Capsid expansion follows the initiation of DNA packaging in bacteriophage T41

Article No. mb982179

J. Mol. Biol. (1998) 284, 661±672

Capsid Expansion Follows the Initiation of DNA Packaging in Bacteriophage T4 Paul J. Jardine and David H. Coombs* Department of Biology University of New Brunswick Fredericton, New Brunswick Canada, E3B 6E1

Most bacteriophages undergo a dramatic expansion of their capsids during morphogenesis. In phages lambda, T3, T7 and P22, it has been shown that expansion occurs during the packaging of DNA into the capsid. The terminase-DNA complex docks with the portal vertex of an unexpanded prohead and begins packaging. After some of the DNA has entered, the major head protein undergoes a conformational change that increases both the volume and stability of the capsid. In phage T4, the link between packaging and expansion has not been established. We explored the possibility of such a connection using a pulse-chase protocol and high resolution sucrose gradient analysis of capsid intermediates isolated from wild-type T4-infected cells. We show that the ®rst particle appearing after the pulse is an unexpanded prohead, which can be isolated in vitro as the ESP (empty small particle). The next intermediate to appear is also unexpanded, but contains DNA. This new intermediate, the ISP (initiated small particle), can also be isolated on agarose gels, permitting con®rmation of both its expansion state and DNA content (10 kbp). It appears, therefore, that 58% of the T4 genome enters the head shell prior to expansion. Following packaging of an undetermined amount of DNA, the capsid expands, producing the ILP (initiated large particle), which is ®nally converted to a full head upon the completion of packaging. An expanded, empty prohead, the ELP (empty large particle), was also observed during 37 C infections, but failed to mature to phage during the chase. Thus the ELP is unlikely to be an intermediate in normal head assembly. We conclude by suggesting that studies on assembly bene®t from an emphasis on the processes involved, rather than on the structural intermediates which accumulate if these processes are interrupted. # 1998 Academic Press

Keywords: Bacteriophage T4; capsid; morphogenesis; expansion; DNA packaging

*Corresponding author

Introduction The question of the packaging and capsid genesis is central to mechanisms involved

timing or coupling of DNA expansion in virus morphoour understanding of the in both processes. Since the

Present address: P. J. Jardine, Department of Oral Science, University of Minnesota, 18-246 Moos Tower, Minneapolis, MN 55455, USA Abbreviations used: ESP, empty small particle; ISP, initiated small particle; ILP, initiated large particle; ELP, empty large particle; gp, gene product; PI, postinfection; 9AA, 9-aminoacridine; MOI, multiplicity of infection. E-mail address of the corresponding author: [email protected] 0022±2836/98/480661±12 $30.00/0

early discovery that petite lambda, a small, rounded capsid precursor, could be packaged in vitro, yielding the larger, angular, ®lled capsid (Hohn & Hohn, 1974; Hohn et al., 1974; Kaiser et al., 1974), there have been numerous reports of the apparent coupling of expansion with DNA packaging during phage morphogenesis (see Earnshaw & Casjens, 1980 for a review). For phage lambda, gene A terminase mutants produced a petite lambda particle indistinguishable from that seen in wild-type infections, while Dÿ mutants in the capsid stabilizing protein yielded expanded, partially ®lled capsids. When the packaging ef®ciency of these particles was compared in vitro, the unexpanded petites had at least ten times more activity than the expanded ones (Hohn et al., 1975). Finally, Hohn (1983) convincingly demonstrated the link # 1998 Academic Press

662 between packaging and expansion when she showed that packaging a 5.5 kb (11%) left end restriction fragment of lambda DNA failed to expand petite lambda during an in vitro packaging reaction, while a 22.2 kb (44%) left end fragment did. The re®ned phage T3 in vitro packaging system (Hamada et al., 1986; Shibata et al., 1987) has permitted higher resolution of the coupling of packaging and expansion. In the presence of g-S-ATP, the terminase-DNA-procapsid complex forms, but fails to package. Upon the addition of ATP to the reaction, packaging begins at a linear rate. One minute, 40 seconds later, after about 25% of the chromosome has been packaged, the capsid expands (Shibata et al., 1987). If g-S-ATP is added to the reaction after expansion has occurred, the chromosome exits until 30% remains inside, apparently in a stable interaction with the inner surface of the capsid. The addition of more ATP drives the chromosome back into the capsid, demonstrating that expansion is coupled with, but not required, for DNA packaging. Evidence from phages T7 and P22 supports this view. Masker & Serwer (1982) puri®ed unexpanded and expanded empty T7 capsids and showed that only the unexpanded one was packagable in vitro. Poteete et al. (1979) assayed expanded and unexpanded P22 capsids puri®ed on a gradient with similar results. The consensus reached with these phage is that the unexpanded prehead is the substrate for the terminase-DNA packaging complex, and that expansion occurs after a signi®cant fraction of the chromosome has entered the capsid. The relationship between packaging and expansion in T4 is more tenuous. In vitro studies of the packaging reaction (Black, 1981; Rao & Black, 1985, 1988) have been hampered by low ef®ciency (10ÿ4 to 10ÿ5), but in these studies, DEAE-puri®ed expanded capsids are better packaging substrates than unexpanded ones, raising the possibility that expansion is not coupled to DNA packaging, and might even precede it in vivo. Thus, we examined the relationship between expansion and packaging in T4 in two stages. In the accompanying paper (Jardine et al., 1998), we demonstrated that the unexpanded capsid is the substrate for the packaging reaction in vivo. Here, we extend this result using a pulse-chase protocol similar to that of Laemmli & Favre (1973), but with two important new analytical tools: highresolution density gradient fractionation and agarose gel electrophoresis of whole capsids. We report for the ®rst time the isolation from wildtype infected cells of a transient capsid intermediate that contains DNA but has not yet expanded. This demonstrates that in all phage systems where it is observed, expansion is coupled to DNA packaging.

