Autoregulation ensures vertical transmission of the linear prophage GIL01

Abstract

Betatectiviruses are prophages consisting of linear extrachromosomal genomes without obvious plasmid modules. It remains unclear how betatectiviruses are maintained in low-copy numbers in host cells and how they are vertically transmitted. Phage GIL01 is a model betatectivirus that infects the mosquito pathogen Bacillus thuringiensis serovar israelensis. Previous studies identified two closely spaced promoters, P1 and P2, responsible for the expression of GIL01 genes required for prophage replication and the switch from the lysogenic to lytic cycle. Here, we report that the GIL01-encoded 58-amino acid long gp1 protein forms a large nucleoprotein complex that represses its transcription from the strong promoter P2. Notably, ectopic expression of gp1 resulted in the loss of GIL01 in exponential cultures and immunized cells against infection with GIL01, indicating that gp1 plays a repressive role in the phage cycle. This finding is consistent with mutations in gp1 committing GIL01 to the lytic cycle and we show that maintenance of this phage variant in the bacterial population is contingent on the accumulation of deletions in the P1-P2 region. The fact that gp1 is conserved across most sequenced betatectiviruses suggests that the regulatory mechanism of gp1 that controls prophage maintenance is widespread among these bacteriophages.

Introduction

Bacillus thuringiensis serovar israelensis is the most widely used larvicide to control mosquito populations¹. Most isolates of this serovar carry the betatectiviral prophage GIL01². In the lysogenic cycle, the 15-kbp GIL01 genome resides in the host cell as an autonomous linear replicon with terminal proteins covalently attached to both 5′-ends³. The terminal proteins appear to associate with the bacterial nucleoid, which is important for efficient replication of the phage DNA⁴. The 30 ORFs of the GIL01 genome are clustered into two transcriptional and functional domains (Fig. 1)⁵: tandem promoters P1 and P2 are located upstream of the left arm (~4.8 kbp) of the phage genome associated with replication and transcriptional regulation, and promoter P3 lies upstream of the right arm (~10.2 kbp) encoding capsid and cell lysis proteins⁶.

Fig. 1: Schematic representation of the phage GIL01 genome.

The 15-kbp genome of phage GIL01 comprises 30 ORFs, all transcribed in the same left-to-right direction and grouped into two functional modules: on the left, a 4.8-kbp module containing genes required for the regulation and replication of the GIL01 genome, and on the right, a 10.2-kbp module containing structural and lytic genes. Each ORF is color-coded to indicate a function described in the legend. Transcription of the left module is controlled by the P1 and P2 tandem promoters, and that of the right module is controlled by the P3 promoter. All three promoters are indicated with red arrowheads. The regulatory region containing P1 and P2 is enlarged below the schematic of the GIL01 genome. The P1 and P2 transcription start sites are indicated with angled arrows, the -10 and -35 promoter elements are denoted with black boxes, the LexA binding sites—canonical dinBox1 and non-canonical dinBox1b—are denoted with blue boxes, the gp1 nucleation site is shown in purple, and a partial ORF1 is represented in red. Below the -10 promoter elements are the nucleotide substitutions (highlighted in red) that form derivatives of the wild-type P1⁺P2⁺ promoter region with inactivated P1 (named P1^-P2⁺) and inactivated P2 (named P1⁺P2^-), both used in promoter–lacZ fusions for measurement of ß-galactosidase activity (Fig. 3). Below the sequence of the gp1 nucleation site (operator 1; purple box) is the sequence of the Δop1 deletion mutant, with displaced nucleotides in red, that was used along with the wild-type sequence in lacZ fusions, DNase I footprinting (Fig. 2A, F), surface plasmon resonance (SPR) (Fig. 2D), and electrophoretic mobility shift assay (EMSA; Fig. 2E) experiments. The positions and GIL01 genome coordinates (GenBank ID, AJ536073.2) of the probes used in these experiments are indicated. This figure was created with BioRender.com.

The presence of closely spaced promoters such as P1 and P2 is common in bacterial genomes; however, their characterization remains challenging owing to their proximity⁷. To silence gene expression during the lysogenic cycle, GIL01 utilizes the LexA repressor protein, a key regulator of the host DNA damage response. LexA binds to canonical (dinBox1) and non-canonical (dinBox1b) target sites downstream of the weak promoter P1⁶. However, the intrinsic properties of LexA do not allow it to stably bind to the low-affinity site dinBox1b. To effectively repress P1 activity, the GIL01-encoded 50-amino acid gp7 must form a complex with LexA and serve as a scaffold to position the repressor in a DNA-binding conformation that enables LexA to also occupy dinBox1^8,9. Thus, GIL01 uses LexA as an indicator of host cell viability and gp7 to stabilize LexA binding to target sites on promoter P1 and prevent premature phage reactivation^6,8,9.

Genome sequencing coverage experiments and quantitative real-time PCR analysis found prophage GIL01 to be present at 5–15 copies per cell^10,11. In contrast to circular phage-plasmids that appear to have arisen from the fusion of a plasmid and a prophage^12,13,14, GIL01 essentially encodes phage structural, replication, and regulatory proteins. GIL01 and other betatectiviruses lack recognizable genome maintenance and copy number control functions, which are classically encoded by the plasmid modules¹³. In the present study, we demonstrate that the ~6.5-kDa protein gp1 is crucial for maintaining the lysogenic state and for the stable vertical inheritance of prophage GIL01. We found that gp1 is a repressor of the strong promoter P2 and likely regulates the expression kinetics of GIL01 replication and regulatory genes through a negative feedback loop. Moreover, we confirmed that only the weak promoter P1, which is controlled by LexA and gp7⁸, is induced by DNA damage. Furthermore, we showed that GIL01 evolves rapidly to circumvent the mutation in gp1 that prevents lysogeny, by deleting or weakening P2 through homologous recombination. Our findings contribute to the understanding of the regulatory mechanisms controlling the maintenance of prophages in bacteria.

