of the DivIVA-based polar organelle in streptomycetes in parallel to but not redundantly with Scy and FilP, two coiled-coil proteins known to influence polarisome properties. In summary, PsmA interacts with DivIVA and modulates the integrity of the growth zones at hyphal tips.

A pseudokinase version of a hybrid histidine kinase is identified as part of the protein assembly that directs polar growth in Streptomyces bacteria. This protein, PsmA, interacts with the essential DivIVA and modulates polar growth and cell shape.

Introduction

The ability to assort cellular components in relation to an axis (polarisation) gives cells a plethora of abilities, such as motility, asymmetric cell division and cell differentiation. A hallmark of cellular polarisation is directional growth seen in some cell types, enabling them to acquire complex shapes through precise placement of the growth machinery. A familiar example is the hyphal growth in fungi^1,2. Here a mega-complex called the ‘polarisome’ controls growth and morphogenesis at the hyphal tips (or budding sites in yeast) and directs the polarity and organisation of both the actin cytoskeleton and the secretory apparatus involved in assembly of the cell wall^2,3.

In an intriguing similarity to the fungal polar growth, filamentous bacteria of the genus Streptomyces also build their cell walls in restricted zones at hyphal tips. This apical zone is organised by an assembly of proteins together referred to as the polarisome (or the tip-organising complex) in analogy to the fungal polarisome^4–6. Central to the polarisome is the essential protein DivIVA. The DivIVA-dependent mechanism to organise polar assembly of the peptidoglycan (PG) cell wall is shared with other members of the phylum Actinomycetota, including both filamentous forms reminiscent of streptomycetes and rod-shaped bacteria, like corynebacteria and mycobacteria^6–8. DivIVA is broadly conserved among Gram-positives and a few other bacterial lineages and acts as a cell polarity determinant that is targeted to cell poles and septation sites⁹. Structurally it consists primarily of coiled-coils, with an N-terminal membrane-targeting domain. DivIVA preferentially localises to negatively curved membranes in the cell^10–12. Lipid affinity may also contribute to subcellular localisation in some DivIVA orthologs^13,14. In investigated Actinomycetota, DivIVA is essential for growth and is required to concentrate cell wall assembly to the cell poles^5,15–17.

Polar growth is observed in at least two other clades of bacteria, the alpha-proteobacteria and planctomycetes, though both employ distinct mechanisms to control growth polarity compared to actinomycetes¹⁸. In other rod-shaped bacteria, cell walls are enlarged by intercalating new PG into the lateral wall, leaving cell poles relatively inert. This non-polar mode of growth is directed by polymers of actin-like MreB protein, which organise a complex of cell wall assembly proteins referred to as the elongasome^19–21. New cell poles originate by cytokinesis orchestrated by the tubulin-like FtsZ and the divisome complex. Occasionally, FtsZ also contributes to cell elongation^19–21. Notably, actinomycetes do not require MreB proteins for their polar mode of growth^17,22,23 and the hyphal growth of streptomycetes is independent of cell division and FtsZ^24,25.

In rod-shaped actinomycetes like mycobacteria, DivIVA (Wag31) is recruited to cell division sites and then remains at the new poles generated by division, where it is required for establishing polar growth^8,16. The hyphal growth of streptomycetes imposes additional complexity since the establishment of new zones of growth occurs via hyphal branching and not cell division⁶. The branches arise from small foci of DivIVA along the lateral walls of hyphae¹⁵. The majority of these small polarisome seeds form by scission from the main apical polarisomes at the extending hyphal tips and deposition along the lateral wall. These daughter polarisomes gradually enlarge before giving rise to new lateral branches²⁶. Thus, in addition to orchestrating DivIVA-driven cell elongation at the cell pole, the Streptomyces polarisomes exhibit dynamics that is crucial for polarisome scission, cell polarity re-establishment and formation of new branch initials.

Studies of unicellular and filamentous fungi have been pivotal in uncovering the mechanisms that govern cell polarity and polarised growth, processes that are widely conserved among eukaryotic organisms²⁷. However the mechanism of DivIVA-driven growth in bacteria is not comprehensively understood⁷. While a few components of the Streptomyces polarisome have been identified, the underlying regulatory mechanisms remain poorly understood. The two coiled-coil proteins FilP and Scy have been recognised to interact directly with DivIVA and affect polar growth^4,28,29. Both proteins share features with and are suggested to have evolved from DivIVA³⁰. Scy co-localises with DivIVA in Streptomyces coelicolor and strongly affects hyphal shape^4,31. FilP (possessing intermediate filament-like features) forms a relatively dynamic localisation pattern, reminiscent of a subapical gradient adjacent to the DivIVA and Scy cluster^29,31. Mutants lacking filP have altered cell shapes but with only subtle phenotypic defects compared to the prominent phenotype of scy mutants^28,29,31. An additional coiled-coil protein that associates with polarisomes, SepIVA, is not associated with any detectable phenotypic effects³². DivIVA is also subject to Ser/Thr protein phosphorylation, but it is not known how phosphorylation affects the function of the protein^33,34. The central role of the polar growth apparatus in streptomycetes and the complex morphologies these organisms can acquire suggest that additional proteins are involved in controlling its function.

In this study we set out to identify new polarisome members and report here the identification of a previously uncharacterised protein, PsmA, that associates with DivIVA and forms a part of the polarisome complex in Streptomyces venezuelae (S. venezuelae). PsmA bears similarity to hybrid histidine kinases (HKs) but is functionally a pseudokinase. We show that PsmA interacts directly with DivIVA and strongly affects cellular morphology and hyphal branching, including altering the shape and distribution of apical DivIVA foci. PsmA contributes to the stability of polar growth zones, and it does so in parallel to but independently of Scy and FilP. Our findings indicate that PsmA tunes the dynamics of apical polarisomes to control hyphal branching and prevent excessive tip bifurcations and hyperbranching.

Results

Identification of PsmA as an interaction partner of DivIVA

To identify proteins associated with polar growth, we immunoprecipitated FLAG-tagged DivIVA from S. venezuelae. A protein of about 150 kDa was found to co-precipitate and was identified as Vnz18495 by mass spectrometry (Fig. 1a and Supplementary Data File 1). Vnz18495 (UniprotKB entry F2REC4) has been annotated as a predicted PAS domain S-box protein and has weak similarity to histidine kinase A family (HisKA). A domain search using ScanProsite showed that it consists of an N-terminal part with tandem PAS domains (residues 19–133 and 157–263), followed by dimerisation and phosphotransfer (DhP) domain (301–381) and a catalytic and ATP-binding (CA) domain (382–526), together forming the HK domain (Fig. 1b). In addition, Vnz18495 carries a C-terminal phosphoreceiver (REC) domain (1227–1344), which is connected to the HK domain by a linker (696 amino acid residues) that is predicted to be an intrinsically disordered region (IDR) and is rich in proline residues (Fig. 1b). Alphafold modelling gave results consistent with the bioinformatic predictions of domain structure (Fig. 1c, d). The N-terminal part of the S. venezuelae PsmA protein modelled as a dimer typical of HKs (Fig. 1e; Supplementary Videos 1 and 2). Overall, the domain organisation is characteristic for hybrid HKs. Based on the phenotype of the mutants that will be presented below Vnz18495 is herein named PsmA (polar growth and shape modulator A).

