Genetics of cosQ, the DNA-Packaging Termination Site of Phage : Local Suppressors and Methylation Effects

Corresponding author: Douglas J. Wieczorek, University of Iowa, 3-315 BSB, Iowa City, IA 52242., wieczorekd{at}mail.medicine.uiowa.edu (E-mail)

Communicating editor: G. R. SMITH

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The cos site of the bacteriophage chromosome contains the sites required for DNA processing and packaging during virion assembly. cos is composed of three subsites, cosQ, cosN, and cosB. cosQ is required for the termination of chromosome packaging. Previous studies have shown cosQ mutations to be suppressed in three ways: by a local suppressor within cosQ; by an increase in the length of the chromosome; and by missense mutations affecting the prohead's portal protein, gpB. In the first study reported here, revertants of a set of cosQ mutants were screened for suppressors, and cis-acting suppressors of cosQ mutations were studied; these included second-site cosQ point mutations, base-pair insertions within cosQ, and an additional genome-lengthening suppressor. The 7-bp-long cosQ, with the sequence 5'-GGGTCCT-3', coincides exactly with the recognition site for the EcoO109I restriction/methylation system, which has the consensus sequence 5'-PuGGNCCPy-3'. In a second study, EcoO109I methylation was found to strongly interfere with the residual cosQ function of leaky cosQ mutants. cis-acting suppressors that overcome methylation-associated defects, including a methylation-dependent suppressor, were also isolated. Models of cosQ suppression are presented.

MANY double-stranded DNA viruses have replication and recombination pathways that produce concatemers, i.e., end-to-end multimers of virus chromosomes. For a subset of these viruses, including many tailed bacteriophages and the herpes viruses, the concatemers are cut at specific sites to generate unit-length virion chromosomes (FUJISAWA and HEARING 1994 ). A virally encoded enzyme, terminase, carries out the cutting reaction, which is coordinated with packaging of the DNA into an empty protein shell.

Phage chromosomes are 48.5-kb duplexes with 12-base-long, single-stranded extensions at the 5' ends of the strands. These extensions, called cohesive ends, are complementary and enable the chromosome to cyclize in an infected cell. Late during infection, concatemers produced by rolling circle replication and recombination are cut by terminase and packaged into empty shells called proheads (FEISS 1986 ; BECKER and MURIALDO 1990 ; CATALANO et al. 1995 ). The site at which terminase introduces staggered nicks is cosN; cosN is located between two other sites, cosQ and cosB. These cos subsites, which are located in a 200-bp segment, orchestrate the recognition, processing, and packaging of DNA (Fig 1). In addition to cosN, the adjacent site cosB is required for cutting at cosN to initiate DNA packaging. Terminase consists of a large subunit, gpA, which contains the endonuclease, and a small subunit, gpNu1, which binds cosB to anchor gpA during cosN cutting. After a concatemer's cosN is cut, terminase remains bound to the resulting cosB-containing DNA end, which is the left end of the chromosome to be packaged. The terminase-DNA complex binds to the portal vertex of a prohead, and translocation of the DNA into the shell ensues. Translocation moves the DNA-packaging complex along the DNA until the next cos is encountered and the terminase docked at the portal vertex recognizes and cuts the downstream cos. Following cleavage, terminase undocks from the newly filled head and remains bound to the left end of the next chromosome along the concatemer, sponsoring its packaging. cosQ, although not required for initiation of DNA packaging, is required for cleavage of the downstream cos. Because cosQ mutants fail to cut the downstream cos and fail to stop translocation at cos, the shell is filled to capacity, and because the protruding DNA prevents tail attachment, the cosQ defect is lethal. The bypassed downstream cos in cosQ mutants is properly nicked on the top strand of cosN, but the bottom strand is not nicked. The depolarization model proposes that cosQ acts in presenting a gpA subunit to the bottom strand of cosN by forming a bend in the region of DNA between cosQ and cosN, forming a loop for cosQ and cosN to be aligned in the same orientation. A second version of the model proposes that cosQ is needed for a pause in the packaging process to recruit the second of two gpA subunits from solution to the cosN site for the nicking of the bottom strand of DNA (CUE and FEISS 1998 ; WIECZOREK and FEISS 2001 ).

View larger version (19K): In this window In a new window Download PPT slide   Figure 1. Structure of the cohesive end site and relative virus yields of the cosQ mutational analysis. The cos region of bacteriophage extends across the region of DNA from base pairs 48,473 to 166 and is composed of three subsites: cosN, cosB, and cosQ. cosN is the site at which staggered nicks are introduced by terminase to generate the cohesive ends of virion DNA and represents the junctions of individual chromosomes in the concatemer. cosN contains a 22-bp element, extending from base pairs 48,498 to 17, in which 10 of the base pairs show twofold rotational symmetry. cosN can be divided into two half-sites on the basis of this center of symmetry, denoted by the dot (•): the right half-site, cosNR, with a nicking site at N1, and the left half-site, cosNL, with a nicking site at N2. cosB is located between cosN and the gene Nu1 and is required for the initiation of chromosome packaging. It is composed of the three R sequences, R1, R2, and R3, which are bound by the gpNu1 subunit of terminase. Between R3 and R2 is I1, a binding site for E. coli integration host factor. cosQ is required for the termination of chromosome packaging. The cosQ region has been defined as the 7-bp sequence from base pairs 48,473 to 48,479, as denoted by the box. The genes near cos are also shown. The Nu1 and A genes encode the small and large subunits of terminase, W encodes a head-tail joining protein, and B encodes the portal protein.

Three classes of suppressors of cosQ mutations have been identified (CUE and FEISS 1997 ; WIECZOREK et al. 2002 ). The first class consists of local suppressors affecting base pairs within cosQ. The second suppressor class increases the length of the phage chromosome to near the capacity of the head. It has been proposed that the rate of translocation is dependent upon the length of DNA packaged and that the rate likely slows as more DNA is packaged into the prohead. Such slowing of the rate of translocation has recently been documented for phage 29 (SMITH et al. 2001 ). CUE and FEISS 1997 proposed that the resulting increase in chromosome length further slows the rate of translocation so that the translocation complex can more efficiently recognize the mutant cosQ site, leading to more efficient termination. The third class of suppressors maps to the B gene, which encodes the prohead's portal protein. It has been proposed that the portal protein acts as a sensing mechanism either to measure the rate of translocation or to identify the cosQ site (CUE and FEISS 1998 ; WIECZOREK and FEISS 2001 ). These suppressors may slow the rate of translocation, allowing the mutant cosQ site to be more efficiently recognized by terminase or by the portal protein itself.