Results and Discussion

GIL01-encoded gp1 forms a large nucleoprotein complex with promoter P2

In a previous study, we found that mutations in GIL01-encoded gp1 committed the phage to the lytic cycle⁶, leading to speculating that gp1 is involved in the regulation of P1-P2 activity and its own downstream transcription (Fig. 1). To investigate the binding of gp1 to the P1-P2 region, we purified gp1 and performed in vitro DNA binding experiments. DNase I footprinting showed that gp1 forms a large nucleoprotein complex by protecting seven sites over a 65-bp segment overlapping with the -35 element of P2 and the downstream transcribed region (Figs. 1 and 2A). Examination of the interaction of gp1 with the P2 region revealed 11 DNase I hypersensitive sites (Fig. 2A, B), indicating local bending of the DNA helix at these sites as a result of a widened minor groove and increased susceptibility to DNase I¹⁵.

Fig. 2: Gp1 forms a large nucleoprotein complex with promoter P2.

A, F DNase I footprint analysis of His₆-gp1 binding to the GIL01 P1-P2 promoter region. The negative strand of DNA probes encompassing the wild-type gp1 nucleation site (215 bp) (A) or Δop1 mutant (207 bp) (F) (GenBank ID, AJ536073.2; genome coordinates 136 to 350; Fig. 1) was labeled, incubated with increasing concentrations (200 to 3200 nM in two-fold increments; lanes 2 to 6) of His₆-gp1, and digested with DNase I. The cleavage products were separated on an 8% sequencing gel. G + A lanes contain purine cleavage products corresponding to the respective ³²P-labeled DNA probes and the first marked lanes contain DNA probes without added protein. The positive strand sequences of the wild-type gp1 nucleation site (A) and the Δop1 mutant (F) are shown to the left of the footprints. The P1 and P2 transcription start sites are indicated with angled arrows. The -10 and -35 sequences of P2 are shown in grey and the -10 promoter element of P1 is indicated with a number. The LexA-binding sites (dinBox1 and dinBox1b) are denoted in blue. Arrows to the right of A indicate DNase I hypersensitive sites. B Nucleotide sequence protected by His₆-gp1 in DNase I footprint analysis. The P2 transcription start site is indicated with an angled arrow, and the -10 and -35 elements of P2 are highlighted in black. The region protected by His₆-gp1 is underlined, and the gp1 nucleation site is highlighted in purple. Vertical arrows above the sequence indicate the positions of DNase I hypersensitive sites. C Model of the heptamer gp1 filament interacting with the 65-bp DNA fragment protected in DNase I footprint analysis (shown in gray and purple) flanked by 6 nucleotides at each extremity (shown in black). The gp1 nucleation site (operator 1) is shown in purple and the P2 transcription start site (TSS) is indicated with an angled arrow. The figure was created using program ChimeraX²². D Surface plasmon resonance (SPR) analysis of the interaction of His₆-gp1 with the 81 bp DNA fragment of the P2 promoter region (GenBank ID, AJ536073.2; genome coordinates 236 to 317; Fig. 1) carrying the wild-type gp1 nucleation site or Δop1 mutant. Gp1 was injected over each DNA fragment immobilized on the chip (50–55 RU) for 90 s at a flow rate of 30 µL min^-1 in concentrations ranging from 4 to 64 nM in two-fold increments. The sensorgram shows the concentration-dependent responses of His₆-gp1 interaction with the wild-type gp1 nucleation site (black lines) or the Δop1 mutant (red lines). The obtained association rate constants (k_a) of three independent experiments (n = 3) and standard deviations are shown in respective colors. Nucleation site sequences are shown to the right of the sensorgram, with displaced nucleotides in red. Experiments were performed in triplicate and a representative sensorgram is shown. E Electrophoretic mobility shift assay of His₆-gp1 binding to the GIL01 P1-P2 promoter region. ³²P-labeled DNA fragments of the P1-P2 promoter region (GenBank ID, AJ536073.2; genome coordinates 151 to 350; Fig. 1) carrying the wild-type gp1 nucleation site (200 bp, left panel) or the Δop1 mutant (192 bp, right panel) were used. Nucleation site sequences are shown above the gel, with the displaced sequence in red. Lanes 1 and 7 is labeled DNA without added protein. His₆-gp1 concentrations range from 2 to 32 nM (in two-fold increments) in lanes 2 to 6 and lanes 8 to 12. The black arrows to the left indicate the bands corresponding to DNA and gp1–DNA complexes.

To gain insights into the structure of the gp1 nucleoprotein complex, we generated an in silico structural model of the gp1-DNA complex using AlphaFold 3¹⁶. Using a DNA sequence corresponding to the 65-bp long site protected by gp1 in the footprint analysis (Fig. 2A, B) and flanked by 6 nucleotides at each extremity, we obtained a model in which gp1 binds DNA through a winged helix-turn-helix (wHTH) motif to form a heptameric filament (Fig. 2C). We next used surface plasmon resonance (SPR) to test the binding of gp1 to the 81-bp long DNA fragment comprising P2 in addition to the downstream gp1 binding sites (Fig. 2D). The maximal response obtained with gp1 bound to P2 (Fig. 2D), and the molecular masses of the recombinant gp1 protein and chip-immobilized DNA, further validated DNase I footprinting and modeling findings of seven gp1 molecules in the nucleoprotein complex (Fig. 2C).