Fig. 1. Hybrid histidine kinase-like PsmA interacts with DivIVA in S. venezuelae.

a Co-immunoprecipitation with FLAG-tagged DivIVA from S. venezuelae cell extracts. Eluted proteins shown on Coomassie-stained SDS-PAGE. M molecular weight marker (kDa), C controls with empty vector, IP lane with proteins co-immunoprecipitating with FLAG-DivIVA. Asterisks indicate IgG antibody bands common to all lanes. Arrowheads close to 37 kDa indicate 3x-FLAG-DivIVA and co-precipitating untagged DivIVA (lower band). An arrow indicates the co-precipitating protein identified as PsmA (Vnz18495). b Schematic illustration of domain structure of PsmA (1350 amino acids), with two PAS domains and histidine kinase (HK) domain, a 696 residue intrinsically disordered region (IDR) and a C-terminal phosphoreceiver domain (REC). The red line in the graph above shows predicted disorder of the protein structure using IUPred2A and the blue line represents the prediction from ANCHOR2. c Structure prediction of PsmA using Alphafold2 with folded domains indicated, while the IDR is predicted to be unfolded. d The predicted alignment error of the model in (c). e Prediction of structure of a dimer of the N-terminal part of PsmA using Alphafold2. Pdb files of models provided in Supplementary Data Files 2 and 3. f Assays of protein-protein interactions using a bacterial two-hybrid system. E. coli cells carrying plasmids with protein fusion to the T18 and the T25 domains of a bacterial adenylate cyclase are spotted on MacConkey maltose agar plates. PsmA was fused as full-length protein or split into either N-terminal (PAS-PAS-HK domains) or C-terminal (IDR-REC domains) parts. pKT25 and pUT18c indicate empty vector controls and Zip indicates a positive control with a self-interacting leucine zipper domain. Red colonies indicate interaction and reconstitution of active adenylate cyclase. g Recruitment of PsmA to subcellular clusters of DivIVA in E. coli. PsmA-mCherry and DivIVA-mNeonGreen were produced as fusion proteins (plasmids pKF959 and pKF961, respectively), separately or co-produced in the same strain. Representative images are shown as overlays of fluorescence and phase-contrast. Demographs showing focus positions along cell length in cell populations were generated with MicrobeJ/ImageJ. PsmA-mCh produced alone yielded disperse signal and only few putative foci were detected in these cells (top left graph). Scale bar, 5 μm.

We verified the interaction of PsmA and DivIVA using bacterial two-hybrid assays (BACTH). HKs typically show self-interaction in the kinase domain and consistently, the N-terminal part of PsmA self-interacted in the BACTH system (Fig. 1f). Further PsmA interacted directly with DivIVA in both plate assays (Fig. 1f) and semi-quantitative β-galactosidase activity assays (Supplementary Fig. S1). Specifically, the N-terminal dimerising region interacted with DivIVA, while the C-terminal part with the IDR and receiver domains did not (Figs. 1f and S2). PsmA did not show interaction with the polar growth-related coiled-coil proteins Scy or FilP (Figs. 1f; S1; S3 and S4).

The interaction between DivIVA and PsmA was further validated by a cytological assay in E. coli. Similar to other DivIVA orthologs^12,35, an S. venezuelae DivIVA-mNeonGreen (mNG) fusion accumulated at cell poles and in longer cells also to mid-cell positions (Fig. 1g). In contrast, PsmA tagged with the fluorescent protein mCherry (mCh) showed diffuse fluorescence throughout the cell (Fig. 1g). When DivIVA-mNG and PsmA-mCh were co-produced, PsmA was recruited to the polar DivIVA foci (Fig. 1g). In summary, we show that PsmA (Vnz18495) is a DivIVA-interacting hybrid HK-like protein.

Deletion of psmA strongly affects hyphal growth and morphology

The role of psmA in S. venezuelae was investigated by replacing the gene with an apramycin resistance marker (apra). The genotype of the ΔpsmA::apra deletion mutant (LUV282) was verified by genome sequencing. The deletion of psmA affected growth and led to smaller colonies with a characteristic pitted appearance (Fig. 2a, b). A clean deletion of psmA was also generated using a CRISPR/Cas9-based system resulting in a mutant (LUV293) with similar phenotype.

Fig. 2. Deletion of psmA affects growth and hyphal morphology.

a Colony appearance of psmA mutant and complemented strains on MYM agar, 96 h. Wild-type parent and mutant strain LUV282, with and without plasmids for in trans complementation of the ΔpsmA::apra mutation. The psmA gene is expressed from its own promoter (pKF788) or from the constitutively active kasO*p promoter (pKF781). b Single colonies arising from wild type and ΔpsmA mutant (LUV282) spores imaged after 72 h on MYM agar. c Mycelial microcolonies of wild type and ΔpsmA strains grown on cellophanes on MYM agar and imaged after 16 h of incubation at 30 °C in a phase-contrast microscope. The dotted yellow lines show how the circumference of the areas covered by microcolonies were drawn by connecting the outermost hyphal tips with lines. Scale bar, 50 µm. d Microcolony sizes of the wild type (n = 26) ΔpsmA mutant (n = 25) and mutant complemented in trans from pKF788 (native promoter; n = 27) and pKF781 (kasO*p; n = 25), respectively. The areas covered by microcolonies were estimated as shown in (c). The difference between the mean for each strain and that of the wild type was tested with Welch’s ANOVA test, followed by Dunnett’s multiple comparisons test. **** represents P < 0.0001. ns means no significant difference (P > 0.05). e Representative images of vegetative mycelial shapes of wild type and psmA mutant (LUV282). Mycelia were grown in a microfluidic cell perfusion system. Inset shows magnified images of hyphal tips from the boxed regions. Scale bars, 10 μm. Corresponding time-lapse movies are shown in Supplementary Video 3.

Growth of the vegetative mycelium was studied in mycelial microcolonies on cellophane membranes placed on MYM agar medium. Congruent with the macroscopic colony phenotype, the psmA mutant formed clearly smaller and denser microcolonies than the parent (Fig. 2c). Quantification of the effect confirmed that microcolonies of the psmA mutant traverse a much smaller area compared to the wild type after a similar time of growth (Fig. 2d). The defects in the size of both colonies and microcolonies were complemented by psmA expression in trans driven from its native promoter region (200 bp upstream, see Supplementary Fig. S5), as well as by expressing psmA from the heterologous kasO* promoter, which is frequently used to drive constitutive expression of genes in S. venezuelae³⁶ (Fig. 2a, c, d).

To characterise the effect of psmA deletion on cell shape we examined hyphal morphology in a microfluidic cell perfusion system. The mutant hyphae differed from the wild type by exhibiting irregular shape and aberrant width compared to the wild type (Fig. 2e). The hyphal tips of the psmA mutant had a rounder shape and higher mean hyphal width compared to the wild type (Supplementary Fig. S6). Timelapse microscopy of growing hyphae further revealed that the psmA mutant exhibits frequent bifurcations of the hyphal tips, leading to a striking hyperbranching morphology. This apical branching largely explains the compactness and poor spreading of mycelial microcolonies exhibited by the mutant (Fig. 2e and Supplementary Video 3).

The developmental life cycle of streptomycetes includes formation of arial mycelium and spores, but these processes are not overtly affected by psmA. The grey-green colony pigmentation was similar between the psmA mutant and the wild type (Fig. 2a) and abundant spores could be harvested from confluent lawns of the mutants. We further used strains producing mNG-tagged FtsZ to visualise sporulation septation and monitored the sporogenesis with fluorescence microscopy. Both psmA mutant and wild type sporulated similarly and Z rings appeared at similar times. The Z rings in the mutant strain were often more irregularly oriented, which is probably due to the irregular hyphal morphology (Supplementary Fig. S7 and Supplementary Video 4). Staining of sporulation septa and spore nucleoids revealed spore chains with regular septation and no clear differences between mutant and wild type (Supplementary Fig. S7). In conclusion, deletion of psmA affects primarily vegetative growth and hyphal morphology and does not directly influence spore formation.

PsmA co-localises with DivIVA and Scy at the hyphal tips

To determine subcellular localisation, we created a functional C-terminal fusion of PsmA to mNG and examined the fluorescence signal in S. venezuelae. Complementation with the construct restored the growth defects of the psmA mutant when expressed in trans. PsmA-mNG showed distinctive signals from foci at hyphal tips, and the low and diffuse signal along the hyphae was only marginally above the autofluorescence background (Figs. 3a; S8; S9; and Supplementary Video 5). This clear apical localisation is consistent with the observed interaction between PsmA and DivIVA.