In a previous study, we used saturation mutagenesis to determine that cosQ is 7 bp long, composed of base pairs 48,473–48,479 with the sequence 5'-GGGTCCT-3' (WIECZOREK and FEISS 2001 ). The 7-bp sequence is conserved in the related lambdoid phages 21, 80, and N15 (SMITH and FEISS 1993 ; RAVIN et al. 2000 ). In this article we report two studies dealing with how local changes affect cosQ function. First, we describe suppression of cosQ mutations by local mutations that alter cosQ function. Second, we describe how methylation of cosQ impacts cosQ function. The second study stems from the observation that the 7-bp-long cosQ segment coincides exactly with the recognition site for the EcoO109I restriction/methylation system, which has the consensus sequence 5'-PuGGNCCPy-3'. The restriction enzyme cuts between the G residues, and the inner C residues are methylated by the methylase (KITA et al. 2001 ). We report experiments indicating that EcoO109I methylation has an impact on cosQ function. Methylation especially affects the residual cosQ function of leaky cosQ mutants. In addition, we describe cis-acting suppressors that overcome methylation-associated defects, including a methylation-dependent suppressor.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media:
Media were prepared as described by WIECZOREK and FEISS 2001 with the exception that chloramphenicol was added at a final concentration of 30 µg/ml for pACYC184-based vectors.

Strains:
Strains used in this study are described by WIECZOREK and FEISS 2001 . The standard bacterial host used was MF1427, a galK derivative of the Escherichia coli C strain C1a (SIX and KLUG 1973 ).

General recombinant DNA techniques:
General recombinant DNA techniques are described by WIECZOREK et al. 2002 . Clones of the EcoO109I methylase were generously provided by New England Biolabs (Beverly, MA). Phage DNA was purified by CsCl centrifugation and phenol extraction as described by ARBER et al. 1983 .

Sequence designations:
All references to sequence are based on the numbering convention described by DANIELS et al. 1983 . Numbering of the sequence begins with the first base of the left cohesive end and continues along the top strand in a 5' to 3' direction. The position of each restriction cut site is given as the first nucleotide of the recognition sequence.

Phage yield determinations:
Phage yield determinations are described by WIECZOREK et al. 2002 .

Identification and isolation of plaque-forming revertants:
Lysogens of non-plaque-forming cosQ mutants were induced and lysates titered on MF1427. Also, lysogens of cosQ T_48,479C were induced and lysates titered on MF1427 expressing the EcoO109I methylase. Plaque-forming revertants were selected and single plaque purified. Lysogens of the revertants were constructed by infecting MF1427 with phages isolated from plaques. PCR amplification followed by restriction enzyme analysis and DNA sequencing were performed to determine the cosQ sequence.

Introduction of B suppressor mutations into the genome:
The introduction of B suppressor mutations into the genome is described by WIECZOREK et al. 2002 with the exception that the mutations were introduced, by transformation, into MF1427 lysogenic for cosQ G_48,473A.

E. coli mutD mutagenesis:
The method of E. coli mutD mutagenesis is described by WIECZOREK et al. 2002 .

PCR mutagenesis:
PCR mutagenesis of the B gene is described by WIECZOREK et al. 2002 with the exception that for each lysogen, 4000 transformed colonies were scraped from the plates and resuspended in Luria broth. The pooled transformed lysogens were grown overnight, prophages were induced, and lysates were titered on MF1427 as described previously.

DNase protection assay:
A 100-µl aliquot of each lysate was incubated with 5 units of DNase, 20 µl of RNase A (500 µg/ml), 6 mM MgCl₂, and 10 mM CaCl₂ at room temperature for 30 min. Ten microliters of 0.5 M EDTA was added for 10 min at room temperature to stop the reaction. Five nanograms of linearized pUC19 was included as a control for DNA recovery. The lysate was extracted twice with phenol-CHCl₃-isoamyl alcohol (25:24:1, v/v) and once with CHCl₃. Samples included 30 of 200 µl (15%) of the extracted DNA. The DNA was subjected to electrophoresis on a 0.8% gel. To quantify the amount of DNA, the DNA was denatured and transferred to a GeneScreen Plus (New England Nuclear) membrane. DNA hybridization was performed using [-³²P]dCTP-labeled (Amersham, Buckinghamshire, UK) linearized pUC19 and whole-length DNA as probes. An autoradiogram was obtained by exposure of the membrane to a Fuji Super RX film for 8 hr at -70°. The recovery of pUC19 DNA and the amount of DNA packaged in the methylated and unmethylated lysates was determined by phosphorimaging on a Packard Instantimager. The packaging ratio is the yield of packaged phage DNA per induced lysogen in the presence of the methylase relative to the yield of packaged phage DNA per induced lysogen in the absence of the methylase. For example, the effect of methylation on the packaging of wild-type DNA was calculated by dividing the counts per minute per induced lysogen from lane 8 by the counts per minute per induced lysogen from lane 3 (Fig 2). The yields were further adjusted to account for the percentage recovery of pUC19 as an indicator of the overall recovery of packaged DNA. More than 87% of the control pUC19 DNA was recovered for each sample. Portions of each culture were removed prior to induction, diluted 1:10,000 in 10 mM MgSO₄, and plated on tryptone agar. Plates were incubated overnight at 31° to determine the number of viable lysogens.

View larger version (49K): In this window In a new window Download PPT slide   Figure 2. Effects of methylation on the yields of packaged DNA for cosQ T48,479C and its pseudorevertants that are able to grow in the presence of the EcoO109I methylase. Packaged unmethylated and EcoO109I methylated phage DNA was extracted from 100-µl aliquots of each phage lysate. Five nanograms of linearized pUC19 was included as a control for DNA recovery. Lanes 1 and 2: 5 ng pUC19 and control DNAs, respectively. Lanes 3 and 8: unmethylated and methylated wild-type DNA. Lanes 4 and 9: unmethylated and methylated cosQ T48,479C DNA. Lanes 5 and 10: unmethylated and methylated Methrev1 cosQ T48,479C DNA. Lanes 6 and 11: unmethylated and methylated Methrev3 cosQ T48,479C DNA. Lanes 7 and 12: unmethylated and methylated Methrev21 cosQ T48,479C DNA. The packaging ratio is the yield of packaged phage DNA per induced lysogen in the presence of the methylase relative to the yield of packaged phage DNA per induced lysogen in the absence of the methylase. For example, the effect of methylation on the packaging of wild-type DNA was calculated by dividing the counts per minute per induced lysogen from lane 8 by the counts per minute per induced lysogen from lane 3. The yields were further adjusted to account for the percentage of recovery of pUC19 as an indicator of the overall recovery of packaged DNA. Samples included 30 of 200 µl (15%) of the extracted DNA. For each sample, >87% of the control pUC19 DNA was recovered.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Local suppressors in pseudorevertants of cosQ mutants
Local suppressors in spontaneous pseudorevertants of cosQ mutants: cosQ is 7 bp long, so there are 21 possible base pair changes in cosQ. Earlier, we constructed phages with these 21 cosQ mutations and classified them by phenotypic severity (WIECZOREK and FEISS 2001 ). Of the 21 point mutations in cosQ, 8 were severe lethals, reducing the virus yield to 1.0 virion/induced lysogen, and 3 were moderate lethals with yields between 1.0 and 5.5 (see WIECZOREK et al. 2002 , Table 1). Recently we reported on trans-acting suppressors of cosQ mutations (WIECZOREK et al. 2002 ). Here we report a characterization of the nature of local changes in pseudorevertants of cosQ mutants.