Packaging Precedes Expansion in T4

The role of T4 packaging enzymes As with most other phages, T4 codes for a large and a small terminase subunit which attack the concatemeric DNA substrate and guide it to the portal vertex of the capsid. Gene product 17 (gp17), the large subunit, has a single-stranded DNA binding domain that binds to random, transcriptionally active sites in T4 DNA to initiate packaging, as well as two ATPase domains that are presumably active in the ATP-dependent packaging reaction (for a recent review, see Franklin et al., 1998). Gp17 also possesses the nuclease activity required to cut the chromosome from the concatemer at the beginning and end of the packaging reaction. There is genetic evidence for contact between gp17 and the portal protein, gp20 (Hsiao & Black, 1977). The small subunit, gp16, plays an accessory role since some DNA is packaged in its absence (Wunderli-Allenspach, 1977). Recent evidence shows that it promotes site-speci®c recombination between sequences in genes 16 and 19, so it may act as a pac site recognition factor during packaging (Wu & Black, 1995). Mutants in genes 16 and 17 produce a mixture of unexpanded empty (ESP) and expanded empty (ELP) capsids at 37 C, while only ESPs are produced during infection at 20 C (Rao & Black, 1985; Carrascosa & Kellenberger, 1978; Hsiao & Black, 1977; Jardine et al., 1998). Another key protein in the packaging reaction is gp49 (endonuclease VII), which is required to remove the branches created by a DNA replication mechanism that is driven by invasive recombination (Minagawa & Ryo, 1978; Mizuuchi et al., 1982; Mosig, 1994). Packaging stalls in 49ÿ mutants presumably because the branches cannot enter the capsid, leaving the large DNA concatemer decorated with partially ®lled heads (Kemper & Brown, 1976). Despite the block, however, virtually all capsids are expanded (Jardine et al., 1998).

Results Pulse-chase experiments on 10ÿ  49ÿ Laemmli & Favre (1973) investigated the assembly of the T4 head using pulse-chase protocols and rate-zonal gradient analysis of head assembly intermediates. They identi®ed four major stages in head assembly: formation of the core-containing prehead (400 S); cleavage of the core yielding an essentially empty prohead capsid (350 S); packaging of DNA into this empty, preformed shell yielding intermediate, partially ®lled (550 S) capsids; and, ®nally, completed full heads (1000 S). While they were aware of the change in size of the capsid during morphogenesis, the full details of prohead expansion and its possible involvement in packaging were not discussed. We have returned to pulse-chase gradient analysis to investigate the timing of packaging and

Packaging Precedes Expansion in T4

expansion left unresolved by Laemmli & Favre, but are now equipped with the knowledge that the terminase-DNA complex targets the unexpanded prohead (Jardine et al., 1998), and a new high resolution gradient fractionator capable of resolving 15 S differences in the 285± 340 S capsid region of gradients. If expansion occurs prior to entry of DNA into the head, then the unexpanded prohead (the ESP) should be the ®rst intermediate to appear after the pulse, followed by the empty, expanded ELP and, ®nally, by the partially ®lled, expanded intermediate (the ILP; initiated large particle). Conversely, if expansion occurs only after some DNA has been packaged into the head, then an unexpanded, partially ®lled intermediate will appear in the sequence and the ELP will not. Our ®rst choice for analysis was the 49ÿ mutant which progresses through all the early stages of capsid morphogenesis, stalling on an unresolved branch in the DNA after a signi®cant portion of the chromosome has been packaged and expansion has occurred (Luftig & Ganz, 1972; Laemmli & Favre, 1973; Wagner & Laemmli, 1976; Kalinski & Black, 1986; Jardine et al., 1998). Although 49ÿ does not produce ®lled heads, it also fails to accumulate ELPs during infection (Jardine et al., 1998). If the ELPs are transient intermediates, then they might be detectable in gradients following a pulse-chase. The 49ÿ mutation was combined with a 10ÿ mutation (am10-am49): the 10ÿ mutation blocks tail assembly, therefore halting the production of tailed, empty capsids (ghosts) which would confound the interpretation of results. Cells were infected and super-infected with 10ÿ  49ÿ phage at 37 C and then transferred to a 20 C water bath at nine minutes post-infection (PI), after the onset of late protein synthesis. The low infection temperature (20 C) is permissive for wild-type head assembly and slows the ¯ow of label through the intracellular pool of viable intermediates, thus facilitating their detection. At 28 minutes PI, cultures were pulsed with [35S]methionine for two minutes followed by an unlabeled methionine chase to halt further incorporation of the radioisotope. Samples were taken at intervals and analyzed using high resolution gradient analysis. ESPs are the ®rst particle to appear in sucrose gradient pro®les at two minutes post-chase (Figure 1), migrating at 42 mm (300 S) in the calibrated gradients. They reach their maximum between ®ve and ten minutes, and then recede from the pro®le. DNA-containing ILPs, the end point of 49ÿ infection, begin to accumulate at ®ve minutes post-chase and migrate below ESPs (Figure 1). Note the absence of detectable ELPs (at 40 mm, 285 S) in all samples. Closer examination of the 45 ±50 mm region of the gradients reveals two peaks. At ®ve minutes post-chase, a small peak can be seen at 48 mm (340 S) in the gradients (Figure 1). A second peak, corresponding to the 325 S ILP , appears shortly thereafter, migrating at 46 mm (Figure 1, ten min-

663

Figure 1. 35S gradient pro®les of am10-am49 infections at 20 C which were pulsed at 28 minutes PI for two minutes and then chased. Samples were taken at 0, 2, 5, 10, 20 and 40 minutes after the chase, respectively. All gradients are at the same scale and were aligned on the x-axis using co-sedimented 3H-labeled capsids from am10-am13 as an internal standard (not shown).

utes). By 20 minutes, the ESP peak has declined and the ILP and 48 mm peaks continue to increase, suggesting that the ESPs are precursors to these faster particles. The ultimate fate of the 48 mm peak cannot be determined since it is eventually overwhelmed by the accumulating ILP peak. The ILP peak gradually becomes broader and sedi-

664

Packaging Precedes Expansion in T4

ments faster as the amount of DNA in these heads increases in variability and size (Jardine et al., 1998). Since 49ÿ accumulates mainly ILPs, this mutant has been useful in limiting the types of structures which appear in gradients. It is possible, however, that the absence of ELPs and the presence of this new 48 mm peak are artifacts of the 49ÿ phenotype. Further, since ILPs accumulate in 49ÿ, the ultimate fate of the new 48 mm particle is obscured. To address these concerns, pulse-chase experiments were conducted using a tail-®berless phage mutant which provided a wild-type phenotype for head assembly and completion. Pulse-chase experiments with wild-type phage Pulse-chase experiments on wild-type phage provide two important advantages over those described above. Firstly, the ILP is an intermediate in wild-type infection rather than a ®nal product, so it should not accumulate and obscure the 48 mm intermediate. Secondly, the appearance of the 48 mm intermediate in wild-type would eliminate the possibility that this particle is an artifact of 49ÿ infection. An am34-am37 tail ®berless mutant permits normal head and tail assembly and joining. The absence of tail ®bers eliminates the possibility that progeny phage will infect cell debris after lysis, producing ghosted phage which would sediment in the capsid area of the gradient. Pulse-chase experiments were ®rst conducted on am34-am37 phage at 20 C (Figure 2). Once again, the low temperature is intended to slow the process of head assembly so that the assembly intermediates might be more easily observed. Gradient pro®les of pulse-chase samples reveal that the ®rst particle to appear is the 300 S ESP, migrating at 42 mm (Figure 2). Between seven and ten minutes after the chase, a 340 S peak at 48 mm and then 325 S ILPs at 46 mm begin to appear. By 15 minutes post-chase, the ESPs begin to recede, followed by a reduction in the 48 mm peak after 25 minutes. ILPs begin to recede by 40 minutes as they are converted to completed heads and phage which sediment to the bottom of the gradient. As with the am10-am49 experiments, signi®cant numbers of ELPs do not appear. Phage assembly is thought to be normal at 20 C (Yanagida et al., 1984). It is nevertheless possible that low temperature can in¯uence the timing of packaging or expansion. To eliminate temperature as a variable, am34-am37 pulse-chase experiments were conducted at 32 C. Although the events of assembly occur much faster at higher temperature, a clearer precursor/product relationship was observed between ESP ! 48 mm peak ! ILP (Figure 3). The higher temperature also gave improved separation of the ILP from the 48 mm peak; the rise and fall of the 48 mm peak can now be seen to precede that of the ILPs. The order of appearance and subsequent disappearance of the