Fig. 3. Co-localisation of PsmA with polarisome proteins at hyphal tips.

a Localisation of PsmA-mNeonGreen, DivIVA-mCherry, FilP-mCherry and mNeonGreen-Scy at apical regions of growing hyphae. Representative fluorescence images (top) and their overlay on phase contrast images (below) are shown. The apical localisation is indicated by arrows. Plasmid used for production of the fusion proteins in S. venezuelae wild type were pSS204 (DivIVA-mCh), pKF769 (mNG-Scy), pNA1258 (FilP-mCh) and pKF788 (PsmA-mNG). Comparison to the background signal in wild-type hyphae is shown in Supplementary Fig. S8. b Co-localisation of PsmA-mNG and DivIVA-mCh. Strain LUV282 carrying both pKF789 and pSS204 was used for co-expression. A representative example of the localisation pattern at a growing hyphal tip is shown. Note overlap of the localisation but also distinct size and shape of the PsmA and DivIVA foci. c Localisation of N-terminal part of PsmA (N-PsmA) to the apical growth region while C-terminal part (C-PsmA) does not show clear signal at hyphal tips, when produced as fusions to mNeonGreen. Wild-type S. venezuelae strain was transformed with plasmid pKF808 (encoding N-PsmA-mNG) or pKF809 (encoding C-PsmA-mNG). d Subcellular localisation of N-PsmA-mNG (pKF808) and C-PsmA-mNG (pKF809) when produced in ΔpsmA mutant LUV282. Fluorescence signal shown in greyscale (right) or overlaid on phase-contrast images (left). The arrow indicates apical localisation of N-PsmA-mNG while no such clear localisation was seen with C-PsmA-mNG. Scale bars, 2 µm.

Co-production of PsmA-mNG and DivIVA-mCh allowed confirmation of their co-localisation at hyphal tips. It also revealed a subtle difference in position and shape of the foci, with PsmA-mNG forming smaller and more restricted foci at the apices of the hyphae, while DivIVA showed broader distribution at the cell pole (Figs. 3b and S10).

Both Scy and FilP interact with DivIVA^4,29, but FilP shows sub-apical localisation in actively growing hyphae, adjacent to the apical DivIVA foci³⁷. Consistently, the PsmA foci at the hyphal tips did not co-localise with FilP, which formed more diffuse subapical patterns of localisation (Fig. 3a and Supplementary Video 6). To observe Scy localisation in S. venezuelae, we generated mNG-scy encoding an N-terminal fusion of mNG to Scy, expressed from the kasO* promoter and found that it complements a scy deletion mutation (Supplementary Fig. S11). In agreement with results for S. coelicolor⁴, mNG-Scy localised to hyphal tips reminiscently of both DivIVA and PsmA foci (Fig. 3a and Supplementary Video 7).

The N-terminal region of PsmA (N-PsmA; residues 1–530), fused to mNG, was found to be sufficient for polar localisation, while the C-terminal region (C-PsmA; residues 531–1350) was less clearly recruited to the growing tip (Figs. 3c; S9 and Supplementary Video 5). Since N-PsmA may localise to the tip by interacting with the native PsmA protein we produced N-PsmA-mNG in a psmA deletion background and found that it showed distinct but less intense polar localisation compared to full-length PsmA-mNG. However it failed to provide complementation of the psmA deletion phenotype. C-PsmA neither showed polar localisation nor complementation (Fig. 3d). In conclusion, PsmA co-localises with DivIVA and Scy to the polarisome and the N-terminal part of PsmA is necessary and sufficient for DivIVA interaction and targeting to hyphal tips.

PsmA is a widely conserved histidine pseudokinase among Streptomyces spp

The phylogenetic distribution of PsmA in Actinomycetota was investigated in a recently published high-quality set of actinomycete genomes³⁸. PsmA orthologs were identified exclusively in the family Streptomycetaceae in this dataset (Supplementary Fig. S12). WebFlaGs³⁹ analysis revealed that psmA is present in a conserved genetic context in most of the genomes (Supplementary Data File 4). In S. venezuelae, the psmA locus is located a little over 100 kbp from the origin of replication and this proximity to the centre of the chromosome was verified in a few selected species (Supplementary Fig. S13).

The PsmA orthologue in S. coelicolor (SCO4009) has been recognised as one of the three hybrid HKs present in this organism. The others are OsaA and SCO7327,  and both are also encoded by the S. venezuelae genome^40–42. It was noted that SCO4009 (PsmA) does not carry the conserved histidine residue in the H-box of the DHp domain⁴³. The PsmA homologue from S. venezuelae contains glutamate in place of the normally conserved histidine in this position, illustrated by alignment with bona fide HKs of HisKA family (Fig. 4a). The corresponding residue is variable in the complete PsmA homologues that we examined, being for example aspartate in S. coelicolor, glutamine in S. avermitilis, valine in S. leeuwenhoekii and tyrosine in S. scabies (Supplementary Fig. S14). Thus, PsmA lacks the catalytic histidine residue normally found in HKs. Examination of the CA domain shows that key residues in the ATP-binding motifs typical for HKs are not conserved in PsmA and structure prediction using Alphafold indicates that the region corresponding the ATP lid in other kinases has a fold that appears incompatible with ATP binding (Supplementary Fig. S15). Consistent with these observations, we did not detect autokinase activity when the N-terminal HK (1–530) domain of S. venezuelae PsmA was subject to an autophosphorylation assay. Unlike the bona fide HK VanS from S. venezuelae (used as positive control), the PsmA HK domain did not incorporate γ-phosphate from ATP (Fig. 4c). In conclusion, PsmA is a pseudokinase that lacks the ability to catalyse histidine phosphorylation in the H-box. In addition, the H-box in PsmA (XSLRGP; Fig. 4b) lacks the D/EXXT/N motif that is crucial for phosphatase activity of HisKA family members^44,45, therefore it likely also lacks the ability to dephosphosphorylate response regulator proteins.

Fig. 4. PsmA is a pseudokinase.

a Alignment of the H-box motif in the catalytic core of PsmA with examples of bona fide histidine kinases of Streptomyces venezuelae (PhoR, OsaA, VanS and KdpD) shows the absence of catalytic histidine at position 323 in PsmA (position highlighted red in the alignment and by an arrow). Weblogo was used for graphical illustration. b Illustration of consensus sequence in the H-box region of PsmA homologues identified from actinomycetota genomes (Bioproject PRJNA747871), created using Weblogo. The position of the normally conserved histidine residue in bona fide histidine kinases is indicated (arrow) while the residues immediately following the position show high degree of conservation among the PsmA sequences. c Assay of autokinase activity N-PsmA (60 kDa; bottom panel). The kinase activity of the cytoplasmic region of the bona fide histidine kinase VanS (32 kDa) was used as a positive control (top panel). Equal amounts of the purified proteins were incubated with γ-³²P-labelled ATP and the reaction stopped at indicated time points. Incorporation of radioactivity in proteins, separated on SDS-PAGE was assessed in a phosphorimager. An arrowhead points to the phosphorylated protein that matched the predicted molecular mass of VanS cytoplasmic domain. Molecular weights of marker proteins are indicated. The uncropped raw image is provided as Supplementary Data File 8. d Schematic representation of the VanS and N-PsmA proteins used in kinase assay.

PsmA function does not rely on canonical hybrid histidine kinase phosphorelay

The conserved aspartate that is normally subject to phosphorylation in response regulators is present in the PsmA REC domain (D1276). To test whether phosphorylation of this residue is important, we mutated it to phosphoablative alanine or glutamine or phosphomimic glutamate and the function of mutant alleles was assayed by testing their ability to complement a chromosomal psmA deletion, quantified as in Fig. 2c. None of these mutations led to loss of complementation (Fig. 5a), showing that phosphorylation of the conserved REC domain aspartate is not critical for the observable function of PsmA.

