Bacteriophage Mu is a paradigm coliphage studied mainly because of its use of transposition for genome replication. However, in extensive nonsense mutant screens, only one lysis gene has been identified, the endolysin gp22. This is surprising because in Gram-negative hosts, lysis by Caudovirales phages has been shown to require proteins which disrupt all three layers of the cell envelope. Usually this involves a holin, an endolysin, and a spanin targeting the cytoplasmic membrane, peptidoglycan (PG), and outer membrane (OM), respectively, with the holin determining the timing of lysis initiation. Here, we demonstrate that gp22 is a signal-anchor-release (SAR) endolysin and identify gp23 and gp23.1 as two-component spanin subunits. However, we find that Mu lacks a holin and instead encodes a membrane-tethered cytoplasmic protein, gp25, which is required for the release of the SAR endolysin. Mutational analysis showed that this dependence on gp25 is conferred by lysine residues at positions 6 and 7 of the short cytoplasmic domain of gp22. gp25, which we designate as a releasin, also facilitates the release of SAR endolysins from other phages. Moreover, the entire length of gp25, including its N-terminal transmembrane domain, belongs to a protein family, DUF2730, found in many Mu-like phages, including those with cytoplasmic endolysins. These results are discussed in terms of models for the evolution and mechanism of releasin function and a rationale for Mu lysis without holin control.
IMPORTANCE Host cell lysis is the terminal event of the bacteriophage infection cycle. In Gram-negative hosts, lysis requires proteins that disrupt each of the three cell envelope components, only one of which has been identified in Mu: the endolysin gp22. We show that gp22 can be characterized as a SAR endolysin, a muralytic enzyme that activates upon release from the membrane to degrade the cell wall. Furthermore, we identify genes 23 and 23.1 as spanin subunits used for outer membrane disruption. Significantly, we demonstrate that Mu is the first known Caudovirales phage to lack a holin, a protein that disrupts the inner membrane and is traditionally known to release endolysins. In its stead, we report the discovery of a lysis protein, termed the releasin, which Mu uses for SAR endolysin release. This is an example of a system where the dynamic membrane localization of one protein is controlled by a secondary protein.


For release of progeny virions, the Caudovirales use “multigene lysis” (MGL) systems which involve at least two proteins, the holin and the endolysin (1). The holin initiates lysis by forming holes in the cytoplasmic membrane, which allows the endolysin, a muralytic enzyme, to attack the peptidoglycan (PG). Other MGL lysis proteins have been identified, including spanins and disruptins to attack the outer membrane and antiholins to regulate the timing of holin function (2, 3). Among the paradigm phages used to study most of the fundamental processes of prokaryotic biology, phage Mu stands out because its lysis system has not been delineated (4, 5). Extensive amber-mutant hunts generating more than 500 conditional lethal alleles have allowed the identification of 28 essential genes (6, 7). However, only one, designated lys, was associated with a defect in lysis independent of DNA replication (4, 5). In 2002, Morgan et al. (8) sequenced and annotated the complete Mu genome and unambiguously identified lys as gene 22. No other lysis genes were identified. The presence of a distinctive catalytic triad identified its product, gp22, as a member of the “true lysozyme” (muraminidase/glycosidase/glycosyl hydrolase [9]) type of endolysins.
Subsequent bioinformatic analyses pointed to gp22 as a SAR (signal-anchor-release) endolysin (10, 11). Unlike soluble endolysins, which are released to the periplasm through micron-scale holes formed by canonical holins, SAR endolysins are exported in an inactive form by the Sec translocon. Tethering to the inner membrane (IM) depends on the eponymous SAR domain, a special type of N-terminal transmembrane domain (TMD) thought to require the proton motive force (PMF) for retention in the bilayer (1). Crucially, the membrane-tethered form of the SAR endolysin is catalytically inactive. Normally, the release of the SAR domain occurs suddenly and quantitatively due to holin triggering, but it can also occur at a low rate spontaneously when a holin is absent (12). The result is lysis in both cases, although the spontaneous release pathway generally occurs later and over a broader time spectrum than the normal holin triggering time. Most phages that use a SAR endolysin also encode a pinholin, which triggers to form nanometer-scale holes in the membrane rather than forming the micron-scale lesions characteristic of canonical holins (13).
Here, we report an experimental analysis of the Mu lysis system. The results not only demonstrate the existence of a novel lysis pathway in Mu, but also establish Mu as the first known tailed phage to lack a holin. The evolutionary implications for this lysis strategy in double-stranded DNA (dsDNA) phages are discussed.


Lysis genes of phage Mu.

In many phage genomes, the lysis genes are clustered (14). As noted above, the endolysin can be identified by the presence of one of the muralytic enzymatic domains, as in the case of gene 22 (lys). Generally, genes encoding the holin, antiholin and the two subunits of the spanin complex are nearby and co-transcribed. All four of these MGL proteins have transmembrane domains (TMDs) or membrane-targeting lipoprotein signals. At first inspection, Mu seems to fit this model. Indeed, in the Mu genome there are only four coding sequences (CDS) encoding proteins with at least one TMD (described in Methods), and all are near 22: genes 19, 20, 23, and 25 (Fig. 1 and Fig. S1 [in the supplemental material] for GenBank Accession no. AF083977). Gene 23 encodes a polypeptide with an N-terminal TMD and a periplasmic domain comprised of alpha-helical regions. We identified an overlapping CDS, 23.1, not annotated in the original Mu genome, that encodes a short outer membrane (OM) lipoprotein rich in Pro residues (Fig. 1D). Indeed, amber mutations in these genes also demonstrate the classic spherical-cell phenotypes associated with the phage λ RzRz1 spanin defect in outer-membrane disruption (Fig. 2A) (15). Identification of the endolysin and spanin genes left only the holin and possible antiholin to complete the traditional lysis cluster. Of the three remaining membrane proteins, only gp25 is expressed from the late transcript, which is characteristic of all experimentally confirmed holins (16). Samanta et al. (10) proposed that gp19 and gp20 constituted the Mu holin-antiholin pair, based on predicted membrane topology matching other putative holins and antiholins. However, genes 19 and 20 are middle genes, rather than late genes. Moreover, previous genetic analysis had shown that both could be inactivated without a significant effect on lysis (17).
FIG 1 Mu membrane proteins and lysis genes. (A) Mu middle and late regions drawn to scale showing lysis genes in genomic context with nucleotide coordinates given below. The locations of middle- (Pm) and late-transcription (Plys) promoters are indicated. Historic and new gene numbers are given inside the boxes. Functional assignments for lysis genes are listed above and predicted topologies are shown below in (B). (C to E) Primary structures of all identified Mu lysis proteins and transmembrane domain-containing proteins are shown in the boxes. Predicted transmembrane domains (TMDs) are shown as gray lines. Catalytic residues E-D-T in gp22 are marked with an asterisk. The Cys in the lipobox of the o-spanin is marked with an arrowhead. The added C-terminal epitope tags used in this study are shown in a blue box. Circled residues correspond to the predicted periplasmic domain of gp25.
FIG 2 Mu lysis gene phenotypes. (A) Lysis genes in pBAD vectors were induced with arabinose in MG1655 cells grown in the presence of 10 mM Mg2+ in the following combinations: empty pBAD33 + pBAD18-Kan-Mu-gp22-gp25 (gray squares), empty pBAD33 + pBAD18-Kan-Mu-gp22-gp25-gp23.1am (“23.1am,” filled black circles), and λ spanins in pBAD33-λ-RzRz1 +pBAD18-Kan Mu-gp22-gp25-gp23.1am (“23.1am + RzRz1,” open black circles). At 20, 45, and 60 min (marked 1, 2, and 3) 23.1am samples were observed at 100× magnification for cellular morphology. (B) Mu lysis gene candidates cloned into pBAD vectors were induced with arabinose in MG1655 cells as follows: empty vectors (gray squares), pBAD24 gp25-his (25, blue star-squares), pBAD24 λ-S105 (holins, black filled circle, solid line) and pBAD24 21-S68 (holins, black open circle, dashed line), pBAD24-gp19 (open green diamond, dashed line), pBAD33-gp20 (open green diamond, dotted line), or together (19 + 20, filled green diamond, solid line). A 2-mM 2,4-dinitrophenol (DNP) treatment was applied at 60 min. A 5-mL aliquot of cells was pelleted at 70 min, then resuspended into fresh growth medium in new flasks to follow recovery in the absence of DNP. (C) MDS12 lysogens containing λ (pink circles), λ hy21 (mustard diamonds), λ Δ(SRRzRz1) (gray squares), or Mucts (black star-squares) were induced at 42°C. At 30 min, cells were treated with dimethyl sulfoxide (DMSO; solid shapes) or 1 mM DNP (open shapes, dashed line). (D) Lysis genes in pBAD vectors were induced with arabinose in MG1655 cells in the following combinations: empty pBAD33 and pBAD24 vectors (“empty,” gray squares), pBAD33-gp22-c-myc (“22,” filled black star square), pBAD24-21-S68 (“S2168,” filled green diamonds), pBAD24-21-S68 + pBAD33-gp22-c-myc (“22 + S2168,” open green diamonds), pBAD24-21-S68am (“S2168am,” filled blue circles), and pBAD24-21-S68am + pBAD33-gp22-c-myc (“22 + S2168am,” open blue circles).
A definitive feature of holin function is that lethal premature triggering can be imposed by chemical depolarization of the membrane (16). This is true for both the canonical phage λ holin gene S105 and the phage 21 pinholin S68, which are devoid of their native antiholin/antipinholin regulatory components S107 and S71, respectively. To attempt to induce holin triggering, we cloned each of these holin and putative holin genes into an expression vector under arabinose control, grew the respective transformants until mid-logarithmic phase, induced with arabinose and then subjected the induced cultures to treatment with 2 mM 2,4-dinitrophenol (DNP) (18). Upon resuspension of the induced, DNP-treated cells in fresh medium, the cultures carrying the 19, 20, and 25 constructs showed unimpaired viability (Fig. 2B). In contrast, DNP-triggering of isogenic constructs with the well-characterized holin and pinholin genes of phage λ and phage 21 was lethal. Indeed, an additional isogenic construct carrying all the Mu holin candidates showed no lethality under the same conditions (this construct includes an inactivating E37A catalytic residue mutation in gp22, demonstrated to be inactive in Fig. 2B). These results suggest that Mu does not encode a holin. This conclusion is supported by DNP treatment with induced lysogens (Fig. 2C), where λ lysogens carrying either the canonical S105 holin or the phage 21 pinholin both exhibit premature lysis upon addition of DNP. In contrast, DNP actually delays Mu lysis.

