Coliphage lambda proteins Rz and Rz1 are the inner membrane and outer membrane subunits of the spanin complex—a heterotetramer that bridges the periplasm and is essential for the disruption of the outer membrane during phage lysis. Recent evidence suggests the spanin complex functions by fusing the inner and outer membrane. Here, we use a genetics approach to investigate and characterize determinants of spanin function. Because Rz1 is entirely embedded in the +1 reading frame of Rz, the genes were disembedded before using random mutagenesis to construct a library of lysis-defective alleles for both genes. Surprisingly, most of the lysis-defective missense mutants exhibited normal accumulation or localization in vivo, and also were found to be normal for complex formation in vitro. Analysis of the distribution and nature of single missense mutations revealed subdomains that resemble key motifs in established membrane-fusion systems, i.e., two coiled-coil domains in Rz, a proline-rich region of Rz1, and flexible linkers in both proteins. When coding sequences are aligned respective to the embedded genetic architecture of Rz1 within Rz, genetically silent domains of Rz1 correspond to mutationally sensitive domains in Rz, and vice versa, suggesting that the modular structure of the two subunits facilitated the evolutionary compression that resulted in the unique embedded gene architecture.

Most phages of Gram-negative hosts encode spanins for disruption of the outer membrane, the last step in host lysis. However, bioinformatic analysis indicates that ∼15% of these phages lack a spanin gene, suggesting they have an alternate way of disrupting the OM. Here, we show that the T7-like coliphage phiKT causes the explosive cell lysis associated with spanin activity despite not encoding spanins. A putative lysis cassette cloned from the phiKT late gene region includes the hypothetical novel gene 28 located between the holin and endolysin genes and supports inducible lysis in E. coli K-12. Moreover, induction of an isogenic construct lacking gene 28 resulted in divalent cation-stabilized spherical cells rather than lysis, implicating gp 28 in OM disruption. Additionally, gp 28 was shown to complement the lysis defect of a spanin-null λ lysogen. Gene 28 encodes a 56-amino acid cationic protein with predicted amphipathic helical structure and is membrane-associated after lysis. Urea and KCl washes did not release gp 28 from the particulate, suggesting a strong hydrophobic membrane interaction. Fluorescence microscopy supports membrane localization of the gp 28 protein prior to lysis. Gp 28 is similar in size, charge, predicted fold, and membrane association to the human cathelicidin antimicrobial peptide LL-37. Synthesized gp 28 behaved similar to LL-37 in standard assays mixing peptide and cells to measure bactericidal and inhibitory effects. Taken together, these results indicate that phiKT gp 28 is a phage-encoded cationic antimicrobial peptide that disrupts bacterial outer membranes during host lysis and thus establishes a new class of phage lysis proteins, the disruptins. Significance We provide evidence that phiKT produces an antimicrobial peptide for outer membrane disruption during lysis. This protein, designated as a disruptin, is a new paradigm for phage lysis and has no similarities to other known lysis genes. Although many mechanisms have been proposed for the function of antimicrobial peptides, there is no consensus on the molecular basis of membrane disruption. Additionally, there is no established genetic system to support such studies. Therefore, the phiKT disruptin may represent the first genetically tractable antimicrobial peptide, facilitating mechanistic analyses.

Leviviruses are bacteriophages with small single-stranded RNA genomes consisting of 3-4 genes, one of which (sgl) encodes a protein that induces the host to undergo autolysis and liberate progeny virions. Recent meta-transcriptomic studies have uncovered thousands of leviviral genomes, but most of these lack an annotated sgl, mainly due to the small size, lack of sequence similarity, and embedded nature of these genes. Here, we identify sgl genes in 244 leviviral genomes and functionally characterize them in Escherichia coli. We show that leviviruses readily evolve sgl genes and sometimes have more than one per genome. Moreover, these genes share little to no similarity with each other or to previously known sgl genes, thus representing a rich source for potential protein antibiotics.

The final step of lysis in phage infections of Escherichia coli is mediated by the spanins Rz and Rz1. These proteins form a complex that bridges the cell envelope and that has been proposed to cause fusion of the inner and outer membranes. Accordingly, mutations that block spanin function are found within coiledcoil domains and the proline-rich region, motifs essential in other fusion systems. To gain insight into spanin function, pseudorevertant alleles that restored plaque formation for lysis-defective mutants of Rz and Rz1 were selected. Most second-site suppressors clustered within a coiled-coil domain of Rz near the outer leaflet of the cytoplasmic membrane and were not allele specific. Suppressors largely encoded polar insertions within the hydrophobic core of the coiled-coil interface. Such suppressor changes resulted in decreased proteolytic stability of the Rz double mutants in vivo. Unlike the wild type, in which lysis occurs while the cells retain a rod shape, revertant alleles with second-site suppressor mutations supported lysis events that were preceded by spherical cell formation. This suggests that destabilization of the membrane-proximal coiled coil restores function for defective spanin alleles by increasing the conformational freedom of the complex at the cost of its normal, all-ornothing functionality.IMPORTANCE Caudovirales encode cell envelope-spanning proteins called spanins, which are thought to fuse the inner and outer membranes during phage lysis. Recent genetic analysis identified the functional domains of the lambda spanins, which are similar to class I viral fusion proteins. While the pre-and postfusion structures of model fusion systems have been well characterized, the intermediate structure(s) formed during the fusion reaction remains elusive. Genetic analysis would be expected to identify functional connections between intermediates. Since most membrane fusion systems are not genetically tractable, only few such investigations have been reported. Here, we report a suppressor analysis of lambda spanin function. To our knowledge this is the first suppression analysis of a class I-like complex and also the first such analysis of a prokaryote membrane fusion system.