Fig. 5. Mutational analysis of PsmA to test whether specific domains or amino acid residues are required for the function of the protein.

The mutant versions of PsmA are schematically depicted below the graphs. Functionality of the mutant alleles were tested by complementation of psmA deletion and monitoring of microcolony size, as described in Fig. 2c. a Effect of specific mutations in psmA affecting either the conserved aspartate in the REC domain (D1276) or residues in the H-box in the histidine kinase (HK) domain. Plasmids pKF817 (wild-type psmA; n = 27), pKF896 (E323H mutation; n = 24), pKF897 (ESLRG > HDLRT mutation in the H box; residues 323–327; n = 26) pKF901 (D1276E mutation; n = 26), pKF892 (D1276A mutation; n = 27) and pKF893 (D1276Q mutation; n = 25) were introduced into the psmA mutant strain LUV282 (n = 26). b Effect of domain deletion mutations in psmA on its ability to complement morphological defect the psmA mutant. Plasmids used for complementation are pKF781 (wild-type psmA; n = 26), pKF783 (N-PsmA; n = 26), pKF797 (C-PsmA; n = 25), pKF867 (ΔIDR; n = 26), pKF898 (ΔREC; n = 25), pKF917 (ΔPAS1; n = 27) and pKF918 (Δ(PAS1-PAS2); n = 26). c Effect of shortening of the intrinsically disordered region (IDR) on the function of PsmA. Full-length psmA was expressed from pKF817 (n = 30) and the truncations of psmA removing parts of the IDR by shortening from the C-terminal end expressed from pKF904 (residue A1055–P1225 removed; n = 27), pKF905 (residue P802–P1225 removed; n = 26) and pKF906 (residue P688–P1225 removed; n = 27), introduced into the psmA mutant strain LUV282 (n = 24). The difference between the mean for each strain and that of the psmA mutant LU282 (a) or the LUV282 strain with a plasmid expressing wild type psmA (b, c) was tested with Welch’s ANOVA test, followed by Dunnett’s multiple comparisons test. **** represents P < 0.0001; **P = 0.0034.

Then we tested whether reintroduction of the normally conserved catalytic histidine residue into the DHp domain of PsmA can affect the phenotype and function of the protein. We mutated the H-box residues to swap histidine for glutamate (E323H) or replace it with the consensus DHp domain H-box residues (ESLRGP > HDLRTP) (Fig. 5a). Notably, this change also restores the HisKA phosphatase motif D/EXXT/N. The substitutions in this motif did not detectably affect the growth phenotype and both mutated versions were able to complement the psmA deletion. We conclude that PsmA function does not rely on canonical hybrid HK phosphorylation or phosphatase activities.

Individual domains of PsmA are required for functioning of PsmA

The same type of complementation tests was used for deletion analyses aimed at testing whether specific domains are necessary for the function of PsmA. However, it can not be excluded that tested deletions also may affect the stability of the protein. The assays confirmed the lack of function of the N-terminal fragment of the protein (Fig. 5b). Similarly, expression of only the region C-terminal to HK domain did not allow complementation. Deletions of only PAS1 or both PAS1 and PAS2 domains also abolished function of PsmA (Fig. 5b). Further, loss of either the REC domain or the IDR domain led to the loss of complementation by the truncated protein. While removal of the entire IDR leads to loss of function, we tested to what extent this region can be shortened without affecting the ability to complement. For this purpose, three mutant psmA alleles were generated that decrease the length of the IDR (696 residues long, spanning residue 531–1227) by removing residues 1054–1225 (leaving 524 residue IDR), 804–1225 (leaving 275 residue IDR) and 687–1225 (leaving 158 residue IDR). The results show that the IDR can be truncated from residues 1054 to 1225 (shortening by 25%) without loss of function, but the longer deletions were not tolerated (Fig. 5c). Taken together, PAS, IDR and REC domains of PsmA are essential for the function of the protein.

No clear effect of PsmA on transcription of polar growth-related genes

Given the strong phenotypic effect of psmA deletion, we asked to what extent this is associated with and may be caused by altered patterns of gene expression. RNA was extracted from exponentially growing wild-type and psmA mutant (LUV282) cultures and subjected to RNA-seq transcriptome analysis. We found that psmA deletion led to significant changes in expression with bias towards genes with decreased expression in the mutant (200 downregulated vs 18 upregulated (Supplementary Fig. S16 and Supplementary Data File 5). Pathway analysis of these genes did not reveal any clear relation to polar growth, cell wall synthesis, or cell shape (Supplementary Fig. S17), suggesting that changes in gene expression may not directly mediate the observed effects of PsmA on polar growth and cell shape. Notable among the 200 genes that are downregulated in the psmA mutant are 38 related to transporters and 22 genes annotated as GNAT N-acetyltransferases, but otherwise we do not recognise patterns suggesting for example specific stress responses elicited in the mutant.

PsmA affects the distribution of DivIVA, polarisome splitting and bifurcations of hyphal tips

Since PsmA has large effects on hyphal shape and branching patterns, we investigated whether the psmA deletion causes altered localisation of fluorescently tagged polarisome-associated proteins. In psmA mutant cells, mNG-Scy showed similar localisation at hyphal tips as it did in the wild type (Supplementary Fig. S18 and Supplementary Video 7). FilP-mCh is known to accumulate in a diffuse subpolar pattern just below the apices of extending hyphae in wild-type cells and a similar pattern was observed in the psmA knockout cells, despite the perturbed cell shape (Supplementary Fig. S18 and Supplementary Video 8). Accumulation of FilP-mCh signal was also observed at the forks of bifurcated tips similarly to what has been observed in scy mutants³¹. However we conclude that psmA deletion does not cause major changes in FilP and Scy localisation.

DivIVA-mCh localised to hyphal apices in both strains often in crescent-like shape, reflecting the shape of the hyphal tip. When comparing, we found that the DivIVA-mCh foci in the psmA mutant were more diffuse and irregular than the narrow and more consistently shaped foci in the wild-type hyphae (Figs. 6a; S19a and Supplementary Video 9). An analysis of the shape of the foci showed that polar DivIVA-mCh assemblies showed significantly lower circularity in the psmA mutant compared to the wild type, while mean focus area did not differ (Supplementary Fig. S19b, c). The effect on the shape of apical DivIVA-mCh assemblies was also clear when psmA was conditionally expressed from a tetracycline-inducible promoter in the ΔpsmA mutant. The cells were first grown without inducer when they showed a hyphal morphology characteristic of the psmA mutant with broad and irregularly shaped apical DivIVA foci (Fig. 6b; Supplementary Videos 10 and 11). After switching to a medium with anhydrotetracycline to induce expression of psmA-mNG the hyphae smoothened and tips became uniformly shaped, less branched and concurrently the DivIVA patterns at the tips became more consistent and narrower (Supplementary Fig. S19d).

Fig. 6. psmA deletion causes altered shape and frequent splitting of DivIVA polarisomes.

a Effect of psmA deletion on distribution of DivIVA-mCh at hyphal tips. Fusion protein was produced from plasmid pSS204 in wild-type and LUV282 strains. Fluorescence (DivIVA-mCh) and phase contrast (phase) micrographs are shown. Polar localisation of DivIVA-mCh indicated by arrows. Difference in shape and distribution of apical DivIVA clusters can be noted. See also Supplementary Fig. S19 and Supplementary Video 9. Scale bars, 2 µm. b The ΔpsmA strain expressing divIVA-mCh from pSS204 was modified to conditionally express psmA-mNG from an anhydrotetracycline (ATC)-inducible promoter using plasmid pKF852. The strain was grown in CellASIC microfluidic system without ATC for 8 h from seeding the spores and then exposed to 100 ng/mL of ATC and followed for another 2 h. A representative example of an individual hyphal tip was followed over time and used to create the montage. The top panel shows the phase contrast image of a tip and its corresponding fluorescence for PsmA (green) and DivIVA (red) are in panels below. A white arrow indicates a tip at the time of induction, where PsmA-mNG will appear. A movie from which the example was extracted are shown as Supplementary Video 10. Scale bar, 2 μm. c Quantification of polarisome splitting events. Polarisomes of ΔpsmA and wild type (WT) were followed by time-lapse imaging using fluorescently labelled DivIVA-mCh (pSS204) during growth of vegetative hyphae. Apical DivIVA-mCh foci were followed using the trackmate program in ImageJ. The rate of splitting of the polarisome tracks (number of splits per distance travelled) was measured in the wild type (n = 21) and compared to the ΔpsmA mutant (n = 9) by setting the parameters for splitting of the DivIVA-mCh as described in the methods. The graph shows mean and standard deviations. The difference between the distribution of values was tested using a Mann–Whitney test. **** represents P < 0.0001.