Mu gp22 is a SAR endolysin.

Previously, gp22 was recognized as the Mu endolysin because its product contains the canonical lysozyme catalytic triad motif E-X8-D-X5-T (19) (Fig. 3A). Closer inspection of the endolysin amino acid sequence revealed an N-terminal extension with the characteristics of a SAR domain: a weakly hydrophobic stretch of 20 residues beginning at position 6, of which 10 residues have little or no hydrophobic character (four Ala, five Gly, one Thr) (Fig. 1E). The definitive test for a SAR endolysin is whether it can be released from the IM by a pinholin. Fig. 2D demonstrates this for gp22 with the well-characterized phage 21 pinholin S2168.
FIG 3 Gp22 relies on gp25 for release. (A) In MDS12 cells (filled squares) with a Mucts lysogen (open squares), Mucts 22::cam (red circles) or Mucts 25::cam (blue diamonds) are induced to replicate by a shift to 42°C at T = 0 min. (B) In the same experimental setup as used in panel A, 1% chloroform (dashed lines with open markers) was added to MDS12 (filled square), Mucts 22::cam (red circle), and Mucts 25::cam (blue diamond) at 70 min. (C) Mu lysis genes were cloned into pBAD vectors and induced in MG1655 cells as the following: empty vectors (black squares), pBAD24 gp25-his (25, blue diamonds) and pBAD33 gp22-c-myc (22, red circles) separately, or the latter two together (22 + 25, purple star squares). (D) Protein covariance calculated as significant intramolecular evolutionary coupling with EVcomplex was used as constraints for the gp22 catalytic domain and gp25 cytoplasmic domain folding. Catalytic residues of gp22 are shown in green as stick models. Charged residues in the predicted cytoplasmic regions are colored.

Mu gene 25 required for SAR endolysin release.

Our results indicated that none of the Mu membrane proteins had holin properties and that genes 22, 23, and 23.1 encoded the endolysin and spanin proteins. This raised the possibility that Mu relies on degradation of the PG following spontaneous activation of the SAR endolysin gp22 and disruption of the OM by the gp23/23.1 spanin complex. From this perspective, we wondered whether, despite its clustering with the other lysis genes and its predicted membrane localization, gp25 had no essential role in lysis. In the parental Mu lysogen, wild-type lysis occurs as a sharply defined event at ~40 min after thermal induction (Fig. 3A). Observation of individual induced lysogenic cells shows a classic slight rounding due to SAR endolysin activity prior to explosive cell lysis (Video S1) (15). However, when we compare Mu lysogens with deletions of 22 and 25 to the parental genotype, both deletion mutants continue to grow unabated (Fig. 3A). Eventually, at about 3 h after induction, bulk culture lysis is observed with the 25::cam mutant but not the isogenic 22::cam mutant. However, gp22 can be artificially released from the membrane with chloroform and function to effect lysis (Fig. 3B). The simplest interpretation of this is that gp25 is required for gp22 release and activation, at least until gp22 levels reach much higher levels those obtained during the normal vegetative cycle. To address this possibility, we constructed two plasmids encoding gp22 and gp25 with C-terminal c-myc and His-tags, respectively. When induced under conditions where spanin activity was not required (15), neither pBAD33-gp22-c-myc nor pBAD24-gp25-his alone caused lysis (Fig. 3C). In contrast, induction of cells carrying both plasmids led to lysis at ~50 min. Since gp25 is not required for gp22 maturation or activity, these results indicate that gp25 provides an ancillary function in SAR endolysin release which is not not currently provided by another known lysis protein, enabling the SAR endolysin gp22 to escape the IM. In view of this novel function, we designated gp25 as the Mu “releasin.”

Basic residues in the cytoplasmic domain of gp22 confer dependence on gp25.

gp25 is predicted to be a 99-aa (amino acid) type I (N-out, C-in) integral membrane protein with only four residues exposed in the periplasm and a 73-aa C-terminal cytoplasmic domain (Fig. 1E and 3D). The topology of gp22 is essentially the opposite, with the catalytic domain in the periplasm and only the short N-terminal extension in the cytoplasm. Most SAR endolysins have one or two N-terminal basic residues (12). Given our previous demonstration that adding basic residues at the cytoplasmic N terminus of a SAR domain can lock it in the bilayer (12), we hypothesized that the presence of basic residues, K6, K7, and K9, at the N terminus of gp22, confers dependence on gp25. To test this, we constructed plasmids carrying a panel of K→A substitution alleles of gp22-c-myc. First, the single Ala substitutions were tested for lytic function with or without the presence of gp25 (Fig. 4A and C). All three mutant gp22 alleles were found to be fully lytic with the co-expression of gp25 (Fig. 4C). Strikingly, both gp22K6A and gp22K7A also supported efficient lysis in the absence of gp25 (Fig. 4A). In contrast, gp22K9A alone retained the parental non-lytic phenotype. No significant increase in gp22 accumulation was detected for any of the missense changes; in fact, alleles with apparently slightly lower expression levels compared to the parental were still functional. Thus, it is the presence of lysine residues at both positions 6 and 7 that confers gp25 dependence on the escape of gp22 from the membrane. Moreover, the gp25 dependence was unchanged when either or both the lysine 6 and lysine 7 residues were substituted by arginines (Fig. 4B and D). These results indicate that it is the presence of positively charged residues at positions 6 and 7 that confers gp25 dependence for gp22 release.
FIG 4 Cytoplasmic residue charge determines gp22 dependence on gp25. Gp22-c-myc alleles in pBAD33 with K→A (A and C) or K→R (B and D) mutations were assayed for lysis in MG1655 cells + 0.4% Ara. Gp22-c-myc alleles were assayed alone (A and B) and in the presence of pBAD24-gp25-his (C and D). Protein levels at 50 min post-induction are shown as detected by Western blotting.

gp25 is a general “releasin” of SAR endolysins.