To further investigate the role of gp1 at P2, we analyzed the promoter regions of several GIL01 variants that spontaneously formed clear plaques on lawns of GBJ002, a B. thuringiensis serovar israelensis strain cured of GIL01. One such lytic variant, cp31, had a deletion of nine nucleotides (Δop1) that form a perfect direct repeat sequence, named operator 1 (5′-GGGACAAAcGGGACAAA), downstream of the -10 element of P2 (Figs. 1 and 2B). DNase I footprinting, SPR, and electrophoretic mobility shift assay (EMSA) showed significant differences in gp1 binding to the wild-type operator 1 and mutant Δop1 (Figs. 2A, 2D–F), such as fewer DNA bending sites and reduced binding affinity to Δop1 probes. These results suggest that the direct repeat motif serves as a nucleation site from which gp1 assembles into a larger nucleoprotein complex covering the P2 promoter. The gp1-DNA structural model (Fig. 2C) further highlighted similarities with the Streptomyces protein BldC, a 68-amino acid transcription factor with a DNA-binding MerR-like wHTH motif but lacking the coiled-coil dimerization and effector binding domains characteristic of MerR¹⁷. Similar to BldC, which forms a homomeric 11-mer nucleocomplex¹⁷, gp1 likely undergoes asymmetric head-to-tail oligomerization at the direct repeat of operator 1 (Fig. 2C). This may enable gp1 to spread and bind adjacent DNA, a feature reported for the non-homologous Apl protein of coliphage 186. Apl functions as both an excisionase and a transcriptional regulator, binding cooperatively to several target sites and subsequently to adjacent non-specific DNA¹⁸. Because most of the gp1 sequence is a predicted DNA-binding domain (residues 4–45 of 58 amino acids, InterPro), it likely does not harbor a regulatory effector-binding domain¹⁹, further suggesting that its activity is determined by its auto-regulated expression and concentration inside the cell.

Gp1 represses the promoter P2

To elucidate the specific role of gp1 in regulating promoters P1 and P2, we generated P1-P2 fusions with the Escherichia coli lacZ reporter gene and examined LacZ expression in the cured host, GBJ002, and the lysogen, GBJ002(GIL01). We first studied the wild-type P1-P2 promoter region (P1⁺P2⁺) and found that it was strongly active in the absence of GIL01 determinants (Fig. 3A). In contrast, in GBJ002(GIL01), P1⁺P2⁺ activity was significantly repressed (up to 16-fold) and repression was alleviated by three-fold in response to the DNA damage-inducing agent mitomycin C (MMC; Fig. 3A). We attribute this moderate induction to derepression of P1 as a consequence of LexA proteolysis and a decline in LexA-gp7 complexes under MMC treatment^6,8. Next, we investigated the effect of gp1 on P1-P2 activity. For this, we ectopically expressed gp1 from an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible promoter in strain GBJ002(pDG1). Gp1 repressed transcription from P1⁺P2⁺ to similar levels as observed in GBJ002(GIL01); however, in contrast to the lysogen, gp1 overexpression overrode induction in the presence of MMC (Fig. 3A).

Fig. 3: Gp1 represses SOS-independent promoter P2.

Bar charts illustrating the ß-galactosidase activity in GIL01-cured GBJ002, GIL01 lysogen GBJ002(GIL01), or gp1-expressing GBJ002(pDG1), carrying the following promoter–lacZ fusions: A wild-type P1⁺P2⁺ promoter region, B P1⁺P2⁺ (Δop1) with partial deletion of the gp1 nucleation site, C P1⁺P2^- with mutated -10 promoter element in P2, and D P1^-P2⁺ with mutated -10 promoter element in P1. Isopropyl β-D-1-thiogalactopyranoside (IPTG; 0.1 mM) was added to the strains harboring the pDG148 plasmid derivatives to induce the expression of gp1 upon inoculation, and the activity of the different promoter–lacZ fusions was measured in the absence (white) or presence (gray) of MMC (50 ng mL^-1) 1 h after its addition to bacterial cultures in early exponential phase. Data are presented as bar charts with data points overlap of three independent replicates (n = 3). Upper line of the bar chart represents the mean value and the error bar represents the data range within the 1.5 interquartile range. The statistical significance of the data was assessed using Fisher’s Least Significant Difference (LSD) test.

To confirm that the direct repeat sequence to which gp1 binds in vitro is important for repression, we assayed the activity of a P1⁺P2⁺(Δop1) lacZ fusion in GBJ002, GBJ002(GIL01), and GBJ002(pDG1). The Δop1 deletion effectively abolished repression of P1⁺P2⁺ by the GIL01 prophage (Fig. 3B), and LacZ expression remained elevated in GBJ002(GIL01), both with and without DNA damage. In contrast, LacZ expression in strain GBJ002(pDG1) overexpressing gp1 was repressed five-fold under both conditions (Fig. 3B), suggesting that gp1 continues to affect P1⁺P2⁺ despite the deletion in operator 1. This observation is consistent with the fact that gp1 continues to bind to Δop1 in vitro (Fig. 2D–F).

To gain a deeper understanding of transcriptional regulation and the involvement of gp1 in the activity of P1 and P2, we used the promoter fusion P1⁺P2^- in which the -10 element of P2 was inactivated⁸, generated a P1^-P2⁺ fusion by mutating the -10 element of P1 as previously described, and used a P1^-P2^- lacZ fusion as a promoter-free control⁸. These promoter fusions allowed us to examine the relative contributions of P1 and P2 to downstream gene expression with minimal alterations to the P1-P2 region (Figs. 1, 3C, D). Our results showed that P2 is approximately 10–17-fold stronger than P1 under standard growth conditions and is not markedly induced by MMC. As expected, the activity of both promoter constructs was strongly repressed in the lysogenic strain GBJ002(GIL01). However, only P1⁺P2^- was markedly induced (10-fold) in response to DNA damage in GBJ002(GIL01), whereas the induction of P1^-P2⁺ was significantly lower (1.4-fold), confirming that only P1 is SOS-inducible⁸. While ectopically expressed gp1 moderately repressed P1⁺P2^- (0.5-fold repression; Fig. 3C), probably as a result of gp1 binding to the P2 promoter region, it repressed P1^-P2⁺ to similar levels as observed in GBJ002(GIL01) (Fig. 3D). Collectively, our results indicate that gp1 essentially regulates P2 and repression is independent of SOS regulation of P1.

