Download HOLINS: The Protein Clocks of Bacteriophage Infections

P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
Annu. Rev. Microbiol. 2000. 54:799–825
c 2000 by Annual Reviews. All rights reserved
Copyright HOLINS: The Protein Clocks of Bacteriophage
Infections
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Ing-Nang Wang,1 David L. Smith,2 and Ry Young1
1Department of Biochemistry and Biophysics, Texas A & M University, College Station,
Texas 77843-2128; e-mail: [email protected]; [email protected]; 2VAGLAHS,
Lipid Research, Los Angeles, California 90073; e-mail: [email protected]
Key Words lysis, endolysins, membranes, peptidoglycan
■ Abstract Two proteins, an endolysin and a holin, are essential for host lysis by
bacteriophage. Endolysin is the term for muralytic enzymes that degrade the cell wall;
endolysins accumulate in the cytosol fully folded during the vegetative cycle. Holins
are small membrane proteins that accumulate in the membrane until, at a specific time
that is “programmed” into the holin gene, the membrane suddenly becomes permeabilized to the fully folded endolysin. Destruction of the murein and bursting of the
cell are immediate sequelae. Holins control the length of the infective cycle for lytic
phages and so are subject to intense evolutionary pressure to achieve lysis at an optimal
time. Holins are regulated by protein inhibitors of several different kinds. Holins constitute one of the most diverse functional groups, with >100 known or putative holin
sequences, which form >30 ortholog groups.
CONTENTS
OVERVIEW: Strategies for Lysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
LESSONS FROM THE LYSIS PARADIGM IN LAMBDOID PHAGES . . . . . . . . .
Lysogenic Induction and Lysis Phenotypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Saltatory Nature of Lysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Evolutionary Forces Shaping Holin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Mosaicism of Lysis Cassettes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sλ and S21 Represent the Two Major Classes of Holins . . . . . . . . . . . . . . . . . . . . .
Dual-Start Motifs: Setting the Clock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Probing the Structure of Sλ in the Membrane . . . . . . . . . . . . . . . . . . . . . . . . . . .
Genetic Analysis of Sλ Function and Regulation . . . . . . . . . . . . . . . . . . . . . . . . .
Oligomerization of the Holin and the Nature of the Hole . . . . . . . . . . . . . . . . . . .
Physical and Biochemical Analysis of the Holin . . . . . . . . . . . . . . . . . . . . . . . . .
What Makes the Lysis Clock Tick? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
HOLINS IN NONLAMBDOID SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Extragenic Holin Regulation in Coliphages . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Functional Tests of Holins in Other Phages . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Biotechnology and Holin/Endolysin Lysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
0066-4227/00/1001-0799$14.00
800
801
801
802
803
803
804
805
806
807
809
809
810
812
812
814
815
799
P1: FUI
August 11, 2000
800
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
SURVEY OF HOLINS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Preliminary Identification of Holins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Distribution of Holins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Prevalence of the Dual-Start Motif . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Organization of Holin and Endolysin Genes . . . . . . . . . . . . . . . . . . . . . . . . . . . .
EVOLUTIONARY CONSIDERATIONS AND APPROACHES . . . . . . . . . . . . . . .
PERSPECTIVE AND SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
815
815
816
817
817
818
819
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
OVERVIEW: Strategies for Lysis
Except for filamentous phages, which extrude from the host, bacteriophages are
lytic. There are at least two general strategies for host lysis (95). Some simple phages of Gram-negative hosts, like single-stranded DNA and single-stranded
RNA phages, accomplish lysis with a single lysis gene that does not encode a muralytic enzyme activity. Recent results suggest that, in at least one of these systems,
the lysis protein is an inhibitor of murein synthesis (8a). To achieve lysis, doublestranded DNA phages of eubacteria resort to active degradation of the peptidoglycan with a muralytic enzyme or endolysin (93, 94). The endolysin, which has one
or more of four different muralytic activities but no secretory signal sequence, accumulates in a fully folded and active state in the cytosol. At minimum, lysis requires
another gene that encodes a small membrane protein called a holin. At a time that
is “programmed” into the structure of the holin gene, the endolysin suddenly gains
access to its substrate, the cell wall; destruction of the murein and bursting of the
cell are immediate consequences. The interesting questions are, “What does the
holin do to the membrane?” and “How is holin function scheduled so precisely?”
It is important to define terms carefully. Holin function is associated with a
collapse of the membrane potential and permeabilization of the membrane (93).
Throughout this review, the term hole is used as a convenient substitute for “holinmediated permeabilizing lesion,” and the term hole formation is used for the saltatory holin-dependent permeabilization event. However, we emphasize that it is
not known whether endolysin actually passes through a defined hole or channel;
all that is known is that, before lysis, the bulk of endolysin activity is found in
the soluble fraction, and chemical permeabilization of the bilayer by CHCl3 can
substitute for holin function. In at least one case, it has been reported that holin
function can activate a muralytic enzyme that is already exported from the cytosol
but remains inactive for unknown reasons (27; see Functional Tests of Holins in
Other Phages below). Thus it is important to emphasize that the hole is completely
uncharacterized at the molecular level. In any case, the holin certainly controls
the functional access of the endolysin to the murein and, perforce, the length of
the vegetative cycle. From this perspective, the holin can be considered the simplest molecular clock. Since the last reviews (93, 94) there has been an explosion
of new information about the sequence, function, and regulation of holins. This
work organizes and summarizes these advances. For more comprehensive listings,
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
801
please refer to Table 2, a catalog of all holin genes and their characteristics, and
Table 3, which lists the amino acid sequences of all known or putative holins and
holin inhibitors [see the Supplemental Materials section of the Annual Reviews
home page (http://www.AnnualReviews.org)].
LESSONS FROM THE LYSIS PARADIGM
IN LAMBDOID PHAGES
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Lysogenic Induction and Lysis Phenotypes
The induction of a thermosensitive λ lysogen offers a useful system for studying lysis because it allows the experimenter to initiate the vegetative cycle synchronously
in an entire culture. Under exactly specified physiological conditions of aeration
and culture density, a cIts lysogen will undergo a sharply defined lysis event at
50 min after thermal induction (Figure 1), liberating a burst of 100 virions. Lysis
Figure 1 Lysis phenotypes and virion accumulation in an induced λ lysogen. Cell density
is measured as absorbance at 550 nm (A550) at times after induction of a λ lysogen with
various S and R lysis gene alleles. Symbols: open triangle, S+ R+; open circle, S− R+; closed
diamond, S− R+, with CHCl3 added at 70 min; open diamond, S+ R−; closed box, S+ R+
with 5 mM KCN added at 30 min; open box, Sa52g R+ (93); star, approximate intracellular
content of virions, as determined previously for a thermally induced S− lysogen with the
same normal lysis time (70).
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
802
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
requires two genes, S and R, which are the first two cistrons of the sole late transcriptional unit of λ (Figure 2, see color insert). The late promoter p0 R is activated
∼8 min after infection, and thus the lysis genes and all late genes are expressed
constitutively for the rest of the vegetative cycle.
R encodes the endolysin, in this case an 18-kDa soluble muralytic transglycosylase (9). Rz and Rz1 are nested genes that encode outer membrane proteins that
are auxiliary lysis factors. When either Rz or Rz1 is defective and the medium
contains millimolar concentrations of divalent cations, cell lysis stops at a spherical cell stage, apparently maintained by the stabilized outer membrane. It has
been proposed that Rz and Rz1 are involved in cleaving the oligopeptide linkages between the peptidoglycan and the outer membrane lipoprotein Lpp (98).
Whatever the Rz/Rz1 function is, it is not required for lysis under standard laboratory conditions and is restricted to some phages of Gram-negative hosts (47,
96, 98).
S spans 107 codons and encodes two proteins—the holin and the holin inhibitor,
designated Sλ105 and Sλ107, respectively, for their lengths in amino acid residues
(see Dual Start Motifs: Setting the Clock below). A null allele of S has a distinctive
phenotype that is diagnostic for a holin gene; for >2 h after the normal lysis time,
culture mass continues to increase, and the number of intracellular virions increases
to1000/cell (70). Holin action can be complemented chemically, by the addition
of CHCl3, which directly permeabilizes the membrane, resulting in immediate and
total lysis of the culture and demonstrating that the R endolysin lacks access only
through the membrane to its murein target (Figure 1). In contrast, a null R allele
is characterized by cessation of culture growth and respiration at the normal lysis
time, without detectable lysis. Thus, the endolysin can be thought of as a reporter
gene for holin function.
The Saltatory Nature of Lysis
Bacteriophage λ is a paradigm system for studying holin function because the
induced culture continues to grow right up until lysis. Recently, microscopy of
motile Escherichia coli cells after induction of the λ lysis genes from a plasmid vector revealed that the cells are fully motile until, suddenly, swimming and
tumbling stops, and then, within a few seconds, the cell bursts and disappears
as a refractile body (A Gründling, M Manson, R Young, manuscript in preparation). Moreover, under carefully controlled conditions, almost all of an induced
lysogenic culture undergoes lysis within minutes of lysis onset. This precise
temporal regulation likely involves the energy state of the membrane because
adding an energy poison (i.e. cyanide or dinitrophenol) sufficiently late in the
infective cycle can instantly trigger premature lysis with λ (Figure 1) and in other
holin-dependent systems (28). Any model that is presented to explain the timing
function of holins must also explain the premature lysis that is effected by energy
poisons.
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
803
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Evolutionary Forces Shaping Holin
If phage lysis is considered from first principles, the tight scheduling of holin
action is not unexpected. In the bacteriophage λ vegetative cycle, it takes ∼20 min
before the first progeny virion has been assembled, which means that, in the last
30 min of the vegetative cycle, intracellular virions increase 100-fold. If, instead
of triggering at 50 min, the holin did not trigger until 3 h, there would be a
further 10-fold increase (70; Figure 1). Assuming that, after lysis, each progeny
virion immediately infects a new host, a single phage with the 50-min holin would
multiply to a level of >106 virions, which is >103-fold higher than the phage
with the 3-h holin. It is clear that wastage of progeny virions and the stochastic
aspects and kinetics of absorption to new hosts would reduce this advantage, but
the evolutionary pressure to curtail the infective cycle is also clear. It is equally
disadvantageous to undergo lysis too early, while virion assembly is still maximally
robust. The contrasting phenotypes of two S alleles, Sλa52g and Sλa52v, illustrate
this point perfectly at the extremes of holin phenotypes. Neither allele permits
plaque formation, for opposite reasons. The latter is an absolute lysis-defective
allele, whereas the former causes saltatory lysis at 20 min; thus, in neither case is
there a productive release of progeny virions. Therefore, the hole-formation and
clock functions of the holin are equally essential.
