SPO1 gene 44 and the truncated constructs were PCR amplified from SPO1 genomic DNA using pET EK/LIC primers for ligation independent cloning and cloned into pET-46 EK/LIC for adding N terminal His-tag and pDE2 vector for C-terminal His-tag or tag free constructs. Three Gp44 constructs: (1–55), (1–122), and full length (1–237) were made by nickel affinity purification from E. coli strain BL21(DE3). The culture of BL21(DE3) cells contains pET-46 Gp44 construct, was grown at 37°C to OD₆₀₀ of ∼0.6 and induced by 1 μM of IPTG. The cells were left to continue growing at 37°C for 4 h before harvesting. The cell pellet was re-suspended in binding buffer (50 mM NaH₂PO₄, pH 8.0, 0.3 M NaCl, and 5 mM DTT) containing cocktail of protease inhibitors and lysed by sonication. The cleared cell lysate was loaded onto a His-Trap HP column (GE Healthcare Life Sciences), which was connected to AKTA pure chromatograph machine. The purified protein was eluted over a 50 ml gradient of 0–100% Elution buffer (Binding buffer + 1 M imidazole pH 8.0) according to manufacturer’s instructions. The purified protein was dialyzed into storage buffer (50 mM NaH₂PO_4, pH 6.8, 250 mM NaCl, 1 mM DTT) and concentrated for NMR studies. Expression and purification of β and β′ subunits were using pDE1 vector and followed the same protocol described above. Expression and purification of recombinant B. subtilis RNA polymerase followed the protocol described by Yang and Lewis (2008).

For His-tag protein pull down assay, the Ni-NTA column was first equilibrated with five column volumes of lysis buffer (10 mM imidazole) then the His-tagged protein was applied into the column. The column was subsequently washed with five column volumes of lysis buffer before applying supernatant of B. subtilis (strain 168) cell lysate which was harvested at 0.8 of OD₆₀₀. The column was washed with 10 column volumes of lysis buffer before samples were eluted with 50 μl of Laemmli 2x concentrate SDS Sample Buffer. Samples were then boiled for 5 min prior to analysis by SDS-PAGE.

Nuclear magnetic resonance spectra were collected at 310K on Bruker DRX600 and DRX800 spectrometers equipped with cryo-probes. Spectral assignments were completed using our in-house, semi-automated assignment algorithms and standard triple-resonance assignment methodology. H_α and H_β assignments were obtained using HBHA (CBCACO)NH and the full side-chain assignments were extended using 3D HCCH-TOCSY and (H)CCH-TOCSY experiment. Three-dimensional ¹H-¹⁵N/¹³C NOESY-HSQC (mixing time 100 ms at 800 MHz) experiments provided the distance restraints used in the final structure calculation. The ARIA protocol was used for completion of the NOE assignment and structure calculation. The frequency window tolerance for assigning NOEs was ±0.025 ppm and ±0.03 ppm for direct and indirect proton dimensions and ±0.6 ppm for both nitrogen and carbon dimensions. The ARIA parameters p, Tv, and Nv were set to default values. 110 dihedral angle restraints derived from TALOS were also implemented. The 10 lowest energy structures had no NOE violations greater than 0.5 Å and dihedral angle violations greater than 5°. The structural statistics are shown in Supplementary Table 1.

One equivalent of β or β′ subunit of either E. coli or B. subtilis was added the unlabeled full length Gp44 to perform the 1D titration. Maximal twofolds of protein were added to Gp44 in order to reach the end of the reaction. For the DNA titration, double stranded DNA fragments was added to ¹⁵N labeled Gp44 constructs according to stoichiometric ratio to perform NMR titration. Maximal fivefold dsDNA was added to Gp44 in order to reach the end of the reaction.

Isothermal titration calorimetry (ITC) experiments were performed on a MicroCal PEAQ-ITC Instrument (Microcal) at 25°C using the dialysis buffer described above. β or β′ subunit 30 μM in the cell was titrated with 300 μM Gp44 in the syringe via 19 injections with 2 μL each at 120 s interval. The raw data were integrated, normalized for the molar concentration and analyzed using MicroCal PEAQ-ITC Analysis Software.