To investigate whether the mechanism of gp25 is specific to gp22, we cloned three other SAR endolysin genes from phages encoding holins and lacking genes resembling 25 into the isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible pZE12 vector. Characteristic of SAR endolysins in general, all three have an N-terminal region with 2 to 3 basic residues followed by a TMD consisting of primarily weakly hydrophobic residues (Fig. 5A). Other than the presence of N-terminal basic residues, the SAR domains in the panel shared no apparent sequence patterns or motifs.
FIG 5 gp25 is a general releasin for SAR endolysins. (A) Alignment of the N-terminal regions of the four SAR endolysins, including the positively charged residues (boxed in blue) and the weakly hydrophobic residues of the SAR domain. All residue backgrounds are colored with the Kyte-Doolittle scale based on a 13-residue window in CLC Genomics Workbench. (B) SAR endolysins in pZE12 induced with isopropyl-β-D-thiogalactopyranoside (IPTG) at T = 40 were assayed with pBAD33 or pBAD33-gp25-his in MG1655 cells + 0.4% Ara at T = 0 min. Cell morphologies for ≥100 cells were observed before (at T = 30) and 10 min after endolysin induction (at T = 50). Scale bar is 5 μm. Example cell views are shown above the quantification averaged across three biological replicates. Samples with a significant difference by t test with P < 0.0001 are shown by an asterisk (*).
We then co-transformed each SAR endolysin plasmid with and without an arabinose-inducible pBAD33 vector encoding gp25. To prevent cell lysis after PG degradation, 10 mM Mg2+ was added to stabilize the outer membrane; under these conditions, lysis requires the presence of spanins. Consequently, if PG degradation occurs, cells are converted into spherical forms (15). At SAR endolysin expression levels achieved from the pZE vector, a small fraction of cells carrying only a SAR endolysin plasmid were converted to spherical forms (Fig. 5B). In contrast, a significantly higher proportion of cells co-expressing a SAR endolysin with gp25 were spherical, demonstrating that the ability of gp25 to facilitate SAR endolysin release is not specific to gp22.

Bioinformatic analysis of endolysin-releasin pairs in phages.

Since there were few similarities shared among the SAR endolysins released by gp25, we wondered whether gp25-like proteins were genetically paired to endolysins with specific characteristics. SAR endolysins, however, are difficult to find by BLAST analysis since alignment results tend to favor similarity in the catalytic domain and localization signals are ignored as low-complexity features. Moreover, SAR domains are weakly hydrophobic and often are not recognized as TMDs by hydropathy-based algorithms. Therefore, we took advantage of the fact that the entire length of gp25, including the predicted N-terminal TMD, is defined by a conserved domain of unknown function, DUF2730. Thirty-four Caudovirales genomes carrying DUF2730 were found, all of which were Mu-like myophages or siphophages, with 31 gamma- and 3 alphaproteobacterial hosts (Fig. 6, Table S1). All the phages had a nearby endolysin gene, except Vibrio phage Martha 12B12, where synteny suggests that a soluble protein with weak similarity to L,d-transpeptidase, as assessed by HHPred, is the endolysin (20).
FIG 6 Alignment of SAR-endolysin-DUF2730 (releasin) pairs in phages. All active Caudovirales phages with the Mu gp25 DUF2730 domain found in InterPro were inspected for a paired endolysin. Protein sequences for (A) DUF2730 proteins and (B) endolysins were aligned and labeled according to phage and host (by genus) source, including host classification within the classes Gammaproteobacteria (γ, pink) or Alphaproteobacteria (α, green), and by endolysin catalytic type: I muramidase, II transglycosylase, and III NAM amidase, with letter designations for similarity groupings. SAR domains within the endolysins are boxed in black. The DUF2730 Profile HMM logo for the Pfam PF10805 domain is shown split similar to the alignment to illustrate collinearity (69, 70). Further information on phage, hosts, and conserved protein domains with accession numbers is detailed in Table S1 in the supplemental material.
When each genome was inspected for the cognate endolysin gene, nine distinct sequence families could be discerned. Counting the Martha endolysin, four enzymatic classes of endolysin were represented, including four in the muramidase class, such as Mu gp22. Importantly, only two of the nine endolysin families had SAR domains, the Mu muramidase type and a transglycosylase type from Pseudomonas Mu-like siphophages. Thus, in diverse Mu-like phages, the DUF2730 proteins are paired with a great diversity of endolysins, with and without secretion signals.


Why does the gp22 SAR endolysin require another protein for release?

Our results show that the positively charged side chains at positions 6 and 7 confer gp25 dependence on the release and activation of gp22; replacement of either one with an Ala residue removes this dependency (Fig. 4). In contrast, the Lys residue at position 9 is irrelevant. Thus, this dependency is not due to the total positive charge in the cytoplasmic domain, which is presumably 4, counting the protonated N-terminal amino group and the three Lys residues. Instead, it is due to the arrangement of the charged residues. Unfortunately, nothing is known about the molecular details of SAR release. In the two cases where structural information was available, the details of the packing of the SAR domain into the globular periplasmic domain were vastly different (12, 21). Nevertheless, it is clear that extensive refolding of the endolysin occurs. The energetics and/or the kinetic pathway for these refolding events may influence the rate and extent of extraction of the SAR domain from the membrane. It is also possible that host factors interact with and influence the release of SAR endolysins. This certainly is true for the first SAR endolysin characterized, P1 Lyz, where the SAR domain undergoes covalent modification by the periplasmic Dsb system (12). SAR domains have not been systematically studied, so it is unclear what determines the relative stability in the bilayer, except that small, weakly hydrophobic residues are found at significantly higher rates in SAR domains compared to those in standard TMDs (12, 22, 23). One difference between the SAR domain of gp22 and other SAR endolysins found in late-transcript lysis cassettes carrying holin genes is that there is a stretch of seven positions containing six Gly residues (11, 12). No other SAR domain that has been characterized has more than two Gly residues. It is possible that the forces exerted from the periplasm which favor spontaneous release of the SAR domain may be decreased or dissipated by two full turns of easily extended Gly residues. In this regard, it is worth noting that the Mu SAR endolysin has a longer predicted cytoplasmic domain than other well-studied phage lysis systems (11).

How does the gp25 releasin work?

This is the first example of a SAR endolysin which is dependent on a non-holin phage-encoded protein. The membrane topologies of the two proteins place conceptual constraints on models for gp25 function. gp25 consists of a predominantly anionic cytoplasmic domain attached to the membrane with an N-terminal TMD and has virtually no periplasmic component, whereas gp22 has the classic SAR endolysin topology, with its enzymatic domain in the periplasm and only a short N-terminal domain in the cytoplasm.
From these considerations, several scenarios can be envisioned for releasin function (Fig. 7). One general category would be covalent modification of the cytoplasmic N terminus of gp22 to reduce the net positive charge of the first 7 residues. The simplest idea would have the cytoplasmic domain of the releasin act as a protease that would remove some or all of the cytoplasmic N-terminal segment of gp22; certainly, proteolytic enzymes which target basic residues are well known (24). Moreover, the cytoplasmic domain of gp25 has an abundance of acidic (12 of 72) and serine (9 of 72) residues, as well as 2 histidine residues. This is consistent with the catalytic and specificity-determining motifs of serine proteases that are specific for cleavage at basic residues. Other phages are known to utilize membrane-associated proteolytic cleavage at various stages in their life cycle. For example, the T4 prohead is matured at the membrane (25).
FIG 7 The gp22-gp25 membrane release model. (A) The SAR domain of gp22 is in the membrane and has three positively charged lysine residues. When lysis begins, gp22 and gp25 may interact via one of the three proposed models. The cytoplasmic DUF2730 gp25 domain releases the gp22 SAR domain from the membrane: (i) through proteolytic cleavage, (ii) an indirect, unknown partner, or (iii) hydrophilic channel formation. After gp22 is released from the IM, it folds into its active state, leading to degradation of the peptidoglycan and lysis.
A non-covalent alternative to the protease scenario can be envisioned in which the cytoplasmic domain of gp25 would complex with the N-terminal oligopeptide of gp22 and, by virtue of closely positioned acidic residues, lower the pKa of Lys6 and Lys7. Since Arg substitutions at positions 6 and 7 do not affect the gp25 requirement, the role of gp25 in this scenario would be to provide an acidic microenvironment for deprotonation of the basic N terminus of gp22. Another general model would have the releasin act as a membrane-transit chaperone by forming some sort of channel in the bilayer, allowing the charged N terminus to bypass the hydrophobic core. In this scenario, the fact that gp25 has no apparent effect on cell growth or the holin triggering makes it unlikely for a permanent pore which affects cell energetics to be formed, although this is under active investigation. Instead, gp25 would have to function in a mode operationally similar to that suggested for the TAT (twin arginine translocase) secretion apparatus, in which a channel is assembled around the secreted substrate protein (26). Interestingly, gp25 has a conserved twin arginine or arginine-lysine motif at the junction of its transmembrane and cytoplasmic domains (Fig. 6A). Although the topology of the canonical TAT signal (arginines followed by hydrophobic signal) is reversed, the gp25 twin arginine motif is, nevertheless, also at the cytoplasmic membrane interface, raising the possibility that Mu hijacks part of the TAT system for the export of gp22. Other schema can be imagined, including a catch-all idea of the participation of an unknown host factor in the release of gp22. In any case, our results suggest that gp25 can facilitate the release of highly diverse SAR endolysins, although perhaps less efficiently than for gp22 (Fig. 5). Given the topological differences between gp25 and SAR endolysins and the lack of obvious similarity in the cytoplasmic and transmembrane domains of the latter, specific interaction between the releasin and the endolysin may not play an important role in this phenomenon, which would rule out the protease model.