Figure 2. 35S gradient pro®les of am34-am37 infections at 20 C which were pulsed at 28 minutes PI for two minutes and then chased. Samples were taken at 4, 7, 10, 15, 25 and 40 minutes, respectively. All gradients are scaled and aligned as in Figure 1.

prohead intermediates re¯ected by the changes in peak maxima between samples is: the ESP, then the 48 mm peak and ®nally the ILP (Figure 4). This indicates that the 48 mm peak is an assembly intermediate between the ESP and the ILP. Unlike the previous 20 C experiments, ELPs appear almost immediately during pulse-chase of am34-am37 at 32 C. These particles persist throughout the sampling regime and never chase into

665

Packaging Precedes Expansion in T4

Figure 4. Kinetics of capsid intermediates from am34am37 pulse-chased at 32 C. Maximum peak heights from pulse-chase gradient pro®les (Figure 3) plotted against time reveal the order of appearance and disappearance of head intermediates: ESPs appear ®rst (peaking at 1 minute post-chase), followed by the 48 mm peak (peaking between two and three minutes), and ®nally ILPs (peaking between three and four minutes). ELPs appear to accumulate over time.

Expansion state of head assembly intermediates

Figure 3. 35S gradient pro®les of am34-am37 infections at 32 C which were pulsed at 19 minutes PI for one minute and then chased. Samples were taken at 0, 1, 2, 3, 4 and 5 minutes respectively. All gradients are scaled and aligned as in Figure 1.

phage (Figures 3 and 4), suggesting that ELPs are abortive structures. We then sought to determine the expansion state and DNA content of the particles in the 48 mm peak. Understanding the structure of this new intermediate is crucial to answering the question of when expansion occurs during DNA packaging in phage T4.

Thus far, we have observed four particles in these gradients. In order of sedimentation, they are the ELP, the ESP, the ILP and the new 48 mm intermediate. The expansion state of these particles was evaluated ®rst using a chymotrypsin sensitivity assay (Jardine et al., 1998). Unexpanded heads have a chymotrypsin-sensitive site, Phe154-Ser155, in gp23* which is inaccessible in expanded heads. Quanti®cation of the resultant 38 kDa peptide gives an indication of the expansion state of a capsid sample. Samples from a pulse-chased culture of an am34-am37 infection were collected at two minutes post-chase at 32 C, isolated by sucrose gradient sedimentation and the pooled peak fractions were subjected to the chymotrypsin assay. Both ELP and ILP appear to be refractile to chymotrypsin, con®rming that they are expanded, whereas both the ESP and 48 mm peak show a markedly higher sensitivity to proteolysis, indicating that they are unexpanded (Figure 5). The unexpanded 48 mm particle appears after ESPs in pulse-chase (Figures 1, 2 and 3), so it is probably derived from the ESP. Since both particles are unexpanded, the simplest explanation for the faster sedimentation rate of the 48 mm particle is that it contains DNA. Attempts to label the DNA in this particle with [3H]thymidine failed because of poor label uptake, prompting the use of an

666

Figure 5. Relative chymotrypsin sensitivity of am34am37 35S-labeled particles generated by pulse-chase. The sample was prepared from a 32 C infection at two minutes post-chase as in Figure 3. Fractions through the capsid banding area were pooled and assayed for chymotrypsin sensitivity (*) and plotted against the relative speci®c activity in the gradient pro®le (*).

alternative method of demonstrating both the expansion state and DNA content of this particle: whole particle agarose gel electrophoresis. Agarose gel electrophoresis and separation of whole T4 capsids Serwer & Pichler (1978) had demonstrated that agarose gel separation of whole phage T7 head structures could resolve particles according to their expansion state. They demonstrated that migration through an agarose matrix in an electric ®eld was dependent not on the internal content of the capsid (i.e. core or DNA), but on its size and possibly net surface charge. To test whether phage T4 capsid structures were amenable to such separation, a variety of head structures of differing expansion state and DNA content were isolated on 0.8% agarose gels. Figure 6(a) shows the relative migration of puri®ed heads in agarose by visualization with both Coomassie staining (top gel, shows protein) and ethidium bromide staining (lower gel, shows the presence of DNA). All head particles which are known to be expanded migrate at the same position in the gels, regardless of their DNA content. This includes expanded capsids isolated from am10-am13 mutants which have been ®lled with DNA but have retained only a small fraction of it (Figure 6(a), lane a), a mixture of expanded ELPs and ILPs isolated from high temperature am10am16 infection (Figure 6(a), lane c), and empty

Packaging Precedes Expansion in T4

Figure 6. Expansion state and DNA content of gradient puri®ed capsids analyzed by agarose gel electrophoresis of whole particles. (a) Capsids from 37 C infection of am10-am13 (lane a), am10-am16 (lane c), and am10-am17 (lane d) and from 20 C infection of am10am16 (lane b), run in 0.8% agarose gel buffered with 100 mM Tris and stained with Coomassie (top panel) and ethidium (bottom panel). All expanded particles, including ILPs (lanes a and c) and ELPs (lanes c and d) migrate slower than unexpanded ESPs (lanes b± d). DNA in ILPs is visible by ethidium staining (lower panel). (b) Control lanes include am10-am13 ILPs (lane a) and am10-am16 ESPs (lane b) run in 0.8% agarose buffered with TAE and stained with Coomassie (top panel) and ethidium (bottom panel). ESPs from am10-am49 (see the text) migrate with control ESPs and contain no DNA (lane c), whereas both expanded ILPs and unexpanded particles from 48 mm contain DNA (lane d).