Similar considerations explain the suddenness of the lytic event at the level of
the individual cell. The ideal holin would exert no toxic effect on the physiology
of the host until the time to trigger lysis, thus maximizing the rate of progeny
production. Then, at the time that is programmed into the structure of holin
by natural selection (90), the ideal holin should trigger and release an excess of
endolysin, thereby ensuring prompt release of accumulated progeny. If the holin
permeabilized the membrane and then allowed slow release of endolysin, energy
metabolism would be compromised, and the progeny phage would in effect be
trapped in a rotting corpse until murein degradation is at last effected. This fact
would surely handicap the phage, compared with another with a more decisive
lysis mechanism.
The Mosaicism of Lysis Cassettes
All lambdoid phages have their four lysis genes arranged like those in λ, that is,
SRRzRz1. All of the auxiliary RzRz1 genes belong to the same ortholog family,
but there are three ortholog families of holins and two ortholog families of endolysins (Figure 2, see color insert). The lysis cassettes of λ, P22, 21, and PS3
represent four of the six possible combinations of these holin and endolysin gene
families, and one of the other two combinations, S21 with Rλ, has been shown to
be functional (15). This mosaicism indicates that holin function is nonspecific
for endolysin, a notion that has now been tested many times because complementation of the plating or lysis defect of λSam7 has become a diagnostic test
for identifying holins (see Functional Tests of Holins in Other Phages below).
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
804
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
Four different muralytic-enzyme activities are found in endolysins. There are
glycosidase (also muramidase or lysozyme) and transglycosylase activities that
attack glycosidic bonds and amidase and endopeptidase activities directed against
the peptide cross-links (95). Crystal structures for the T4 gene e glycosidase,
T7 gene 3.5 amidase, and the λ gene R transglycosylase reveal nothing in common
structurally other than that all three are monomeric, globular, and, at ∼18 kDa,
relatively small (23, 30, 92). Recently, mosaicism within endolysin genes has been
documented in a Staphylococcus phage, where different murein-binding domains
and two muralytic activities were found in endolysins ranging up to 70 kDa in
size (31, 49, 61). Despite this heterogeneity in endolysin activity and structure,
every holin-endolysin pair tested has proven to be lytically competent, even if the
sources of holin and endolysin genes are phages of hosts from different bacterial
kingdoms. The simplest interpretation is that there is no essential interaction between the holin and the endolysin. It is not known what other proteins may escape
by way of the hole, mainly because of the technical difficulties imposed by the
existence of the murein-linked outer membrane in E. coli, where almost all of the
functional studies of holins have been done. (For one report of the release of a
heterologous protein from a Gram-positive host, see Functional Tests of Holins in
Other Phages below.)
Sλ and S21 Represent the Two Major Classes of Holins
The Sλ and S21 holins share no sequence similarity but have some resemblance in
terms of distribution of polar, charged, and hydrophobic residues, and predicted
secondary structure (Figure 3A, see color insert). Both sequences have C-terminal
domains that are rich in basic residues, both have short polar amino-terminal sequences that precede a hydrophobic domain, and both have a short sequence that
is strongly predicted to be a β-turn domain between two putative helical transmembrane (TM) domains (15). However, Sλ has three domains that are predicted
to have no net charge on side chains and are long enough to span the membrane
(Figure 3A, see color insert), whereas S21 has only two. Many nonorthologous
holin sequences have been identified in recent years, to various levels of certainty;
Figure 3B lists a single representative of >30 apparently unrelated holin families.
Most holins can be assigned to one of two classes, based on the number of putative
TM domains: class I, like Sλ, with three potential TM domains [entries 1–7 and
14–20 (Figure 3B, see color insert)], and class II, like S21, with only two [entries
8–13 and 21–26 (Figure 3B, see color insert)]. [See also Table 1 and, on the
Annual Reviews home page (http://www.AnnualReviews.org), Tables 2 and 3.]
There are a number of interesting exceptions, including the t holin of phage T4
[Figure 3B (see color insert), entry 28]. In any case, the apparent distinction based
on inspection of the primary structure of class I and class II holins now has compelling experimental support (see below). Why there are so many different kinds
of holins (with only four different kinds of endolysins), why there are at least
two fundamentally different classes of holins, and what functional differences are
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
805
TABLE 1 Summary of holins
Hosta type
Classb
Gram negative
I
II
Unknown
7
6
2
15
20
4
5
8
0
Gram positive
I
II
Unknown
5
8
5
21
29
9
8
5
0
1
7
0
34
105
26
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Unconventional
Total
Familiesc
Sequences
Dual-start motifsd
a
Type of host of the phage, prophage, or cryptic prophage in which the holin gene is identified; unconventional,
the LrgA family [see Tables 2 and 3 (http://AnnualReviews.org)]
b
Class I and II, distinctions are based on whether the primary structure has 3 or 2 predicted transmembrane domains,
respectively. Unknown, the predicted number of transmembrane domains within one ortholog family is neither 2
nor 3
c
Number of holin sequence families
d
A sequence is recognized as having a dual-start motif if it has the M–Xn–M... structure at its N terminus, where
n is arbitrarily chosen to be <8, and at least one of the intervening residues is arginine (R) or lysine (K)
implied by the topological distinctions of classes I and II holins are all unanswered
questions.
Dual-Start Motifs: Setting the Clock
A remarkable feature of Sλ is that it encodes two proteins with opposing functions:
Sλ105, the actual lethal holin, from translational initiations at codon 3; and Sλ107, a
holin inhibitor, from translational starts at codon 1. This feature has been recently
reviewed (13) and thus is only briefly treated here. The inhibitory character is
derived from the extra positively charged residue (Lys2) in Sλ107. The proportion
of S105 to S107, normally ∼2:1, is determined by an RNA-stem loop structure at
the ribosome-binding site of Sλ. Mutations in this region change the proportion
and also the lysis time in the predicted direction [i.e. more S107 gives slower
onset of lysis, whereas more S105 accelerates the lysis time (11, 22)]. The S107
inhibitor, which normally cannot spontaneously trigger lysis, is converted to a
functional holin when the membrane is depolarized (11). However, an Sλ allele
that produces only Sλ105 still has precisely defined lysis timing, albeit earlier
than that of Sλ+, and this allele still can be triggered by the addition of CN−. The
dual-start motif thus appears to be a fine-tuning system for the scheduling of lysis.
Although it is attractive to think that this regulatory motif might be responsive to
growth conditions, it has not yet been demonstrated that the proportion of holin to
holin inhibitor is ever altered. Apparent dual-start motifs are also found in almost
all class II lambdoid holin sequences and in many other holins [Table 1; also see
columns 10 and 11 in Table 2 (http://www.AnnualReviews.org)]. Recently, it was
demonstrated that the dual start of S21 also serves the same general functions as in
P1: FUI
August 11, 2000
806
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
Sλ, resulting in the expression of a holin–holin-inhibitor pair (in this case, S2168
and S2171, respectively) again at an ∼2:1 ratio in favor of the holin.
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Probing the Structure of Sλ in the Membrane
Regardless of class, it was obvious early on that the most conserved feature of
holins is the highly charged C-terminal domain, which, with its multiple basic
residues, seemed likely to be disposed into the cytoplasm. Gene fusion and protease accessibility experiments confirmed this prediction (12). Based on this result, the membrane topology of Sλ mandates either a periplasmic location for the
N terminus (three TM domains) or a cytoplasmic location (two TM domains)
(Figure 4, see color insert). The existence of dual starts in both class I and class II
holin genes lent support to the latter model because one would expect a common
regulatory feature to operate from the same side of the bilayer. Recently, however,
a landmark experiment by Graschopf & Bläsi has provided strong evidence for
the 3-TM model (36). In this work the secretory signal sequence (SS) of the coat
protein of M13 was fused to the N terminus of the S λ105 sequence. It is significant
that the resultant SS-Sλ105 retained lytic function but acquired leader peptidase
(Lep) dependence. Moreover, lysis triggering was shown to be concomitant with
the accumulation of the processed chimera in the membrane. It is remarkable that
the isogenic SS-Sλ107 chimera also exhibited Lep-dependent lysis and processing.
These data strongly suggest that the N terminus of the λ holin must be externalized
during the pathway to lysis and that, accordingly, the functional form of the holin
has three TM domains with N out and C in (Figure 4, see color insert). Also, the fact
that the inhibitor character of Sλ107 was abolished by the signal sequence suggests
that it is the penetration of the membrane that is differentially blocked by the presence of the extra positive charge on the longer protein. The requirement that the
N terminus of Sλ107 must traverse the bilayer explains the effect of adding an
energy poison because it has already been shown that the energized membrane
opposes the export of the N termini of other membrane proteins with analogous
topology (24).
Direct probing of the membrane topology by extensive cysteine-scanning mutagenesis of Sλ and chemical modification studies with a thiol-specific reagent have
provided convincing support to the three-TM model (37). Moreover, these studies
have suggested that the highly charged C terminus is bound to the inner surface
of the membrane by Coulombic interactions. phoA, lacZ, and bla gene fusion
studies, the traditional methods for investigating membrane topology, have been
less useful with S because the small size of holins militates against the construction
of chimeras that lack the cognate C-terminal topology determinants. However, a
new “sandwich” fusion that carries a bifunctional phoA-lacZα reporter activity (4)
has been used on Sλ, confirming that the TM2-TM3 connector loop is periplasmic
(I-N Wang & R Young, manuscript in preparation).
Similar fusion gene analysis has shown that the C terminus of S21 is in the
cytosol (I-N Wang & R Young, manuscript in preparation), but otherwise S21 is
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
HOLINS AS PROTEIN CLOCKS
807
uncharacterized. However, its simple primary structure makes it inconceivable
that it can have anything but the 2-TM structure, with N in and C in (Figure 4,
see color insert). The S21 inhibitor is different in several respects from that
of Sλ (7). Expressed from a multicopy plasmid in trans to the holin, an allele of S21 that produces only the S2171 inhibitor retards the lysis time of the
wild-type holin gene by only ∼10 min, whereas, in analogous conditions, the
Sλ107 protein is able to abolish lysis entirely (A Gründling, DL Smith, U Bläsi,
R Young, manuscript in preparation). Moreover, cyanide does not trigger lysis
with an allele that elaborates only S2171, nor does replacement of the
N-terminal Lys residue of the inhibitor convert it to a holin effector. These differences are consistent with the idea that the class II inhibitor exerts its lysisretarding effect in a manner that is distinctly different from Sλ107. Thus, despite
the apparently unifying similarity in primary structure, the dual-start motifs in
class I and class II holins are fundamentally different regulatory adaptations, presumably arising from the extraordinary evolutionary pressure to optimize holin
function.