DUF2730: a diverse family of Mu proteins.

gp25 is the prototype protein for a long-standing domain of unknown function, DUF2730 (Pfam entry PF10805, InterPro entry IPR020269). At the time of writing (December 2021), there are nearly 1,000 proteins harboring the DUF2730 signature in public databases, of which 38 are encoded in Mu-like (transposable) myophages and siphophages and the rest are in presumably transposable prophages in a wide range of Gram-negative hosts. The distinguishing hallmarks of transposable phages include the transposase, the protein of unknown function GemA, the Mor regulatory protein, and the portal protein (27, 28). Due to a lack of careful prophage annotation and varying levels of complete closed assemblies of host genomes containing the presumed prophage contigs, we cannot say with certainty that every instance of a DUF2730 protein is associated with a transposable phage genome. However, when we focus on the 34 genomes from active phages (usually) isolated as plaque-formers; where, in contrast to the uncharacterized prophage elements, confidence can be placed for lytic function, and we can carefully reannotate complete genomes; we see that DUF2730 proteins occur only in transposable phages. Like the Mu releasin gene, in all cases the DUF2730 gene mapped within 5 genes downstream of the endolysin cistron. Like the Mu releasin, all the phage-borne DUF2730 proteins have N-terminal TMDs with a predicted N-out, C-in topology based on the positive-inside rule (Fig. 6A) (29). However, there are three clearly distinct families as judged by alignment, only one of which (type I) has significant sequence similarity with the Mu gp25 releasin. In addition, compared in the context of the endolysin for each phage, two of the DUF2730 protein families, II and III, were associated with entirely different enzymatic classes of endolysins, transglycosylases and amidases (Fig. 6B). Moreover, other than the gp22 type, only the endolysins of type II have the weakly hydrophobic, charge-free N-terminal segments that are characteristic of N-terminal SAR domains. Thus, the operational models for the releasin proposed above would be applicable to endolysins associated with both Mu type I and II DUF2730 proteins, but not to the remaining families, which are associated with endolysins lacking SAR domains. The simplest notion would be that both the type I and II DUF2730 proteins are required for lysis, since both are associated with SAR endolysins, albeit of different enzymatic types. An important next step is to determine whether the remaining types of DUF2730 proteins are required for lysis in the context of these other Mu-like phages. Certainly, additional genetic and biochemical experiments with DUF2730 proteins are required in a wider range of systems to clarify their mechanism more broadly.

Mu is a holin-less phage.

Heretofore, MGL systems for phages of Gram-negative hosts have been found to be tripartite, defined by the components targeting the IM, PG, and OM (Table S1 in the supplemental material). The central component of MGL systems is the holin, because holin triggering at an allele-specific time terminates morphogenesis (by collapsing the PMF) and initiates the lysis pathway involving the PG and OM components (30). Holins exhibit high genetic malleability in that single missense changes can dramatically and unpredictably advance or retard the triggering time, allowing a phage to evolve through “lysis timing space” to find the optimal length of the infection cycle (31). In addition, holin triggering can be advanced or delayed in real-time. Superinfection by most Caudovirales is thought to involve transient depolarization of the membrane associated with passage of the gDNA from the virion through the cytoplasmic membrane (32, 33). If a holin-mediated infection is already in progress, this would cause premature triggering, resulting in the immediate release of already matured virions and abortion of superinfection at the time of entry. In contrast, homotypic superinfection of T4-infected cells is blocked before membrane penetration and instead leads to lysis inhibition by activating the T4 antiholin (34, 35). From this perspective, the homotypic superinfection event serves as a signal of host depletion in the environment, best accommodated by postponement of lysis in favor of continued intracellular accumulation of virions (30). The work reported here shows that Mu uses a dramatically different lysis system without a traditional holin protein (Fig. 2B and C).
It is worth considering why, of all the well-studied model phages of E. coli and Salmonella, Mu might evolve a lysis system where lysis control is exerted at the SAR endolysin level. One rationale is that Mu has experienced less selection pressure for holin evolution due to its unique replication mechanism. Unlike other characterized phages, Mu copies its genome via replicative transposition during the lytic cycle (36). Typical of transposons, Mu uses transposition immunity mechanisms to avoid self-insertion (3739). Transposition immunity in Mu is strict, within 5 kb of the genome ends, and gradually decays until becoming undetectable beyond 25 kb (40). This has significant implications for Mu’s theoretical maximum reproductive potential relative to that of non-transposing phages. Assuming a 4.6-Mb E. coli genome and perfect transposition spacing (~5 kb host DNA between each Mu genome), the maximum expected burst size for Mu should be about 460 particles. Nonetheless, the actual spacing between Mu genomes will be less uniform, as genome insertion is mostly random (41) and inhibited up to 25 kb from the ends (40). As Mu’s burst size under laboratory conditions is generally measured at ~100 to 200 particles (42), this suggests that Mu might only be capable of increasing its burst size by about 3-fold at most. In contrast, phages such as T4, λ, and N4 can delay their holin-triggering time and increase their burst sizes by as much as 10- to 100-fold (4345). Thus, if Mu’s maximum reproductive potential is biologically constrained in this manner, it may gain little fitness benefit from the rapid temporal adaptability of holin proteins. Perhaps this has allowed Mu (and other transposable phages) to explore an unusual alternative within the evolutionary landscape of phage lysis strategies.


Growth and induction of phages and bacterial strains.

Phages and bacterial strains used in this study are listed in Table 1. When plating bacterial strains, we used standard LB-agar plates supplemented with the appropriate antibiotics (ampicillin, 100 μg/mL; chloramphenicol, 10 μg/mL; kanamycin, 40 μg/mL) (46, 47). Bacteria were incubated at 30°C for Mucts and λ cI857 lysogens and 37°C for non-lysogenic E. coli strains. AraR strains of MC4100 were isolated from the surviving colonies after plating cells on agar plates containing 1% (vol/vol) arabinose and re-streaking twice on 1% arabinose plates (48, 49). Strains harboring the pBAD plasmid were induced by adding arabinose to a final concentration of 0.4% (vol/vol), or repressed with 0.4% (vol/vol) glucose (50, 51). pZ-series plasmids were induced with 1 M IPTG to 1 mM. Subcultures were prepared by diluting overnight cultures inoculated from single colonies 1:250 to A550 ~ 0.03 in LB with antibiotics and grown at 30°C or 37°C with aeration. Bacterial growth at A550 was monitored using a Gilford Stasar III sipping spectrophotometer (Gilford Instrument Laboratories, Inc., Oberlin, OH), as described previously (47). Mucts and λ lysogens were induced by shifting the incubation temperature from 30°C to 42°C for the remainder of the experiment. For plasmid inductions in the absence of spanin function, lysis curves were recorded in the absence of divalent metal cations with vigorous agitation, unless an addition of 10 mM Mg2+ is stated (15). Where indicated, cultures were treated with 100% chloroform or 1 M DNP dissolved in dimethyl sulfoxide (DMSO) at the final concentrations indicated. Absorbance data were plotted with a custom “lysis-curve” package for the Python graphing library Plotly (52, 53).
TABLE 1 Bacteriophages and bacterial strains
Bacteriophages and bacterial strainsGenotype and featuresSource
    MuctsPhage Mu carrying a temp-sensitive cts62 repressor protein that is inactivated at 42°CLaboratory stock (71)
    Mucts 22::camPhage Mu 22::camThis study
    Mucts25::camPhage Mu 25::camThis study
    λPhage λ cI857 bor::kan 
    hy21λ cI857 hy21(SRRzRz1)21 bor::kan 
    λ Δ(SRRzRz1)λ cI857 stf + Δ(SRRzRz1) bor::spec 
Bacterial strains  
    MG1655E. coli K-12 F λ ilvG rfb50 rph1 lacIqlacY tonA- AraRLaboratory stock
    MC4100E. coli K-12 F araD139 Δ(argF-lac)U169 fhuA rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsRLaboratory stock
    MC4100 AraRE. coli K-12 F araD139 ΔargF-lac U169 fhuA rpsL150 relA1 flbB5301 deoC1 ptsF25 rbsR AraRThis study
    MC4100 MuctsMC4100 lysogenized with Mucts62This study
    MC4100 Mucts 22::camMC4100 lysogenized with Mucts22::camThis study
    MC4100 Mucts25::camMC4100 lysogenized with Mucts25::camThis study
    MC4100 Mucts AraRAraR derivative of MC4100 MuctsThis study
    MC4100 Mucts 22::cam AraRAraR derivative of MC4100 Mucts 22::camThis study
    MC4100 Mucts25::cam AraRAraR derivative of MC4100 Mucts 25::camThis study
    MDS12MG1655 with 12 deletions totaling 376,180 nt, including cryptic prophages, fhuA::Tn10 lacIq1Laboratory stock (72)
    MDS12 MuctsMDS12 lysogenized with Mucts 62This study
    MDS12 Mucts 22::camMDS12 lysogenized with Mucts 22::camThis study
    MDS12 Mucts25::camMDS12 lysogenized with Mucts 25::camThis study
    MDS12 λMDS12 lysogenized with λ cI857 bor::kanThis study
    MDS12 hy21MDS12 lysogenized with λ cI857 hy21(SRRzRz1)21 bor::kanThis study
    MDS12 λ Δ(SRRzRz1)MDS12 lysogenized with λ cI857 stf + Δ(SRRzRz1) bor::specThis study

