Crystal and cryoEM structural studies of a cell walldegrading enzyme in the bacteriophage �29 tailYe Xiang†, Marc C. Morais†‡, Daniel N. Cohen§, Valorie D. Bowman†, Dwight L. Anderson§, and Michael G. Rossmann†¶

†Department of Biological Sciences, Purdue University, West Lafayette, IN 47907; and §Departments of Diagnostic/Biological Sciences and Microbiology,University of Minnesota, Minneapolis, MN 55455

Contributed by Michael G. Rossmann, April 28, 2008 (sent for review March 20, 2008)

The small bacteriophage �29 must penetrate the �250-Å thickexternal peptidoglycan cell wall and cell membrane of the Gram-positive Bacillus subtilis, before ejecting its dsDNA genomethrough its tail into the bacterial cytoplasm. The tail of bacterio-phage �29 is noncontractile and �380 Å long. A 1.8-Å resolutioncrystal structure of gene product 13 (gp13) shows that this tailprotein has spatially well separated N- and C-terminal domains,whose structures resemble lysozyme-like enzymes and metallo-endopeptidases, respectively. CryoEM reconstructions of the WTbacteriophage and mutant bacteriophages missing some or mostof gp13 shows that this enzyme is located at the distal end of the�29 tail knob. This finding suggests that gp13 functions as atail-associated, peptidoglycan-degrading enzyme able to cleaveboth the polysaccharide backbone and peptide cross-links of thepeptidoglycan cell wall. Comparisons of the gp13� mutants withthe �29 mature and emptied phage structures suggest the se-quence of events that occur during the penetration of the tailthrough the peptidoglycan layer.

hydrolase � infection � structure � zinc ion

Bacterial cells, including Bacillus subtilis, are surrounded by apeptidoglycan-containing cell wall. The peptidoglycan con-

sists of alternating N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) residues, cross-linked with oligopep-tides at each NAM (1). This highly cross-linked peptidoglycanlayer is not only essential for the integrity of bacteria in the faceof the osmotic pressure created by the high concentration ofmolecules in the cell, but also serves as a physical barrier thatrestricts the passage of macromolecular complexes. In Gram-negative bacteria, the peptidoglycan layer is within the periplas-mic space and is �25–75 Å thick (2), whereas in Gram-positivebacteria the peptiglycan layer is outside the cytoplasmic mem-brane and can be as much as 250 Å thick (3). Most bacterio-phages have a tail that penetrates the host cell wall and mem-branes to deliver the genome. Bacteriophage tails are largemacromolecular complexes that contain components for recog-nizing and puncturing the host cell and have a channel wideenough to provide passage for ejection of the viral genome (4–7).The peptidoglycan layer is a major barrier that must be spannedby a phage tail to infect a host cell. Unlike most eukaryoticviruses, infection of a host by tailed bacteriophages usuallyrequires only one virion per bacterium, indicating that tailedbacteriophages have evolved mechanisms that easily surmountthe peptidoglycan barrier. Virus-encoded peptidoglycan degrad-ing enzymes, such as endopeptidases, N-acetylmuramyl-L-alanine amidases, N-acetylglucosaminidase, lysozymes, and lytictransglycosylases, have been found widespread in tailed bacte-riophages (8, 9). These enzymes are known to play a significantrole in facilitating tailed bacteriophage genome entry throughlocalized peptidoglycan degradation or rearrangement (9, 10)and are frequently structural components of the virion. Theirlocation within the virions is varied, but they are located mostlywithin the tail (8, 11, 12).

Bacteriophage �29 infects the Gram-positive bacterium, B.subtilis, and is one of the smallest known tailed phages (13). The

mature �29 particle contains a prolate head and a short non-contractible tail. The head and part of the tail are filled with the19-kbp DNA genome and the terminal protein gene product 3(gp3) (7) that is covalently attached to both 5� ends of the DNAgenome (14). Previous studies show that the head consists ofthree proteins: the capsid protein (gp8), the head fibers (gp8.5),and the portal protein (connector) (gp10) (13, 15, 16). The tailhad been found to consist of three proteins: gp11, gp12* (cleavedfrom the gp12 precursor), and gp9, that form the lower collar,appendages, and knob, respectively (Fig. 1) (7, 17). Biochemicaldata (8) have suggested that the peptidoglycan-degrading activ-ity associated with �29 lies in the DNA terminal protein gp3,although the crystal structure of gp3 does not support such afunction (18). However, as gp9 is located at the distal end of thetail, it might be associated with the cell wall-degrading enzymeused by �29.