We first examined spontaneous revertants of cosQ mutants. Efforts were taken to minimize siblings among the revertants by using multiple induced lysates in the screening process. We screened plaque-forming revertants in unmutagenized lysates of phages with the eight severe lethal cosQ mutations: G_48,473T, G_48,473C, G_48,474A, G_48,474T, G_48,474C, C_48,478A, C_48,478T, and C_48,478G (see Table 1 in WIECZOREK et al. 2002 ). Plaque-forming revertants were screened by sequencing cos to identify true revertants and revertants with local, cis-acting suppressors in cosQ. Of 423 revertants screened, 15 were pseudorevertants carrying a local suppressor plus the original cosQ mutation. All 15 suppressors were suppressors of only two cosQ mutations, both of which affected the first cosQ base pair, cosQ G_48,473T and cosQ G_48,473C. Furthermore, all 15 suppressors were insertions of either an A:T base pair or a T:A base pair between cosQ base pairs 48,475 and 48,476 (see WIECZOREK et al. 2002 , Table 2). That is, Rev12 of cosQ G_48,473C (5'-CGGTCCT-3') contained an insertion of an A between base pairs 48,475 and 48,476 to give the sequence 5'-CGGATCCT-3'. Rev28 and Rev15 of cosQ G_48,473T contained insertions of A or T, respectively, also between base pairs 48,475 and 48,476. These nucleotide insertions appeared to suppress the original cosQ mutation by shifting the original mutation one position to the left, resulting in a new 7-bp cosQ site. The novel cosQ sites created by the insertion suppressors begin at base pair 48,474 instead of 48,473, with the inserted base pair representing the only mutation in cosQ in the third base pair of cosQ.

Our previous work showed that mutants with an A or T in the third cosQ base pair (5'-GGATCCT-3' or 5'-GGTTCCT-3') were viable (WIECZOREK and FEISS 2001 ). To test the proposal that the insertion mutations generated new cosQ sites with sequences identical to those of cosQ G_48,475A and cosQ G_48,475T, we compared the yields of the insertion-containing revertants with those of cosQ G_48,475A and cosQ G_48,475T. First, we compared the yields of Rev12 of cosQ G_48,473C and Rev28 of cosQ G_48,473T with that of cosQ G_48,475A, which has the same 7-bp cosQ sequence. The yields of Rev12 and Rev28, at 15 and 19% that of wild type, respectively, agree closely with the 19% value for cosQ G_48,475A (Table 1). Second, Rev15 of cosQ G_48,473T, with a relative yield of 33% that of wild type, had a yield in close agreement with that of cosQ G_48,475T, with a relative virus yield of 35%. The results strongly support the proposal that the insertion suppressors create new cosQs.

Local suppressors in pseudorevertants of cosQ mutants subjected to E. coli mutD mutagenesis: The vast majority of revertants of cosQ mutants were true revertants, making the isolation of pseudorevertants difficult. To vary the strategy, we searched for pseudorevertants in lysates of mutD-mutagenized cosQ mutants with the moderate mutations G_48,475C and T_48,479G and with the severe mutations G_48,473T, G_48,473C, G_48,474C, G_48,474A, G_48,474T, C_48,478A, C_48,478T, and C_48,478G. The mutD mutation inactivates , the proofreading exonuclease subunit of DNA polymerase III (MARUYAMA et al. 1983 ). Lysates of phages grown on E. coli mutD had an 500-fold increase in the number of plaque-forming revertants identified when compared with nonmutagenized control lysates. Lysogens of revertants were first screened by using PCR to generate cos-containing segments and then digested with the EcoO109I restriction enzyme. Since cosQ⁺ is identical to the EcoO109I recognition sequence, digestion with EcoO109I indicates reversion to cosQ⁺.

As with spontaneous revertants, most mutD-induced cosQ revertants were true revertants (see WIECZOREK et al. 2002 , Table 2). A number of local suppressors within cosQ were also identified (Table 1). CUE and FEISS 1997 previously identified G_48,473A as a suppressor of C_48,477T; single mutants cosQ G_48,473A and cosQ C_48,477T have more severe defects than the cosQ G_48,473A C_48,477T double mutant. We also recovered C_48,477T as a suppressor of our cosQ G_48,473A mutant (not shown). C_48,477T was also identified as a suppressor of the two other point mutations at position 48,473 of cosQ: cosQ G_48,473C (Rev16) and cosQ G_48,473T (Rev41). Thus, all three mutations at position 48,473 of cosQ, G A, G C, and G T, are suppressed by C_48,477T, or vice versa, resulting in large increases in virus yield (Table 1). A local suppressor of the G_48,475C mutation was also identified as the T_48,479C mutation (Rev8), affecting the last base pair of cosQ (Table 1). Curiously, in all four cases involving these local suppressors, the distance between the original mutation and the suppressor within the cosQ site is 4 bp. Given that all three changes at 48,473 were suppressors of the C_48,477T mutation, we were curious to see if multiple examples of suppression involving base pairs 48,475 and 48,479 were possible.

Suppression study of base pairs 48,475 and 48,479: To determine which combinations of mutations affecting base pairs 48,475 and 48,479 showed mutual suppression, we constructed all possible combinations of cosQ changes affecting base pairs 48,475 and 48,479. Of the nine double mutants constructed, the only viable combination recovered was that found previously, G_48,475C and T_48,479C (GGCTCCC; yield = 16 phage/induced lysogen). An additional pair, the combination of two viable cosQ mutations, G_48,475T and T_48,479C (GGTTCCC), produced a significant yield (0.6 phage/induced lysogen), although the yield was insufficient for plaque formation.