Cloning and DNA constructs.

Plasmids used and constructed in this study are listed in Table 2, and primers are given in Table 3. Mu lysogens and lysates were used as the templates for cloning Mu genes. Mu genes 19, 20, 19 + 20, 22, and 25 were cloned in pBAD24 between its unique EcoRI and HindIII sites, or in pBAD33 between its unique KpnI and HindIII sites, with their native ribosome binding Shine-Dalgarno (S-D) sequence or with a stronger S-D sequence (AGGAGG) (50, 54). Alleles encoding gp22 or gp25 were constructed with the c-myc or His6 oligopeptide tags at the 3′ end of the gene, respectively. Gene 22 and 25 mutants were constructed by using a Quick-Change site-directed mutagenesis kit (Agilent, Santa Clara, CA) on pBAD33-gp22-c-myc or pBAD24-gp25-his, respectively.
TABLE 2 Plasmids used in this study
pKD46(λ) Red recombinase expression systemLaboratory stock (73)
pBAD33pACYC184-derived plasmid with PBAD promoter; araC catLaboratory stock (50)
pBAD24pBR322-derived plasmid with PBAD promoter; araC blaLaboratory stock (50)
pBAD18-KanpBR322-derived plasmid with PBAD promoter; araC neoLaboratory stock (50)
pZE12pZE12-lucLaboratory stock (74)
pBAD33-gp22-c-mycpBAD33 with Mu gp22-c-myc inserted between the KpnI and HindIII sites after Shine Dalgarno AGGAGGThis study
pBAD33-gp22-c-myc-K6AK6A mutation in gp22-c-mycThis study
pBAD33-gp22-c-myc-K7AK7A mutation in gp22-c-mycThis study
pBAD33-gp22-c-myc-K9AK9A mutation in gp22-c-mycThis study
pBAD33-gp22-c-myc-K6RK6R mutation in gp22-c-mycThis study
pBAD33-gp22-c-myc-K7RK7R mutation in gp22-c-mycThis study
pBAD33-gp22-c-myc-K9RK9R mutation in gp22-c-mycThis study
pBAD33-gp22-c-myc-K6A-K7AK6A and K9A mutations in gp22-c-mycThis study
pBAD33-gp22-c-myc-K7A-K9AK7A and K9A mutations in gp22-c-mycThis study
pBAD33-gp22-c-myc-K6R-K7RDerived from pBAD33-gp22-c-myc-K6R, added K7R mutations in gp22-c-mycThis study
pBAD24-gp19pBAD24 with Mu gp19 under Shine Dalgarno AGGAGGThis study
pBAD33-gp20pBAD33 with Mu gp20 under Shine Dalgarno AGGAGGThis study
pBAD24-gp19+gp20pBAD24 with Mu gp19 and gp20 inserted with native RBSThis study
pBAD24-gp25-hispBAD24 plasmid with Mu gp25-his cloned into EcoRI and HindIII sites under Shine Dalgarno AGGAGGThis study
pBAD33-gp25-hispBAD33 plasmid with Mu gp25-his under Shine Dalgarno AGGAGGThis study
pBAD18-Kan-Mu-Pm-gp25pBAD18-Kan with the Mu middle promoter Pm through gp25This study
pBAD18-Kan-Mu-gp18-gp25Derived from pBAD18-Kan-Mu Pm-gp25, Pm deletedThis study
pBAD18-Kan-Mu-gp18-gp25-ΔPlysDerived from pBAD18-Kan-Mu gp18-gp25, Plys deleted, placing all genes under PBAD controlThis study
pBAD18-Kan-Mu-gp18-gp25-gp22-E37ADerived from pBAD18-Kan-Mu-gp18-gp25-ΔPlys, inactivating catalytic domain mutation E37A in gp22This study
pBAD18-Kan-Mu-gp22-gp25Derived from pBAD18-Kan-Mu-gp18-gp25-ΔPlys, genes 18–21 deletedThis study
pBAD18-Kan-Mu-gp22-gp25-gp23.1amDerived from pBAD18-Kan-Mu-gp22-gp25, W65 amber mutation in gp23.1This study
pBAD33-λ-RzRz1pBAD33 with λ spanins RzRz1 subcloned into the KpnI and HindIII sitesLaboratory stock
pBAD24-21-S68pBAD24 with phage 21 S68 pinholinThis study
pBAD24-21-S68ampBAD24-21-S68 with amber allele at V43This study
pBAD24-λ-S105pBAD24 with λ S105 holinThis study
pZE12-gp22pZE12 with Mu gp22 insertThis study
pZE12-21-R-c-mycpZE12 with phage 21 SAR endolysin insert with a C-terminal myc tag (EQKLISEEDL)Laboratory stock (11)
pZE12-933W-RpZE12 with phage 933W SAR endolysin insertLaboratory stock (11)
pZE12-Bcep22-RpZE12 with phage Bcep22 SAR endolysin insertLaboratory stock (11)
TABLE 3 Oligonucleotides used in this study
Primer name5′–3′ SequencePurpose
gp25 C-his Rev HindIIICCCAAGCTTTCAATGATGATGGTGGTGGTGACCTCCTGClone gp25 with C-terminal his tag
Mu gp22 pBAD33 For KpnIGGGGTACCTTCAGGGAAGCATGATGGClone gp22 with ssd
gp22 KpnI Strong SD ForGGGGTACCAGGAGGAATTCATGGClone gp22 with osd
pBAD24 gp22 C-myc QC ForCAGAAGAGGATCTGTAAAAGCTTGGCTGTTTTGGClone gp22 with C-terminal c-myc tag
pBAD24 gp22 C-myc QC RevAGATGAGTTTTTGTTCCTGGCGAAGGTCATCCGClone gp22 with C-terminal c-myc tag
K7R gp22 ForwardGGATACCAAAAAGACTGAAAGCCGCgp22 mutagenesis
K9R gp22 ForwardAAAAAAACTGAGAGCCGCACTGCTGGgp22 mutagenesis
Mu gp22 K6R K7R FCGTAGACTGAAAGCCGCACgp22 mutagenesis
Mu gp22 ssd P5 RTGGTATCCCCGCCATGAATTCgp22 mutagenesis
Mu gp20 ssd KpnI FacgtggtaccaggaggaattcatgtacagaaaattcagtgClone gp20
Mu gp20 HindIII RagccaagcttttaacagacagaaagatacaccacggcClone gp20
Mu Pm nt9163 FttctgtaaacagtaaagccggttaatccggcClone Mu18–25
Mu gp25 end RtcatgaattactccggtttactgcgClone Mu18–25
Mu gp21 end RTTATTTTGGGGTTTCAGGTGAGAACAGADelete Mu Plys in gp18-gp25 ΔPm clone
Mu gp22 pBAD FAGGGAAGCATGATGGCGGGGATACDelete Mu Plys in gp18-gp25 ΔPm clone
Mu gp22 pBAD FAGGGAAGCATGATGGCGGGGATACDelete Mu gp18-gp21 in gp18-gp25 ΔPm ΔPlys clone
pBAD18-Kan RCCGAGCTCGAATTCGCTAGDelete Mu gp18-gp21 in gp18-gp25 ΔPm ΔPlys clone
Mu Mup23.1 W65am FCCCCTGACGTAGGGAGCATCCClone gp22-gp25 23.1-am
Mu Mup23.1 Q61 RCTGTGGCATGGGCGGAACClone gp22-gp25 23.1-am
Mu gp22 E37A FATCCACCTGGCAAATATCGCTTATATGgp22 catalytic domain knockout E37A
Mu gp22 L33 RCAGAGACTGACGGGTCATgp22 catalytic domain knockout E37A
Construction of the λ lysogen lacking a lysis cassette relied on recombination between λ stf cI857 and a pLambchops plasmid. The pLambchops plasmid was missing the entire λ lysis cassette but retained upstream and downstream homology to the phage late region. pLambchops was generated by inverse PCR on a spectinomycin-resistant pRE RzRz1 vector generated from the combination of a commercially synthesized gene block with the spanin-containing vector.