Genetic Analysis of Sλ Function and Regulation
Sλ and, potentially, all holins have especially facile genetics for several reasons:
(a) One can select for and against holin function by using phage or plasmid vectors;
(b) holin genes are small and easily sequenced; and (c) the holin and the endolysin
are the only essential genes that can be mutated without affecting virion structure. An extensive collection of Sλ mutants has been generated and mapped to its
321-base-pair reading frame (68, 69; Figure 3A, see color insert). Most S mutations that were selected for loss of lethality map to TM1, TM2, and the connecting
cytoplasmic loop, and TM2 appears to be especially crucial for Sλ function. For
example, three Ala residues, Ala-48, Ala-52, and Ala-55, which would all be on
the same face of TM2, give rise to lysis-defective alleles as a result of changes that
increase side-chain bulk. Two adjacent residues, Cys-51 and Ala-52, give rise to
early lysis phenotypes as a result of changes that decrease side-chain mass by 16
[Cys-51 Ser (37)] and 14 [Ala-52 Gly] Daltons, respectively. As noted above, in
the latter case, the timing is so early that production of progeny virions is abolished
(44). When transferred back to the phage context, the phenotypes ranged from absolute lysis defective to lysis delayed, which indicates that a particular holin variant
might not be lethal at the lower expression level of Sλ in the plasmid context but
still be capable of causing lysis at the normal levels of expression of the vegetative
phage. Presumably, some of these mutations change the critical concentration in
the membrane at which Sλ exerts its saltatory lethal function, whereas others may
cause reduced stability and thus retard lysis simply by requiring a longer time to
accumulate.
To form a membrane lesion that is capable of allowing escape of endolysin
molecules, Sλ or any holin must oligomerize, and thus it was not surprising to
find that both dominant and recessive alleles were present in the mutant collection.
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
808
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
What was surprising was that some alleles exhibit “antidominance ”(earlier defined
as early dominance) (Figure 3A, see color insert). This phenotype was discerned
in inductions of double lysogens. If a double lysogen with two S+prophages is
induced, lysis occurs earlier than if an S+, Sam double lysogen is induced, because
the lysis clock depends on the rate of accumulation of functional holin. The antidominant alleles are all lysis-defective mutations when tested alone, but in the
presence of the wild-type allele in double-lysogen tests, they accelerate lysis as
well as a second wild-type allele or better. Originally, it was suggested that this
phenotype reflects differential effects on the S105 holin and S107 holin-inhibitor
gene products, so that in the dominance tests, the effector form of the mutant might
titrate out the wild-type inhibitor and thus accelerate the onset of lysis. However,
the same behavior is seen in alleles that produce only Sλ105 (A Gründling &
R. Young, manuscript in preparation). We favor the idea that the antidominant mutant proteins are defective in the ability to undergo a critical conformational change that leads to hole formation. This change would have to be
propagated within an oligomeric aggregate of S proteins, and the antidominant
proteins, although unable to nucleate the change, would be capable of propagating it.
Genetics has clarified the role of the N-terminal and C-terminal domains of
the Sλ holin. Informed by the inhibitory properties of Sλ107, Steiner & Bläsi
(85) found that, in general, alterations of the N-terminal sequence of the holin
gene, resulting in more net anionic charge, result in faster lysis times. Rigorous interpretation of these results is difficult in the absence of information on
synthesis rates. Nevertheless these data are consistent with the proposed model
in which the N terminus must be externalized for holin function (Figure 4, see
color insert). In Sλ, the highly charged C-terminal domain is nonessential for hole
formation but instead constitutes a regulatory feature. The first evidence for this
came from a second-site mutation isolated as an intragenic suppressor to the earlylysis Ala-52 Gly mutation (44). The suppressor turned out to be a frame-shift
in the C terminus, which scrambled the sequence and increased the net positive
charge by +4 (12). Moreover, a nonsense mutant lacking all but one of the basic
residues at the normal C terminus of Sλ (Figure 4, see color insert) was found to
be fully lytic but, similarly to Ala-52 Gly, triggered lysis so early that burst size
and plaque-forming ability were compromised (12).Thus probably the N-terminal
and certainly the C-terminal hydrophilic domains appear to be regulatory decorations, in which the number of positive charges somehow modulate the lysis
clock.
As more physical and structural information becomes available, the phenotypic
richness of this extensive mutation collection should help narrow the choices of
models to explain holin function. Meanwhile, the mutants already provide reagents
that are invaluable for biochemical and physical analysis (see below). It seems
clear, then, that genetics should be the starting point for investigation of other
holins, and the excellent genetics and small size of holin genes should make
similar mutational analyses a rewarding undertaking.
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
809
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Oligomerization of the Holin and the Nature of the Hole
Treatment of membranes with oxidizing agents results in nearly quantitative conversion of Sλ into covalent dimers by disulfide bond formation at the single Cys
residue at position 51, near the middle of TM2. Cys-scanning mutagenesis and
studies of disulfide bond formation in vitro have provided strong evidence that
Sλ dimers are the building blocks of hole formation and that a dimerization interface extends along the face of the TM2 helix (A Gründling & R Young, manuscript
in preparation). Treatment of S-containing membranes with cross-linking reagents
revealed homomeric Sλ ladders up to n = 6 (97). Similar experiments have shown
that Sλa52v has a specific molecular defect; it has a normal propensity for dimer
formation but does not form higher oligomers in the membrane (A Gründling &
R Young, manuscript in preparation).
So what is the hole? All evidence indicates that it is a homo-oligomer of S,
but the state of oligomerization in the hole is unknown. Gel-filtration analysis
of octyl-glucoside extracts of Sλ from membranes suggests that 8-mer to 10-mer
aggregates of S survive detergent extraction (J Deaton, unpublished data). The
simplest Sλ hole, if it is an aqueous pore walled by all three TM helices of the
holin, would need at least an 8-mer to achieve the minimum 5-nm diameter required
by the folded R protein. As emphasized above, the structure of the hole is still
unknown, or even whether there is a single structure or a population of different
structures, related stoichiometrically or dynamically. Holins can also be thought of
as membrane-disrupting proteins, conjuring up the notion of a soaplike function.
However, the purified Sλ protein appears to be an unremarkable integral membrane
protein with three helical transmembrane domains, which is difficult to model
into micellar structures, especially at the low levels (∼103/cell) at which holeformation is triggered. Thus, in the absence of a definition of “disruption,” the
hole concept, poorly defined as it still is, seems preferable. The hole must be
quite large to accommodate native endolysins ≥70 kDa (61). Recent results with
gene fusions between R and lacZ revealed that endolysins with masses of >100 kDa
are efficiently released by Sλ (I-N Wang & R Young, manuscript in preparation).
This would require the inside diameter of the channel to be ≥10–12 nm, comparable
with the aqueous channels formed from the outside of cell membranes by the
streptolysin O/listerialysin O cytotoxins (19, 76).
Physical and Biochemical Analysis of the Holin
Although Sλ is lethal at 1–3 × 103 molecules/cell (22), attempts to overproduce
Sλ with a T7-based hyperexpression system have been surprisingly successful,
achieving 50-fold overproduction during the 10 min after induction, before complete death of the culture (81). Moreover, all of the protein was localized in the
membrane and was uniformly extractable with nonionic or mild zwitterionic detergents. This has made biochemical and physical analysis of Sλ feasible, and
the lesson is likely to be applicable to other holin systems. Although multiple
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
810
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
different positions within Sλ had to be tested first, ultimately it was found that an
oligohistidine tag could be inserted between codons 94 and 95 without unacceptable change of the holin phenotype. This permitted facile purification, yielding
1–3 mg of purified tagged Sλ105/liter of induced culture (83).
The CD spectrum of the purified protein was found to be detergent dependent. In
octyl glucoside, the holin was unstably soluble and had ∼40 residues in α-helical
conformation, consistent with two TM domains (80). However, in a relatively
new, mild zwitterionic detergent, Empigen-BB, Sλ is stably soluble and has ∼60
residues in α-helix, consistent with three TM domains. Neither the Ala-52 Val nor
the Ala-55 Thr lysis-defective proteins showed any defect in helical content, even
when the latter, which has a temperature sensitive lysis defect (68), was examined at
the permissive and nonpermissive temperatures (J Deaton & R Young, manuscript
in preparation).
The availability of purified protein has permitted the first steps in biochemical
analysis of holins. In one assay, the holin diluted out of a chaotrope was shown
to cause release of a fluorescent dye from liposomes (82). That this in vitro hole
formation reflects the same properties as those required for hole formation in vivo
is shown by the fact that the Ala-52 Val variant is defective in this dye release, and
the Ala-55 Thr variant is defective at 42◦ C, but not 30◦ C (J Deaton, unpublished
data). These data constitute the first unambiguous proof that the Sλ holin, on its
own, is capable of permeabilizing membrane bilayers. Further analysis of the
stoichiometry of these liposome holes and the kinetics of their formation should
provide mechanistic insights into holin function. Other biochemical approaches
are possible with holins. A powerful method for studying hole-forming proteins
is to measure the electrical currents across a “black-lipid membrane” (46). Recently, relatively stable single-channel currents have been obtained when purified
holin in detergent was diluted into the cis chamber of a black-lipid membrane
apparatus. No channels were obtained when the Ala-52 Val variant was added at
the same concentration (Z Yong, O Braha, H Bayley, personal communication).
This promising methodology is probably the best approach for determining the
size of the holin-mediated lesion. Ultimately, however, despite the powerful genetics and physiology of holins and the promising new biochemical approaches,
progress in understanding holin function at the molecular level will remain slow
until significant structural information is available. Recently, protein crystals of the
purified Ala-52 Val variant have been obtained (J Deaton & J Sacchettini, personal
communication), perhaps making a crystal structure of the solubilized Sλ holin a
realistic goal. Another promising approach is NMR, especially in view of the recent
landmark success with the detergent-solubilized dimer of the glycophorin
TM domain (55).
What Makes the Lysis Clock Tick?
We now know several parameters that set the Sλ lysis clock. As discussed above,
the inhibitor-effector ratio and the charges at the N and C termini influence lysis
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
HOLINS AS PROTEIN CLOCKS
811
time, and this is also true for the rate of expression of the holin gene, as has been
demonstrated by cloning the lysis cassette into various transcriptional contexts
(12, 22, 32). However, the major determinant is still in the hydrophobic core,
where single missense changes can alter the spontaneous triggering time from
20 min (after only 12 min of late gene expression) to 120 min (44, 69). Even
without knowledge of the structure of the hole, we know that holins must exist in
at least two conformations—a chronic or prehole state, in which the holin protein
accumulates in the membrane, and an acute state, the hole itself. It is also clear that
the conversion from prehole to hole is opposed by the energized membrane, which
makes holin function inherently saltatory, because the first hole that is formed will
collapse the membrane potential and thereby trigger all of the remaining holes to
be formed. The question thus reduces to what determines when the first hole is
formed.