Next we monitored the dynamics of apical polarisomes in relation to the hyperbranching phenotype of the psmA mutant. Apical DivIVA-mCh foci (indicating polarisomes) were tracked in time-lapse movies using the trackmate function in Fiji software. The tracking allowed detection of polarisome splitting events that lead to splitting of the growth zone and tip bifurcation. In contrast the normal branching mechanism in wild-type strains involves budding off of small daughter foci from the apical DivIVA clusters, but these small daughter foci stay on the lateral wall and do not mature into growth zones and lateral branch initials until much later²⁶. Such small foci leading to subapical branching were not detected as bifurcations and tip-splitting events in trackmate. Notably and as previously shown in S. coelicolor²⁶, a very small fraction of wild-type hyphae showed splitting of the apical polarisome into two similarly sized structures, leading to a new branch at minimal distance from the tip (Fig. 6c). In sharp contrast, the psmA mutant exhibited a significantly higher frequency of polarisome splitting (Fig. 6c). In summary, psmA deletion leads to broader and more inconsistently shaped DivIVA foci at the tips and to significantly increased number of tip splitting or apical branching events compared to the wild type, associated with splitting of apical DivIVA clusters. The results suggest that PsmA is required for maintaining the integrity of the apical growth zones and prevention of splitting of polarisomes.

Both PsmA and Scy/FilP contribute to maintain stability of polar growth zones

The conspicuous phenotype of psmA mutants, with irregular shape and high rate of tip bifurcations, is reminiscent of the previously described morphology of scy mutants^4,31. This notion raised the possibility that scy and psmA work together in the same pathway, which would lead to a double mutant showing a similar phenotype to that of the single mutants. Alternatively if psmA and scy have redundant functions, the prediction would be a synthetic effect with much stronger defect in the double deletion compared to what would be expected from combining the single mutations. To distinguish between these possibilities, we generated a scy psmA double mutant. The mycelial morphology and microcolony area were monitored as described above. Combined deletion of psmA and scy led to significant reduction of microcolony size compared to the individual mutants (Fig. 7a). This phenotype was restored to be similar to the scy phenotype upon expression of psmA in trans, including the emergence of some regularly shaped hyphae growing out from the microcolonies, typical of scy mutants (Fig. 7a). Live imaging of the double mutant strain confirmed the strong effect on growth and morphology with hyperbranching hyphae forming extremely dense mycelial clumps (Supplementary Video 12, compare to psmA mutant in Fig. 2c and Supplementary Video 3). Using the relative decrease of the one-dimensional expansion rate of microcolonies (i.e. radius of mutant microcolony calculated from the mean area after 16 h of growth, relative to the wild type) as a quantitative measure of the phenotype, the double mutant phenotype was only somewhat more severe (32% of wild type radius) than what was expected from the product of the two mutant phenotypes (43%). We interpret the results to mean that there is no substantial redundancy and no strong genetic interaction and that the genes contribute to a large extent independently to prevent apical branching and to maintain the integrity of the growth zones at hyphal tips.

Fig. 7. Double mutant phenotypes indicating that PsmA controls apical branching and hyphal morphology in parallel to and independently of Scy and FilP.

a, b Phenotypes arising from combined deletions of psmA and scy and psmA and filP, respectively. Representative images of the microcolonies grown on cellophane membranes on MYM agar medium are shown. The phenotypes of the mutants with respect microcolony size were estimated as described in Fig. 2c. The difference between the mean for each strain and that of the psmA scy or psmA filP double mutants was tested with Welch’s ANOVA test, followed by Dunnett’s multiple comparisons test. Astrisks represent P values of <0.0001 (****); P = 0.0003 (***); or P = 0.0166 (*). a LUV296 (ΔpsmA Δscy; n = 19) microcolonies are compared to NA1255 (Δscy; n = 25) and LUV282 (ΔpsmA; n = 26). LUV296 was complemented with pKF817 (psmA⁺) and the microcolony size (n = 22) compared to NA1225. Partial restoration of the phenotype of LUV296 to the Δscy mutant level can be noted in the presence of pKF817, both in microcolony size and by the emergence of apparently normally shaped hyphae emanating from mycelial clumps, characteristic of scy mutants. b LUV295 (ΔpsmA ΔfilP; n = 27) microcolonies are compared to NA1225 (ΔfilP; n = 10) and LUV282 (ΔpsmA; n = 28). LUV295 was complemented with pKF817 and the microcolony size (n = 17) compared to NA1225. It can be noted that complementation of psmA in the double mutant leads to restoration of colony area to the ΔfilP mutant size. Scale bar, 100 µm.

Given the relatedness of FilP and Scy, we asked whether FilP also affects the psmA mutant phenotype. Combined deletion of filP and psmA led to reduced microcolony size compared to the two single mutants (Fig. 7b and Supplementary Video 13) and this effect was restored to the microcolony size of the filP mutant on complementation with psmA. The pronounced phenotype of the psmA filP mutant compared to the psmA single mutant suggests that also filP contributes to stabilising growth zones at hyphal tips.

Finally, psmA was deleted in the Δ(scy-filP) background to further test for possible redundant or overlapping roles. Mutants with deletion of all three genes were viable, showed similar microcolony sizes as the scy psmA mutant (Supplementary Fig. S20), and had similarly aberrant morphology and extremely dense and hyperbranching mycelia (Supplementary Video 14). The similarity of the triple ΔpsmA Δ(scy-filP) mutant to the psmA scy mutant is in line with the previously reported epistasis of scy on filP³¹. Overall, we conclude that PsmA has a critical role in maintaining and tuning the stability of the DivIVA-based polarisomes and that this role is largely independent from and additive to the role of Scy and FilP in controlling hyphal tip stability and hyphal branching.

Discussion

We have identified a new component of the complex polar organelle that directs cell polarity and polar growth in Streptomyces bacteria. In contrast to the previously known polarisome proteins DivIVA, Scy, FilP and SepIVA, which consist mainly of coiled-coil structure and may be related to each other³⁰, PsmA is a conserved pseudokinase with a domain organisation typical for hybrid HKs. As summarised in Fig. 8, PsmA strongly affects the polarisome and polar growth zone in ways that have direct consequences for growth, cell shape and hyphal branching.

Fig. 8. Schematic summary of how PsmA affects the polar growth zones at hyphal tips.

PsmA and Scy interact with DivIVA and independently affect the polarisome stability and cell wall synthesis complex (shown in yellow). In the absence of PsmA DivIVA localisation is affected, which affects cell wall synthesis machinery. This effect can be observed in the change in cell morphology and the distribution of DivIVA. Formation of lateral branches by budding-off of polarisome complexes in the wild type is disturbed the mutant, observed as irregular localisation and dynamics of DivIVA leading to irregular morphology and splitting of the tip in a way that is reminiscent of scy mutants. Loss of both scy and psmA leads to additive impairment of proper DivIVA localisation, leading to strong hyper-branching phenotype.