The results show that in all observed cases of mutual suppression one of the base pairs involved is one of the symmetrically disposed pairs of base pairs at 48,475 and 48,477 (base pairs G_48,473A/T/C + C_48,477T and G_48,475C + T_48,479C). However, the combinations of suppressing base pairs are not symmetric. That is, the mutations G_48,475C and C_48,477T affect symmetrically disposed base pairs, but while the C_48,477T mutation is suppressed by any change of the first cosQ base pair at 48,473, only the T_48,479C change in the last cosQ base pair (not T_48,479A or T_48,479G) suppresses the G_48,475C mutation.

Segment-specific PCR mutagenesis: In a previous study we employed PCR mutagenesis to identify suppressors of cosQ mutations (WIECZOREK et al. 2002 ). An obvious candidate was the B gene, which encodes the portal protein and was already known to contain a variety of cosQ suppressors (CUE and FEISS 1997 ; WIECZOREK and FEISS 2001 ). This strategy produced several missense mutations in B that act as cosQ suppressors. Among plaque-forming variants of cosQ T_48,479A obtained using this strategy was an unusual phage, Rev19. Rev19 retained the T_48,479A mutation and had no suppressor within cosQ or B. Previous work had shown that the C_48,477T mutation was suppressed by an increase in phage chromosome length (CUE and FEISS 1997 ). Accordingly, a restriction enzyme analysis of prophage DNA was carried out and the results showed the presence of an 3-kb duplication within the head genes of involving A, W, parts of Nu1, and B (data not shown).

Effects of cosQ methylation on cosQ function
EcoO109I methylation of cosQ: The 7-bp region of cosQ with the DNA sequence 5'-GGGTCCT-3' corresponds to the target sequence of the EcoO109I restriction and modification enzymes that have the recognition sequence 5'-PuGGNCCPy-3' (MISE and NAKAJIMA 1985 ). Our strain of has two EcoO109I restriction sites: at base pairs 48,473 (cosQ) and 2815. To see if cosQ methylation affects cosQ function, we determined the effects of EcoO109I methylation on phage yield. A lysogen of wild type was transformed with a clone of the EcoO109I methylase in the vector pACYC184, and, following induction, the phage yield was determined (Table 2). The expression of EcoO109I methylase during wild-type phage production resulted in a virus yield of 42 (±3) in comparison to the unmethylated wild-type control transformed with the pACYC184 vector alone with a yield of 136 (±9), roughly a threefold decrease in phage yield. The virus yield in the absence of the pACYC184 vector was 87 (±3). It is unclear why the virus yield was higher in the presence of the pACYC184 vector, which served as a positive control.

View this table: In this window In a new window   Table 2. Effect of EcoO109I methylation of cosQ mutants on phage yield

EcoO109I methylation of cosQ mutants: We noted that some cosQ mutations left the EcoO109I target sequence intact. To study further the effects of cosQ methylation, we studied the growth of two phages bearing cosQ mutations that retained the EcoO109I recognition site, namely cosQ T_48,476G and cosQ T_48,479C. As controls, we also used three phages with cosQ mutations that inactivated the EcoO109I site, cosQ G_48,475A, cosQ G_48,475T, and cosQ C_48,477A. All five cosQ mutations used are nonlethal. The mutants with methylatable cosQ sites had severe decreases in burst sizes when grown in the host expressing the EcoO109I methylase (Table 2). That is, cosQ T_48,476G and cosQ T_48,479C had 97- and 76-fold decreases in virus yields [relative to the pACYC184 (+) control], respectively, when grown in the presence of the EcoO109I methylase, and consequently were unable to form plaques. In contrast, the phages with nonmethylatable cosQ sites had mild decreases in yield when grown in the presence of the methylase (1.3- to 2.5-fold), decreases comparable to that of wild type (3-fold). Since these mutants retained only the single EcoO109I site at 2815, it is possible that methylation of the base pair 2815 site mildly decreases the virus yield. Since only cosQ mutants with methylatable cosQ sites showed severe growth defects, we concluded that these severe growth defects were due to methylation of the mutant cosQ site. Because EcoO109I methylation had only mild effects on cosQ⁺'s yield, we further concluded that phages with weakened cosQ sites are particularly sensitive to methylation. We speculate that the cosQ mutations interfere with recognition of cosQ by some component of the packaging machinery and that recognition is further weakened when cosQ is methylated.

Suppression of EcoO109I methylation defects of cosQ by suppressors in gene B: Numerous suppressors of cosQ mutations have previously been identified as missense mutations in gene B, which encodes the portal protein (CUE and FEISS 1997 ; WIECZOREK and FEISS 2001 ; WIECZOREK et al. 2002 ). It is proposed that the changes in gpB slow the rate of DNA packaging, thus enhancing recognition of mutant cosQ sites by the packaging machinery. We asked whether these B suppressors are able to suppress the packaging defects associated with EcoO109I methylation of cosQ. The G_48,473A mutation retains the ability to be recognized by the EcoO109I methylase. cosQ G_48,473A is able to form tiny plaques in the absence of the methylase, but is unable to form plaques on MF1427 expressing the EcoO109I methylase. Brev98, Brev22, Brev33, and Brev31 are all mild suppressors that increase the virus yield of cosQ T_48,479A three- to eightfold, and they are general suppressors of other cosQ mutations (WIECZOREK and FEISS 2001 ). Since the T_48,476G and T_48,479C mutations are much more severely affected by methylation, a three- to eightfold increase would most definitely not allow plaque formation by such mild suppressors. Thus, G_48,473A was selected for this study, since relatively mild suppression could be observed.

First we tested the abilities of four B suppressors for suppression of G_48,473A in the absence of the methylase. Prophages bearing the G_48,473A mutation and several of the B suppressors were constructed, and virus yields were determined. cosQ G_48,473A Brev98, cosQ G_48,473A Brev33, cosQ G_48,473A Brev22, and cosQ G_48,473A Brev31 had virus yields increased by 1.5- to 5-fold in comparison to cosQ G_48,473A, indicating that each of the B suppressors of T_48,479A is also able to mildly suppress the G_48,473A mutation in the absence of methylation, as expected for general suppressors (Table 3). The relative strengths of suppression of G_48,473A are similar to that previously shown for T_48,479A (WIECZOREK et al. 2002 ).