Gp1 overexpression results in loss of GIL01 prophage through generations

Considering that gp1 regulates the P1-P2 promoter region upstream of replication and regulation genes (ORFs 1–8)⁵, we hypothesized that gp1 is responsible for the maintenance of prophage GIL01 in the cell. To test this hypothesis, we overexpressed gp1 from the multicopy plasmid pDG1 in GBJ002(GIL01). We determined the effect of gp1 expression on the maintenance of GIL01 in strain GBJ002(GIL01) grown in the presence of IPTG for 3 h until mid-exponential phase. Fig. 4A and B show that ectopic expression of gp1 resulted in significant loss of GIL01 in B. thuringiensis cultures (83%), whereas all the cells that did not overexpress gp1 retained the GIL01 replicon. These results indicate that gp1 prevents the multiplication and vertical transmission of the prophage. To further confirm this, we examined the ability of phage GIL01 to infect and multiply inside GBJ002, both in the presence and absence of gp1 expression. We expected that following GIL01 genome injection into the host cell, gp1 repression of promoter P2 would prevent phage replication and host lysis. No plaques were detected on lawns of GBJ002 overexpressing gp1, indicating that gp1 conferred immunity to GIL01 infection (Fig. 4C), in contrast to visible plaques on lawns of GBJ002 (Fig. 4D). Our results underscore the critical role of gp1 in effectively controlling the activity of P2 and the inheritance of linear prophage GIL01 across generations. The conservation of ORF1 in most betatectiviruses (Supplementary Fig 1) indicates a fundamental role of this gene in their biology and suggests a common regulatory mechanism for gp1 in these bacteriophages. It is worth noting that the genomes of only six members of the genus Betatectivirus have been sequenced, implying that these viruses are rare in nature. However, a recent report of alphatectiviruses of gram-negative bacteria noted that linear genomes of tectiviruses seem to be missed in metagenomic studies, while culture-based methods show that they are abundant in nature²⁰.

Fig. 4: Ectopic expression of gp1 prevents vertical transfer of prophage GIL01 and infection by phage GIL01.

A Percentage of GIL01-positive colonies of Bacillus thuringiensis strains GBJ002(pDG) and GBJ002(pDG1) 4 h after addition of 0.1 mM IPTG to induce expression of gp1. One hundred individual colonies of each strain were screened for the presence of GIL01 by transferring them to a lawn of recipient strain GBJ002. GIL01-positive colonies formed turbid plaques on GBJ002 lawns. B Analysis of gp1 expression in Bacillus thuringiensis strains GBJ002(pDG) and GBJ002(pDG1). The first unmarked lane of the SDS-PAGE gel contains a protein ladder, and lanes 1 and 2 contain 20 µL of GBJ002(pDG) and GBJ002(pDG1) culture lysates obtained 4 h after addition of 0.1 mM IPTG. The band corresponding to gp1 (~6.5 kDa) is indicated with an arrow. IPTG-induced strains GBJ002(pDG1) (C) and GBJ002(pDG) (D) infected with GIL01 in soft agar layers.

Bacteriophage GIL01 bypasses the need for a functional gp1 by modifying gp1-regulated P2

In a previous study, we isolated a class of GIL01 mutants in ORF1, which formed clear to semi-turbid plaques on lawns of susceptible host bacteria⁶. Three of the six ORF 1 clear plaque (cp) mutants carried a single nucleotide substitution in the DNA-binding domain, and cp33, carrying an A11V mutation in gp1, was selected for further studies. We purified the gp1 (A11V) mutant and examined binding to the P2 promoter region using SPR. This confirmed the loss of DNA-binding capacity of gp1 (A11V) except for unstable non-specific interactions observed at higher concentrations (Supplementary Fig. 2A). The structural model of gp1 indicates that the alanine at position 11 is not in direct contact with the DNA (Supplementary Fig. 2B), suggesting that the A11V mutation affects the structure of the protein.

Because wild-type gp1 is required for the lysogenic cycle of GIL01⁶, we used cp33 to test the potential reversal of the ORF1 mutation or the accumulation of additional mutations that negate the effects of the first one. For this purpose, we collected the center of a cp33 plaque using a sterile needle and plated it out on Luria broth (LB) agar. Growing colonies were analyzed for the presence of GIL01 by PCR, and the phage DNA from two GIL01-positive colonies was sequenced. Sequencing revealed that both derivatives of cp33 retained the original ORF1 mutation and had acquired additional changes in the P1-P2 promoter region. Cp33 derivative cp33.1 lost both dinBox sites and the entire P2 sequence (172Δ355, Fig. 5A) and retained only P1 with its original transcription start site. The cp33.14 variant lost an 11-nucleotide sequence between the P2 promoter elements -35 and -10 (252Δ262, Fig. 5A). In both variants, DNA rearrangements occurred at short identical sequences delineating the start and the end of the excised DNA (ATGAGT in cp33.1 and GGTAA in cp33.14), indicating that the cp33 derivatives likely arose through homologous recombination over short repeats.

Fig. 5: Characterization of GIL01 double mutants.