The simplest timing model presumes nothing other than the idea that, as the
holin accumulates in the membrane, it forms higher oligomeric states. Genetic
and biochemical data cited above indicate that, for Sλ, the incremental unit is
the dimer. We can postulate that dimers form and then tetramers, hexamers, and
higher n-mers, until the crucial N-mer is formed, where N is the size required to
make the hole. In this simplest timing mechanism, the N-mer will, within a certain
average lifetime, undergo a spontaneous, concerted conformational change that
results in hole formation. A prediction of this N-mer model is that, if different
holins are accumulating within the membrane and populating independent N-mer
pools, the lysis time will be determined by whichever of the two holins populates
the critical N-mer state first. That is, the timing of different, noninteracting holins
should be essentially independent, as long as the accumulation of the two holins
is independent. This prediction was tested by using Sλ and T4 t, which are about
as unlike as two holins can be in terms of primary structure, and the result was
clear. Expression of the two holin genes simultaneously resulted in much faster
onset of lysis than did expression of either one independently (E Ramanculov &
R Young, submitted for publication). Assuming that the two holins do not interact,
this result effectively rules out the simple N-mer population model.
Is the assumption of noninteraction valid? After all, the holins accumulate in the
two-dimensional arena of the cytoplasmic membrane, and there are no useful rules
yet for defining what determines or limits specific versus nonspecific interactions
in the bilayer. (In fact, holins may be an ideal system for determining such rules!)
One experiment has been done that addresses this subsidiary question. When the Sλ
and S21 holins were tested for cross-sensitivity to the inhibitor form of each holin,
no lysis delay was detectable in either case (M Barenboim & R Young, unpublished
data). This result suggests that either the holins do not interact or the interaction
has no effect on the timing of holin function. In either case, this result supports
the validity of the assumption underlying the t/Sλ experiment and reinforces the
conclusion that timing involves more than formation of a critical N-mer.
What then is the next simplest model for holin timing? The t/S λ experiment
suggests that, in the prehole state, both holins are doing something to the cell and
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
812
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
that the effects are additive. Holins are clearly not enzymes, and the simplest
notion would be that the effect on the cell has to be some effect on the integrity
of the cytoplasmic membrane. Unless we are misled by the effects of detergent,
the business part of Sλ appears to be three α-helical TM domains with short interhelical linkers. This would seem to exclude the formation of mysterious holes
like those associated with the incomplete pores of the cholesterol-dependent cytolysins, which perforate the membrane with β-strands (8, 77). All known cellular
ion and proton channels in the cytoplasmic membrane are helical bundles. Thus a
simple model is that the holins, in one or more oligomeric states, cause a leakage
of protons and that, as the holin population builds up, ultimately the ability of
the cell to resist this drainage and maintain the membrane potential is exhausted.
This would lead to a precipitate collapse in the membrane potential and triggering
of all the holins into the hole conformation. This scheme is particularly attractive because it puts the energetic state of the cell into the equation for timing.
Ultimately, the evolutionary fitness of the holin is thus determined by how well
the critical timing achieved by this leakage catastrophe mechanism matches the
optimum time for virion accumulation. Note that it would be critical in this kind
of mechanism for the effects of the holin on the energy metabolism of the cell
not to be manifested until after the rate of virion assembly in the cytoplasm is no
longer maximal. Moreover, remembering that holin accumulation does not cause
a gradual slowing of the flagellar motor, it seems likely that the host compensates
for the leakage and maintains the membrane potential for most of the time leading
up to the lytic catastrophe.
The leakage catastrophe model makes a number of predictions, chiefly that the
prehole state leaks protons. Ideally, a mutant holin blocked in the final prehole-tohole conversion could be purified and tested in an in vitro system (i.e. black-lipid
membranes or liposomes). Clearly, then, genetic analysis of holin function is
still an essential component for further understanding of the timing mechanism,
especially considering that only Sλ has been studied to any serious degree. There is
a need for intensive genetic study of other holins among the >100 now identified.
The phage or prophage of origin is apparently irrelevant, because either one can
be imported into the λ system as a substitute for Sλ, where superb genetic handles
are provided by λ and its well-characterized host. The opportunity is especially
great for the simple class II holins, some of which are so small that they could
realistically be synthesized in vitro.
HOLINS IN NONLAMBDOID SYSTEMS
Extragenic Holin Regulation in Coliphages
In contrast to the intragenic holin regulation of lambdoid phages, extragenic regulation is featured in three well-studied coliphages: P1, P2, and T4. In P1, separate
loci define the lysis genes (75). Gene 17 encodes the endolysin, a muramidase,
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
HOLINS AS PROTEIN CLOCKS
813
and the distant lydAB locus encodes a class I holin, LydA, and LydB (formerly
gene 2). Nonsense mutations in lydB cause disastrously early lysis. In P2 the holin
(Y) and endolysin (K) genes are adjacent and just upstream of lysA, where amber
mutations cause lysis somewhat earlier than in the wild type (99). In both cases,
the accelerated lysis phenotype of the nonsense mutants was attributed to the loss
of a holin antagonist. The dual lysine residues at the N terminus of LysA, which
has several predicted TM domains, is reminiscent of the lambdoid antiholin. In
contrast, P1 LydB does not appear to be a membrane protein, based on hydropathy analysis and distribution of charges. As a caveat, however, it remains to be
demonstrated that lysA and/or lydB can inhibit the holin in trans without affecting
holin gene expression. Also, it is interesting that, among the four other P2-related
phages and cryptic prophages including the well-studied phage 186, the structure
of the lysis gene region is always holin-endolysin-Rz/Rz1, so the lysA lysis inhibitor gene appears to have been inserted between the holin and endolysin genes
soon after P2 diverged (67). That such closely related phages as P2 and 186 could
have such a fundamentally different mode of timing the lytic cycle is consistent
with the idea that P2 and 186 have evolved to fill primarily lytic and lysogenic
niches, respectively (67).
The oldest and most advanced phage genetics work is that with T4, and the
earliest genetic analyses of phage biology centered on the study of lysis inhibition
(LIN), a complex and venerable phenomenon that is reviewed in detail elsewhere
(93). The LIN state, in which lysis is blocked and phage continue to accumulate
in the infected cell for hours, results from secondary infections by T4, presumably
to avoid further dispersal of progeny into a host-deficient environment. Somehow, LIN is established by a signal transduction process in which the attempted
injection of the superinfecting phage DNA is detected and relayed to the lysis
system by products of the famous phage r genes (41). The LIN state eventually
collapses if new superinfection events stop occurring, or it can be subverted by
the addition of CN−, which indicates that LIN depends on inhibition of the holin.
LIN-defective r (rapid-lysis) mutants are easy to isolate as large, distinct plaque
formers, compared with the small, indistinct plaques of the LIN+ wild-type phage,
and LIN− mutations were mapped to the genes rI, rIIA, rIIB, rIII, and rV (3).
The endolysin gene, e, of T4 is the canonical lysozyme (86). In contrast, the
holin gene t, although phenotypically identified long ago (45), was revealed to
have a decidedly noncanonical sequence that is >200 residues long, with only one
identifiable TM domain and lacking any of the hallmarks of other holins [58, 72;
Figure 3B (see color insert), entry 28]. Doubts about t as a bona fide holin were
alleviated when a plasmid clone of t was shown to restore lysis to induced λSam7
lysogens. Moreover, the t protein was identified as a membrane protein species
that could be converted into oligomers to about n = 4 by chemical cross-linking
(54). Recently, the sequencing of some ancient mutants revealed that rV is allelic
to t, which demonstrates that the recipient of the r-mediated signal ultimately is
the t holin (29). Abedon (1, 2) and Paddison et al (65) have formulated a detailed
model for LIN. According to this model, the key role is played by the rI protein,
P1: FUI
August 11, 2000
814
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
which is thought to be located in the periplasm and to be activated by the injection
of the DNA of superinfecting T4. The activated rI is proposed to interact with the
holin t to inhibit its hole-forming ability. Recently, t has been substituted for S in
the λ lysis cassette and shown to support very saltatory lysis, as well as to undergo
LIN by superinfection with T4 or, more significantly, by induced expression of a
cloned rI gene (E Ramanculov & R Young, manuscript in preparation). These developments suggest that T4 LIN, one of the most venerable problems of molecular
genetics, may at last be amenable to molecular analysis, and they illustrate again
the evolutionary forces mandating the control of holin function.
Functional Tests of Holins in Other Phages
Since the last review of holins, many new holin sequences have been identified
but not all have actually been characterized for their lysis phenotype. In those
cases in which experimental data were reported [see Garcia et al (31)], usually
the most convincing functional tests were done in E. coli [column 17 in Table 2
(http://AnnualReviews.org)], because suitable tightly regulated, inducible plasmid
vector systems are not often available in other hosts (10, 21, 40, 50, 63, 64, 100). In
these cases, the usual result is inducible lethality when the holin is expressed alone
and lysis when the holin is expressed with its cognate endolysin or with a different
endolysin. The tandem lysis genes cph1 and cpl1 of Streptococcus phage Cp-1
were also tested in the cognate host (57). Relying on an apparent growth phase
induction of the lysis genes cloned into a low-level constitutive shuttle plasmid,
Martin et al (57) found the holin-endolysin gene pair to cause efficient lysis in
the Gram-positive host. They also reported that isogenic constructs carrying only
the cph1 holin gene did not cause lethality, but expression data that would allow
interpretation of this observation were not provided. In any case, this work also
addressed the question of whether other proteins can escape through the holinpermeabilized membrane. Pneumolysin, a soluble 52-kDa hemolytic toxin (59),
is released from the Streptococcus cells carrying the cph1 plasmid without the
endolysin gene. However, no assessment was made of what fraction of the total
pneumolysin is released, and so it is not clear whether the released protein comes
from a minority of the host cells undergoing slow “rotting” after the lethal effect of
the holin. Obviously, if a sufficiently well-controlled inducible-expression system
could be established in a Gram-positive host, the absence of the outer membrane
might make it simpler to address the question of what else besides the endolysin
escapes via the hole.
Diaz et al (27) described lytic induction in E. coli and Pseudomonas putida with
plasmid clones of the tandem holin-endolysin genes ejh-ejl of the pneumococcal
phage EJ-1. A surprising finding was that lawns of an E. coli plating strain carrying
the inducible holin plasmid could be grown at limiting levels of inducer and used
in plating tests with λSam7, in which nearly unit-plating efficiency was observed.