Construction of Mu knockout mutants.

Construction of Mucts deletion-substitutions was performed as described previously (48, 55). Briefly, flanking regions of the target genes (19, 20, 22, and 25) were attached to the cat gene via PCR using primers listed in Table 3. Each construct retained the bases encoding ~10 residues at the ends so as not to disrupt overlapping upstream and downstream features. Next, 10 to 100 ng of purified linear dsDNA PCR products was digested with DpnI and transferred into MC4100 AraR cells carrying Red helper plasmid pKD46 by electroporation. Prior to electroporation, MC4100 AraR cells were grown at 30°C in LB supplied with Amp and l-arabinose (Ara) and then resuspended in 10% glycerol. Cam-resistant colonies were selected. The cat insertion was confirmed by PCR and Sanger sequencing with Eton Biosciences (San Diego, CA).
Mu phage stocks collected from induced prophages were sterilized and used to infect MDS12 cultures at a low multiplicity of infection, then selected for Cam resistance via plating at 30°C or tested for temperature sensitivity for the Mucts strain. The lysogens were subjected to next-generation sequencing to determine the location and number of insertions.

Microscopy and visualization of cells.

To visualize each sample, 1 μL of culture was placed on a glass slide under a coverslip. Snapshots were acquired with an Axiocam 702 mono camera mounted on a Zeiss Axio Observer 7 inverted microscope equipped with plan-neofluar 4×/0.75 NA Ph2 and alpha plan-apochromat 100×/1.46 oil (UV) Ph3 oil M27 objectives. A glass slide was not used when recording videos of motile cells: cell culture was applied directly to a coverslip and left exposed to allow for air exchange and free cell movement. Images and time-series videos were processed by minor brightness/contrast adjustments within the Carl Zeiss Zen 2.3 imaging software.

Recovery of 2,4-dinitrophenol-treated cells.

MG1655 cells containing pBAD plasmids were treated with DNP at a final concentration of 2 mM 60 min after plasmid induction. Ten min later, a 5-mL aliquot of cells was transferred to a 50-mL conical tube and centrifuged at 5,000 × g for 10 min. After the supernatant was discarded, cells were resuspended in fresh LB with the appropriate antibiotics and 0.4% arabinose and transferred to a 250-mL flask. Recovery incubation proceeded at 37°C in a water bath with aeration.

Protein structural models.

Protein covariance was assessed locally with the EVcouplings Python pipelines using the UniRef100 database (February 2020 release) for alignment and PSIPRED BLAST (56, 57). Evolutionary couplings for gp22 (both full-length and range restricted to SAR residues 5 to 30) and gp25 (full-length or cytoplasmic residues 25 to 99) were calculated in the monomer pipeline, using default parameters. Protein models were visualized using UCSF ChimeraX (58). Transmembrane topology diagrams were generated using TOPO2 (Johns S.J., TOPO2, transmembrane protein display software available from http://www.sacs.ucsf.edu/TOPO2/).

SDS-PAGE and Western blotting.

Whole protein samples were prepared from culture samples taken at specific time points after induction by addition to ice-cold 10% (vol/vol) trichloroacetic acid. Pelleted protein samples were then washed with cold acetone before resuspension in nonreducing Laemmli buffer. Electrophoresis on Novex 16% Tricine gels (Thermo Fisher Scientific, Waltham, MA) was performed according to manufacturer recommendations. Proteins were transferred to a 0.2-μm polyvinylidene difluoride membrane by the iBlot2 system with a pre-programmed protocol (20 V for 1 min, 23 V for 4 min, 25 V for 1 to 2 min). After blocking with 4% (wt/vol) milk–Tris-buffered saline; His-tagged and c-myc tagged proteins were detected separately. Antibodies and substrate were acquired from Thermo Fisher Scientific (Waltham, MA). The mouse α-His primary antibody was used at a 1:2,000 dilution, and the secondary goat-anti-mouse Alexa Fluor 680 was used at a 1:20,000 dilution. α-c-myc (9E10) mouse monoclonal was used at a dilution of 1:1,000 overnight, followed by secondary antibody (goat-anti-mouse horseradish peroxidase) at 1:1,000, with SuperSignal West Femto Maximum Sensitivity Chemiluminescent Substrate (Thermo Fisher) for detection. Blots were scanned on Amersham Imager 600 RGB (GE Healthcare, Chicago, LA) and Bio-Rad-ChemiDoc machines, respectively, and analyzed using LI-COR ImageStudioLite version 4.0.21 software.

Cell shape counting assay.

MG1655 cells were subcultured 1:250 from an overnight into 250-mL flasks containing 25 mL LB, appropriate antibiotics, and 10 mM Mg2+. Flasks were incubated at 37°C with aeration in a water bath. At A550 ~ 0.2, the pBAD vector was induced with addition of arabinose to 0.4% (T = 0). Thirty minutes after pBAD vector induction (T = 30), aliquots were taken and placed on ice for microscopy visualization. Forty minutes after pBAD vector induction (T = 40), 1 M IPTG was added to a final concentration of 1 mM to induce the pZE12 vector. At T = 50, aliquots were taken and placed on ice for microscopy visualization. Counting of “round”- versus “rod”-shaped cells was accomplished with assistance from a custom analysis program within the proprietary Zen 2.3 software.

Protein bioinformatic analyses.

The Mu gp25 amino acid sequence was searched in the InterPro 86.0 June 2021 database release and analyzed by BLASTp against bacteria and phage sequences in the NCBI nonredundant database (59, 60). Thirty-eight Caudovirales phages are listed with a protein containing the DUF2730 domain (equivalent to Pfam entry PF10805, InterPro entry IPR020269). Of these, 3 metagenome entries and 1 incomplete genome were excluded, for 34 complete records sequenced from active phages retrieved from the NCBI genome record and inspected in Artemis or the Center for Phage Technology Galaxy and Apollo web instance (https://cpt.tamu.edu/galaxy-pub) (61, 62). Full information on these phages, including accession numbers, taxonomic classification, and their hosts, is given in Table S1. In 33 cases, the phage endolysins was identified within 10 genes up- or downstream by analyzing InterProScan for domains known to be associated with endolysins. The endolysin for Vibrio phage Martha 12B12 was not identifiable by this bioinformatic method; instead, that genome was aligned with Vibrio phage 1.028.O._10N.286.45.B6 using EasyFig (63), then by using HHPred through their web interface (20). Manual inspection of synteny and analyses were conducted with TMHMM, Phobius, and InterProScan in the Center for Phage Technology Galaxy instance to predicts lysozyme and transmembrane domains for phage Martha (6467). Full protein sets were aligned in CLC Genomics Workbench (Qiagen, Redwood City, CA). In addition to the tools mentioned above, predicted secondary structures for all lysis proteins were inspected using HHPred in the MPI Bioinformatics Toolkit (20).

Figure preparation.

Graphics for all figures were combined for display in Inkscape 1.0 (68).


This work was supported by Public Health Service grants GM27099 and GM136396 to R.Y. and by the Center for Phage Technology, which is jointly supported by Texas A&M AgriLife Research and Texas A&M University. We thank the Young laboratory members, past and present, for their helpful criticisms and suggestions. We also acknowledge the work of previous lab members Gabriel Kuty, Russell Moreland, and Ashley Holt for their construction of unpublished plasmid reagents used in this study. Finally, we thank Deborah Jacobs-Sera and Graham Hatfull for their review and update of the Mu genome annotation.
Conceptualization, J.R., J.S.C. Y.C., K.H.T., J.L.C., and R.Y.; Data Curation, J.S.C., J.R., Y.C.; Formal Analysis, J.S.C. and C.R.; Investigation, J.R., J.S.C., Y.C., L.T.M., K.H.T., and R.Y.; Methodology, J.S.C., J.R., Y.C., K.H.T.; Resources, R.Y.; Supervision, J.R., Y.C., J.L.C.; Writing –Original Draft, J.S.C., J.R., Y.C., and R.Y.; Writing –Review & Editing, J.R., J.S.C., Y.C., L.T.M., C.R., J.L.C., and R.Y.; Validation, J.S.C., J.R., and L.T.M.; Visualization, J.R. and J.S.C.; Project Administration, J.R. Y.C., J.L.C., and R.Y.; Funding Acquisition, R.Y.