A Nucleotide sequence of the P1-P2 promoter region and the gene encoding gp1. The P1 and P2 transcription start sites are indicated with angled arrows, -35 and -10 promoter elements are underlined, canonical dinBox1 and non-canonical dinBox1b LexA-binding sites are highlighted in blue, gp1 nucleation site is highlighted in purple, and ORF1 start and stop codons are shown in green and red, respectively. The nucleotide substitution in ORF1, translating into gp1 (A11V) in cp33, is shown in red as a boxed codon. The 172Δ355 deletion found in the double mutant cp33.1 is double underlined, and the 252Δ262 deletion of double mutant cp33.14 is boxed. B Bar charts illustrating the ß-galactosidase activity in cured GBJ002 or in the GIL01 lysogen, GBJ002(GIL01), carrying the P1-P2 promoter regions of cp33.1 (172Δ355) or cp33.14 (252Δ262) fused to the reporter gene lacZ. Activity of the different promoter–lacZ fusions was measured in the absence (white) or presence (gray) of MMC (50 ng mL-1) 1 h after its addition to bacterial cultures in early exponential phase. Data are presented as bar charts with data points overlap of three independent replicates (n = 3). Upper line of the bar chart represents mean value and error bar represents the data range within 1.5 interquartile range. The statistical significance of the data was assessed using Fisher’s Least Significant Difference (LSD) test. C Percentage of GIL01-positive colonies in strains GBJ002(GIL01), GBJ002(cp33.1), and GBJ002(cp33.14). At least one hundred individual bacterial colonies of each strain were screened for the presence of GIL01 by transferring them to a lawn of recipient strain GBJ002. GIL01-positive colonies formed turbid plaques on GBJ002 lawns.

Fusions of the cp33.1 and cp33.14 altered promoters to lacZ showed that P1 in cp33.1 was no longer controlled by GIL01 (Fig. 5B) and exhibited similar expression to induced P1⁺P2^- (Fig. 3C). In cp33.14, the activity of the modified P1-P2 region was reduced by approximately half compared with wild-type P1-P2 and remained regulated by GIL01 (Figs. 3A and 5B, C). The binding of purified gp1 to the modified P1-P2 regions was also studied in vitro using SPR. Our results confirmed that gp1 binding to the P1-P2 region of cp33.1 was abolished with the deletion of P2 (Supplementary Fig 2C) whereas the interaction with the P1-P2 region of cp33.14 was comparable to the wild-type (Fig. 2E and Supplementary Fig 2D).

To study the stability of cp33.1 and cp33.14 carriers, single colonies were grown to exponential phase in liquid cultures and plated. Only ~55% of growing colonies still harbored cp33.1 (Fig. 5C), indicating that cp33.1 carriers tend to lose the phage; in contrast, cp33.14 was retained in ~96% of the cells. These results concorded with the loss of P1 regulation in cp33.1 and the ability of GIL01 to continue repressing the modified P1-P2 in cp33.14. Collectively, these results confirm the critical role of gp1 control of P2 for the stable maintenance of prophage GIL01 in the host (Fig. 6). Moreover, the formation of double mutants that can maintain themselves to varying degrees within the host population highlights the remarkable adaptability of GIL01, suggesting a high degree of genomic plasticity that enables dynamic responses and adaptations to the ever-changing environment.

Fig. 6: Model of regulation of GIL01 gene expression.

A Upon GIL01 infection, the host transcription factor LexA alone is not sufficient to repress transcription from the P1 promoter. This is because LexA requires complex formation with gp7 to simultaneously occupy the target sites dinBox1 and dinBox1b⁶. Because gp1 is not yet synthesized, the P2 promoter is also derepressed and active. Consequently, genes responsible for replication and regulation of GIL01 are transcribed from P1 and P2, leading to expression of ORFs 1-8⁵ and accumulation of the regulatory proteins gp1, gp6, and gp7, among others. Transcription from the internal lytic promoter P3 is blocked because its activation requires LexA inactivation and the intracellular accumulation of the transcription activator gp6²³. B In a fully repressed state, gp1 represses transcription from promoter P2 and LexA-gp7 specifically targets the weak promoter P1 and the lytic promoter P3. C Autorepression of gp1 leads to a decrease in gp1 levels and enables the production of DNA polymerase and terminal protein to ensure the replication of GIL01 and its vertical transfer during the lysogenic cycle. Expression of gp7 enables further repression of P1 and P3 to prevent entry into the lytic cycle. D Following persistent DNA damage, LexA is inactivated and its intracellular concentration falls, leading to the derepression of promoters P1 and P3. Expression from P1 leads to substantial accumulation of gp6, resulting in P3 activation and the expression of the downstream genes responsible for virion production and cell lysis.

GIL01 differs from prophage lambda in that it utilizes host LexA to establish the lysogenic cycle instead of coding for its own SOS-cleavable repressor. By pairing LexA with gp7, GIL01 adds an additional layer of control to modulate the timing of its activation in response to stress, which in lambda is accomplished through the slower cleavage of CI to only trigger phage multiplication if the DNA insult warrants escaping the cell²¹. Similarly to lambda, GIL01 also uses multiple transcription regulators to orchestrate the establishment and maintenance of lysogeny and the switch to the lytic cycle. In GIL01, establishment of lysogeny is likely achieved through LexA binding to P1 and P3 to prevent the expression of lytic genes while allowing the production of phage-coded repressors (gp1 and gp7). In our model, gp1 is responsible for maintaining lysogeny by keeping DNA replication low through its ability to regulate its own expression. Gp7 ensures that LexA remains tightly bound to its sites until significant DNA damage is inflicted. In response to prolonged stress, the lytic cycle is activated and P1 takes over. The interplay between P1 and P2 is not fully understood but it is likely to exist to allow GIL01 to fine-tune gene expression and adapt it to its different life cycles and environmental cues. Despite its small size, GIL01 efficiently harnessed host and phage functions to exist as an autonomous unit.