The gp13 of bacteriophage �29 consists of 365 amino acidresidues. Although gp13 had not been previously detected in thevirion (19), it was known to be an essential morphogenetic factorthat functions in the phage tail assembly stage and is indispens-able for producing infectious particles (19, 20). In vitro comple-mentation experiments showed that gp13 is required after gp11has assembled onto the phage head. Furthermore, gp13 has beenshown to interact with gp9 in vivo to yield infectious particles(19). Re-examination of gp13 using specific antisera in Westernblots showed that it is a structural component of WT �29,emptied �29, free tails, and the mutant sus13(330) virus (20, 21).This mutant virus has a nonsense stop codon that results inproduction of gp13 that is missing the last 34 aa. On the otherhand, gp13 was not detected by Western blotting in the tail forthe sus13(342) mutant virus (21), which has a nonsense stopcodon that results in the production of a vestigial gp13 consistingof only the 32 N-terminal residues. These results demonstratethat gp9, gp11, gp12, and gp13 are all components of �29tails (21).

Here, we report cryoEM image reconstructions that suggestgp13 to be located at the end of the tail knob. We also report thatthe crystal structure of gp13 has two domains that resemblelysozyme-like enzymes and metallo-endopeptidases, showingthat it probably functions as a tail-associated peptidoglycan-degrading enzyme.

Author contributions: Y.X. designed research; Y.X. and M.C.M. performed research; D.N.C.,V.D.B., and D.L.A. contributed new reagents/analytic tools; Y.X. analyzed data; and Y.X.and M.G.R. wrote the paper.

The authors declare no conflict of interest.

Data deposition: The atomic coordinates have been deposited in the Protein Data Bank,www.pdb.org (PDB ID codes 3CSQ, 3CSR, 3CSZ, 3CT0, 3CT1, and 3CT5). The cryoEM mapshave been deposited in the Electron Microscopy Data Bank, www.ebi.ac.uk/msd-srv/docs/emdb (accession nos. EMD1506 and EMD5010).

‡Present address: Department of Biochemistry and Molecular Biology, University of TexasMedical Branch, Galveston, TX 77555.

¶To whom correspondence should be addressed. E-mail: mr@purdue.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0803787105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

9552–9557 � PNAS � July 15, 2008 � vol. 105 � no. 28 www.pnas.org�cgi�doi�10.1073�pnas.0803787105

Results and DiscussionThe Structure of gp13. The full-length gp13 (residues 1–365)formed very thin needle crystals, too small to detect a diffractionpattern. Hence, the protein was degraded with trypsin and themajor fragment (residues 1–347, gp13TD) was repurified andcrystallized. The crystal structure of gp13TD [supporting infor-mation (SI) Text] showed that gp13TD had two spatially wellseparated domains, although the C-terminal domain was poorly

defined. Therefore, three constructs, gp13NTD (residues 1–159,the N-terminal domain), gp13CTD (residues 166–365, the C-terminal domain), and gp13�C (residues 1–334, missing the endof the C-terminal domain), were made to study the domainsindependently. The gp13CTD construct failed to crystallize.Crystals of gp13�C were isomorphous with gp13TD, but pro-duced a better electron density map. The resultant refined modelof gp13�C (Fig. 2a) gave final Rworking and Rfree factors of 23.8%

Fig. 1. Diagram of bacteriophage �29 showing its structural components. (Left) A surface-rendered presentation. (Right) A central section.

Fig. 2. The structure of gp13. (a) Ribbon stereo diagram of gp13�C with the N- and C-terminal domains shown in green and purple, respectively. Two flexibleregions that were not clearly visible in the electron density maps are shown in coral. The active sites of the N- and C- terminal domains are indicated by arrows.(b) Stereo diagram showing surface charge distribution of the gp13 N-terminal domain. Sugar substrates are shown in a ball-and-stick representation with N,C, and O atoms colored blue, gray, and green, respectively. Sugar residues at sites A, B, C, and D (black bonds) are from the crystal structure of gp13NTD complexedwith (NAG)4, whereas sugar residues at sites E and F (gray bonds) are based on homology model building. (c) Stereo diagram showing the interaction betweensugar substrates and protein residues. Parts of the protein C� backbone are shown in a ball-and-stick presentation with C� atoms colored red. Sugar substratesand protein residues are shown in a ball-and-stick presentation and colored the same as in b.