This article is a direct contribution from Ryland F. Young, a Fellow of the American Academy of Microbiology, who arranged for and secured reviews by William Margolin, McGovern Medical School; Ing-Nang Wang, Department of Biological Sciences, University at Albany; and John Dennehy, Queens College CUNY.

Supplemental Material

File (mbio.00813-22-s0001.xlsx)
File (mbio.00813-22-s0002.avi)
File (mbio.00813-22-s0003.svg)
ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.


Cahill J, Young R. 2019. Phage lysis: multiple genes for multiple barriers. Adv Virus Res 103:33–70.
Kongari R, Rajaure M, Cahill J, Rasche E, Mijalis E, Berry J, Young R. 2018. Phage spanins: diversity, topological dynamics and gene convergence. BMC Bioinformatics 19:326.
Holt A, Cahill J, Ramsey J, Martin C, O’Leary C, Moreland R, Maddox LT, Galbadage T, Sharan R, Sule P, Cirillo JD, Young R. 2022. Phage-encoded cationic antimicrobial peptide required for lysis. J Bacteriol 204:e00214-21.
Symonds N, Toussaint A, Howe M. 1987. Phage Mu. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.
Faelen M, Toussaint A. 1973. Isolation of conditional defective mutants of temperate phage Mu-1 and deletion mapping of the Mu-1 prophage. Virology 54:117–124.
Howe MM. 1973. Prophage deletion mapping of bacteriophage Mu-1. Virology 54:93–101.
Howe MM, O'Day KJ, Schultz DW. 1979. Isolation of mutations defining five new cistrons essential for development of bacteriophage Mu. Virology 93:303–319.
Morgan GJ, Hatfull GF, Casjens S, Hendrix RW. 2002. Bacteriophage Mu genome sequence: analysis and comparison with Mu-like prophages in Haemophilus, Neisseria and Deinococcus. J Mol Biol 317:337–359.
Vermassen A, Leroy S, Talon R, Provot C, Popowska M, Desvaux M. 2019. Cell wall hydrolases in bacteria: insight on the diversity of cell wall amidases, glycosidases and peptidases toward peptidoglycan. Front Microbiol 10:331.
Samanta S, Sharma A, Saha A, Singh M, Das A, Bhattacharya M, Saha R, Lee S-S, Chakraborty C. 2021. The bacteriophage Mu lysis system: a new mechanism of host lysis? Biocell 45:1175–1186.
Kuty GF. 2011. SAR endolysin regulation in dsDNA phage lysis of Gram-negative hosts. Doctoral Dissertation. Texas A&M University, College Station, TX.
Xu M, Struck DK, Deaton J, Wang I-N, Young R. 2004. A signal-arrest-release sequence mediates export and control of the phage P1 endolysin. Proc Natl Acad Sci USA 101:6415–6420.
Park T, Struck DK, Dankenbring CA, Young R. 2007. The pinholin of lambdoid phage 21: control of lysis by membrane depolarization. J Bacteriol 189:9135–9139.
Young R. 2014. Phage lysis: three steps, three choices, one outcome. J Microbiol 52:243–258.
Berry J, Rajaure M, Pang T, Young R. 2012. The spanin complex is essential for lambda lysis. J Bacteriol 194:5667–5674.
Young R. 2013. Phage lysis: do we have the hole story yet? Curr Opin Microbiol 16:790–797.
Mathee K, Howe MM. 1993. The bacteriophage Mu middle operon: essential and nonessential functions. Virology 196:712–721.
Pang T, Fleming TC, Pogliano K, Young R. 2013. Visualization of pinholin lesions in vivo. Proc National Acad Sci USA 110:E2054-63.
Babu K, Arulandu A, Sankaran K. 2018. The structure of DLP12 endolysin exhibiting alternate loop conformation and comparative analysis with other endolysins. Proteins 86:210–217.
Zimmermann L, Stephens A, Nam S-Z, Rau D, Kübler J, Lozajic M, Gabler F, Söding J, Lupas AN, Alva V. 2018. A completely reimplemented MPI bioinformatics toolkit with a new HHpred server at its core. J Mol Biol 430:2237–2243.
Sun Q, Kuty GF, Arockiasamy A, Xu M, Young R, Sacchettini JC. 2009. Regulation of a muralytic enzyme by dynamic membrane topology. Nat Struct Mol Biol 16:1192–1194.
Oliveira H, Melo LDR, Santos SB, Nóbrega FL, Ferreira EC, Cerca N, Azeredo J, Kluskens LD. 2013. Molecular aspects and comparative genomics of bacteriophage endolysins. J Virol 87:4558–4570.
Gontijo MTP, Vidigal PMP, Lopez MES, Brocchi M. 2021. Bacteriophages that infect Gram-negative bacteria as source of signal-arrest-release motif lysins. Res Microbiol 172:103794.
Vandermarliere E, Mueller M, Martens L. 2013. Getting intimate with trypsin, the leading protease in proteomics: trypsin in proteomics. Mass Spectrom Rev 32:453–465.
Black LW, Showe MK. 1983. Bacteriophage T4, p 219–245. In Matthews CK, Kutter EM, Mosig G, Berget PB (ed), Morphogenesis of the T4 head. American Society for Microbiology, Washington, DC.
Frain KM, Robinson C, van Dijl JM. 2019. Transport of folded proteins by the Tat system. Protein J 38:377–388.
Hulo C, Masson P, Mercier PL, Toussaint A. 2015. A structured annotation frame for the transposable phages: a new proposed family “Saltoviridae” within the Caudovirales. Virology 477:155–163.
Toussaint A, Gijsegem FV. 2018. Extension of the transposable bacterial virus family: two genomic organisations among phages and prophages with a Tn552-related transposase. Res Microbiol 169:495–499.
von Heijne G. 1992. Membrane protein structure prediction. Hydrophobicity analysis and the positive-inside rule. J Mol Biol 225:487–494.
Young R. 2002. Bacteriophage holins: deadly diversity. J Mol Microb Biotechnol 4:21–36.
Zheng Y, Struck DK, Dankenbring CA, Young R. 2008. Evolutionary dominance of holin lysis systems derives from superior genetic malleability. Microbiology (Reading) 154:1710–1718.
Labedan B, Letellier L. 1981. Membrane potential changes during the first steps of coliphage infection. Proc Natl Acad Sci USA 78:215–219.
Letellier L, Boulanger P. 1989. Involvement of ion channels in the transport of phage DNA through the cytoplasmic membrane of E. coli. Biochimie 71:167–174.
Ramanculov E, Young R. 2001. An ancient player unmasked: T4 rI encodes a t-specific antiholin. Mol Microbiol 41:575–583.
Hays SG, Seed KD. 2020. Dominant Vibrio cholerae phage exhibits lysis inhibition sensitive to disruption by a defensive phage satellite. Elife 9:e53200.
Harshey RM. 1984. Transposition without duplication of infecting bacteriophage Mu DNA. Nature 311:580–581.
Darzins A, Kent NE, Buckwalter MS, Casadaban MJ. 1988. Bacteriophage Mu sites required for transposition immunity. Proc Natl Acad Sci USA 85:6826–6830.
Adzuma K, Mizuuchi K. 1988. Target immunity of Mu transposition reflects a differential distribution of Mu B protein. Cell 53:257–266.
Ge J, Lou Z, Harshey RM. 2010. Immunity of replicating Mu to self-integration: a novel mechanism employing MuB protein. Mob Dna 1:8.
Manna D, Higgins NP. 1999. Phage Mu transposition immunity reflects supercoil domain structure of the chromosome. Mol Microbiol 32:595–606.
Walker DM, Harshey RM. 2020. Deep sequencing reveals new roles for MuB in transposition immunity and target-capture, and redefines the insular Ter region of E. coli. Mob Dna 11:26.
Waggoner BT, González NS, Taylor AL. 1974. Isolation of heterogeneous circular DNA from induced lysogens of bacteriophage Mu-1. Proc Natl Acad Sci USA 71:1255–1259.
Reader RW, Siminovitch L. 1971. Lysis defective mutants of bacteriophage lambda: genetics and physiology of S cistron mutants. Virology 43:607–622.
Doermann AH. 1948. Lysis and lysis inhibition with Escherichia coli bacteriophage. J Bacteriol 55:257–276.
Schito GC, Molina AM, Pesce A. 1967. Lysis and lysis inhibition with N4 coliphage. Giornale di Microbiologia 15:229–244.
Moussa SH, Lawler JL, Young R. 2014. Genetic dissection of T4 lysis. J Bacteriol 196:2201–2209.
To KH, Dewey J, Weaver J, Park T, Young R. 2013. Functional analysis of a class I holin, P2 Y. J Bacteriol 195:1346–1355.
Bolhuis A, Bogsch EG, Robinson C. 2000. Subunit interactions in the twin-arginine translocase complex of Escherichia coli. FEBS Lett 472:88–92.
Englesberg E, Anderson R, Weinberg R, Lee N, Hoffee P, Huttenhauer G, Boyer H. 1962. L-Arabinose-sensitive, l-ribulose 5-phosphate 4-epimerase-deficient mutants of Escherichia coli. J Bacteriol 84:137–146.
Guzman LM, Belin D, Carson MJ, Beckwith J. 1995. Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol 177:4121–4130.
Wang I-N, Deaton J, Young R. 2003. Sizing the holin lesion with an endolysin-beta-galactosidase fusion. J Bacteriol 185:779–787.
Inc. PT. 2015. Collaborative data science publisher: Plotly Technologies Inc., Montréal, Quebec, Canada. Available from https://plot.ly.
Chamblee J. 2021. jakechamblee/lysis-curve: 1.12 (1.12.). Zenodo. Available from https://doi.org/10.5281/zenodo.5514286.
Bläsi U, Young R. 1996. Two beginnings for a single purpose: the dual-start holins in the regulation of phage lysis. Mol Microbiol 21:675–682.
Grundling A, Smith DL, Bläsi U, Young R. 2000. Dimerization between the holin and holin inhibitor of phage lambda. J Bacteriol 182:6075–6081.
Hopf TA, Green AG, Schubert B, Mersmann S, Schärfe CPI, Ingraham JB, Toth-Petroczy A, Brock K, Riesselman AJ, Palmedo P, Kang C, Sheridan R, Draizen EJ, Dallago C, Sander C, Marks DS. 2019. The EVcouplings Python framework for coevolutionary sequence analysis. Bioinformatics 35:1582–1584.
Hopf TA, Colwell LJ, Sheridan R, Rost B, Sander C, Marks DS. 2012. Three-dimensional structures of membrane proteins from genomic sequencing. Cell 149:1607–1621.
Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. 2004. UCSF Chimera: a visualization system for exploratory research and analysis. J Comput Chem 25:1605–1612.
Camacho C, Coulouris G, Avagyan V, Ma N, Papadopoulos J, Bealer K, Madden TL. 2009. BLAST+: architecture and applications. BMC Bioinformatics 10:421.
Mitchell AL, Attwood TK, Babbitt PC, Blum M, Bork P, Bridge A, Brown SD, Chang H-Y, El-Gebali S, Fraser MI, Gough J, Haft DR, Huang H, Letunic I, Lopez R, Luciani A, Madeira F, Marchler-Bauer A, Mi H, Natale DA, Necci M, Nuka G, Orengo C, Pandurangan AP, Paysan-Lafosse T, Pesseat S, Potter SC, Qureshi MA, Rawlings ND, Redaschi N, Richardson LJ, Rivoire C, Salazar GA, Sangrador-Vegas A, Sigrist CJA, Sillitoe I, Sutton GG, Thanki N, Thomas PD, Tosatto SCE, Yong S-Y, Finn RD. 2018. InterPro in 2019: improving coverage, classification and access to protein sequence annotations. Nucleic Acids Res 47:D351–D360.
Ramsey J, Rasche H, Maughmer C, Criscione A, Mijalis E, Liu M, Hu JC, Young R, Gill JJ. 2020. Galaxy and Apollo as a biologist-friendly interface for high-quality cooperative phage genome annotation. PLoS Comput Biol 16:e1008214.
Carver T, Harris SR, Berriman M, Parkhill J, McQuillan JA. 2012. Artemis: an integrated platform for visualization and analysis of high-throughput sequence-based experimental data. Bioinformatics 28:464–469.
Sullivan MJ, Petty NK, Beatson SA. 2011. Easyfig: a genome comparison visualizer. Bioinformatics 27:1009–1010.
Krogh A, Larsson B, von HG, Sonnhammer ELL. 2001. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol 305:567–580.
Jones P, Binns D, Chang H-Y, Fraser M, Li W, McAnulla C, McWilliam H, Maslen J, Mitchell A, Nuka G, Pesseat S, Quinn AF, Sangrador-Vegas A, Scheremetjew M, Yong S-Y, Lopez R, Hunter S. 2014. InterProScan 5: genome-scale protein function classification. Bioinformatics 30:1236–1240.
Afgan E, Baker D, Batut B, van den Beek M, Bouvier D, Čech M, Chilton J, Clements D, Coraor N, Grüning BA, Guerler A, Hillman-Jackson J, Hiltemann S, Jalili V, Rasche H, Soranzo N, Goecks J, Taylor J, Nekrutenko A, Blankenberg D. 2018. The Galaxy platform for accessible, reproducible and collaborative biomedical analyses: 2018 update. Nucleic Acids Res 46:W537–W544.
Käll L, Krogh A, Sonnhammer ELL. 2004. A combined transmembrane topology and signal peptide prediction method. J Mol Biol 338:1027–1036.
Project I. Inkscape 1.1. Available from https://inkscape.org.
Schuster-Böckler B, Schultz J, Rahmann S. 2004. HMM Logos for visualization of protein families. BMC Bioinformatics 5:7.
Wheeler TJ, Clements J, Finn RD. 2014. Skylign: a tool for creating informative, interactive logos representing sequence alignments and profile hidden Markov models. BMC Bioinformatics 15:7–9.
Casadaban MJ. 1976. Transposition and fusion of the lac genes to selected promoters in Escherichia coli using bacteriophage lambda and Mu. J Mol Biol 104:541–555.
Kolisnychenko V, Plunkett G, Herring CD, Fehér T, Pósfai J, Blattner FR, Pósfai G. 2002. Engineering a reduced Escherichia coli genome. Genome Res 12:640–647.
Datsenko KA, Wanner BL. 2000. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97:6640–6645.
Lutz R, Bujard H. 1997. Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements. Nucleic Acids Res 25:1203–1210.