Materials and Methods

Protein expression and purification

GIL01 protein gp1 (coordinates 356 to 532; GenBank ID AJ536073.2) was expressed from vector pQE-30 (Qiagen) generated in a previous study⁸. The gp1 (A11V) mutant with a single nucleotide substitution (C32T), leading to translation of the amino acid valine instead of alanine, was expressed from the pET29b(+) expression vector (obtained from Twist Bioscience) using DNA synthesis. Wild-type gp1 and the A11V mutant were expressed with an N-terminal His₆ tag (His₆-gp1; referred to as gp1 in the main text) in E. coli strains M15 [pREP4] and BL21(DE3), respectively, by adding 0.8 mM IPTG to a culture grown at 37 °C when OD₆₀₀ reached 0.9. Cultures were grown in LB broth containing ampicillin (100 µg mL^-1) and kanamycin (25 µg mL^-1) for wild-type gp1 expression and kanamycin (50 µg mL^-1) for mutant gp1 (A11V) expression. At 2 h after IPTG induction, cultures were centrifuged (8000 × g, 20 min, 4 °C) and pellets were resuspended in 2 mL buffer A (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole; pH 8) per g wet weight with added proteinase inhibitors (cOmplete™ Protease Inhibitor Cocktail, Roche) and Benzonase nuclease (5 U mL^-1). Bacterial cells were homogenized on ice using ultrasound (tapered tip, 40% amplitude, cycles of 5 s ON and 10 s OFF) until complete lysis. After centrifugation (20 000 × g, 30 min, 4 °C), the supernatant was incubated at 4 °C with slow stirring and Ni-NTA agarose (Invitrogen; 1.5 mL per 8 g initial weight of bacteria) in buffer A (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole; pH 8). The suspension was then loaded onto a gravity column and washed with buffer A and buffer B with a higher imidazole concentration (50 mM NaH₂PO₄, 300 mM NaCl, 20 mM imidazole; pH 8). Proteins were eluted with buffer C (50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole; pH 8) with three volumes of Ni-NTA agarose. Eluted proteins were dialyzed against buffer D (20 mM HEPES, 140 mM NaCl, pH 7.4), and protein concentrations were determined using the Pierce™ BCA Protein Assay Kit.

DNase I footprint analysis

Oligonucleotides flanking the P1-P2 promoter region of GIL01 (136 F, EMSA6; Supplementary Table 1) were used to amplify DNase I footprinting probes (coordinates 136 to 350; GenBank ID AJ536073.2; Fig. 1) carrying the wild-type gp1 nucleation site (215 bp) or the Δop1 mutant (207 bp). PCR fragments were gel purified using the QIAquick Gel Extraction Kit (Qiagen), and the negative strand of each probe was labeled at the 5′-end with [γ-32P] ATP and T4 polynucleotide kinase (NEB). His6-gp1 was then incubated with 0.25 nM of each DNA probe at concentrations ranging from 200 to 3200 nM (in two-fold increments) in binding buffer (80 mM NaCl, 25 mM Tris [pH 7.5], 5 mM EDTA, 2 mM DTT, 0.1 mg/ml BSA, 0.05 μg/μl poly(dI-dC)) in a total volume of 20 µL for 30 min at room temperature before DNase I (0.1 U) was added. The reactions were stopped by the addition of 20 mM EDTA, 50 mM NaCl, 0.1 mg/ml tRNA and 1% SDS. The DNA products were precipitated with ethanol, resuspended in loading buffer (50% formamide, 50% ddH2O, 0.01% xylene cyanol, 0.01% bromophenol blue) and heated at 90°C for 1–2 min before loading onto 8% sequencing gels (9 M urea). The G + A ladder was prepared by formic acid modification and piperidine cleavage of the DNA probes. The bands were visualized on autoradiography film.

Bioinformatics

The structural model of the gp1 heptamer filament was created using AlphaFold 3¹⁶. To create the model, we used a DNA sequence corresponding to the 65-bp long site protected by gp1 in the DNase I footprint analysis (Fig. 2A, B), flanked by six nucleotides at each extremity, and seven gp1 monomers. Visualization of the structures was performed using the ChimeraX program²².

Surface plasmon resonance assays

Biacore T200 (GE Healthcare) was used to perform SPR measurements at 25 °C on the streptavidin sensor chip SA (GE Healthcare). The chip was equilibrated in SPR buffer (20 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.1 mg mL^-1 bovine serum albumin (BSA), 0.005% P20; pH 7.4). A flow of 2 µL min^-1 was used to immobilize approximately 60 response units (RU) of the 3′-biotinylated S1 primer (Supplementary Table 1) on the flow cells of the streptavidin chip. Complementary primer pairs (SPR_wt_L/S, SPR_Δop1_L/S, SPR_172Δ355_L/S, and SPR_252Δ262_L/S; Supplementary Table 1) were used to assemble 81-bp double-stranded DNA fragments of the P1-P2 region containing the wild-type gp1 nucleation site or Δop1 mutant (Fig. 1), and corresponding regions of the double mutants cp33.1 and cp33.14 (Fig. 5A). The long and short complementary primers were mixed at a ratio of 1:1.5 (mol: mol, respectively) and annealed with a temperature gradient of 94 °C to 4 °C (∼1.5 h) in a PCR machine (Eppendorf). The obtained double-stranded DNA probes had a 15-nucleotide 5′-overhang complementary to the S1 primer immobilized on the streptavidin chip. At 5 µL min^-1, the DNA probe was immobilized on each flow cell to approximately 58–74 RU. Dilutions of His₆-gp1 in SPR buffer were injected at 30 µL min^-1 for 90 s over DNA probes at concentrations ranging from 4 to 64 or 512 nM (in two-fold increments) followed by 350 s of dissociation. After each cycle, 50 mM NaOH was injected at 30 µL min^-1 for 8 s to regenerate the sensor surface. Sensorgrams were double-referenced for the untreated flow cell 1 surface response and the flow cell 2 buffer response. The suitable data were fitted using a one-to-one kinetics model with Biacore T200 Evaluation Software to obtain the respective association rate constants (k_a) describing the rate of complex formation. Responses were normalized to compensate for differences in the extent immobilization of DNA fragments used. The experiment was performed in triplicate.