The absence of the conserved H-box histidine residue in PsmA orthologues and the finding that the aspartate in the receiver domain is not required for the detectable function suggests that PsmA does not fulfil the canonical phosphorylation-based functions of bona fide hybrid HKs⁴⁶. It remains to be tested whether the CA domain of the protein can bind ATP and function as an ATPase, but it lacks key residues normally involved in ATP-binding and its predicted structure does not appear compatible with ATP-binding, further supporting that PsmA should be classified as a pseudoenzyme. Further, there is no histidine phosphotransferase domain in PsmA and no phosphotransferase encoded by nearby genes that could be predicted to transfer a phosphate group from the PsmA receiver domain to another response regulator. Thus, available data show that PsmA is unlikely to be involved in typical relaying of phosphate groups. The limited effects on the transcriptome in the psmA mutant with no genes potentially involved in polar growth being clearly affected, also suggest that PsmA is not mainly involved in regulatory pathways leading to transcriptional regulation. In the context of the four classes of mechanisms that pseudoenzymes are involved in, as suggested by Murphy et al.⁴⁷ there are no indications that PsmA would affect or interplay with other signalling enzymes or regulators, mainly because there is no information about any such proteins being directly involved in polarisome function. However, based on the interaction with the main polarisome protein DivIVA, we hypothesise that PsmA has evolved to play roles as scaffold or partner in the assembly of protein complexes, involving protein–protein interactions and that the distinctive mutant phenotype is associated with these roles.

PsmA lacks a predicted transmembrane region, which leaves the N-terminal PAS domains in the cytoplasm. PAS domains are versatile in their binding of ligands and co-factors and they are common as signal-sensing or signal-transducing domains in sensor HKs⁴⁸. The PAS domains in PsmA are essential for complementation of the psmA deletion phenotype, but given that PsmA is not involved autophosphorylation and relaying of a phosphate group, it is possible that the PAS domains have other roles than as sensory domains. In the pseudokinase DivL in Caulobacter crescentus the signal transduction flow is reversed, with the PAS domains acting as the effector domains. According to the proposed model, when the response regulator DivK is phosphorylated by DivJ in the stalked compartment, DivK~P interacts with the catalytic core of DivL, leading to conformational change that affects the interaction of the PAS domains of DivL with the hybrid HK CckA and affects its activity at the new cell pole^49,50. It is also notable that the PAS domains of HK PleC mediate the interaction with an intrinsically disordered domain in the polar scaffold protein PodJ in C. crescentus⁵¹. Since it is the N-terminal part of PsmA that interacts with DivIVA it should be investigated whether the PAS domains are mediators of this interaction.

The IDR (696 residues in S. venezuelae) that connects the HK and REC domains in PsmA is a peculiar but common feature in all PsmA homologues identified in this study and it is essential for the function of the S. venezuelae protein. IDRs can have a range of different roles for example related to cell signalling and cell biological functions^52,53. It remains unclear whether the IDR in PsmA is a flexible spacer between the HK and receiver domains, acts in protein–protein interaction, could contribute to phase-separation behaviour, or have other functions. The fungal polarisome provides interesting parallels. Scaffolding the polarisome to give structural rigidity and interaction sites are important for proper fungal morphogenesis. The coiled-coil protein Spa2 shares some appreciable features with PsmA with high proportion of intrinsically disordered regions providing sites of attachment of polarisome-associated proteins⁵⁴. PsmA may function analogously to Spa2 and provide scaffolding for polar growth-related proteins.

An important feature of polar growth in streptomycetes is that it involves remodelling of the apical DivIVA clusters so that they bud off and deposit smaller daughter polarisomes that are the seeds for initiation of new branches²⁶. This process likely requires tight control of the nature and dynamics of the polarisome and tuning of a certain degree of instability of polarisomes to allow budding-off of smaller daughter foci. We speculate that excessive dynamics and a tendency to split into larger foci will lead to too unstable polarisomes and frequent hyphal bifurcation, which may explain the hyperbranching phenotypes of the psmA and scy mutants. Therefore we hypothesise that PsmA, as well as Scy and FilP, are involved in stabilising and tuning the dynamics of the apical polarisomes in streptomycetes to allow hyphal branching and proper control of cell shape and direction of growth.

Further investigations into the nature of the DivIVA-based apical clusters and how their properties change upon interaction with PsmA, Scy and FilP will clarify the functioning of the polar growth organelle. Equally informative will be to identify additional interaction partners of these polarisome members.

Methods

Bacterial strains, plasmids, primers and growth conditions

All bacterial strains used in the study are listed in Supplementary Table S1 and plasmids are listed in Table S2. Oligonucleotide primers are listed in Supplementary Table S3. The S. venezuelae strains were grown at 30 °C on maltose-yeast extract-malt extract medium (MYM) as described by Bush et al.⁵⁵ or soy flour mannitol medium⁵⁶, unless otherwise indicated. E. coli DH5α or Top10 strains were used for cloning and maintenance of plasmids. E. coli strain DY380 was used for Lambda red-based recombineering. E. coli ET12567/pUZ8002 was used for the mobilisation of plasmids and cosmids containing oriT into S. venezuelae and strain BT340 allowed expression of yeast Flp recombinase. E. coli strain BTH101 was used for bacterial two-hybrid assays.

Generation of psmA mutants

To delete psmA from the S. venezuelae chromosome the gene was first replaced by a resistance marker on cosmid Sv-3-G09 using Lambda red-based recombineering as per our established protocol³². Primers KF1854 and KF1855 were used for amplification of the apramycin resistance cassette (apra) from the plasmid pIJ773 and recombineering of cosmid Sv-3-G09 for creating pKF811. The ΔpsmA::apra gene deletion/replacement on the mutated cosmid was then transferred into S. venezuelae as described previously³².

To enable CRISPR-based deletion of psmA from the genome pCRISPOmyces-2 system was used to target psmA in S. venezuelae as described by Cobb et al.⁵⁷. Briefly complementary primers KF1880/KF1881 spanning 24 bases (nucleotides 3849–3868 of psmA) were annealed together to create the guide RNA sequence with required 4 base compatible overhangs and cloned into pCRISPOmyces-2 using BbsI-based Golden gate assembly, as described⁵⁷, to create pKF821. Primers KF1882/KF1883 were used to amplify the psmA::FRT mutation with flanking sequence from cosmid pKF812, digested with NheI and ligated in the compatible XbaI site of pKF821 to create final plasmid pKF822a.

Generation of plasmids

Q5® high-fidelity DNA polymerase (NEB) was used for all DNA amplifications. All restriction and ligation enzymes used were purchased from NEB. Details of the construction of specific plasmids are provided in the Supplementary Methods.

Bacterial two-hybrid assay

The bacterial two-hybrid system was used according to previously described protocols⁵⁸. E. coli BTH101 cells were co-transformed with pairwise combinations of relevant plasmids and transformants selected using carbenicillin and kanamycin. Both spot assays on plates and assays of β-galactosidase activity were performed as described previously³². Experiments were performed in biological triplicates and technical duplicates and measured in Miller units. Data obtained was plotted using Graphpad Prism (version 8.0.2).

Microscopy

Live imaging was carried out using the CellASIC ONIX2 microfluidic system and B04A-03 microfluidic plates (Merck Millipore), as described previously⁵⁹. Imaging was carried out either on Zeiss AxioObserver.Z1 with thermostated stage enclosure maintained at 30 °C or Nikon Ti Eclipse equipped with a stage incubator. Seeded spores were perfused with indicated medium. For induction of sporulation the bacteria were exposed to spent MYM medium after 6 h of growth in fresh MYM, as described previously⁵⁹. For induction of expression from tetracycline-driven promoter MYM with 50 ng mL⁻¹ anhydrotetracycline was pumped into the growth chambers at indicated time.