View this table: In this window In a new window   Table 3. Effects of cosQ methylation and B suppressors on the yield of cosQ G48,473A

We next tested the ability of the B suppressors of cosQ G_48,473A to suppress the defects associated with EcoO109I methylation. The EcoO109I methylase expression plasmid was used to transform lysogens of cosQ G_48,473A containing the various B suppressors. Lysogens with and without the EcoO109I methylase were induced and virus yields were determined (Table 3). cosQ G_48,473A grown in the presence of the methylase had a yield 5.3% of its yield in the absence of the methylase. In contrast, cosQ G_48,473A Brev98, cosQ G_48,473A Brev33, cosQ G_48,473A Brev22, and cosQ G_48,473A Brev31 grown in the presence of the methylase had yields of 11, 11, 31, and 14%, respectively, of their yields in the absence of the methylase. Overall, the presence of the B suppressors resulted in two- to sixfold increases in virus yield in the presence of the EcoO109I methylase when compared to cosQ G_48,473A alone. Thus, the B mutations resulted in mild suppression of the defects associated with EcoO109I methylation of cosQ. The B mutations were effective to similar extents in suppressing cosQ mutations and cosQ methylation, both resulting in plaque formation, indicating that effects of cosQ methylation on cosQ function are similar to the effects of cosQ mutations.

Pseudorevertants of EcoO109I-methylated cosQ mutants: cosQ T_48,476G and cosQ T_48,479C carry moderate cosQ mutations and are able to form plaques on the host E. coli strain MF1427 in the absence of the EcoO109I methylase but are unable to form plaques on MF1427 expressing the methylase. We looked for pseudorevertants among plaque-forming revertants of cosQ T_48,476G and cosQ T_48,479C for suppressors. We assumed we would identify cis-acting suppressors that would alter cosQ and prevent recognition by the EcoO109I methylase. Lysates of cosQ T_48,476G and cosQ T_48,479C were plated on MF1427 cells carrying the EcoO109I methylase-expressing plasmid [pACYC184-EcoO109IM]. Among 38 plaque-forming revertants of the two mutants, 35 were true revertants and 3 were pseudorevertants of cosQ T_48,479C that carried the original cosQ mutation and an additional cosQ change (Table 4). Methrev1 and Methrev21 each contained an additional transition mutation within cosQ, while Methrev3 contained a single base-pair deletion of the original T_48,479C mutation, in effect, generating a wild-type cosQ sequence shifted 1 bp to the right, i.e., toward cosN.

View this table: In this window In a new window   Table 4. Analysis of pseudorevertants of EcoO109I-methylated cosQ T48,479C

Lysogens of the pseudorevertants that contained the EcoO109I methylase expression plasmid [pACYC184-EcoO109IM] were induced to prepare lysates. We determined the efficiency of plating of the pseudorevertants on MF1427 and MF1427 [pACYC184-EcoO109IM] (Table 4). While the cosQ T_48,479C parent plated with an efficiency of 10^-5 on MF1427 [pACYC184-EcoO109IM], relative to the titer on MF1427, the plating efficiencies of Methrev3 and Methrev21 were only mildly reduced. Clearly, these phages are able to grow well in the presence of the EcoO109I methylase. Interestingly, Methrev1 exhibited an 8000-fold higher titer on MF1427 [pACYC184-EcoO109IM] than on MF1427; i.e., Methrev1 was methylation dependent.

Packaged DNA is resistant to attack by DNase I (BODE and GILLIN 1971 ). To confirm the results of our virus yield assays (Table 4), we performed a DNase protection assay to directly measure the amount of packaged DNA in phages grown in the absence and presence of EcoO109I methylase (Fig 2). The wild type, cosQ T_48,479C, Methrev1, Methrev3, and Methrev21 were studied. When methylated wild type, packaged 81% of DNA compared to unmethylated wild type. Recall that the burst size data of cosQ⁺ showed a threefold reduction in phage yield when grown in the presence of the methylase (Table 2); the discrepancy is likely due to DNA recovery errors inherent in the DNase protection assay. Methylation of cosQ T_48,479C reduced DNA packaging to 10%, reflecting that phage's inability to produce plaques in the presence of the methylase. Methylated Methrev3 and Methrev21 packaged 53 and 121%, respectively, of the amount of DNA packaged in the absence of the methylase. The methylation-dependent virus Methrev1 packaged >6-fold more DNA when grown in the presence of the methylase than when grown in the absence of the methylase. Thus, Methrev1 is dependent on EcoO109I methylation for efficient phage production. These data are in reasonable agreement with the results of the plaque-forming assays. The 8000-fold increase in plating efficiency of Methrev1 in the presence of the methylase is due to the requirement for plaque formation in the plating assay. While mature phages are produced in the absence of the methylase, the yield is less than the yield (>5.5/cell) required for plaque formation.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

We have carried out a detailed pseudoreversion study of cosQ mutants, using a previously isolated complete set of cosQ point mutants (WIECZOREK and FEISS 2001 ). Previously, we reported on trans-acting cosQ suppressors that mapped to B, the gene for the virus's portal protein (WIECZOREK et al. 2002 ). Here we report on cis-acting suppressors of cosQ mutations. Also, we report on effects of EcoO109I methylation on cosQ function and on the nature of suppressors of methylation-dependent cosQ defects.

Local cosQ suppressors: frameshift mutations:
Two suppressors of the cosQ point mutations G_48,473C (found in Rev12) and G_48,473T (found in Rev28 and Rev15) were identified as insertions of an A or a T between base pairs 48,475 and 48,476 within the cosQ site (Table 1). These nucleotide insertions shift the original mutation one position to the left and result in a new 7-bp cosQ site, with novel cosQ sites beginning at base pair 48,474 instead of at base pair 48,473. Thus, the inserted base pair now represents the third base pair of cosQ and is the only mutation present in cosQ. Consistent with this interpretation, we previously found that cosQ mutants with an A or a T at the third cosQ base pair were viable (WIECZOREK and FEISS 2001 ).