ELPs produced by high temperature infection of am10-am17 (Figure 6(a), lane d). Unexpanded particles migrate faster in agarose: ESPs from both low and high temperature am10-am16 infection (Figure 6(a), lanes b and c, respectively) and from high temperature am10am17 infection (Figure 6(a), lane d) all migrate at the same rate. These results show that, regardless of the DNA content, agarose gel separation of head particles is entirely dependent on expansion state. If the 48 mm peak seen in the pulse-chase experiments is unexpanded and contains DNA, then agarose gel isolation of these particles will show a band migrating at the unexpanded rate that contains DNA detectable by ethidium bromide staining.

667

Packaging Precedes Expansion in T4

Discussion Expansion follows DNA packaging

Figure 7. DNA isolated from capsids recovered from agarose gel separation. Gradient puri®ed capsids were isolated on agarose gels, eluted from agarose, and their DNA extracted and run on 0.3% agarose gel. DNA from am10-am13 (lane c) is slightly shorter than DNA from am10-am49 ILPs (lane d) and ISPs (lane e). Control lanes include lambda HindIII partial digest (lanes a, f) and T4 phage DNA (lane b).

To obtain suf®cient quantities of 48 mm particles for gel analysis, am10-am49 extracts were prepared at 45 minutes PI at 20 C. The low temperature and short duration of infection were intended to slow the assembly process and provide 48 mm particles which had not yet been dominated by accumulating ILPs. The gradient pro®le of such a preparation revealed only two large peaks, one at 41 mm, the ESP banding area, and a broader peak spanning the banding position of the 48 mm peak and ILP (data not shown). The Coomassie-stained agarose gel of these pooled peaks (Figure 6(b), upper gel) showed the presence of unexpanded particles in the ESP section of the capsid region (40 ± 42 mm; Figure 6(b), lane c) and a mixture of unexpanded and expanded particles sedimenting in the ILP48 mm region (44 ± 48 mm; Figure 6(b), lane d). Ethidium staining (Figure 6(b), lower gel) revealed that the unexpanded particles sedimenting with the ILPs in these gradients also contain DNA (Figure 6b, lane d). These results con®rm the existence of an unexpanded, partially ®lled intermediate. We designate this particle the 340 S ISP (initiated small particle). DNA extracted from the ILP and ISP region of agarose gels reveals that the fragment length of the DNA in ISPs, 10 kbp, is the same as that retained in ILPs (Figure 7, lanes d and e, respectively). DNase treatment during preparation of extracts prior to gradient puri®cation, and the uniform length of the isolated DNA effectively excludes the possibility that this DNA is somehow associated with the outside of the head shell. Since no steps were taken to retain DNA in the capsid, this length is a minimal estimate of the amount that enters the head prior to expansion.

Our discovery of an unexpanded, partially ®lled capsid intermediate in wild-type T4 infection accomplishes two things: it aligns T4 capsid morphogenesis with that of the other tailed phages, and it extends previous in vitro work to the infected cell. There is now a general consensus that the unexpanded prehead is targeted at the portal vertex by a terminase/concatemeric DNA complex and that packaging commences before expansion occurs. Two improvements which facilitated our study were the use of physiological glutamate-based buffer in extract preparation, which stabilizes intermediate structures (Kuhn & Kellenberger, 1985), and the improved resolution of intermediates offered by the recently developed piston gradient fractionator (D.H.C. unpublished results). We probed the link between packaging and expansion using the same pulse-chase strategy used by Laemmli & Favre (1973) to follow T4 capsid morphogenesis. There is an important conceptual quali®er concerning the intermediate particles we observed. During the sampling and sedimentation procedures, no steps were taken following cell lysis to prevent continued proteolytic maturation of the capsid proteins or release of packaged DNA. Therefore, the band of ESPs isolated on gradients could represent capsids which, at the time of lysis, ranged from preheads just beginning proteolysis to those which are entirely cleaved. Likewise, the DNA remaining in the capsid is the minimum amount packaged which is stable under the conditions of isolation. Thus, the ESP contains no DNA but could theoretically have packaged up to 10 kbp (the amount found in ISPs) before sampling. The still unexpanded ISP contains 10 kbp but could have packaged a larger fraction of the chromosome, up to the amount needed to induce expansion. The ILP also contains 10 kbp, but since no intermediates with larger packaged chromosomes are seen on gradients even though they must exist, we speculate that they represent the fall-back of a full spectrum of packaging intermediates ranging from 10 kbp to full size (166 kbp). This nearly complete emptying of even full heads to a discrete fragment size is also observed in gene 13 mutants (Coombs & Eiserling, 1977; Hamilton & Luftig, 1972; Jardine et al., 1998). That all these possible combinations of expansion state and amount of DNA packaged reduce to only three discrete species, the ESP, ISP and ILP, is an obvious bene®t experimentally, but one must not forget their origins when interpreting the results of the pulse-chase protocol. If, as we suspect, the coalescence of a range of particles into an observable few is an artifact of isolation, then typical assembly pathways showing a series of intermediates that are transformed by processes such as cleavage, expansion or packaging,

668 fail to re¯ect the potential interplay or overlap between these processes. There is some evidence in T4, for example, that the target of the DNA packaging complex is not only unexpanded, but uncleaved as well (Wunderli-Allenspach, 1977). Our experiments could not have resolved such an intermediate if, for example, the maturation cleavages were rapid and less than 10 kbp of DNA was packaged, since these particles would still appear in our gradients as cleaved, unexpanded ESPs. An alternative way of presenting the T4 head morphogenesis pathway is to describe it as a series of maturation processes from which stable, but not necessarily natural, intermediates are derived (Figure 8). In the absence of detergent during cell lysis to release the membrane-bound preheads, the ®rst intermediate to appear in gradients during the wild-type pulse-chase is the cytoplasmic, cleaved, unexpanded prohead, the ESP. Next, the terminase-DNA complex docks and packaging commences, eventually yielding the unexpanded ISP, continuing later into the expanded ILP. Since both the ISP and ILP contain the same amount of DNA, we cannot identify the amount of DNA packaged at the moment of expansion. It is clear, however, that some DNA enters the capsid before expansion occurs and that packaging and expansion overlap. The ILPs are the last particles to appear in our pulse-chase experiments. They accumulate in vivo behind an am49 block in packaging at the DNAsubstrate level (Figure 1). When a ts49 mutant is shifted to the permissive temperature, the ILPs can be chased into completed phage (Luftig & Ganz, 1972). In wild-type infection, ILPs are transient intermediates which chase into ®nished heads with a half-life of two minutes at 32 C (data not shown) and so are derived from true intermediates in the head pathway. In the in vitro phage T3 packaging system, there is no stable equivalent to the ISP. Packaging of 25% of the chromosome leads to expansion, and to the stabilization of the DNA-capsid complex against a high salt challenge (Shibata et al., 1987). If g-S-ATP is added to the packaging reaction after