Use of a partially suppressing supE host for the plating tests burdens the interpretation of these unexpected results. Moreover, the authors’ explanation that, in the
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
HOLINS AS PROTEIN CLOCKS
815
infected cells, the λ R endolysin contributes to the lethality of the constitutively
expressed subcritical levels of the Ejh holin seems unlikely because R does not
contribute to Sλ lethality (33). Another interesting observation by these authors for
Ejh and by Martin et al (57) for Cph is that the LytA amidase, the major autolysin
of Streptococcus species, can serve as the endolysin for these heterologous holins.
A report that LytA actually accumulates in an inactive state on the outside face
of the cytoplasmic membrane of E. coli (26) led Diaz et al (27) to the interesting
proposal that holins can activate LytA directly by an unspecified membrane disturbance. However, it is unclear how LytA, which has no signal sequence or other
distinguishing feature than numerous basic residues, could be localized in such a
way in E. coli. Moreover, the data suggesting this unusual localization are based
on an unprecedented pattern of subcellular fractionation in an osmotic-shock protocol, which may not be unambiguously applicable to murein-binding proteins.
Nevertheless, testing the holin-endolysin system of the lactococcal phage φadh in
E. coli, Henrich et al (40) also found that the endolysin (in this case, a muramidase
ortholog) was associated weakly with the total membrane fraction. Thus, it is unresolved whether the unexpected localization of these muralytic activities reflects
an artifact of endolysin synthesis in a heterologous (Gram-negative) host or a clue
to at least one pathway for holin-mediated activation of murein degradation.
Biotechnology and Holin/Endolysin Lysis
Biotechnological applications have played a major role in stimulating the study of
holin/endolysin functions in Gram-positive bacteria. These efforts have led to the
identification or cloning of lysis genes for a number of bacteriophages involved in
food production; for example, Streptococcus thermophilus (18, 79), Lactococcus
lactis (25, 73), and Oenococcus species (34). Dairy product ripening involves the
late release of enzymes from fermentative bacteria. To this end, the holin/endolysin
gene pairs from phages φUS3 and rlt have been cloned under inducible promoters
in plasmids of L. lactis and shown to cause efficient lysis not only in liquid (74)
but in model cheeses (25).
SURVEY OF HOLINS
Preliminary Identification of Holins
Table 2 (see http://www.AnnualReviews.org) contains our current compilation
of putative holin genes. Basically, sequences that were annotated as holins or
holinlike lysis proteins were collected by key word searches. At a minimum, all
such submissions found to date have at least one predicted TM domain, and many
putative holin genes are found adjacent to endolysin genes. Whenever a resulting
open reading frame (ORF) shows significant sequence identity to the known endolysin sequence, it is tentatively considered to encode an endolysin function. Our
analysis suggests that every completed DNA sequence of a double-stranded DNA
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
816
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
eubacterial phage has been found to contain one or more of the four known types
of endolysins. Frequently, an adjacent or nearby ORF, usually upstream, can be
assigned as a holin gene if it is small (<150 residues), has one or more clearly discernible TM domains with no net predicted charge, and has a hydrophilic, highly
charged C-terminal sequence. Obviously, there is bias in this search strategy. For
example, a potential holin gene that is not linked to an endolysin cistron, as in P1
and T7, would certainly be missed. Homology searches with known holins have
been useful, although the number of orthologous groups already known suggests
that this is a limited approach. Holin sequences occur in phages and prophages
but also in bacterial chromosomes, either in cryptic prophages or in isolated loci.
Clearly, the greatest confidence in holin identifications is placed on those with
an experimentally determined lytic phenotype. We have the least confidence in
sequences that appeared only in the bacterial chromosome and have not been
phenotypically characterized (e.g. the LrgA family). By combining both search
strategies, we have identified 105 unique and complete holin sequences [>110,
including partial sequences, in various databases (as of November, 1999; see
Table 2 at http://www.AnnualReviews.org)]. Based on the sequence homology
determined by application of the gapped BLAST and PSI-BLAST algorithms
(5, 6), these holin sequences are classified into 34 families [Table 1; see also column 6 of Table 2 (http://www.AnnualReviews.org)], of which one representative
of each is shown in Figure 3B.
Distribution of Holins
Of these 105 identified holin genes, the majority are from bacteriophages
or prophages [82/105 (78%)], some are from cryptic prophages [5/105 (5%)],
and the remaining are found in bacterial chromosomes [18/105 (17%)], possibly
remnants of cryptic prophages. Of the families, 19 have more than one sequence
member, whereas the rest have only one sequence so far. The largest of these holin
ortholog families is that of 80α [Table 3 (http://www.AnnualReviews.org)], which
contains 12 orthologs. These holin families are distributed almost equally between
Gram-positive and Gram-negative bacteria (Table 1). There is not a confirmed
holin from an archaeal phage, although several lytic phages have been identified,
and the endolysin of the phage psi-M2 Methanobacterium thermoautotrophicum
was shown to be a pseudomurein endoisopeptidase encoded by peiP (66). Within
the eubacteria, all of the holins identified so far are either from phages or chromosomes of the gamma subdivision of proteobacteria or the low-G+C Gram-positive
bacteria, except for the holins from φC31, a Streptomyces phage, and the mycobacteriophages D29 and L5. Clearly, the distribution of holin sequences is not even
and reflects both the historic legacy and economic and medical importance of the
hosts.
As a criterion for determining whether a holin sequence belongs to class I or
class II, the TM hidden Markov model was used (84; for exceptions, see legend
to Figure 3). In general, this algorithm determined that holins within the same
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
817
ortholog family had the same topology. There are almost equal numbers of families
in the two different classes of holins (Table 1). In addition, there are several
examples of holins for which the TM hidden Markov model algorithm predicts
only one TM domain, including T4 t.
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
Prevalence of the Dual-Start Motif
The dual-start motif, leading to the synthesis of the holin and holin inhibitor,
is found in all lambdoid holins, regardless of class, except for two interesting
exceptions: the S21 orthologs of the lambdoid prophages, which carry the determinants for Shiga-like toxin secretion. In these prophages, the toxin gene
complex is inserted between p0 R and S, and there is some evidence that efficient release of the toxin may involved the lytic functions of the prophages (62).
Many other holin sequences have two N-terminal Met residues separated by at
least one positive charge (see Table 1; see also Table 2, columns 10 and 11
(http://www.AnnualReviews.org)]. However, in only one case outside the lambdoid holins is there some experimental evidence for a dual-start, the class I holin
gene 14 of the Bacillus subtilis virulent phage φ29. Gene 14 begins with MetLys-Met. . ., and both start codons are used in vitro and in vivo (87). Ablation of
codon 1 accelerates lysis in a plasmid clone, which indicates that the short form,
gp14129, is the effector and the long form, gp14131, is the inhibitor, although
inhibition has not yet been demonstrated in trans. It seems strange that the nearly
identical holin sequence of the closely related phage PZA lacks a basic residue
between the two start codons.
Organization of Holin and Endolysin Genes
For lambdoid phages, the essential lysis genes are clustered and transcribed in the
order holin-endolysin. This is also true for the majority of the holin/endolysin
pairs presented here (86%), but it is certainly not required. In many cases, the
genes are not linked (i.e. T4), and, in T7, the endolysin gene is also separated
in time, being expressed as an early gene (39). No holin-like sequence precedes
the Mu endolysin, and functional analysis will be required to see if any of several
downstream reading frames encode a holin (GenBank accession AF083977). An
unusual holin-endolysin arrangement is found in many phages of S. thermophilus.
Sequence and Southern blot analyses reveal two holin-like sequences, a class I
and a class II, respectively, immediately upstream of the endolysin gene (18, 79).
High-level expression of either of these putative holin genes in E. coli is lethal and
causes some leakage of the cytoplasmic marker enzyme isocitrate dehydrogenase.
An intriguing notion is that the two holins respond to different environmental
conditions, but the simplest interpretation is that one of the two proteins is an
antiholin, as is thought for the P2 LysA protein, and that it is lethal under these
conditions because of the level of overexpression (79). An even more confusing
picture has emerged from the study of the lysis genes of the PBSX, an inducible
prophage of B. subtilis, which appears to have a canonical type II holin gene, xhlB,
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
818
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
adjacent to an amidase-type endolysin gene, xlyA; however, deletion analysis in
B. subtilis indicates that XhlB requires a protein, XhlA, encoded upstream, to
allow endolysin release and also that there is at least one other endolysin gene,
unlinked to xlyA, in the PBSX chromosome (48, 53). Probably the most unusual
holin-endolysin gene structure has been described in the Staphylococcus phage
187. Loessner et al (49) have presented evidence that the holin is encoded in a
short 57-codon reading frame (hol187) that is embedded in the +1 register within
the N-terminal region of the endolysin gene. Moreover, these authors report that
hol187 can complement the Sam7 null allele for plaque formation. Although again
the use of a supE strain muddies the interpretation, nevertheless it is tempting to
think that this might represent a holin gene in the process of evolving from a
fortuitous membrane-toxic reading frame. Another unexpected result came from
sequence analysis of the Bacillus cereus phage TP21, in which Loessner et al (51)
suggest that the endolysin gene has a signal sequence. In support of this heretical
idea, recent studies with the endolysin of oenoccocal phage fOg44 found that its
N terminus engages the SecA-dependent secretory system and is processed in
E. coli (C São-José, R Parreira, G Vieira, & M Santos, submitted for publication).
This is very puzzling, especially, as the authors note, considering the close similarity of the putative signal sequence of the fOg44 endolysin with the N termini of
the endolysins of the lactococcal phages Tuc2009 and φLC3, which have canonical holins. Considering the apparent universality of the holin-endolysin story, the
idea that endolysins can evolve the ability to accomplish properly timed lysis is
intriguing and worth serious study.
EVOLUTIONARY CONSIDERATIONS AND APPROACHES
Mathematical modeling shows that, under any combinations of host quantity and
quality, there exists an optimal lysis time that maximizes the fitness (90). That
is, given enough time and the persistence of conditions, only one clone of the
phage will eventually dominate any local scene. It seems clear that the decision
of “when to lyse” is constantly under tremendous evolutionary pressure to track
the optimal time. From the example of λ, we know that the timing of lysis is
entirely dependent on the holin activity. As discussed above, the lysis time can be
widely varied, according to the quantity and the structure of the holin. The logical
hypothesis is that the holin is the target of evolutionary tinkering to achieve the
optimal lysis time. There are two ways to test this idea: the first is to demonstrate
the potential for selection among various holin alleles, and the second is to show
the presence of a population genetic signature left on the sequences of the holin
gene.