Xiang et al. PNAS � July 15, 2008 � vol. 105 � no. 28 � 9553

BIO

CHEM

ISTR

Y

and 28.2%, respectively. Crystals of gp13NTD had a differentcrystal form and their structure was determined by molecularreplacement. The rmsd between equivalent C� atoms in thegp13�C and gp13NTD structures was 0.4 Å.

The N-terminal domain of gp13�C had a structure closelysimilar to hen egg white lysozyme (Dali Z-score of 6.4) (22) andthe C-terminal domain was similar to the active form of LytM(23) [Protein Data Bank (PDB) ID code 2B0P] a metallo-endopeptidase (Z score of 11.3) in the M23 family (24). Thelinker polypeptide between the two domains consists of sixsequential glycine residues (160–165) defined by rather poorelectron density in the gp13�C map, indicative of flexibility (Fig.2a). Structure and sequence comparisons provided informationabout the putative substrate binding sites in the two gp13domains (Figs. 2a and 3). A long groove (Fig. 2a) on one side ofthe C-terminal domain forms its putative substrate binding site.

The N-Terminal Lysozyme-Like Domain of gp13. Comparisons of �29gp13NTD with T4 lysozyme (PDB ID code 3LZM), bacterio-phage lambda lysozyme (PDB ID code 1AM7), hen egg whitelysozyme (PDB ID code 3LZT), and goose egg white lysozyme(PDB ID code 153L) showed that �29 had the closest structuralsimilarity to lambda lysozyme (Table 1). Although lambda

lysozyme has a similar fold and substrate binding sites as theother lysozymes, unlike the other lysozymes it does not requirea water molecule in its enzymatic action and hence should bereferred to as a transglycosylase.

The substrate binding site of lysozymes is a groove that canaccommodate a polysaccharide with up to six single sugarresidues at sites A through F (25, 26) with the cleavage siteoccurring between sites D and E. Lysozymes cleave the �(1–4)glycosidic bonds between NAM at site D and NAG at site E(27). The catalytic center includes an invariable glutamic acidresidue and a less conserved aspartic acid residue. The latterprobably stabilizes the cleavage intermediate in some ly-sozyme-like proteins (28). A structural comparison ofgp13NTD with other lysozymes shows that Glu-45 in gp13corresponds to the catalytically essential glutamic acid residue.The aspartic acid residue in the active center, present in bothhen egg white lysozyme (Asp-52) (25) and T4 lysozyme(Asp-20) (29), corresponds to Gly-90 in gp13. A glycine occursat the corresponding position in bacteriophage lambda ly-sozyme (30) and goose egg white lysozyme (31). The structuresof gp13NTD complexed with (NAG)x (where 3 � x � 6)showed that the side chain of Asn-54 is 4.3 Å from the C1carbon of the sugar at substrate site D. Thus Asn-54 might playa role in stabilizing the substrate during catalysis.

Fig. 3. Structural-based sequence alignment of gp13, hen egg lysozyme [PDB ID code 3LZT (35)] and active form of lytM [PDB ID code 2B0P (23)]. The secondarystructural elements are shown above the alignments. The glycine residues linking the two domains of gp13 are indicated by stars. Residues essential for catalysisare indicated by triangles. Completely conserved residues are shown in white on a gray background.

Table 1. Structure and sequence comparison of gp13NTD and lysozyme-like proteins

CharacteristicsHen egg white lysozyme

(PDB ID code 3LZT)Goose egg white lysozyme

(PDB ID code 153L)Phage lambda lysozyme

(PDB ID code 1AM7)T4 lysozyme

(PDB ID code 3LZM)

No. of residues aligned† 102 (129) 109 (185) 112 (154) 71 (164)Identical residues among

aligned residues, %12 14 10 11

rmsd, Å 3.3 3.0 3.2 3.9Secondary structure

similarity‡

�1, �2, �3, �4, �5, �6, �7,�1, �2, �3

�1, �2, �3, �4, �5, �6, �7,�1, �2, �3

�1, �2, �3, �4, �5, �6, �7,�1, �2, �3

�1, �2, �3, �4, �5, �6, �7,�1, �2, �3

†The complete amino acid sequence length of the corresponding lysozyme-like protein are given in parentheses.‡� 1–7 and � 1–3 represent the secondary structural elements of gp13NTD. Equivalent secondary structures present in lysozyme-like proteins are shown in bold.

9554 � www.pnas.org�cgi�doi�10.1073�pnas.0803787105 Xiang et al.