Packaging Precedes Expansion in T4

expansion, DNA exits the expanded capsid until 30% (13 kbp) remains inside, perhaps in a stable interaction with the inner surface of the capsid. This capsid resembles our ILP. As T4 packaging comes to completion, the packaging apparatus must somehow sense that the head is full in order release the capsid from the concatemer. This so-called headful termination reaction leads to the circular permutation of the chromosome (Streisinger et al., 1967) and circularization of the genetic map (Streisinger et al., 1964) since the prolate capsid can hold 102% of the genome (Black et al., 1994). Casjens et al. (1992) have shown that mutants in the phage P22 portal protein in¯uence the amount of DNA packaged, so the portal protein and the terminase subunits may collaborate on the termination of packaging. Following DNA cleavage, the concatemer and the packaging complex must depart so that gp13 and gp14 can assemble and make the head ready to accept the tail (Coombs & Eiserling, 1977). Identifying the unexpanded capsid which serves as the packaging substrate An issue left unresolved by this work is the precise nature of the unexpanded capsid targeted by the terminase-DNA complex. There are two basic types of unexpanded capsids observed during T4 infection. The ®rst is the membrane-bound prehead or tau particle which accumulates in mutants of the prehead proteinase (gene 21) or the penton protein (gene 24; see Black et al., 1994). Since it is anchored to the membrane by the portal vertex, the prehead almost certainly is not the DNA packaging substrate. The second observable unexpanded capsid is the ESP, which has experienced all the morphogenetic cleavages and is devoid of a scaffolding core. ESPs are not all derived from the same capsid intermediate, however, since some can be chased into ®lled heads while others cannot. The ESPs we observed during wild-type infection are derived from a chasable intermediate, as are those which accumulate when cs mutants in gene 20, the portal protein, are

Figure 8. The order and interaction of maturation events in T4 head morphogenesis. Three major events of T4 head maturation are presented as processes in their order of appearance. Interruption of maturation yields a variety of stable particles (derived intermediates). The origin of ELPs is uncertain (gray arrows).

669

Packaging Precedes Expansion in T4

grown at 20 C (Hsiao & Black, 1977), and those produced when wild-type phage are grown in the presence of 9-aminoacridine (9AA; Schaerli & Kellenberger, 1980). A second type of capsid called the epsilon particle is also produced with 9AA. The main difference between the ESP and the epsilon is the presence of staining material inside the latter, which was tentatively identi®ed as protein. In light of our ®ndings, the epsilon might be the partially ®lled ISP, with the staining material being DNA and/or protein. Once the 9AA block was removed, both particles chased to ®lled heads. These ESPs contrast in unknown ways with the unmaturable ESPs which accumulate during 16ÿ and 17ÿ infection in either ts or am mutants. In the ts case, a return to the permissive temperature fails to chase the ESPs already assembled (Black & Showe, 1983; Luftig & Ganz, 1972), and with the am mutants, ESPs assembled early fail to convert to ELPs during prolonged infection at 37 C (Jardine et al., 1998). By the time any of these maturable or unmaturable ESPs have been puri®ed, they are biochemically indistinguishable. Extrapolating backwards, however, one sees that when the terminase-DNA complex is intact (wt, cs20 and 9AA), the capsid remains maturable for extended periods, and when it is not (16ÿ and 17ÿ), the packaging window rapidly decays. A possible explanation is that the complex must dock and initiate packaging before a ``ripening'' process, such as the protease maturation cleavages, progresses too far. When packaging is blocked at the initiation stage (cs20) or after some DNA has entered the capsid (9AA) (Kellenberger, 1980; Schaerli & Kellenberger, 1980; Wagner & Laemmli, 1979), the presence of the complex at the portal vertex may hold the window open. The origin of ELPs During wild-type pulse-chase experiments, ELPs fail to appear at 20 C and remain as a minor, unchasable species when they do appear at 33 C. The available data suggest that they arise when capsids in the act of packaging expand but subsequently lose all their DNA (Figure 8). We reached a similar conclusion in pulse-chase experiments of 16ÿ and 17ÿ-infected cells in the accompanying paper (Jardine et al., 1998), where we found evidence for abortive packaging that accompanied the generation of ELPs at 37 C. Both of these results bear on the packaging ``window'' described above. In the case of the 16ÿ and 17ÿ mutants grown at 37 C and pulse labeled early (12 minutes), the capsids pass through the window without experiencing packaging and are packaging incompetent thereafter. When they are pulsed at late times (35 minutes), a fraction are quickly converted to ELPs (®ve minutes postpulse). Surprisingly, the fraction remains constant for another 35 minutes, indicating that there is no further conversion of ESPs to ELPs. We conclude

that the expansion of ESPs to produce ELPs in either the 16ÿ and 17ÿ mutant or in wild-type infection can only occur in the ®rst few minutes following prehead completion, and that this represents the window of packagability. The nature of expansion and its trigger Expansion is a variation on the process of capsid completion found in most viruses that seals off the interior, angularizes the rounded shell, greatly stabilizes subunit interactions against denaturation, and blocks external proteolytic attack. With the exception of phage ù29, all the tailed phages studied thus far expand during DNA packaging. Although there is no expansion in ù29, the capsid becomes angularized when one third of the DNA has been packaged (Bjornsti et al., 1983). Recent work with in vitro assembly of Herpes virus (Newcomb et al., 1996; Trus et al., 1996) shows that the procapsid also angularizes without expanding during maturation. The lambdoid phage HK97 extends the metaphor by chemically cross-linking all the capsid subunits into a rugged shell after expansion (Duda, 1998; Duda et al., 1995). Thus, expansion is part of a general phenomenon of the molecular rearrangement of the capsid subunits following shell formation. From micrographs showing a wave of expansion moving down the axis of tubular aberrant T4 head structures called polyheads (Kellenberger, 1980; Steven & Carrascosa, 1979; Steven et al., 1976), it is apparent that expansion is a cooperative process. This has led many to speculate that the expansion has a speci®c starting point, likely the portal vertex, from which it propagates as a seamless wave to the opposite end of the capsid. We will refer to this as the ``wave model''. The discovery of P22 mutants in the portal protein that in¯uenced the amount of DNA packaged in the capsid (Casjens et al., 1992) suggests that the portal is sensitive to the amount of DNA packaged, and could conceivably provide the trigger for expansion. An alternative to the wave model is the ``mosaic model'' in which expansion is triggered at multiple sites by the physical interaction of DNA with the inner surface of the capsid. These multiple origins of expansion would eventually fuse into a completely expanded capsid. The primary stimulus for this model is the universal observation that a substantial fraction of the chromosome is packaged before expansion (or angularization) occurs. The size of the fraction roughly corresponds to the amount of DNA needed to coat the inner surface of the capsid with DNA (Bjornsti et al., 1983; Earnshaw & Casjens, 1980; Hohn, 1983; Shibata et al., 1987). It should be possible to distinguish between the two models by isolating partially expanded capsids on gradients and probing them with proteases, detergents and antibodies. For example, treatment of partially expanded T4 capsids with SDS predicts cup-shaped intermediates