A genome-lengthening cosQ suppressor: a tandem duplication:
Restriction analysis of the Rev19 cosQ T_48,479A revertant revealed the presence of an 3-kb duplication within the head genes of , involving W and at least part of Nu1, A, and B (data not shown). EMMONS 1974 discovered spontaneous tandem duplications in ; 90% of these mutations occurred on the left arm of the chromosome. Our identification of this duplication as a suppressor of the T_48,479A mutation validates the findings of CUE and FEISS 1997 who demonstrated suppression of the C_48,477T mutation through plasmid integration into the nonessential b region of the chromosome to increase the length of the cosQ C_48,477T chromosome from 46.2 to 50–51 kb. They showed that suppression did not depend on the particular DNA segment duplicated; that is, the important feature of this type of suppression was chromosome lengthening per se. Thus, increased chromosome length is a general mechanism of suppression of cosQ mutations. Increased chromosome length is proposed to slow the rate of DNA packaging during the late stages of DNA packaging, thus increasing the efficiency of mutant cosQ recognition (CUE and FEISS 1997 ; WIECZOREK and FEISS 2001 ).

cis-acting cosQ suppressors: a symmetric pattern of substitution mutations:
Some pseudorevertants of cosQ mutants contained local suppressors that were substitution mutations within the 7-bp cosQ segment, as follows. Secondary mutations in the cosQ sites of two pseudorevertants of cosQ C_48,477T were G_48,473C (in Rev16) and G_48,473T (in Rev41). An earlier study found that C_48,477T was suppressed by the G_48,473A mutation. Thus, all three mutations at position 48,473 of cosQ, G A, G C, and G T, show mutual suppression with C_48,477T. Of the base pair 48,473 and 48,477 mutations involved in mutual suppression, two, G_48,473T and G_48,473C, are lethal, and the other two, G_48,473A and C_48,477T, are nonlethal but have significant phenotypic effects (WIECZOREK and FEISS 2001 ).

A local suppressor of the G_48,475C mutation was T_48,479C (found in Rev8), which by itself has no phenotype (WIECZOREK and FEISS 2001 ). To ask whether other pairs of cosQ mutations affecting base pairs 48,475 and 48,479 were mutually suppressing, we constructed the eight other possible double-mutant combinations of base pair changes. None of the double mutants formed plaques, indicating that the changes in Rev8 were the only ones showing suppression at these positions (S. GAETH, D. WIECZOREK and M. FEISS, unpublished observations). Curiously, all of the mutually suppressing pairs of mutations we have found are 4 bp apart. Furthermore, the pairs of mutually suppressing cosQ mutations in the first and fifth cosQ base pairs occupy positions that are rotationally symmetric with the third and seventh base pairs that also exhibit mutual suppression (Fig 3).

View larger version (8K): In this window In a new window Download PPT slide   Figure 3. Mutual suppression pattern of cosQ mutations. The wild-type cosQ sequence is shown, and arrows indicate pairs of base pair positions where mutual suppression has been observed. At position 48,473, mutations to A, C, and T are suppressed by the C48,477T cosQ mutation. Similarly, the mild cosQ T48,479C mutation suppresses the G48,475C mutation.

A second observation suggests that cosQ may be a rotationally symmetric element: the 7-bp cosQ segment is coincident with an EcoO109I restriction-modification system target site, 5'-PuGGNCCPy-3'. The EcoO109I site has twofold rotational symmetry and is presumably recognized by symmetrically disposed subunits of the restriction enzyme and methylase. The coincidence of cosQ and the symmetric EcoO109I site also raises the possibility that cosQ itself is symmetric. Third, all the possible mutations affecting the symmetrically disposed cosQ base pairs G_48,474 and C_48,478 are severe lethals, and mutations affecting the symmetric base pairs G_48,475 and C_48,477 impart less severe changes.

There are also asymmetric phenotypic effects of cosQ point mutations. For example, the first position of cosQ, base pair 48,473, requires a G for cosQ function, whereas the seventh position, base pair 48,479, can be a T or a C. It is possible that cosQ is functionally symmetric, with asymmetry imposed by either base pairs flanking cosQ or the other cos subsites that function with cosQ. There are precedents for external imposition of asymmetric effects on a symmetric site. For example, the rate of cleavage of some restriction sites by type II restriction enzymes, such as EcoRI, is strongly affected by the sequences flanking the site (THOMAS and DAVIS 1975 ). Also, symmetrically disposed cosN mutations behave asymmetrically, due in large part to interactions of terminase with cosB (HANG et al. 2001 ), even though cosN is thought to be recognized by symmetrically disposed terminase protomers. It is clear that cosQ, like cosN, acts in conjunction with other cos subsites in sponsoring the cos cleavage that terminates DNA packaging (CUE and FEISS 1998 ). We note that if cosQ is recognized by a symmetric multimer, as has been proposed for gpA recognition of cosN, it would be of interest to construct symmetric mutations within the cosQ site for the identification of allele-specific suppressors.

EcoO109I methylation severely affects the yield of cosQ mutants:
The EcoO109I methylase has at best marginal effects on the growth of cosQ and cosQ mutants with cosQ sites that cannot be methylated by the EcoO109I enzyme. In contrast, two cosQ mutants with methylatable cosQ sites, cosQ T_48,476G and cosQ T_48,479C, showed very strong inhibition, such that the virus yield is reduced to 1–2% of the yield in the absence of the methylase (Table 2). These results indicate that wild-type cosQ functions normally when methylated, but that cosQ recognition already weakened by a cosQ mutation is severely affected by cosQ methylation. We cannot exclude the alternative explanation that the cosQ of wild type is methylated much less efficiently than the cosQ sites of cosQ T_48,476G and cosQ T_48,479C.

cosQ suppressors located in gene B also suppress defects in cosQ function caused by EcoO109I methylation:
Numerous non-allele-specific suppressors of cosQ mutations map to B, the gene for 's portal protein (WIECZOREK et al. 2002 ). We examined the effects of B suppressors on the cosQ mutation (5'-AGGCTTC-3') that retains methylase recognition. In the absence of the EcoO109I methylase, these B suppressors suppressed the G_48,473A mutation, increasing the yield of cosQ G_48,473A 1.5- to 5-fold (Table 3). These modest increases in phage yield are in agreement with other studies (WIECZOREK and FEISS 2001 ). In the presence of the methylase, the B suppressors suppressed the methylation effects on cosQ G_48,473A 2- to 6-fold. We note that, although the B suppressors suppressed the effects of methylation on cosQ G_48,473A to the same extent that they suppressed the effects of cosQ mutations, the extent of suppression was not sufficient to permit plaque formation by cosQ G_48,473A. Nevertheless, the similar extents of suppression of cosQ mutations and cosQ methylation indicate that methylation interferes with cosQ function by the same mechanism and that the B suppressors act by enhancing recognition of the altered cosQ site.

Local suppressors of methylation-induced cosQ defects:
We further studied methylation-induced cosQ defects by isolating pseudorevertants of cosQ T_48,479C, a phage unable to grow on the EcoO109I-expressing cells.

Methrev3 of cosQ T_48,479C: Methrev3 of cosQ T_48,479C contains a single base-pair deletion of the cosQ T_48,479C mutation. The new sequence contains a wild-type cosQ sequence shifted 1 bp closer to cosN (Table 4). Although other deletions between cosN and cosQ have not been studied, we note that addition of a single base pair between cosQ and cosN has little effect on virus yield (WIECZOREK and FEISS 2001 ).