Although the physiological substrate of lysozyme is a polysac-charide of alternating NAG and NAM residues, lysozymes can alsobind and hydrolase chitin-like (NAG)n polysaccharides (27). Coc-rystallizations of gp13NTD with (NAG)x (where 3 � x � 6) showedgood density only at sites B, C, and D. These densities wereinterpreted as sugar rings in chair conformation (Fig. 2 b and c).This result indicated that sugar substrates longer than three residueswere, in general, cleaved during the cocrystallization procedure,leaving only the cleaved product in the substrate-binding pocket.NAG hydrolysis was confirmed by electrospray ionization massspectrometer analyses of sugar substrates before and after beingincubated with gp13NTD (data not shown), showing that gp13NTDhad lysozyme activity.

C-Terminal Domain. The defining feature of the M23 family ofmetallo-peptidases is an antiparallel �-sheet and a conservedHXH motif that binds a metal ion (24) at the active center. Thereis a large electron density peak (8�) in the gp13 C-terminaldomain electron density, corresponding to a bound metal ionassociated with the HXH motif. This metal cation is tetrahe-drally coordinated by two histidine residues (H188 and H280), anaspartic acid residue (D195), and a water molecule (O59) (Fig.S1). Crystal structures of M23 metallo-endopeptidodases have a

zinc ion bound at their metal binding center even when noexogenous zinc ions had been added. Thus, the ion in the crystalstructure of gp13 was suspected to be zinc. This notion wasconfirmed by an x-ray fluorescence scan that showed the ex-pected absorption maximum at a wavelength of 1.283 Å, near theK-edge of zinc.

LytM is synthesized as a preenzyme in vivo. The zinc ion is notsolvent accessible in inactive LytM (23, 32). A proteolyticmodification is required for activation of the enzyme that cleavesa loop blocking the active center. In the current gp13 structure,the substrate binding site of the C-terminal domain is accessibleto solvent molecules, indicating that the gp13 C-terminal domainis in the active form.

LytM can cleave pentaglycine cross-links in peptidoglycanlayers of staphylococci (32). The structure of active LytM (23)shows that a groove formed by four protruding loops from thecentral �-sheet is the substrate binding site. Comparison of thesubstrate binding groove of active LytM with that of the gp13C-terminal domain shows differences in the lengths and confor-mations of the four loops forming the substrate binding groove,suggesting that the gp13 C-terminal domain may have differentsubstrate binding specificity than that of LytM. This differencecould have been anticipated as B. subtilis has not been shown tohave pentaglycine cross-linkages (1).

Biological Functional Unit Formed by Neighboring Molecules in theCrystal Structure. The largest buried surface area (764 Å2) be-tween adjacent molecules in the gp13�C crystal is between theN- and C-terminal domains belonging to neighboring moleculesrelated by a 21 screw axis along b (Fig. 4). A tetrasaccharide(NAG-NAM-NAG-NAM) built into the N-terminal domain ofone molecule would place a pentapeptide (L-Ala-D-�-Glu-L-Lys-D-Ala-D-Ala) associated with the NAM at site F (Fig. 2c), closeto the peptide-binding groove in the C-terminal domain of the

Fig. 4. Schematic view of the four molecules in the unit cell and theircontacts with symmetry-related neighboring molecules. Molecules A and B,related by a pseudo translation in c, are colored purple. Symmetry-relatedmolecules to A and B are colored green. Molecules C and D, related by apseudo translation in c, are colored blue. Symmetry related molecules to C andD are colored yellow. A potential biological functional unit is outlined inorange. The unit cell is outlined with a black line and the pseudo smallerorthorhombic unit cell is outlined with a dashed black line. The pseudosymmetry elements of the smaller unit cell are not shown. CTD, C-terminaldomain; NTD, N-terminal domain.

Fig. 5. The cone shaped gp13 density at the distal end of the phage tail. (a)Comparison of the cryoEM densities of WT DNA-filled (green; i), WT DNA-emptied (cyan; ii), mutant sus13(330) (blue; iii), and sus13(342) (purple; iv)tails. (b) Construction of the cryoEM density difference maps (iii and iv)[sus13(330) (blue; i)–sus13(342) (purple; ii)] showing the cone-shaped densityat the distal end of the sus13(330) knob is the most significant difference in thetail. (c) Diagrammatic figure showing the WT, sus13(330), and sus13(342) gp13peptides.