Start cultures were grown at 37°C, shaking at 700 rpm. for 6–7 h in a plate incubator by directly inoculating a colony from a freshly transformed Luria agar plate into 200 μl of Luria broth (LB) medium containing 100 μg carbenicillin ml^–1 into a 96-well microtiter plate (Corning). The start cultures were diluted 1:100 in a final volume of 200 μl of fresh LB medium containing 100 μg carbenicillin ml^–1 and incubated at 30°C, shaking at 500 rpm. The expression of Gp44 constructs was induced at OD₆₀₀ of ∼0.2–0.25 by adding 0.1 μm of isopropeyl-β-D-thiogalactoside (IPTG) for E. coli strains DH5α and ATCC35218. The experimental growth curves were also performed in 96-well microtiter plates in a Cytation3 multi-well plate reader (BioTek). At least three biological and technical replicates were performed for each growth curve.

Bacterial two-hybrid assay was performed as previously described (Rao et al., 2009) with minor modifications. Briefly, GP44 (or mutant) and RNAP subunit were first subcloned into pAC and pRB constructs, respectively. KS1 reporter strains cells were then co-transformed with the indicated pAC and pBR derived plasmids and grown in LB containing 100 μg/ml carbenicillin, 35 μg/ml chloramphenicol and 50 μg/ml kanamycin. pAC and pRB empty vectors were used as negative control, while pAC-β flap and pRBL28 co-transformation were used as positive control. When the OD₆₀₀ of the culture reached 0.3, cells were treated with indicated concentrations of isopropeyl-β-D-thiogalactoside (IPTG) for 1 h followed by 50 μM Fluorescein di-β-D-galactopyranoside (FDG) for another 1 h. β-galactosidase activity was assayed by measuring the fluorescence intensity of fluorescein with Cytation 3 plate reader equipped with 485 ± 20 nm excitation and 520 ± 25 nm emission filters.

Two homologous conserved fragments were extracted from Gp44^56–122 using RADAR (Li et al., 2015). The initial disordered 3D structures of the fragments were obtained from I-TASSER, and the starting structures were chosen after energy minimization. The fragments were solvated with water, in cubic box and Na⁺ and Cl^– ions were added to balance the charges. An improved CHARMM36 force field which was optimized to simulate both ordered and disordered proteins (Huang et al., 2017) were employed to model the protein molecules. MD simulations were performed using GROMACS 4.6.7 molecular dynamic engine (Hess et al., 2008). The fragments were constrained, and their energy minimization were carried out, followed by the 1 ns equilibration at NVT and NPT ensembles at 300K and 1 bar, respectively. The fragments were then simulated for 500 ns each, with protein snapshot saved every 500 ps (running parameters are summarized in Supplementary Table 3).

Identification was done locally at instrument analysis center. Samples were digested with trypsin and desalted by a C18 ZipTip. Peptides were analyzed by Q-Exactive plus mass spectrometer (Thermo Fisher Scientific) coupled with an UltiMate 3000 RSLCnano system. Database search were performed by MaxQuant.

To demonstrate Gp44 directly interacts with RNAP, we first expressed recombinant Gp44 fused with Histidine tag (His-tag) in E. coli. A pull-down assay using whole-cell extracts of exponentially growing B. subtilis cells expressing His-tagged Gp44 revealed a band around 120 kDa, which can be recognized by antibody against β subunit of B. subtilis RNAP and mass spectroscopy results confirmed that sample contains both β and β′ subunit (Figure 1A). To further localize the subunit(s) that Gp44 interacts with, we analyzed into the RNAP interacting region of σ⁵⁴, which interacts with both β and β′ subunits (Figure 1B) and postulated that Gp44 might target the β and β′ subunits. We thus used 1D ¹H NMR titration and a bacterial two-hybrid assay (BTH) to verify the hypothesis. In the NMR experiment, we successfully expressed and purified four recombinant subunits (β and β′ subunits of E. coli and B. subtilis) and titrated with Gp44 separately. After addition of any of the four subunits, all ¹H spectra of Gp44 demonstrated peak broadening effects, suggesting a productive interaction (Figure 1C). Furthermore, we used ITC to determine the binding affinities (Supplementary Figure 1A). The K_Ds for the two interactions were determined at 3.46 ± 0.27 μM for Gp44-β and 4.08 ± 0.31 μM for Gp44-β′, respectively. These values were consistent with the peak broadening observed in the NMR titration experiments as both fell into the intermediate NMR timescale. Meanwhile, BTH assay results confirmed the NMR results, showing increases of β-glactosidase activity after IPTG induction of Gp44 and RNAP subunit expression (Figure 1D).