To determine the number of gp1 proteins occupying the P2 promoter region, we analyzed the data obtained for the gp1 binding to the DNA fragment composed of complementary primers SPR_wt_L/S (GenBank ID, AJ536073.2, genome coordinates 237 to 317). The DNA (MW = 54.4 kDa; including half of the sequence of the S1 primer) was immobilized on the SPR chip to 73.5 RU and recombinant His₆-gp1 (MW = 8.0 kDa) was injected over the DNA to reach a maximum response of 77.0 RU at saturated concentration, meaning that about seven molecules of His₆-gp1 are bound to the wt DNA. The maximum possible signal, denoted R_max, in a ligand-analyte pair is determined by the number of binding sites (V_ligand), the amount of immobilized ligand (R_ligand) and the molecular weight of the ligand (Mw_ligand) and analyte (Mw_anylyte) molecules, as shown in the following equation:

$${R}_{\max }={R}_{{ligand}}\left(\frac{{{Mw}}_{{analyte}}}{{{Mw}}_{{ligand}}}\right){V}_{{ligand}}$$

Electrophoretic mobility shift assay

Oligonucleotides flanking the P1-P2 promoter region of GIL01 (EMSA3, EMSA6; Supplementary tab. 1) were used to amplify EMSA probes (coordinates 151 to 350, GenBank ID AJ536073.2, Fig. 1) carrying the wild-type gp1 nucleation site (200 bp) or the Δop1 mutant (192 bp). DNA probes were gel purified, labeled with ³²P and T4 polynucleotide kinase (NEB), and tested for interaction with His₆-gp1 under the same binding conditions as for DNase I footprint analysis. His₆-gp1 was incubated with 0.25 nM radiolabeled DNA probes at concentrations ranging from 2 to 32 nM (in two-fold increments) for 30 min at room temperature. Samples were then separated on native 6% (w/v) polyacrylamide gels at room temperature for 1.5 h at 30 mA with 1× TAE as running buffer. The gels were dried onto blotting paper and then exposed to autoradiographic film.

Isolation of GIL01 double mutants

Cp33-derived double mutants cp33.1 and cp33.14 were isolated, sequenced, and characterized as described below. The double mutants were obtained by infecting a lawn of GBJ002 in the soft agar overlay with 10-µL serial dilutions of clear-plaque mutant cp33 stock. The center of a resulting 10^-1 spot was collected with a sterile needle and streaked onto fresh LB plates. Growing colonies were transferred to fresh plates and were tested for the presence of GIL01 using PCR. Two GIL01-positive colonies, carrying cp33 derivatives, later named cp33.1 and cp33.14, were retained for further studies. Cp33.1 and cp33.14 were amplified with GIL01-specific primer pairs designed to cover the 15-kb genome in segments of 1 kb with 100-bp overlaps⁶. PCR products were purified and sequenced with a BigDye Terminator v3.1 cycle sequencing kit on an ABI Prism 3130xl genetic analyzer instrument (both from Applied Biosystems). Amplicons were sequenced on both strands to provide 3- to 4-fold sequence coverage and were compared with the genome of wild-type pGIL01 (GenBank ID: AJ536073.2).

ß-galactosidase assays

To study the effect of gp1 on the GIL01 P1-P2 promoter region, we used Bacillus strains harboring the low-copy-number shuttle vector pHT304-18Z with reporter lacZ fused to the wild-type P1-P2 promoter or its derivatives with an inactivated -10 promoter element in either P1 or P2^6,8 (Fig. 1) and the P1-P2 region of the GIL01 double mutants cp33.1 and cp33.14 (Fig. 5A). To generate the P1-P2 promoter region of cp31, a GIL01 mutant with a partial deletion of the gp1 nucleation site (Δop1) fused to lacZ in pHT304-18Z, primers pHT1 and pHT4⁶ were used to amplify the region. The resulting product was purified and cloned into the selection vector pCR4-TOPO. The ligation was transformed into competent TOP10 cells, and recombinants were extracted and restricted with HindIII and BamHI enzymes. The gel-purified fragments were then ligated into HindIII- and BamHI-digested pHT304-18Z. The resulting plasmid construct was transformed into competent GBJ002, GBJ002(GIL01) and GBJ002(pDG1) cells. Gp1 was overexpressed from the multicopy plasmid pDG148 (pDG1)⁶ in Bacillus strains carrying lacZ promoter fusions. GBJ002(GIL01) was used as GIL01 lysogen⁶.

Bacillus strains were cultured aerobically in LB broth at 30 °C. To perform β-galactosidase assays, pHT304-18Z, and pDG1 plasmids were maintained by the addition of 12.5 µg mL^-1 erythromycin and 25 µg mL^-1 kanamycin, respectively. Expression of gp1 in strain GBJ002(pDG1), carrying pDG1, was induced with 0.1 mM IPTG 1 h after the inoculation of cells into LB broth (1:100). DNA damage was induced by adding 50 ng mL^-1 MMC 3 h after inoculation to half of the cultures when B. thuringiensis cells were in the exponential phase. The cultures were incubated for 1 h to induce the SOS response. The optical densities (OD) of the cultures were then measured at 595 nm and 20 µL sample from each culture was used to evaluate the β-galactosidase activity. Bacterial cells were permeabilized with 980 μl buffer Z (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄) with freshly added 50 mM β-mercaptoethanol and 10 μl toluene. After 5 min at 37 °C, 200 μl of buffer Z containing 4 mg/ml o-nitrophenyl-β-d-galactopyranoside (ONPG) was added to each sample. The samples were then incubated at 37°C for 20 to 30 min and the reactions were stopped by adding 500 μl of 1 M Na₂CO₃. The samples were centrifuged at 13,000 × g for 7 min and the supernatants were collected. The ODs of the supernatants were measured at 420 nm. The β-galactosidase activity values, expressed in Miller units (MU), were calculated according to the formula below. The experiments were performed in three biological replicates and statistical significances were calculated using Fisher’s Least Significance Difference (LSD) test suitable for pairwise comparisons of several treatment groups.