The simplest and the most direct way to show that different holin alleles result
in different fitness is to perform competition experiments with isogenic phages
that differ only in their holin genes. In this way, it can be demonstrated that, under
a specific and stable environment, differences in lysis time (caused by different
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
HOLINS AS PROTEIN CLOCKS
819
holin alleles) will result in differences in fitness. This approach can be extended
by a long-term evolutionary experiment. One would predict that the phage clone
that emerges after a long series of transfer experiments would possess a mutated
holin gene that lyses at the optimal time.
Besides showing the potential for selection in the laboratory, a “molecular
population genetic” analysis of holin alleles collected in the field could also bear
witness to selection. If the evolution of phage holin is shaped by the dynamics
that we stated in the laboratory condition, then two patterns should be evident.
The first is that there is only one or there are very few holin alleles in any given
geographic area because most are derived from the one allele that outcompeted
the others. Consequently, the sequence variation in this area would be low owing
to purifying selection. Under such circumstances, most of the variation that we
observe would be in the synonymous sites of the codons. The second pattern is
that each geographic area may have its own collection of holin sequences; that is,
different locales usually have different quantity and quality of host populations,
thus resulted in selection of different lysis times and, hence, different holin alleles.
Therefore, by comparing holin sequences from different locales, we would expect
to see that the sequence variation is concentrated at the nonsynonymous sites of
the codons, from which we could conclude that positive selection is driving the
sequence variation.
PERSPECTIVE AND SUMMARY
The explosion of prokaryotic and phage genomic information has revealed an
astonishing variety of putative holin sequences, apparently with many variations
in terms of structure, function, and regulation. Thus, although the concept of
the holin is not yet a decade old, it is already manifest in the largest collection of
nonorthologous functional homologs known in biology. Although genetic analysis
of a few holins is advanced, and many different holin genes have been tested for
lytic function, biochemical analysis has been seriously undertaken in only one
case, and structural studies have not yet borne fruit. We hope that this review will
stimulate interest in these small, lethal membrane proteins that define the clock of
the phage vegetative cycle.
ACKNOWLEDGMENTS
We thank the other members of the Young laboratory, past and present, for their
support and encouragement and members of the entire phage biology community
for their traditional mode of open and interactive science. The clerical support of
Sharyll Pressley is gratefully acknowledged. This work was supported by PHS
grant GM27099 and funds from the Robert A. Welch Foundation and the Texas
Agricultural Experiment Station.
P1: FUI
August 11, 2000
820
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
Visit the Annual Reviews home page at www.AnnualReviews.org
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
LITERATURE CITED
1. Abedon ST. 1989. Selection for bacteriophage latent period length by bacterial density: a theoretical examination. Microbiol.
Ecol. 18:79–88
2. Abedon ST. 1990. Selection for lysis inhibition in bacteriophage. J. Theor. Biol.
146:501–11
3. Abedon ST. 1994. Lysis and the interaction
between free phages and infected cells. In
Molecular Biology of Bacteriophage T4,
ed. JD Karam, JW Drake, KN Kreuzer, G
Mosig, DH Hall, et al. pp. 397–405. Washington DC: Am. Soc. Microbiol.
4. Alexeyev MF, Winkler HH. 1999. Membrane topology of the Rickettsia prowazekii
ATP/ADP translocase revealed by novel
dual pho-lac reporters. J. Mol. Biol.
285:1503–13
5. Altschul SF, Koonin EV. 1998. Iterated
profile searches with PSI-BLAST: a tool
for discovery in protein databases. Trends
Biochem. Sci. 23:444–47
6. Altschul SF, Madden TL, Schaffer AA,
Zhang J, Zhang Z, et al. 1997. Gapped
BLAST and PSI-BLAST: a new generation of protein database search programs.
Nucleic Acids Res. 25:3389–402
7. Barenboim M, Chang C-Y, dib Hajj F,
Young R. 1999. Characterization of the
dual start motif of a class II holin gene.
Mol. Microbiol. 32:715–27
8. Bayley H. 1997. Toxin structure: part of a
hole? Curr. Biol. 7:r763–65
8a. Berhardt TG, Roo WD, Young R. 2000.
Genetic evidence that the bacteriophage
φX174 lysis protein inhibits cell wall
synthesis. Proc. Natl. Acad. Sci. USA
97:4293–302
9. Bienkowska-Szewczyk K, Lipinska B,
Taylor A. 1981. The R gene product of bacteriophage λ is the murein transglycosylase. Mol. Gen. Genet. 184:111–14
10. Birkeland N. 1994. Cloning, molecular
11.
12.
13.
14.
15.
16.
17.
18.
19.
characterization, and expression of the
genes encoding the lytic functions of lactococcal bacteriophage φLC3: a dual lysis
system of modular design. Can. J. Microbiol. 40:658–65
Bläsi U, Chang C-Y, Zagotta MT, Nam
K, Young R. 1990. The lethal λ S gene encodes its own inhibitor. EMBO J. 9:981–
89
Bläsi U, Fraisl P, Chang C-Y, Zhang N,
Young R. 1999. The C-terminal sequence
of the lambda holin constitutes a cytoplasmic regulatory domain. J. Bacteriol.
181:2922–29
Bläsi U, Young R. 1996. Two beginnings
for a single purpose: the dual-start holins
in the regulation of phage lysis. Mol. Microbiol. 21:675–82
Bon J, Mani N, Jayaswal RK. 1997. Molecular analysis of lytic genes of bacteriophage 80α of Staphylococcus aureus. Can.
J. Microbiol. 43:612–16
Bonovich MT, Young R. 1991. Dual start
motif in two lambdoid S genes unrelated to
λ S. J. Bacteriol. 173:2897–905
Boyce JD, Davidson BE, Hillier AJ. 1995.
Sequence analysis of the Lactococcus lactis temperate bacteriophage BK5-T and
demonstration that the phage DNA has
cohesive ends. Appl. Environ. Microbiol.
61:4089–98
Brunskill EW, Bayles KW.
1996.
Identification of LytSR-regulated genes
from Staphylococcus aureus. J. Bacteriol.
178:5810–12
Bruttin A, Desiere F, Lucchini S, Foley S,
Brüssow H. 1997. Characterization of the
lysogeny DNA module from the temperate
Streptococcus thermophilus bacteriophage
φSfi21. Virology 233:136–48
Buckingham L, Duncan JL. 1983. Approximate dimensions of membrane lesions produced by streptolysin S and
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
20.
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
21.
22.
23.
24.
25.
26.
27.
28.
streptolysin O. Biochim. Biophys. Acta
729:115–22
Casjens SR, Eppler K, Parr R, Poteete
AR. 1989. Nucleotide sequence of the bacteriophage P22 gene 19 to 3 region: identification of a new gene required for lysis.
Virology 171(2):588–98
Chandry PS, Moore SC, Boyce JD, Davidson BE, Hillier AJ. 1997. Analysis of the
DNA sequence, gene expression, origin of
replication and modular structure of the
Lactococcus lactis lytic bacteriophage sk1.
Mol. Microbiol. 26:49–64
Chang C-Y, Nam K, Young R. 1995. S
gene expression and the timing of lysis by
bacteriophage λ. J. Bacteriol. 177:3283–94
Cheng X, Zhang X, Pflugrath JW, Studier
FW. 1994. The structure of bacteriophage
T7 lysozyme, a zinc amidase and an inhibitor of T7 RNA polymerase. Proc. Natl.
Acad. Sci. USA 91:4034–38
Dalbey RE, Kuhn A, von Heijne G.
1995. Directionality in protein translocation across membranes: the N-tail phenomenon. Trends Cell Biol. 5:380–83
de Ruyter PG, Kuipers OP, Meijer WC,
de Vos WM. 1997. Food-grade controlled
lysis of Lactococcus lactis for accelerated
cheese ripening. Nat. Biotechnol. 15:976–
79
Diaz E, Garcia E, Ascaso C, Mendez E,
Lopez R, Garcia JL. 1989. Subcellular localization of the major pneumococcal autolysin: a peculiar mechanism of secretion in Escherichia coli. J. Biol. Chem.
264:1238–44
Diaz E, Munthali M, Lünsdorf H, Höltje
J-V, Timmis KN. 1996. The two-step lysis system of pneumococcal bacteriophage
EJ-1 is functional in Gram-negative bacteria: triggering of the major pneumococcal
autolysin in Escherichia coli. Mol. Microbiol. 19:667–81
Doermann AH. 1952. The intracellular growth of bacteriophages. I. Liberation
of intracellular bacteriophage T4 by premature lysis with another phage or with
821
cyanide. J. Gen. Physiol. 35:645–56
29. Dressman HK, Drake JW. 1999. Lysis and
lysis inhibition in bacteriophage T4: rV
mutations reside in the holin t gene. J. Bacteriol. 181:4391–96
30. Evrard C, Fastrez J, Declercq JP. 1998.
Crystal structure of the lysozyme from bacteriophage lambda and its relationship with
V and C-type lysozymes. J. Mol. Biol.
276:151–64
31. Garcia P, Martin AC, Lopez R. 1997.
Bacteriophages of Streptococcus pneumoniae: a molecular approach. Microb. Drug
Resist. 3:165–76
32. Garrett J, Fusselman R, Hise J, Chiou
L, Smith-Grillo D, et al. 1981. Cell lysis
by induction of cloned lambda lysis genes.
Mol. Gen. Genet. 182:326–31
33. Garrett J, Young R. 1982. Lethal action
of bacteriophage lambda S gene. J. Virol.
44:886–92
34. Gindreau E, Lonvaud-Funel A. 1999.
Molecular analysis of the region encoding the lytic system from Oenococcus oeni
temperate bacteriophage φ10MC. FEMS
Microbiol. Lett. 171:231–38
35. Gottlieb P, Metzger S, Romantschuk M,
Carton J, Strassman J, et al. 1988. Nucleotide sequence of the middle dsRNA
segment of bacteriophage φ6: placement
of the genes of membrane-associated proteins. Virology 163:183–90
36. Graschopf A, Bläsi U. 1999. Molecular
function of the dual-start motif in the λ S
holin. Mol. Microbiol 33:569–82
37. Gründling A, Bläsi U, Young R. 2000. Biochemical and genetic evidence for three
transmembrane domains in the class I
holin, λ S. J. Biol. Chem. 275:769–76
38. Hatfull GF, Sarkis GJ. 1993. DNA sequence, structure and gene expression of
mycobacteriophage L5: a phage system
for mycobacterial genetics. Mol. Microbiol. 7:395–405
39. Hausmann R. 1988. The T7 group. In The
Bacteriophages, ed. R Calendar, pp. 259–
89. New York: Plenum
P1: FUI
August 11, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
822
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
40. Henrich B, Binishofer B, Bläsi U. 1995.
Primary structure and functional analysis of the lysis genes of Lactobacillus
gasseri bacteriophage φadh. J. Bacteriol.