Imaging on the Zeiss microscope was carried out with a Plan-Apochromat 100x/1.4 Oil Ph3 objective and phase contrast setting. Fluorescence imaging was done with HXP 120 V light source and appropriate filter sets. Images were captured using ORCA Flash 4.0 LT camera (Hamamatsu). Imaging on the Nikon microscope was carried out with Plan Apo λ 100x Oil Ph3 DM objective with 460 nm light source and an excitation filter of 488–512 nm and emission filter around 550 nm for mNG and WGA-Oregon Green imaging. Nikon DS-Qi2 camera was used to capture the images. mCherry was imaged with 559–585 nm excitation filter and emission filter of 600–690 nm. For 4′,6-diamidino-2-phenylindole (DAPI) fluorescence imaging the light source of 370 nm (CoolLED pE-300^white, Nikon) excitation filter of 379–405 nm; dichroic mirror 409 nm; emission filter of 414–480 nm.

Spore chains and DNA were stained as described previously⁶⁰. Wheat germ agglutinin-oregon Green (WGA-oregon green; molecular probes) was used to stain the cell wall and DAPI was used to stain the nucleoid.

For imaging mycelial microcolonies and their morphology 100 µL of 10⁻⁴ dilutions from a dense spore preparation were spread on MYM agar overlayed with sterilised cellophane. After 16 h of incubation at 30 °C the cellophane was lifted off the plate and mounted on an agarose pad and imaged using phase contrast microscopy with 20× objective. The images were processed in Fiji software. Straight lines were drawn to connect the outer-most hyphal tips and thereby forming an outer circumference of the microcolony and the area of the enclosed space was measured.

Image analysis

To compare the localisation of DivIVA-mCh signals between wild type and LUV282 pSS204 plasmid was introduced into either strain. To study the localisation of DivIVA-mCh depending upon conditional expression of psmA-mNG, pSS204 and pKF852 were introduced into wild-type and LUV293 strains. Cells were cultured in the CellASIC ONIX2 microfluidic system as described above and expression of psmA-mNG was induced at indicated time. Representative image frames from time-lapse sequence were processed using ImageJ. At appropriate time point images of hyphal tips and the corresponding mCh channel images were extracted using a circular region-of-interest (50 pixels across) around the tip, with selection of tips based only on the identification of tip in the phase-contrast channel. Indicated number of tips were isolated from the channels and the distribution of DivIVA-mCh from the mCh channel of these micrographs was used to plot the montages. To compare the shape of DivIVA-mCh foci between strains images from a specific time point in time-lapse sequences like those in Supplementary Video 9 were used and all hyphal tips that were in focus in those image frames were selected and cropped out of the image. The fluorescence foci at the selected hyphal tips (59 for wild type and 66 for psmA mutant) were then identified and analysed for Shape descriptors circularity and area with the MicrobeJ plug-in in ImageJ⁶¹.

For analysis of hyphal tip shape the MicrobeJ plug-in in ImageJ was used⁶¹. Relevant frames from timelapse sequences were exported to ImageJ and images of tips were extracted by drawing a circle at the tip and then cropping the region. A total of 37 tips of psmA mutant and 47 tips of wild type were extracted. The images were used to construct the averaged image of the region in the MicrobeJ plug-in. The average width of the region was also measured and plotted.

For plotting of fluorescence intensity distribution at the tip micrographs were extracted from relevant time point from time lapse sequences of the indicated strains and imported to ImageJ. A line of 15 pixels width was drawn from outside the cell to the inside in phase contrast channel of the image and the fluorescence intensity profile from the corresponding mCh or mNG channels were extracted. The values were imported to GraphPad program (version 8.0.2) and aligned at the highest fluorescence pixel values. Integrated fluorescence intensity was calculated as indicated.

For measuring the frequency of polarisome splitting wild-type or psmA mutant cells expressing DivIVA-mCh were imaged when growing in the CellASIC ONIX2 system, as described above. The interval between images was 10 min. The timelapse sequences were extracted to ImageJ program and the movement of DivIVA tracked using TrackMate program⁶² by using LAP tracker. The parameters for qualifying tip splitting were set by putting track segment splitting to 3 microns. After automated tracking, any missed tracking was corrected and number of split events and track length obtained. The number of splits per micron distance travelled per track was plotted.

Protein production and purification

N-PsmA and VanS cytoplasmic part were produced using E. coli Rosetta-2 cells carrying pKF803 and pKF819, respectively. Overnight culture inoculated from a single colony of cells transformed with the plasmid for protein production was transferred to 250 mL LB and incubated at 37 °C until OD₆₀₀ equalled 0.4–0.6. The production of the protein was then induced with 0.5 mM IPTG for 3 h at 37 °C. The cells were cooled on ice at the end of induction and pelleted (3600 × g, 15 min, 4 °C). Cells were resuspended in 15–20 mL lysis buffer (50 mM Tris-HCl, pH 8.0; 300 mM NaCl; 1% TritonX-100) and sonicated with a 5 s on/7 s off pulse for 20 min total time (VWR sonicator) and centrifuged in conical bottom tube at 50,000 × g for 30 min. The clarified supernatant supplemented with 20 mM imidazole was used for protein purification using Nickel-NTA immobilised affinity chromatography technique. The protein was bound to the Ni-NTA Agarose (Qiagen) for 1 h at 4 °C, washed with wash buffer (50 mM Tris-HCl, pH 8.0; 300 mM NaCl; 40 mM imidazole) for 10 column volumes and then eluted with elution buffer (50 mM Tris-HCl, pH 8.0; 300 mM NaCl; 250 mM imidazole). Samples of eluted fractions were run on 12% Mini-PROTEAN® TGX™ precast protein gels and fractions with protein pooled and dialysed against final buffer (50 mM Tris-HCl, pH 8.0; 300 mM NaCl).

Autokinase assays

Protein that was prepared as described above was subjected to autokinase assay by incubating the protein with radiolabelled γ-³²P ATP (Perkin Elmer: BLU502A250UL) in kinase buffer (5 µg protein, 25 mM Tris-HCl, pH 8.0; 10 mM MgCl₂; 1 mM DTT; 10 µCi mL⁻¹ γ-³²P ATP, total volume 20 µL) at room temperature. Samples of the reaction mixture were taken at the indicated time points and the reaction was stopped by mixing with 2X Laemmli sample buffer (BioRad) and run on BioRad precast stain-free 12% acrylamide SDS-PAGE, as indicated. The gel was sealed in a plastic bag and exposed to phosphorimager screen overnight and was developed using Typhoon FLA 9500 Biomolecular imager (Cytiva).

Pull-down and mass spectroscopy

For pull-down of FLAG-tagged DivIVA wild-type S. venezuelae carrying plasmid pKF752 encoding 3x-FLAG-tagged DivIVA were grown to OD₆₀₀ of 1.0 in 50 mL of MYM medium and then pelleted by centrifugation (3600 × g, 4 °C, 10 min). The pellet was resuspended in 1 mL PBS with cOmplete™ Protease Inhibitor Cocktail (Roche). The samples were mixed with 0.5 mL 0.1 mm Zirconia beads (Biospec products) and lysed with a FastPrep-24 bead beater (MP Biomedicals, 4 cycles, 6 m/s; 30 s; 4 min ice incubation in between). The lysate was spun 2000 x g, 2 min at room temperature to collect the beads. The supernatant was taken and centrifuged 16,000 × g for 15 min at 4 °C. Supernatant was collected and the pellet resuspended in 500 µL of PBS with 1% Triton X-100 at 4 °C for 2 h, re-centrifuged at 16,000 × g for 15 min and supernatants pooled. Pierce™Anti-DYKDDDDK magnetic agarose (25 µL beads per sample) were incubated with the lysate for 1 h at 4 °C and the beads were collected using magnetic rack (Cytiva), washed three times with PBS. The bound protein was eluted with 100 μL of 1.5 mg mL⁻¹ 3x-FLAG peptide (Pierce™ 3x DYKDDDDK Peptide), concentrated in a speedvac vacuum concentrator and analysed using BioRad precast stain-free 12% acrylamide SDS-PAGE.