670 with the wave model and fragments with the mosaic model. In closing, a direct link between expansion and DNA packaging has been con®rmed for all the tailed phages. There has been a great deal of work on both of these complex molecular reactions as separate phenomena. We must now focus on the mechanism that links them.

Materials and Methods Bacterial and phage strains Escherichia coli CR63 was used as a suppressive host for all amber mutants. S/6 and Be were used as wildtype hosts for platings and liquid culture, respectively. Phage strains included am10-am16 (B255-N88), am10am13 (B255-E609), am10-am17 (B255-N56), am10-am49 (B255-E727.1) and am34-am37 (A455-N52). Preparation of phage structures Be was grown at 37 C to 4  108 cells/ml in M9-14aa (50 mg/l of Arg, Asn, Asp, Glu, Gln, Gly, His, Lys, Phe, Pro, Ser, Trp, Tyr and 5 mg/l Cys) for production of both unlabeled and [35S]methionine or [3H]leucinelabeled phage parts. Cells were chilled and concentrated by centrifugation and resuspended in chilled, fresh media. Cells at 2  109 cells/ml were readied for infection by absorption for ®ve minutes on ice with phage at a multiplicity of infection (MOI) of ®ve. Infection was initiated by transferring the culture to 37 C with vigorous aeration. The cultures were superinfected at ®ve minutes PI at an MOI of ®ve. For 20 C infections, cultures were initially infected and superinfected at 37 C and then cooled to 20 C at nine minutes PI. Cultures were treated with DNase (50 mg/ml; Boehringer) and harvested by centrifugation. Pellets were resuspended in chilled G-buffer (250 mM K-glutamate, 1 mM MgSO4, 0.1 mM CaCl2, pH 7.5) with 1 mg/ml of DNase and lysed with chloroform. To prevent proteolysis in the extracts, PMSF (Sigma) was added from a 100 mM stock solution in isopropanol to a ®nal concentration of 3 mM. Cell debris was cleared by three minutes centrifugation at 10,000 g. Where possible, [3H]leucine-labeled empty capsids from am10-am13 infections were included as markers in the gradients. Linear sucrose gradients (5 ± 45% w/v) buffered with G-buffer were formed by tilt tube rotation at 81.5 using a Gradient Master (BioComp Instruments, Fredericton, NB) for one minute 16 seconds at 21 rpm (Coombs & Watts, 1985). Extracts were layered onto gradients and were run for 3900 o2t  105) in an SW41 rotor (Beckman) at 20 C. Gradients were fractionated by piston displacement from the top down (Coombs, 1975) using the BioComp Piston Gradient Fractionator. The pro®les were aligned using a 3H-labeled, ILP internal marker isolated from 13ÿ-infected cells which was added to each sample prior to sedimentation. Gp13 is a head completion protein which helps retain the DNA. In its absence, most capsids lose all but about 10 kbp of their DNA. Pulse-chase experiments Following infection and superinfection of cells, the 20 C cultures were pulsed at 28 minutes for two minutes, while 32 C infections were pulsed at 19 minutes for

Packaging Precedes Expansion in T4 one minute with 50 ± 100 mCi/ml culture of [35S]methionine (NEN, DuPont). Labeling was terminated (chased) by the addition of unlabeled methionine to a ®nal concentration of 1 mM. Samples taken are reported in the text as time after this chase (post-chase). Sampling involved the removal of 1 ml of culture and pelleting infected cells in a microfuge for 15 seconds at 13,000 rpm. Cells were resuspended in 100 ml ice-cold G-buffer containing 3 mM PMSF by vortexing and lysed with chloroform. This sampling procedure takes <60 seconds total. Extracts were stored on ice prior to sucrose gradient analysis. S-value calibration of sucrose gradients Sucrose gradients (5 ± 45%, w/v) in G-buffer were formed in SW41 centrifuge tubes as described above and overlaid with 0.1 ml of ®berless T4 phage. AksiyoteBenbasat & Bloom®eld (1974) calculated an s-value for the same ®berless phage we used at 970 S, while Gordon (1972) measured the s-value of T2 ®bered phage (®bers up) at 1107 S. Our calculations use an average value of 1000 S. Gradients were run for various o2t values between 200 and 2000 (105) using the integrator on the L75 ultracentrifuge (Beckman). The distance from the meniscus to the top of the visible phage band was recorded using the piston fractionator's band illumination system and plotted versus its o2t value. Nearly linear o2t values from 0 ± 1600 (0 ± 52 mm) were ®tted with a second order polynomial curve by Cricket Graph (Computer Associates International Inc., Islandia, NY), giving the formula o2t ˆ 0.031x2 ‡ 27.211x ‡ 5.852 (r2 ˆ 0.998). To determine an unknown particle's s-value, its position (x mm) and the o2t value for the run were noted. The o2t value for T4 phage at this same position was calculated using the formula above and the particle's s-value calculated using the proportion: o2 t…particle†  s…particle† ˆ o2 t…T4†  1000 S solving for s…particle† We have not compensated for the higher density of the G-buffered gradients and the lower density of the capsids compared with the phage standard, so our calculated s-values will be slightly lower than actual sw,20 values. However, the relative s-values are accurate. Thus a particle sedimenting at 285 S sediments at 95% of the rate of a 300 S particle. Protease-expansion assay of capsids Phage structures freshly puri®ed from gradients were proteolytically cleaved by incubating with 62.5 mg/ml chymotrypsin (Sigma) in G-buffer at room temperature for 15 minutes as described previously (Jardine et al., 1998). Cleavage was stopped by adding PMSF to ®nal concentration of 3 mM followed by TCA a to ®nal concentration of 10% (w/v). Samples were chilled and concentrated for electrophoresis by TCA precipitation, then resuspended and electrophoresed in 10% SDSPAGE gels (Laemmli, 1970). For quanti®cation of cleavage of gp23* to 38 kDa, gels were transblotted to nitrocellulose (S&S). Blots were autoradiographed to visualize bands, and strips corresponding to gp23* and 38 kDa in a given lane were excised. Strips were dissolved in Ready Protein cocktail (Beckman) and counted. The percentage of gp23* cleaved was determined by dividing the 38 kDa dpms by the sum of (gp23* ‡ 38 kDa) as determined by [35S]methionine counts, adjusting for 14