177:723–32
41. Hershey AD. 1946. Mutation of bacteriophage with respect to type of plaque. Genetics 31:620–40
42. Howe CW, Smith MC. 1996. Gene expression in the cos region of the Streptomyces
temperate actinophage φC31. Microbiology 142:1357–67
43. Jin S, Chen Y-C, Christie GE, Benedik
MJ. 1996. Regulation of the Serratia marcescens extracellular nuclease: positive
control by a homolog of P2 Ogr encoded by
a cryptic prophage. J. Mol. Biol. 256:264–
78
44. Johnson-Boaz R, Chang C-Y, Young R.
1994. A dominant mutation in the bacteriophage lambda S gene causes premature
lysis and an absolute defective plating phenotype. Mol. Microbiol. 13:495–504
45. Josslin R. 1971. Physiological studies of
the t gene defect in T4-infected Escherichia
coli. Virology 44:101–7
46. Kagan BL, Sokolov Y. 1994. Use of
lipid bilayer membranes to detect pore
formation by toxins. Methods Enzymol.
235:691–705
47. Kedzierska S, Wawrzynow A, Taylor A.
1996. The Rz1 gene product of bacteriophage lambda is a lipoprotein localized in
the outer membrane of Escherichia coli.
Gene. 168:1–8
48. Krogh S, Jorgensen ST, Devine KM.
1998. Lysis genes of the Bacillus subtilis defective prophage PBSX. J. Bacteriol. 180:2110–17
49. Loessner MJ, Gaeng S, Scherer S. 1999.
Evidence for a holin-like protein gene fully
embedded out of frame in the endolysin
gene of Staphylococcus aureus bacteriophage 187. J. Bacteriol. 181:4452–60
50. Loessner MJ, Gaeng S, Wendlinger G,
Maier SK, Scherer S. 1998. The twocomponent lysis system of Staphylococcus
51.
52.
53.
54.
55.
56.
57.
58.
59.
aureus bacteriophage Twort: a large TTGstart holin and an associated amidase endolysin. FEMS Microbiol. Lett. 162:265–
74
Loessner MJ, Maier SK, Daubek-Puza H,
Wendlinger G, Scherer S. 1997. Three
Bacillus cereus bacteriophage endolysins
are unrelated but reveal high homology to
cell wall hydrolases from different bacilli.
J. Bacteriol. 179:2845–51
Loessner MJ, Wendlinger G, Scherer S.
1995. Heterogeneous endolysins in Listeria monocytogenes bacteriophages: a new
class of enzymes and evidence of conserved holin genes within the siphoviral lysis cassettes. Mol. Microbiol. 16:1231–41
Longchamp PF, Mauel C, Karamata
D. 1994. Lytic enzymes associated with
defective prophages of Bacillus subtilis:
sequencing and characterization of the region comprising the N-acetylmuramoyl-Lalanine amidase gene of prophage PBSX.
Microbiology. 140(8):1855–67
Lu M-J, Henning U. 1992. Lysis protein
T of bacteriophage T4. Mol. Gen. Genet.
235:253–58
MacKenzie KR, Prestegard JH, Engelman DM. 1997. A transmembrane helix
dimer: structure and implications. Science
276:131–33
Martin AC, Lopez R, Garcia P. 1996.
Analysis of the complete nucleotide sequence and functional organization of the
genome of Streptococcus pneumoniae bacteriophage Cp-1. J. Virol. 70:3678–87
Martin AC, Lopez R, Garcia P. 1998.
Functional analysis of the two-gene lysis
system of the pneumococcal phage Cp-1 in
homologous and heterologous host cells. J.
Bacteriol. 180:210–17
Montag D, Degen M, Henning U. 1987.
Nucleotide sequence of gene t (lysis gene)
of the E. coli phage T4. Nucleic Acids Res.
15:6736
Morgan PJ, Varley PG, Rowe AJ,
Andrew PW, Mitchell TJ. 1993. Characterization of the solution properties and
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
60.
61.
62.
63.
64.
65.
66.
67.
conformation of pneumolysin,
the
membrane-damaging toxin of Streptococcus pneumoniae. Biochem. J. 296(3):
671–74
Nakayama K, Kanaya S, Ohnishi M, Terawaki Y, Hayashi T. 1999. The complete
nucleotide sequence of φCTX, a cytotoxinconverting phage of Pseudomonas aeruginosa: implications for phage evolution
and horizontal gene transfer via bacteriophages. Mol. Microbiol. 31:399–419
Navarre WW, Ton-That H, Faull KF,
Schneewind O. 1999. Multiple enzymatic
activities of the murein hydrolase from
staphylococcal phage φ11: identification
of a D-alanyl-glycine endopeptidase activity. J. Biol. Chem. 274:15847–56
Neely MN, Friedman DI. 1998. Functional and genetic analysis of regulatory
regions of coliphage H-19B: location of
shiga-like toxin and lysis genes suggest a
role for phage functions in toxin release.
Mol. Microbiol. 28:1255–67
Oki M, Kakikawa M, Nakamura S, Yamamura E-T, Watanabe K, et al. 1997. Functional and structural features of the holin
HOL protein of the Lactobacillus plantarum phage φg1e: analysis in Escherichia
coli system. Gene 197:137–45
Oki M, Kakikawa M, Yamada K, Taketo
A, Kodaira K-I. 1996. Cloning, sequence
analysis, and expression of the genes encoding lytic functions of bacteriophage
φg1e. Gene 176:215–23
Paddison P, Abedon ST, Dressman HK,
Gailbreath K, Tracy J, et al. 1998. The
roles of the bacteriophage T4 r genes in lysis inhibition and fine-structure genetics: a
new perspective. Genetics 148:1539–50
Pfister P, Wasserfallen A, Stettler R,
Leisinger T. 1998. Molecular analysis of
Methanobacterium phage psiM2. Mol. Microbiol. 30:233–44
Portelli R, Dodd IB, Xue Q, Egan JB. 1998.
The late-expressed region of the temperate
coliphage 186 genome. Virology 248:117–
30
823
68. Raab R, Neal G, Garrett J, Grimaila R,
Fusselman R, Young R. 1986. Mutational analysis of bacteriophage lambda lysis gene S. J. Bacteriol. 167:1035–42
69. Raab R, Neal G, Sohaskey C, Smith J,
Young R. 1988. Dominance in lambda
S mutations and evidence for translational
control. J. Mol. Biol. 199:95–105
70. Reader RW, Siminovitch L. 1971. Lysis defective mutants of bacteriophage
lambda: genetics and physiology of S cistron mutants. Virology 43:607–22
71. Regamey A, Karamata D. 1998. The
N-acetylmuramoyl-L-alanine amidase encoded by the Bacillus subtilis 168 prophage SPbeta. Microbiology 144(4):885–
93
72. Riede I. 1987. Lysis gene t of T-even bacteriophages: evidence that colicins and bacteriophage genes have common ancestors.
J. Bacteriol. 169:2956–61
73. Sable S, Lortal S. 1995. The lysins of bacteriophages infecting lactic acid bacteria.
Appl. Microbiol. Biotechnol. 43:1–6
74. Sanders JW, Venema G, Kok J. 1997.
A chloride-inducible gene expression cassette and its use in induced lysis of Lactococcus lactis. Appl. Environ. Microbiol.
63:4877–82
75. Schmidt C, Velleman M, Arber W.
1996. Three functions of bacteriophage
P1 involved in cell lysis. J. Bacteriol.
178:1099–104
76. Sekiya K, Satoh R, Danbara H, Futaesaku Y. 1993. A ring-shaped structure
with a crown formed by streptolysin O
on the erythrocyte membrane. J. Bacteriol.
175:5953–61
77. Shatursky O, Heuck AP, Shepard LA,
Rossjohn J, Parker MW, et al. 1999. The
mechanism of membrane insertion for a
cholesterol-dependent cytolysin: a novel
paradigm for pore-forming toxins. Cell
99:293–99
78. Sheehan MM, Garcia JL, Lopez R, Garcia P. 1997. The lytic enzyme of the pneumococcal phage Dp-1: a chimeric lysin of
P1: FUI
August 11, 2000
824
79.
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
80.
81.
82.
83.
84.
85.
86.
87.
11:18
WANG
¥
Annual Reviews
SMITH
¥
AR110-24
YOUNG
intergeneric origin. Mol. Microbiol. 25:
717–25
Sheehan MM, Stanley E, Fitzgerald GF,
van Sinderen D. 1999. Identification
and characterization of a lysis module
present in a large proportion of bacteriophages infecting Streptococcus thermophilus. Appl. Environ. Microbiol. 65:
569–77
Smith DL. 1998. Purification and biochemical characterization of the bacteriophage λ holin. PhD thesis.Texas A&M
Univ. 246 pp.
Smith DL, Chang C-Y, Young R. 1998.
The λ holin accumulates beyond the lethal
triggering concentration under hyperexpression conditions. Gene Expr. 7:39–
52
Smith DL, Struck DK, Scholtz JM, Young
R. 1998. Purification and biochemical
characterization of the lambda holin. J.
Bacteriol. 180:2531–40
Smith DL, Young R. 1998. Oligohistidine
tag mutagenesis of the lambda holin gene.
J. Bacteriol. 180:4199–211
Sonnhammer EL, von Heijne G, Krogh A.
1998. A hidden Markov model for predicting transmembrane helices in protein sequences. Proc. Int. Conf. Intelligent Syst.
Mol. Biol., 6th, ed. J Glasgow, T Littlejohn, F Major, R Lathrop, D Sankoff,
C Sensen, pp. 175–82. Menlo Park, CA:
AAAI Press
Steiner M, Bläsi U. 1993. Charged aminoterminal amino acids affect the lethal capacity of lambda lysis proteins S107 and
S105. Mol. Microbiol. 8:525–33
Streisinger G, Mukai F, Dreyer WJ, Miller
B, Horiuchi S. 1961. Mutations affecting the lysozyme of phage T4. Cold
Spring Harbor Symp. Quant. Biol. 26:25–
30
Tedin K, Resch A, Steiner M, Bläsi U.
1995. Dual translational start motif evolutionarily conserved in the holin gene
of Bacillus subtilis phage φ29. Virology
206:479–84
88. van Sinderen D, Karsens J, Kok J,
Terpstra P, Ruiters MHJ, et al. 1996.
Sequence analysis and molecular characterization of the temperate lactococcal bacteriophage r1t. Mol. Microbiol. 19:1343–
55
89. Vasala A, Valkkila M, Caldentey J,
Alatossava T. 1995. Genetic and biochemical
characterization
of
the
Lactobacillus delbrueckii subsp. lactis bacteriophage LL-H lysin. Appl.