Methrev21 of cosQ T_48,479C: Methrev21 of cosQ T_48,479C contains a second mutation within cosQ, C_48,477T, and had virus yields of 43 and 19% in the absence and presence of the methylase, respectively, when compared to the wild-type phage (Table 4). The cosQ sequence of this revertant from base pair 48,473 to 48,479 with the C_48,477T mutation is unable to be methylated. Thus, the virus yield is expected to be unaffected by the presence of the methylase. The high yield in the absence of the methylase is inconsistent with the yield expected for cosQ C_48,477T, since a yield 6% that of wild type was found previously for cosQ C_48,477T (WIECZOREK and FEISS 2001 ). Upon further examination of the sequence of Methrev21, we found a second cosQ-related sequence in base pairs 48,474–48,480: 5'-GGTTCCT-3'. This second cosQ sequence is shifted 1 bp closer to cosN [as in the (-1) deletion] and contains only a G T mutation at the third position of the sequence. The G T mutation blocks methylation of the second site. We propose that this nonmethylatable second cosQ site is utilized and that it sponsors efficient phage production by Methrev21. Methrev21 is predicted to have a relative virus yield comparable to cosQ G_48,475T in both the presence and the absence of the methylase, since the effect of the 1-bp shortening of the cosQ-cosN spacing region is expected to have little effect on virus yield. In fact, the cosQ G_48,475T mutant has a yield of 43% that of wild type, in excellent agreement with the yield of 43% of Methrev21 in the absence of the methylase (Table 4).

Methrev1 of cosQ T_48,479C: Methrev1 of cosQ T_48,479C contains a second mutation within cosQ, T_48,476G; this phage was dependent on methylation for viability, showing an 8000-fold increase in plating efficiency on cells expressing EcoO109I methylase, relative to cells lacking the methylase. In addition, Methrev1 packaged >6-fold more DNA when grown in the presence of the methylase than when grown in the absence of the methylase (Fig 2). Thus, Methrev1 is dependent on EcoO109I methylation for efficient phage production.

Upon examination of Methrev1's cosQ, two potential cosQ sites are identified, both of which are methylatable: the 5'-GGGGCCC-3' sequence from base pairs 48,473 to 48,479, containing two mutations, and the sequence 5'-GGGCCCT-3' from base pairs 48,474 to 48,480, a sequence identical to that of cosQ G_48,476C (Table 4). In the absence of the methylase, the 5'-GGGGCCC-3' cosQ sequence is not likely to permit healthy growth due to the presence of the two cosQ mutations. The T_48,476G mutation alone reduces the virus yield to 30% that of wild type, while the T_48,479C mutation alone has little effect on virus yield (see WIECZOREK and FEISS 2001 , Table 1). It is conceivable that the two mutations together are lethal. cosQ G_48,476C, which has the second cosQ sequence, 5'-GGGCCCT-3', has a virus yield 10% that of wild type and forms tiny plaques. In the context of Methrev1, the shift of the second cosQ 1 bp closer to cosN, even if it had only a slight effect, might be enough to prevent plaque formation. So how can we explain the viability of methylated Methrev1?

Structural models of cosQ:
The symmetry properties of the cosQ site and the symmetric aspect of the local cosQ substitution suppressors (Fig 3) suggest that cosQ may be recognized by symmetrically disposed protomers of a component of the packaging machinery, namely a prohead protein or terminase. The absence of allele-specific cosQ suppressors may be because two symmetrically disposed cosQ mutations might be necessary for isolation of allele-specific suppressors affecting the subunits of a symmetric multimer of the recognition factor.

If cosQ is recognized by symmetrically disposed subunits of a component of the translocation complex, how does one explain the types of local substitution suppressors? The suppression pattern is that certain mutations affecting the inner GC base pairs at 48,475 and 48,477 can be suppressed by mutations affecting base pairs 48,479 and 48,473, respectively (Fig 3). If cosQ interacts with symmetrically disposed subunits of a binding protein, then base pairs 48,475 and 48,477 would form equivalent contacts with the subunits. A reasonable but highly speculative model to explain the substitution suppression data can be constructed as follows. If we suppose that a mutation affecting base pairs 48,475 weakens the binding protein's cosQ interaction, then a suppressing change at 48,479 might compensate by strengthening the protein-cosQ interaction. The base pair of this example, 48,475 and 48,479, affect the proposed left and right cosQ half sites. Base pair 48,475 is rotationally symmetric with base pairs 48,477, and if the cosQ symmetry model is correct, it is puzzling that changes at 48,479 were not found as suppressors of mutations at 48,477. Similarly, mutations at 48,473 were found as suppressors of mutations at 48,477 but not as suppressors of mutations affecting 48,475. In each case, the suppressors are located in the cosQ half-site opposite the half-site containing the mutation that is suppressed. Why would the mutation at 48,479 be unable to suppress a mutation at 48,477? Suppose that the R-groups of the amino acids of a structural element of the cosQ-interacting protein, such as an -helix (or a ß-strand), were involved in making contacts with cosQ half-sites. The original mutation and its suppressor might alter the geometry of the cosQ-protein interaction. It is possible that altered R-group/base pair geometry could be accommodated if the two base pair changes were in opposite half-sites, but not if both cosQ changes were in the same half-site. This model, although highly speculative, rationalizes the substitution suppression data (Fig 3).

An alternative explanation for the puzzling suppression and methylation effects is that cosQ might be a DNA element with an unusual structure. Studies by ADELMAN et al. 2001 on herpes simplex virus DNA packaging (HSV-1) suggest that the formation of novel DNA structures in the pac1 motif, a cis-acting sequence required for DNA cleavage and packaging, confers added specificity to the recognition of DNA-packaging sequences by the U_L28 protein, a component of the HSV-1 cleavage and packaging machinery. They showed that these novel single-stranded DNA structures bind U_L28 protein with high affinity, whereas double-stranded DNA with the same sequence remains unbound. cosQ, like pac1, may exhibit a similar DNA structure. DIEKMANN and MCLAUGHLIN 1988 have suggested that regulation by methylation might occur as a result of the alteration of the conformation of the DNA within the binding domain of a protein. Thus, the methylation dependence of Methrev1 may be due to an additional DNA conformational perturbation brought on by methylation restoring the original DNA structure and resulting in suppression of the original mutation(s).