While RNAP of E. coli has five subunits—α₂ββ′ω, that of B. subtilis has seven subunits α₂ββ′δω₁ω₂ representing Gram-positive bacteria (Weiss and Shaw, 2015). As Gp44 is able to interact with both RNAP subunits regardless of structural variations suggests the possible existence of a new RNAP inhibitory mechanism.

To understand the mechanism of Gp44 inhibition further, we employed various structural and biophysical tools for analysis. Despite its resemblance to σ⁵⁴ in the central region (amino acids 56–122 of Gp44 to 44–113 of σ⁵⁴) (Supplementary Figure 1B), the overall amino acid sequence of Gp44 only shares similarity among Bacillus phages in BLAST analyses. No homologous structures have been deposited in Protein Data Bank (PDB) and 3D structural prediction servers failed to produce meaningful predictions. As there are clusters of continuous negatively charged amino acids in its sequence, it is conceivable that Gp44 may not form regular secondary structure due to electrostatic repulsion in these regions. To assess the intrinsic disorder in this region, we use DISOPRED server (Ward et al., 2004) to identify regions in Gp44 that have increased propensity to be disordered. As shown in Figure 2A, the middle of Gp44 ranging from residue 56–122 (Gp44^56–122) has a very high possibility of being disordered. Strikingly, this region consisted primarily of negatively charged amino acids where either aspartate or glutamate making up 57% of amino acids in this region (Figure 2B), which is more extreme than its counterpart in σ⁵⁴ which carries 22% negatively charged amino acids. Notably, no positively charged amino acid residues (arginine, histidine, and lysine) were present in this region. Nevertheless, the percentage of charged amino acids in the rest of the sequence is also high in the rest of the sequence.

To identify the existence of any folded globular regions, we expressed and purified N-terminal histidine tagged (His-tagged) full length and truncated constructs of Gp44 and examined their foldedness by NMR. Dispersion in the ¹H NMR spectrum of full-length Gp44 revealed the clear presence of a folded domain as well as unstructured region (Figure 2C). The spectrum of a C-terminally truncated construct (Gp44^1–122) shows the folded domain is still present with only unstructured residues removed. After further removing central acidic region, the final construct (Gp44^1–55) is fully folded and feasible for NMR structure determination.

Our full length recombinant Gp44, either tag free or carboxy-terminal (CT) His-tagged, fully inhibits the growth of both E. coli and B. subtilis as reported by Wei et al., suggesting the recombinant protein used in our NMR experiments is its natural functional state. As the presence of a functional protein ensures the comprehensive structural study of Gp44 based on the initial NMR structural analysis, we are able to divide it into three distinctive domains—a folded NT domain, a negatively charged middle domain and an unstructured CT domain.

We determined its solution structure of Gp44^1–55 using standard multidimensional NMR spectroscopy (PDB ID: 6L6V and BMRB ID: 36290). A notable feature in the solution structure of Gp44^1–55 is that surface-exposed positively and negatively charged residues are located on opposite sides of the protein (Figure 3A). This type of charge distribution, especially with the positively charged residues on an α helix (α2 in Gp44^1–55) resembles the charge distribution seen in a helix-turn-helix DNA binding motif. Indeed, a search for protein structure similarities using PDBefold (Krissinel and Henrick, 2005) revealed that the overall structure of Gp44^1–55 exhibits a statistically significant similarity to an unusual DNA binding motif found in bacteriophage Lamda Xis protein—a DNA binding excisionase (Abbani et al., 2007). Xis adopts an unusual winged-helix structure in which two α helices are packed against two extended strands and similar positioning of α1 and α2 also found in Gp44^1–55 (Figure 3B). In Xis, the four-residue linker between the two-stranded anti-parallel beta-sheet is called the “wing” (Figure 3B). Although the wing is “missing” in Gp44^1–55, the N terminal region of Gp44^1–55 bends back to α2 forming a loop to potentially mimic the Xis wing. When Xis interacts with DNA, α2 inserts into the major groove, while the wing contacts the adjacent minor groove and phosphodiester backbone. The overlay of Xis and Gp44^1–55 with DNA suggests that Gp44^1–55 could interact with DNA in a similar manner to Xis (Figure 3C).