Packaging Precedes Expansion in T4 methionine residues per copy gp23* and 9 methionine residues in 38K (Parker et al., 1984). Agarose separation of capsid particles Whole capsids were separated on agarose slab gels according to Serwer & Pichler (1978). Gels were buffered in 1  TAE (40 mM Tris-acetate, 2 mM EDTA, pH 8.5) or 100 mM Tris-HCl (pH 7.5) and contained 0.001% ethidium bromide. DNA was visualized by UV trans-illumination of the slab; protein was visualized by Coomassie blue staining in 4% acetic acid, 25% ethanol. Analysis of DNA from capsids Agarose gel puri®ed particles were ®lter-eluted by microfuging gel slices on columns stoppered with siliconized glass wool (Sambrook et al., 1989). DNA was then isolated by phenol/chloroform extraction followed by precipitation with ethanol (Ausubel et al., 1989). Pellets were stored at ÿ20 C and resuspended in 1 X TE buffer, mixed with loading buffer and run on agarose slabs at 2 V/cm (Fisher, Mini-Sub). Gels contained 1 X TAE and 0.001% ethidium bromide. DNA was visualized by UV trans-illumination of the slab.

Acknowledgments This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada. We thank Roger Smith for his excellent photographic and image processing services. Irene Johnston and Marg Morton provided much-appreciated secretarial support. Thanks to Gary Saunders for his excellent editorial comments.

References Aksiyote-Benbasat, J. & Bloom®eld, V. A. (1974). Joining of bacteriophage T4D heads and tails: a kinetic study by inelastic light scattering. J. Mol. Biol. 95, 335± 357. Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K. (1989). Current Protocols in Molecular Biology, Green/Wiley Interscience, Toronto. Bjornsti, M. A., Reilly, B. E. & Anderson, D. L. (1983). Morphogenesis of bacteriophage phi 29 of Bacillus subtilis: oriented and quantized in vitro packaging of DNA protein gp3. J. Virol. 45, 383± 396. Black, L. W. (1981). In vitro packaging of bacteriophage T4 DNA. Virology, 113, 336±344. Black, L. W. & Showe, M. K. (1983). Morphogenesis of the T4 Head. In Bacteriophage T4 (Mathews, C., ed.), pp. 219±245, American Society for Microbiology, Washington, DC. Black, L. W., Showe, M. K. & Steven, A. C. (1994). Morphogenesis of the T4 Head. In Molecular Biology of Bacteriophage T4 (Karam, J., Drake, J. W., Kreuzer, K. N., Mosig, G., Hall, D. H., Eiserling, F. A., Black, L. W., Spicer, E. K., Kutter, E., Carlson, K. & Miller, E. S., eds), pp. 218±258, American Society for Microbiology, Washington, DC. Carrascosa, J. L. & Kellenberger, E. (1978). Head maturation pathway of bacteriophages T4 and T2. III.

671 Isolation and characterization of particles produced by mutants in gene 17. J. Virol. 25, 831± 844. Casjens, S., Wyckoff, E., Hayden, M., Sampson, L., Eppler, K., Randall, S., Moreno, E. T. & Serwer, P. (1992). Bacteriophage P22 portal protein is part of the gauge that regulates packing density of intravirion DNA. J. Mol. Biol. 224, 1055± 1074. Coombs, D. H. (1975). Density gradient fractionation by piston displacement. Anal. Biochem. 68, 95 ± 101. Coombs, D. H. & Eiserling, F. A. (1977). Studies on the structure, protein composition, and assembly of the neck of bacteriophage T4. J. Mol. Biol. 116, 375± 405. Coombs, D. H. & Watts, N. R. M. (1985). Generating sucrose gradients in three minutes by tilted tube rotation. Anal. Biochem. 148, 254± 259. Duda, R. L. (1998). Protein chainmail: catenated protein in viral capsids. Cell, 94, 55 ± 60. Duda, R. L., Hempel, J., Michel, H., Shabanowitz, J., Hunt, D. & Hendrix, R. W. (1995). Structural transitions during bacteriophage HK97 head assembly. J. Mol. Biol. 247, 618± 635. Earnshaw, W. C. & Casjens, S. R. (1980). DNA packaging by the double-stranded DNA bacteriophages. Cell, 21, 319± 331. Franklin, J. L., Haseltine, D., Davenport, L. & Mosig, G. (1998). The largest (70 kDa) product of the bacteriophage T4 DNA terminase gene 17 binds to singlestranded DNA segments and digests them towards junctions with double-stranded DNA. J. Mol. Biol. 277, 541± 557. Gordon, C. N. (1972). Dimensions of the heads of the fast and slow sedimenting forms of bacteriophage T2L. J. Mol. Biol. 65, 435± 445. Hamada, K., Fujisawa, H. & Minagawa, T. (1986). A de®ned in vitro system for packaging of bacteriophage T3 DNA. Virology, 151, 119± 123. Hamilton, D. L. & Luftig, R. B. (1972). Bacteriophage T4 morphogenesis. III. Some novel properties of gene 13-defective heads. J. Virol. 9, 1047± 1056. Hohn, B. (1983). DNA sequences necessary for packaging of bacteriophage lambda DNA. Proc. Natl. Acad. Sci. USA, 80, 7456± 7460. Hohn, B. & Hohn, T. (1974). Activity of empty, headlike particles for packaging of DNA of bacteriophage lambda in vitro. Proc. Natl Acad. Sci. USA, 71, 2372± 2376. Hohn, B., Wurtz, M., Klein, B., Lustig, A. & Hohn, T. (1974). Phage lambda DNA packaging, in vitro. J. Supramol. Struct. 2, 302± 317. Hohn, T., Flick, H. & Hohn, B. (1975). Petit lambda, a family of particles from coliphage lambda infected cells. J. Mol. Biol. 98, 107± 120. Hsiao, C. L. & Black, L. W. (1977). DNA packaging and the pathway of bacteriophage T4 head assembly. Proc. Natl Acad. Sci. USA, 74, 3652± 3656. Jardine, P. J., McCormick, M. C., Lutze-Wallace, C. & Coombs, D. H. (1998). The bacteriophage T4 DNA packaging apparatus targets the unexpanded prohead. J. Mol. Biol. 284, 625± 637. Kaiser, D., Syvanen, M. & Masuda, T. (1974). Processing and assembly of the head of bacteriophage lambda. J. Supramol. Struct. 2, 318± 328. Kalinski, A. & Black, L. W. (1986). End structure and mechanism of packaging of bacteriophage T4 DNA. J. Virol. 58, 951± 954. Kellenberger, E. (1980). Control mechanisms in the morphogenesis of bacteriophage heads. Biosystems, 12, 201± 223.