STERNBERG and COULBY 1988 , STERNBERG and COULBY 1990 found that cleavage of the bacteriophage P1 packaging site (pac) was regulated by adenine methylation of seven 5'-GATC-3' sites within the pac sequence. They hypothesized that replication temporarily generates unmethylated or hemi-methylated DNA that can be bound by the "pacase," the P1 equivalent of terminase, but cannot be cleaved. Cleavage is restored late during infection by the occasional methylation of a hemi-methylated pac site before the site is protected by pacase binding. Thus, the methylation of these sites within the pac sequence may alter the local DNA structure, as has been proposed for cosQ, to allow for DNA cleavage by the P1 pacase.

	ACKNOWLEDGMENTS

We thank our co-workers, Nicole Brogden, Alok Dhar, Carol Duffy, Sara Gaeth, Qi Hang, Jason Luke, Jenny Meyer, and Jean Sippy, for advice and interest during the course of this work. We also thank Shuang-yong Xu and New England Biolabs for generously providing clones of the EcoO109I methylase. This work was supported by National Institutes of Health (NIH) research grant GM-51611 (M.F.), Genetics Research Training grant T32GM08629 (D.W.), and the NIH Iowa Kidney Disease, Hypertension, and Cell Biology Research Training Grant DK07690-10 (D.W.).

Manuscript received January 15, 2003; Accepted for publication April 21, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADELMAN, K., B. SALMON, and J. D. BAINES, 2001  Herpes simplex virus DNA packaging sequences adopt novel structures that are specifically recognized by a component of the cleavage and packaging machinery. Proc. Natl. Acad. Sci. USA 98:3086-3091.[Abstract/Free Full Text]

ARBER, W., L. ENQUIST, B. HOHN, N. E. MURRAY and K. MURRAY, 1983 Experimental methods for use with lambda, pp. 433–466 in Lambda II, edited by R. W. HENDRIX, J. W. ROBERTS, F. W. STAHL and R. A. WEISBERG. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

BECKER, A. and H. MURIALDO, 1990  Bacteriophage DNA: the beginning of the end. J. Bacteriol. 172:2819-2824.[Free Full Text]

BODE, V. C. and F. D. GILLIN, 1971  The arrangement of DNA in lambda phage heads. I. Biological consequences of micrococcal nuclease attack on a portion of the chromosome exposed in tailless heads. J. Mol. Biol. 62:493-502.[Medline]

CATALANO, C. E., D. CUE, and M. FEISS, 1995  Virus DNA packaging: the strategy used by phage . Mol. Microbiol. 16:1075-1086.[Medline]

CUE, D. and M. FEISS, 1997  Genetic evidence that recognition of cosQ, the signal for termination of phage DNA packaging, depends on the extent of head filling. Genetics 147:7-17.[Abstract]

CUE, D. and M. FEISS, 1998  Termination of packaging of the bacteriophage chromosome: cosQ is required for nicking the bottom strand of cosN.. J. Mol. Biol. 280:11-29.[Medline]

DANIELS, D., J. SCHROEDER, W. SZYBALSKI, F. SANGER, A. COULSEN et al., 1983 Completed annotated lambda sequence, pp. 519–676 in Lambda II, edited by R. W. HENDRIX, J. W. ROBERTS, F. W. STAHL and R. A. WEISBERG. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

DIEKMANN, S. and L. W. MCLAUGHLIN, 1988  DNA curvature in native and modified EcoRI recognition sites and possible influence upon the endonuclease cleavage reaction. J. Mol. Biol. 202:823-834.[Medline]

EMMONS, S. W., 1974  Bacteriophage lambda derivatives carrying two copies of the cohesive end site. J. Mol. Biol. 83:511-525.[Medline]

FEISS, M., 1986  Terminase and the recognition, cutting, and packaging of chromosomes. Trends Genet. 2:100-104.

FUJISAWA, H. and P. HEARING, 1994  Structure, function, and specificity of the DNA packaging signals in double-stranded DNA viruses. Semin. Virol. 5:5-13.

HANG, J. Q., C. E. CATALANO, and M. FEISS, 2001  The functional asymmetry of cosN, the nicking site for bacteriophage packaging, is dependent on the terminase binding site, cosB.. Biochemistry 40:13370-13377.[Medline]

KITA, K., J. TSUDA, and R. NISHIGAKI, 2001  Characterization and overproduction of EcoO109I methyltransferase. Biosci. Biotechnol. Biochem. 65:2512-2518.[Medline]

MARUYAMA, M., T. HORIUCHI, H. MAKI, and M. SEKIGUCHI, 1983  A dominant (mutD5) and a recessive (dnaQ49) mutator of Escherichia coli.. J. Mol. Biol. 167:757-771.[Medline]

MISE, K. and K. NAKAJIMA, 1985  Restriction endonuclease EcoO109I from Escherichia coli H709c with heptanucleotide recognition site 5'-PuGGNCCPy. Gene 36:363-367.[Medline]

RAVIN, V., N. RAVIN, S. CASJENS, M. E. FORD, and G. F. HATFULL et al., 2000  Genomic sequence and analysis of the atypical temperate bacteriophage N15. J. Mol. Biol 299:53-73.[Medline]

SIX, E. and C. A. C. KLUG, 1973  Bacteriophage P4: a satellite virus depending on a helper such as prophage P2. Virology 51:327-344.[Medline]

SMITH, M. P. and M. FEISS, 1993  Sites and gene products involved in lambdoid phage DNA packaging. J. Bacteriol. 175:2393-2399.[Abstract/Free Full Text]

SMITH, D. E., S. J. TANS, S. B. SMITH, S. GRIMES, and D. L. ANDERSON et al., 2001  The bacteriophage 29 portal motor can package DNA against a large internal force. Nature 413:748-752.[Medline]

STERNBERG, N. and J. N. COULBY, 1988  Processing of the bacteriophage P1 packaging site (pac) is regulated by adenine methylation. Gene 74:203.[Medline]

STERNBERG, N. and J. COULBY, 1990  Cleavage of the bacteriophage P1 packaging site (pac) is regulated by adenine methylation. Proc. Natl. Acad. Sci. USA 87:8070-8074.[Abstract/Free Full Text]

THOMAS, M. and R. W. DAVIS, 1975  Studies on the cleavage of bacteriophage lambda DNA with EcoRI restriction endonuclease. J. Mol. Biol. 91:315-328.[Medline]

WIECZOREK, D. J. and M. FEISS, 2001  Defining cosQ, the site required for termination of bacteriophage DNA packaging. Genetics 158:495-506.[Abstract/Free Full Text]

WIECZOREK, D. J., L. DIDION, and M. FEISS, 2002  Alterations of the portal protein of bacteriophage suppress mutations in cosQ, the site required for termination of DNA packaging. Genetics 161:21-31.[Abstract/Free Full Text]