To determine whether Gp44^1–55 can indeed interact with DNA, we performed NMR titration experiments with two different double-stranded DNA probes containing strong promoters recognized by the E. coli or B. subtilis RNAP (Supplementary Table 2). Results revealed that Gp44^1–55 binds double stranded DNA without displaying any preference for either DNA probe (Figure 3D). Overall, we conclude that the amino terminal region of Gp44 constitutes a DNA binding motif and build a Gp44NT/DNA complex model based on the published Xis/DNA complex structure and our NMR data (Figure 3E). Another interesting observation is that N-terminal (NT) His-tag block Gp44^1–55 interaction with dsDNA (Figure 3D), perhaps by occluding the positive charged patch on Gp44 and inhibiting the interaction. Consistent with this, NT His-tagged Gp44 does not inhibit bacterial growth, indicating DNA binding is crucial for the function of Gp44 (Supplementary Figure 2).

As the central region of Gp44 is predicted as disordered with significant negative charge and the equivalent region in σ⁵⁴ (44–113) is mostly undefined random coiled yet contacting both β and β′ in the crystal structure of σ⁵⁴-RNAP holoenzyme (Tabib-Salazar et al., 2019), we chose molecular dynamics to simulate this region to see is any interesting conformations are sampled and stable within the ensemble. First, we were able to extract two similarly conserved fragments from this region using RADAR (Li et al., 2015; Figure 4A) for use in simulations. These two fragments were simulated for 500 ns each under the updated CHARMM36 force field, optimized for both ordered and disordered proteins (Huang et al., 2017). The two fragments retain their random coil configuration throughout the simulations, not forming any secondary structure elements at any stage. The repulsion between sidechains of the highly charged segments prevents the formation of tertiary contacts. A representation of the ensembles of disordered conformations populated by fragments A and B during the simulations is shown in Figure 4B. The negatively charged and highly flexible nature of these fragments is comparable to that of single-stranded DNA (poly-thymidine) (Figure 4C), potentially suggesting that, in addition to DNA binding, DNA mimicry, is involved in the mode of action of Gp44. Flexibility is one of the interesting features for Gp44 as all previous reported DMPs do not have DNA-like flexibility due to rigid tertiary structure of proteins. Our attempts to express recombinant Gp44^55–122 to conduct in vitro assays also failed, but BTH assays showed that this region alone is able to interact with β and β’ subunit of B. subtilis (Figure 4D). Interestingly, Gp44^1–122 still has inhibitory effects although not as potent as the full length Gp44 in the growth attenuation assays (Supplementary Figure 2). These observations suggest that the construct with just a DNA binding motif which docks Gp44 on host genome and a DNA mimic region which binds to RNAP is capable of inhibiting host growth.

Our 1D NMR analysis shows that the C-terminal region does not possess any folded domains. Compared to the full length Gp44 NMR spectrum, line widths of folded domain peaks at 0.5 ppm increases and the relative intensity of these peaks compared with unstructured peaks decreases for Gp44^1–122 despite being a shorter construct, suggesting the C-terminus may alleviate aggregation. This region possesses both positive and negative charges in mosaicking blocks, which may play a role in preventing self-tangling (Weiss and Shaw, 2015) keeping the DNA mimic domain accessible (Figure 2C). The inhibitory effect of Gp44ΔCT (Gp44^1–122) is also affected (Supplementary Figure 2), suggesting it is also required for optimal inhibition of the host transcription by this phage protein. Consistent with this, we observed less response of domain 56–122 compared to full length Gp44 in the BTH assay (Figure 4D), also indicating that C-terminal is required for its optimal function.