Information & Contributors


Published In

cover image mBio
Volume 13Number 328 June 2022
eLocator: e00813-22
Editor: Michael T. Laub, Massachusetts Institute of Technology
PubMed: 35471081


Received: 23 March 2022
Accepted: 25 March 2022
Published online: 26 April 2022


  1. bacteriophage lysis
  2. bacteriophages
  3. membranes



Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Yi Chen
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Present address: Yi Chen, MD Anderson, Houston, TX; Kam H. To, Raleigh, NC; Jesse L. Cahill, Sandia National Laboratories, Albuquerque, New Mexico, USA.
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Curtis Ross
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Kam H. To
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Present address: Yi Chen, MD Anderson, Houston, TX; Kam H. To, Raleigh, NC; Jesse L. Cahill, Sandia National Laboratories, Albuquerque, New Mexico, USA.
Jesse L. Cahill
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA
Present address: Yi Chen, MD Anderson, Houston, TX; Kam H. To, Raleigh, NC; Jesse L. Cahill, Sandia National Laboratories, Albuquerque, New Mexico, USA.
Center for Phage Technology, Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas, USA


Michael T. Laub
Massachusetts Institute of Technology


Jake S. Chamblee and Jolene Ramsey contributed equally to this work. Authors are listed in alphabetical order.
The authors declare no conflict of interest.

Metrics & Citations



  • For recently published articles, the TOTAL download count will appear as zero until a new month starts.
  • There is a 3- to 4-day delay in article usage, so article usage will not appear immediately after publication.
  • Citation counts come from the Crossref Cited by service.


If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. For an editable text file, please select Medlars format which will download as a .txt file. Simply select your manager software from the list below and click Download.

View Options

Figures and Media






Share the article link

Share with email

Email a colleague

Share on social media

American Society for Microbiology ("ASM") is committed to maintaining your confidence and trust with respect to the information we collect from you on websites owned and operated by ASM ("ASM Web Sites") and other sources. This Privacy Policy sets forth the information we collect about you, how we use this information and the choices you have about how we use such information.
FIND OUT MORE about the privacy policy