Environ. Microbiol. 61:4004–11
90. Wang I-N, Dykhuizen DE, Slobodkin LB.
1996. The evolution of phage lysis timing.
Evol. Ecol. 10:545–58
91. Ward LJH, Beresford TPJ, Lubbers MW,
Jarvis BDW, Jarvis AW. 1993. Sequence
analysis of the lysin gene region of the prolate lactococcal bacteriophage c2. Can. J.
Microbiol. 39:767–74
92. Weaver LH, Matthews BW. 1987. Structure of bacteriophage T4 lysozyme refined
at 1.7 Å resolution. J. Mol. Biol. 193:189–
99
93. Young R. 1992. Bacteriophage lysis:
mechanism and regulation. Microbiol. Rev.
56:430–81
94. Young R, Bläsi U. 1995. Holins: form and
function in bacteriophage lysis. FEMS Microbiol. Rev. 17:191–205
95. Young R, Wang I-N, Roof WD. 2000.
Phages will out: strategies of host cell lysis. Trends Microbiol. 8:120–28
96. Young R, Way S, Yin J, Syvanen M.
1979. Transposition mutagenesis of bacteriophage lambda: a new gene affecting cell lysis. J. Mol. Biol. 132:307–
22
97. Zagotta MT, Wilson DB. 1990. Oligomerization of the bacteriophage lambda S protein in the inner membrane of Escherichia
coli. J. Bacteriol. 172:912–21
98. Zhang N, Young R. 1999. Complementation and characterization of the nested
Rz and Rz1 reading frames in the genome
of bacteriophage λ. Mol. Gen. Genet.
262:659–67
P1: FUI
August 11, 2000
11:18
Annual Reviews
AR110-24
HOLINS AS PROTEIN CLOCKS
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
99. Ziermann R, Bartlett B, Calendar
R, Christie GE. 1994. Functions involved in bacteriophage P2-induced
host cell lysis and identification of a
new tail gene. J. Bacteriol. 176:4974–84
825
100. Zink R, Loessner MJ, Scherer S. 1995.
Characterization of cryptic prophages
(monocins) in Listeria and sequence analysis of a holin/endolysin gene. Microbiology 141:2577–84
P1: FDS
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
September 11, 2000
14:57
Annual Reviews
AR110-CO
Figure 2 Structure and mosaicism of the lambdoid “lysis cassette”. The lysis genes
of the coliphages λ and 21 (93) and the Salmonella phages P22 (20) and PS3 (Genbank
accession AB008550) are shown, downstream of the late promoter (p’R in λ; arrow) and
named as in λ. The size of each reading frame in codons is shown above each ORF and
unambiguously orthologous relationships are indicated by identical colors. (p’R of PS3 has
not been mapped.)
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
P1: FDS
September 11, 2000
14:57
Annual Reviews
AR110-CO
14:57
Annual Reviews
B. Representative sequences of holin protein families. The predicted amino acid sequence of one member of each orthologous group of known or putative
holins (see text for details) is shown. Class I (Gram-negative hosts): 1. λ S [J02459] (94); 2. Pl LydA (X87674) (94); 3. P2 Y [AF063097] (94); 4.
φCTX ORF9 [AB008550] (60); 5. φCTX ORF10 [AB008550] (60); 6. PRD1 m [M69077] (94); 7. PS3 13 [AJ011579]. Class II (Gram-negative hosts):
8. 21 S [M65239] (94); 9. P. aeruginosa PML14 Hol [AB030826]; 10. N4 ORF63 (LB Rothman-Denes per. comm.); 11. N15 53 [AF064539]; 12. S.
marcescens NucE [U11698] (43); 13. T7 17.5 [V01146] (94). Class I (Gram-positive hosts): 14. A118 Hol118 [X85008] (52); 15. c2 117 [L48605]
(91); 16. Cp-1 Cph1 [Z47794] (57) (56); 17. φ29 14 [X04962] (94); 18. sk1 ORF19 [AF011378] (21); 19. 80α Holin [U72397] (14); 20. BK5-T
ORF95 [L44593] (16); Class II (Gram-positive hosts): 21. Dp-1 Dph [Z93946] (78); 22. L5 ORF11 [Z18946] (38); 23. LL-H ORF107 [M96254] (89);
24. φadh Holin [Z97974] (94); 25. φC31 ORF1 [X91149] (42); 26. rlt ORF48 [U38906] (88); Atypical or unclassified holins or putative holins; 27.
φ6 P10 [M17462] (35); 28. T4 t [AF158101] (94); 29. 187 Hol187 [Y07740] (49); 30. 10MC P98 [AF049087] (34); 31. A511 Hol511 [X85010] (52);
32. PBSX XhlA [L25924] (53); 33. SPbeta BhlA [AF021803] (71); and 34. S. aureus Lrg A [U52961] (17). Color coding: Yellow, predicted TM
domains; Red, acidic residues; Turquoise, basic residues. TM domains predicted using the TMHMM algorithm (84), except for LL-H and T7, which
were determined by inspection.
September 11, 2000
Figure 3 A. Primary structure of prototype class I (Sλ) and class II (S21) holins. Sλ mutants with lysis phenotypes are shown as lower-case single letter
codes above the wild-type sequence of Sλ. The putative TM and connector loop (L1, L2) domains are indicated below the sequence, and acidic/basic
residue colors are as in A. Brown and yellow highlighting indicates severe lysis-defective phenotype. Yellow indicates “anti-dominant” alleles. Green
indicates early-lysis alleles. X indicates nonsense mutation.
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
P1: FDS
AR110-CO
September 11, 2000
14:57
Annual Reviews
Figure 4 Potential membrane topologies of holins. (A) Chimeric holin with signal sequence (red) fused to N-terminus of holin, requiring Lep cleavage
for function and (B) the putative N-terminal externalization of Sλ predicted in the model of Graschopf et al. (36). (C) 3 TM model and (D) 2 TM model
for Sλ. (E) predicted membrane topology of S21 and other class II holins.
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
P1: FDS
AR110-CO
Annual Review of Microbiology
Volume 54, 2000
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
CONTENTS
THE LIFE AND TIMES OF A CLINICAL MICROBIOLOGIST, Albert
Balows
ROLE OF CYTOTOXIC T LYMPHOCYTES IN EPSTEIN-BARR
VIRUS-ASSOCIATED DISEASES, Rajiv Khanna, Scott R. Burrows
BIOFILM FORMATION AS MICROBIAL DEVELOPMENT, George
O'Toole, Heidi B. Kaplan, Roberto Kolter
MICROBIOLOGICAL SAFETY OF DRINKING WATER, U. Szewzyk,
R. Szewzyk, W. Manz, K.-H. Schleifer
THE ADAPTATIVE MECHANISMS OF TRYPANOSOMA BRUCEI
FOR STEROL HOMEOSTASIS IN ITS DIFFERENT LIFE-CYCLE
ENVIRONMENTS, I. Coppens, P. J. Courtoy
THE DEVELOPMENT OF GENETIC TOOLS FOR DISSECTING THE
BIOLOGY OF MALARIA PARASITES, Tania F. de Koning-Ward,
Chris J. Janse, Andrew P. Waters
NUCLEIC ACID TRANSPORT IN PLANT-MICROBE
INTERACTIONS: The Molecules That Walk Through the Walls, Tzvi
Tzfira, Yoon Rhee, Min-Huei Chen, Talya Kunik, Vitaly Citovsky
PHYTOPLASMA: Phytopathogenic Mollicutes, Ing-Ming Lee, Robert E.
Davis, Dawn E. Gundersen-Rindal
ROOT NODULATION AND INFECTION FACTORS PRODUCED BY
RHIZOBIAL BACTERIA, Herman P. Spaink
ALGINATE LYASE: Review of Major Sources and Enzyme
Characteristics, Structure-Function Analysis, Biological Roles, and
Applications, Thiang Yian Wong, Lori A. Preston, Neal L. Schiller
INTERIM REPORT ON GENOMICS OF ESCHERICHIA COLI, M.
Riley, M. H. Serres
ORAL MICROBIAL COMMUNITIES: Biofilms, Interactions, and
Genetic Systems, Paul E. Kolenbrander
ROLES OF THE GLUTATHIONE- AND THIOREDOXINDEPENDENT REDUCTION SYSTEMS IN THE ESCHERICHIA COLI
AND SACCHAROMYCES CEREVISIAE RESPONSES TO OXIDATIVE
STRESS, Orna Carmel-Harel, Gisela Storz
RECENT DEVELOPMENTS IN MOLECULAR GENETICS OF
CANDIDA ALBICANS, Marianne D. De Backer, Paul T. Magee, Jesus
Pla
FUNCTIONAL MODULATION OF ESCHERICHIA COLI RNA
POLYMERASE, Akira Ishihama
BACTERIAL VIRULENCE GENE REGULATION: An Evolutionary
Perspective, Peggy A. Cotter, Victor J. DiRita
LEGIONELLA PNEUMOPHILA PATHOGENESIS: A Fateful Journey
from Amoebae to Macrophages, M. S. Swanson, B. K. Hammer
THE DISEASE SPECTRUM OF HELICOBACTER PYLORI : The
Immunopathogenesis of Gastroduodenal Ulcer and Gastric Cancer, Peter
B. Ernst, Benjamin D. Gold
PATHOGENICITY ISLANDS AND THE EVOLUTION OF
MICROBES, Jörg Hacker, James B. Kaper
DNA SEGREGATION IN BACTERIA, Gideon Scott Gordon, Andrew
Wright
1
19
49
81
129
157
187
221
257
289
341
413
439
463
499
519
567
615
641
681
POLYPHOSPHATE AND PHOSPHATE PUMP, I. Kulaev, T.
Kulakovskaya
ASSEMBLY AND FUNCTION OF TYPE III SECRETORY SYSTEMS,
Guy R. Cornelis, Frédérique Van Gijsegem
PROTEINS SHARED BY THE TRANSCRIPTION AND
TRANSLATION MACHINES, Catherine L. Squires, Dmitry Zaporojets
Annu. Rev. Microbiol. 2000.54:799-825. Downloaded from www.annualreviews.org
by UNIVERSITY OF IDAHO LIBRARY on 01/20/11. For personal use only.
HOLINS: The Protein Clocks of Bacteriophage Infections, Ing-Nang
Wang, David L. Smith, Ry Young
OXYGEN RESPIRATION BY DESULFOVIBRIO SPECIES, Heribert
Cypionka
REGULATION OF CARBON CATABOLISM IN BACILLUS
SPECIES, J. Stülke, W. Hillen
IRON METABOLISM IN PATHOGENIC BACTERIA, Colin Ratledge,
Lynn G Dover
709
735
775
799
827
849
881