$${MU}=\frac{1000\,\times \,{{OD}}_{420}}{{{OD}}_{595}\,\times {{time}}_{\min }\,\times {{volume}}_{{mL}}}$$

Analysis of the vertical inheritance of the GIL01 phage and its double mutants in Bacillus thuringiensis

To investigate the effect of gp1 on the abundance of prophage GIL01 in the bacterial population, we streaked strains GBJ002(GIL01)(pDG) and GBJ002(GIL01)(pDG1) on LB plates supplemented with 50 µg mL^-1 kanamycin and incubated them overnight at 30 °C. We then resuspended the single colonies in 20 mL of fresh LB to an OD of 0.1 at 595 nm. Kanamycin (25 µg mL^-1) and 0.1 mM IPTG were added to the cultures to maintain pDG and pDG1 plasmids and to induce gp1 synthesis, respectively. Cultures were incubated at 30 °C with shaking at 200 rpm for 4 h. Next, a sample from each culture was diluted in fresh LB and plated on LB plates with 50 µg mL^-1 kanamycin, and incubated at 30 °C overnight.

To investigate the effect of the mutations in double mutants cp33.1 and cp33.14 on vertical phage inheritance, we collected centers of GBJ002(cp33.1), GBJ002(cp33.14), and GBJ002(GIL01) colonies with sterile toothpicks and incubated them in 4 mL LB for 4.5 h at 37 °C and 200 rpm. A sample of the culture was then diluted in fresh LB, plated on LB plates and incubated overnight at 30 °C. At least 100 individual colonies of a single strain tested in the above experiments were then analyzed for the presence of GIL01. For this purpose, we prepared a lawn of the recipient strain GBJ002 by mixing 150 µL of the GBJ002 culture in the exponential growth phase with 5 mL of 0.4% soft agar at 48 °C and pouring the mixture onto warm LB plates. When the soft agar was solidified, we used sterile toothpicks to pick the individual colonies and transferred them to the GBJ002 lawns. Plates were incubated overnight at 30 °C and visually examined for lysis, indicative of the presence of phage GIL01 in the transferred colonies.

To evaluate expression of gp1 in strain GBJ002(pDG1), 2 mL samples were collected from cultures GBJ002(pDG1) and GBJ002(pDG) (negative control) at the same time as samples to determine GIL01 prophage abundance in the B. thuringiensis population in the above experiment. Samples were centrifuged (8000 × g, 20 min, 4 °C), and bacterial pellets were resuspended in 500 µL of phosphate-buffered saline containing a dissolved tablet of a cocktail of proteinase inhibitors (cOmplete™ Protease Inhibitor Cocktail, Roche). Resuspended cells were then lysed by applying ultrasound (tapered tip, 40% amplitude, 2 s ON and 4 s OFF) for 3 min. After centrifugation (20 000 × g, 20 min, 4 °C), 20 µL of supernatants were loaded on 4–20% Bis-Tris protein SurePAGE™ gels (GenScript), resolved by SDS-PAGE, and stained with SimplyBlue SafeStain (Invitrogen).

Evaluation of the gp1 effect on GIL01 infectivity of Bacillus thuringiensis

To investigate the effect of ectopically expressed gp1 on phage GIL01 infectivity of B. thuringiensis, we first obtained phage lysate by adding MMC (50 ng mL^-1) to the GIL01 lysogen, GBJ002(GIL01), in the early exponential growth phase (OD₅₉₅ = 0.3–0.4) at 30 °C and 200 rpm. After complete lysis (approximately 2 h after addition of MMC), the culture was harvested, centrifuged at 8000 × g for 20 min at 4 °C, and the supernatant was filtered using a syringe filter with 0.22 µm pores. To determine the number of plaque-forming units (PFU) per milliliter, dilutions of phage lysate prepared in sterile LB were spotted onto a lawn of bacterial cells—prepared by mixing 150 µL of recipient strain HER1410 (harvested in the early exponential growth phase) with 5 mL of 0.4% soft agar, poured over LB plates and incubated overnight at 30 °C. Next, we prepared exponential cultures of strains GBJ002(pDG) and GBJ002(pDG1) for infection with GIL01. Cultures were prepared by 1:100 dilution of overnight cultures in 20 mL of fresh LB containing 25 µg mL^-1 kanamycin and 0.1 mM IPTG to induce the expression of gp1 from the pDG1 plasmid. Cultures were incubated at 30 °C with shaking at 200 rpm until the early exponential growth phase. Next, 100 µL of diluted phage lysate containing ~200 GIL01 particles was mixed with 150 µL of recipient strains GBJ002(pDG) or GBJ002(pDG1) and incubated for 2 min at room temperature. Each mixture was added to 5 mL of 0.4% soft agar and poured over LB plates with 50 µg mL^-1 kanamycin and incubated overnight at 30 °C. The experiment was performed in duplicate.

Statistics and reproducibility

SPR assays were performed in three technical replicates, from which the mean association rate constants (k_a) and standard deviation constants were calculated. DNase I footprinting and EMSA assays were performed in two separate experiments, with representative gels shown. ß-galactosidase assays were performed in three biological replicates, and statistical significances were calculated using Fisher’s LSD test, suitable for pairwise comparisons of several treatment groups. At least 100 individual colonies of a single strain were analyzed for the presence of GIL01 to assess the vertical inheritance of GIL01 phage and its double mutants. Evaluation of gp1 expression in strain GBJ002(pDG1) with SDS-PAGE was performed in two separate experiments, and a representative gel is shown. The experiment in which we evaluated the effect of gp1 on Gil01 infectivity of Bacillus thuringiensis was performed in two biological replicates.

Reporting summary

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