672

Packaging Precedes Expansion in T4

Kemper, B. & Brown, D. T. (1976). Function of gene 49 of bacteriophage T4. II. Analysis of intracellular development and the structure of very fast-sedimenting DNA. J. Virol. 18, 1000± 1015. Kuhn, A. & Kellenberger, E. (1985). Productive phage infection in Escherichia coli with reduced internal levels of the major cations. J. Bacteriol. 163, 906± 912. Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature, 227, 680±685. Laemmli, U. K. & Favre, M. (1973). Maturation of the head of bacteriophage T4. I. DNA packaging events. J. Mol. Biol. 80, 575± 599. Luftig, R. & Ganz, C. (1972). Bacteriophage T4 head morphogenesis. II. Studies on the maturation of gene 49-defective head intermediates. J. Virol. 9, 377± 389. Masker, W. E. & Serwer, P. (1982). DNA packaging in vitro by an isolated bacteriophage T7 procapsid. J. Virol. 43, 1138± 1142. Minagawa, T. & Ryo, Y. (1978). Substrate speci®city of gene 49-controlled deoxyribonuclease of bacteriophage T4: special reference to DNA packaging. Virology, 91, 222± 233. Mizuuchi, K., Kemper, B., Hays, J. & Weisberg, R. A. (1982). T4 endonuclease VII cleaves Holliday structures. Cell, 29, 357± 365. Mosig, G. (1994). Homologous recombination. In Molecular Biology of Bacteriophage T4 (Karam, J., Drake, J. W., Kreuzer, K. N., Mosig, G., Hall, D. H., Eiserling, F. A., Black, L. W., Spicer, E. K., Kutter, E., Carlson, K. & Miller, E. S., eds), pp. 54 ± 82, American Society for Microbiology, Washington, DC. Newcomb, W. W., Homa, F. L., Thomsen, D. R., Booy, F. P., Trus, B. L., Steven, A. C., Spencer, J. V. & Brown, J. C. (1996). Assembly of the herpes simplex virus capsid: characterization of intermediates observed during cell-free capsid formation. J. Mol. Biol. 263, 432± 446. Parker, M. L., Christensen, A. C., Boosman, A., Stockard, J., Young, E. T. & Doermann, A. H. (1984). Nucleotide sequence of bacteriophage T4 gene 23 and the amino acid sequence of its product. J. Mol. Biol. 180, 399±416. Poteete, A. R., Jarvik, V. & Botstein, D. (1979). Encapsulation of phage P22 DNA in vitro. Virology, 95, 550± 564. Rao, V. B. & Black, L. W. (1985). DNA packaging of bacteriophage T4 proheads in vitro, evidence that prohead expansion is not coupled to DNA packaging. J. Mol. Biol. 185, 565±578. Rao, V. B. & Black, L. W. (1988). Cloning, overexpression and puri®cation of the terminase proteins gp16 and gp17 of bacteriophage T4. Construction of a de®ned in vitro DNA packaging system using puri®ed terminase proteins. J. Mol. Biol. 200, 475±488.

Sambrook, J., Fritch, E. F. & Maniatis, T. (1989). Molecular Cloning ± A Laboratory Manual, 2nd edit., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schaerli, C. & Kellenberger, E. (1980). Head maturation pathway of bacteriophages T4 and T2. V. Maturable e-particle accumulating in acridine-treated bacteriophage T4-infected cells. J. Virol. 33, 830±844. Serwer, P. & Pichler, M. E. (1978). Electrophoresis of bacteriophage T7 and T7 capsids in agarose gels. J. Virol. 28, 917± 928. Shibata, H., Fujisawa, H. & Minagawa, T. (1987). Characterization of the bacteriophage T3 DNA packaging reaction in vitro in a de®ned system. J. Mol. Biol. 196, 845± 851. Steven, A. C. & Carrascosa, J. L. (1979). Proteolytic cleavage and structural transformation: their relationship in bacteriophage T4 capsid maturation. J. Supramol. Struct. 10, 1 ± 11. Steven, A. C., Couture, E., Aebi, U. & Showe, M. K. (1976). Structure of T4 polyheads. II. A pathway of polyhead transformations as a model for T4 capsid maturation. J. Mol. Biol. 106, 187± 221. Streisinger, G., Edgar, R. S. & Denhardt, G. H. (1964). Chromosome structure in phage T4. I. Circularity of the linkage map. Proc. Natl Acad. Sci. USA, 51, 775± 779. Streisinger, G., Emrich, J. & Stahl, M. M. (1967). Chromosome structure in phage T4. III. Terminal redundancy and length determination. Proc. Natl Acad. Sci. USA, 54, 1333± 1339. Trus, B. L., Booy, F. P., Newcomb, W. W., Brown, J. C., Homa, F. L., Thomsen, D. R. & Steven, A. C. (1996). The herpes simplex virus procapsid: structure, conformational changes upon maturation, and roles of the triplex proteins VP19c and VP23 in assembly. J. Mol. Biol. 263, 447± 462. Wagner, J. & Laemmli, U. K. (1976). Studies on the maturation of the head of bacteriophage T4. Phil. Trans. Roy. Soc. London ser. B, 276, 15 ± 28. Wagner, J. A. & Laemmli, U. K. (1979). Maturation of the head of bacteriophage T4: 9-aminoacridine blocks a late step in DNA packaging. Virology, 92, 219± 229. Wu, C. H. & Black, L. W. (1995). Mutational analysis of the sequence-speci®c recombination box for ampli®cation of gene 17 of bacteriophage T4. J. Mol. Biol. 247, 604± 617. Wunderli-Allenspach, H. (1977). A maturable, highly fragile head-related particle containing DNA and uncleaved gp23 as a probable intermediate in T4 head maturation. PhD thesis, University of Basel. Yanagida, M., Suzuki, Y. & Toda, T. (1984). Molecular organization of the head of bacteriophage T-even: underlying design principles. Advan. Biophys. 17, 97 ±146.

Edited by J. Karn (Received 7 April 1998; received in revised form 17 August 1998; accepted 26 August 1998)