Mass spectrometry of the identified band was performed using the mass spectrometry facility at Department of Biochemistry and Structural Biology, Lund University. Band of interest was cut after SDS-PAGE separation and submitted to the MS facility and processed as per protocol. Briefly the samples were in-gel digested using trypsin and analysed using MALDI-TOF MS. A MASCOT database search was performed to identify the protein from the peptide fragmentation pattern against a database of S. venezuelae proteins.

Genome sequencing

For sequencing of the genome of wild-type and psmA mutant strain LUV282, genomic DNA was isolated from protoplasts prepared as per previously described protocols⁵⁶, using the nucleospin plant II genomic DNA isolation kit (Macherey-Nagel), as per manufacturer’s instructions. The genome sequencing was carried out using Eurofins Genomics INVIEW resequencing service and Illumina NovaSeq. S. venezuelae reference genome NZ_CP018074.1 was used for alignment of the reads and confirmation of gene deletion. In addition to the expected psmA deletion, two single nucleotide polymorphisms were found at positions 4089233 (G > A) (6 nucleotides upstream of the deleted psmA) and 3857015 (C > T) (synonymous mutation at the Glu80 codon in vnz17470), but they are deemed highly unlikely to affect the phenotype.

Bioinformatics

For the prediction of the intrinsically disordered region in S. venezuelae PsmA, IUPred2⁶³ was used. For studying the synteny of psmA, EDGAR 3.5⁶⁴ was used. The data were exported to GraphPad Prism version 8.0.2 for Windows (GraphPad Software, Boston, Massachusetts, USA, www.graphpad.com) and plotted to highlight the location of psmA. Transcription start site for psmA was obtained from https://streptomyces.org.uk/vnz_tss.html and analysed and plotted as described previously, using Integrated Genome Browser³².

For the identification of the catalytic core of PsmA the protein sequence was aligned with bona fide HKs from S. venezuelae, namely OsaA, PhoR, VanS and KdpD. The H-box in the catalytic core was identified based on the conserved residues of these kinases and critical amino acids used to create the Weblogo (UC Berkley)⁶⁵. For identification of the orthologs of psmA in related actinomycetes, a high-quality genome dataset (Bioproject PRJNA747871)³⁸ was downloaded and searched using BLAST-2.16.0+. PsmA (vnz_18495; WP_014047252.1) from S. venezuelae NRRL B-65442 was used as query. Nine hundred ninety-three assemblies were downloaded and individually searched for the presence of PsmA (e-value of 10⁻³). The top BLAST results were screened with qualifier for sequence length >1000 amino acids, which is commonly the minimum length of PsmA in all identified Streptomyces orthologs. Eight hundered seventy four positive sequence IDs were passed to Entrez of which 853 sequences were downloaded and screened for any false positives (actual HKs) manually using Jalview (Version: 2.11.4.1). Seven sequences were found to be false positives (WUL52868.1, WUV05760.1, WSF07443.1, WUD25642.1, WTY92036.1, WSY14497.1) (Supplementary Data File 6), rest of the 846 sequences were actual orthologs of PsmA. One hundred twenty-six bacteria from the collection show the absence of psmA (Supplementary Data File 6). These sequences were passed through Entrez of which 112 sequence IDs were accepted and were downloaded from Entrez. Seven genomes of Streptomyces and one of Kitasatospora were negative for PsmA ortholog, which were further examined individually using the tblastx program and found to possess genes aligning with psmA, but annotated as pseudogenes due to frameshift, likely arising due to reading error. These 8 sequences were not included in the final plot. The bacteria were grouped by presence or absence of PsmA ortholog and were searched in NCBI taxonomy browser within the phylum Actinomycetota (https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=201174) and used to plot order and families where orthologs are present or absent (Supplementary Fig. S12).

For the identification of the H-box in the catalytic core, the PsmA sequences identified from Bioproject PRJNA747871 were examined. Sequences were aligned using ClustalOmega⁶⁶ and analysed using Jalview (Version: 2.11.4.1). Weblogo 3⁶⁵ was used to create logo for residue conservation around the H-box. The IPR043836 entry for DhP dimerisation and H phosphotransfer domain sequences were downloaded from Interpro and aligned using ClustalOmega and Weblogo 3 was used to generate the catalytic core representation.

The Alphafold modelling for the structure of full-length PsmA and the N-terminus dimer was done with assistance of the LU-fold facility (https://www.medicine.lu.se/research/list-research-infrastructures/lu-fold). Alphafold version 2.3.1 was used for creating PsmA models. Top scoring structure are shown. For visualisation of the modelled proteins, UCSF ChimeraX program was used⁶⁷.

Transcriptomics

For transcriptomics comparison of wild-type and the LUV282 psmA mutant cultures were grown in 25 mL of MYM to OD₆₀₀ of 1.0 and then RNAprotect Bacteria Reagent (Qiagen 76506) was added to preserve the RNA. Cells were harvested by centrifugation (3600 × g, 4 °C, 10 min) and processed for extraction of total RNA using RNAeasy kit (Qiagen) using the bacterial RNA protocol as per manufacturer’s protocol.

Strand-specific RNA-seq was performed by Eurofins INVIEW transcriptome bacteria service using the Illumina NovaSeq platform. This included rRNA depletion, library construction, sequencing and bioinformatics analysis. Raw sequencing data was processed using fastp⁶⁸ software to remove poor quality bases (below Phred Quality 20) using the sliding window approach wherein if the average quality of the bases drops below Q20, those bases are removed from the reads. After quality trimming the program checkes for presence of any adaptors in the reads and removes them. Further shorter reads, which are <30 bp length are also removed to retain only high-quality sequencing reads for each sample in the analysis. Mapping and alignment (to S. venezuelae reference genome NZ_CP018074.1) were done using STAR aligner⁶⁹. Transcript quantification was done using featureCounts program⁷⁰. Differentially expressed genes were quantified using EdgeR program⁷¹. Differentially expressed genes obtained after 2DGE analysis were plotted using ShinyGO⁷² to visualise the pathway enrichment. False discovery rate cutoff used for the search was set at >0.05.

Statistics and reproducibility

Statistical analyses were done using GraphPad Prism version 10.6.1 for macOS (GraphPad Software, Boston, Massachusetts, USA, www.graphpad.com). In comparisons of microcolony areas between strains, the differences of the mean for each strain compared to a control strain were tested by Welch one-way ANOVA, followed by Dunnett’s multiple comparisons test. Alpha was 0.05. In Fig. 6b, the difference between the distribution of values was tested using a two-tailed Mann–Whitney test. In comparison of area and circularity of apical DivIVA foci in Supplementary Fig. S19, the difference between means were tested using an unpaired two-tailed t-test, with confidence level 95%. In Supplementary Fig. S9e, f, Welch one-way ANOVA was done to compare the mean of each strain to all other strains, followed by Dunnett’s multiple comparisons test. Alpha was 0.05. Sample sizes are given in each figure. Numerical data and the details of statistical tests are provided in Supplementary Data File 7. Generally, reproducibility was ensured by repeating each experiment at least twice with consistent result.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

P.S.M. and K.F. designed the study. P.S.M. performed the experiments and analysed the data with inputs from K.F. P.S.M. and K.F. wrote the manuscript.

Peer review

Peer review information

Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary handling editor Tobias Goris.

Funding

Open access funding provided by Lund University.

Data availability

The genome sequence and transcriptome analysis datasets generated and analysed during this study are deposited at the European Nucleotide Archive (ENA) at EMBL-EBI under accession numbers PRJEB92325 for the psmA mutant genome verification and PRJEB92404 for transcriptome analysis. The data for transcription start site mapping that was analysed for the generation of Fig. S5 is available at ArrayExpress (accession number E-MTAB-10690), generated and deposited by Mark Buttner, Govind Candra and Matthew Bush, John Innes Centre, Norwich, UK. Alphafold models are provided as Supplementary files. Numerical data and the details of statistical tests are provided in Supplementary Data File 7. Other data from the current study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09620-z.