Xiang et al. PNAS � July 15, 2008 � vol. 105 � no. 28 � 9555

BIO

CHEM

ISTR

Y

neighboring molecule. The docked peptidoglycan structure cor-responds to the chemical composition of a bacterial cell wall (33).Thus, the docking suggests that the crystallographically observedassociation between neighboring molecules might be a biologi-cally functional unit. The predicted cleavage sites derived fromthe modeling would be between the NAG and NAM residues atsites D and E and between the D-�-Glu and L-Lys residues orbetween the L-Lys and D-Ala residues of the attached peptide.The predicted cleavage sites are consistent with the observationsof the peptidoglycan cleaved product compositions (D.N.C., YukK. Sham, Greg D. Haugstad, Y.X., M.G.R., D.L.A., and DavidL. Popham, unpublished work) that occur when gp13 digests cellwall fragments.

Locating gp13 in the �29 Tail. The structure of the �29 tail has aconical protrusion at its distal end in DNA-emptied WT particles(7). This protrusion appeared to be absent in WT full infectiousparticles, but re-examination has now shown that there is a weakdensity that might correspond to a somewhat disordered conicalprotrusion (Fig. 5ai). Asymmetric cryoEM reconstructions ofthe mutant sus13(330) (containing gp13 missing only the last 34aa) and mutant sus13(342) (which should contain only theN-terminal 32 aa of gp13) particles both show the 5-fold sym-metry of the head and the presence of 12 appendages, featurespresent also in the WT mature and emptied particles. However,the sus13(342) particles were missing the protruding tail cone(Fig. 5aiv), whereas the sus13(330) particles (Fig. 5aiii) had a wellformed, cone-shaped density at the distal end of the tail knob.This finding suggests that the tail cone is an assembly of gp13molecules. A possible explanation is that the gp13 is somewhatdisordered in the WT particle, but becomes more static after thegenome has been ejected or when the carboxyl-terminal 34residues of gp13 are missing. The external nature of gp13 is alsomade manifest by a difference map (Fig. 5biii) between thecryoEM density of the mutant sus13(330) particle (Fig. 5bi) andthe mutant sus13(342) particle (Fig. 5bii), which shows only thecone-shaped density at the tip of the knob. Thus, there is no

evidence for gp13 being within the phage tail. Whether gp13 isreleased from the virion in vivo during genome injection into thehost remains unknown.

Variation in the hinge angle between the two domains observedcrystallographically and the poor density of the hexa-glycine linkershows that the peptide link is extremely flexible. Thus, when fittingthe gp13 molecular structure into the doughnut-shaped densityrepresenting the cone, it cannot be assumed that the two domainsof gp13 have the same spatial relationship as they do in the crystalstructure. The resolution of the cryoEM density was insufficient tobe able to fit the rather spherical N and C domains independently.Instead, it was tentatively assumed that the close association of theN and C domains of neighboring molecules in the crystal structureof gp13�C might be the enzymatically active unit as suggestedabove. Two of these active units could be readily fitted into the2-fold averaged cryoEM density (Fig. 6). This fit showed that twomolecules of gp13 were able to account for the cone density, thusprobably creating a symmetry mismatch with the knob containingabout nine gp9 molecules.

The fit of gp13 into the cone density placed the active sites ofthe two domains onto the outer distal ring of the cone, a locationsuitable for digesting the bacterial peptidoglycan cell wall.Furthermore, the gp13 fit into the tail cone places a lot ofnegative charge into the central channel, much as the gp10dodecameric connector has a negatively charged inner channel(16). This feature may be to repel the negatively chargedphosphates on the DNA, thereby focusing the DNA to the centerof the channel, implying that the gp13 cone is present duringDNA ejection from the virus.

Although the structural results are consistent within them-selves, the location of gp13 in the cone at the distal end of theknob is in conflict with results on the aggregation of �29 particlesin the presence of anti-gp13 polyclonal antibodies (21). Findinga hypothesis that satisfies both the structural data presented hereand the immunological data (21) is a challenge for futureinvestigations.

Materials and MethodsProtein Expression, Purification, Crystallization, Data Collection, and Processing.For details see SI Text.

Structure Determination and Refinement. The structure of gp13TD was deter-mined with a mercury derivative, using the single isomorphous replacementmethod enhanced by anomalous scattering data (SIRAS). The initial processwas simplified by reindexing assuming a smaller unit cell (Fig. 4) so as to omitall reflections with l odd. The program SOLVE found four heavy atom sites inthe asymmetric unit of the smaller pseudo unit cell. The structure had inter-pretable electron density for the N-terminal domain, but the C-terminaldomain density was poorly defined. The partial structure was used as amolecular replacement search model for determining the orientation andposition of each of the four independent molecules in the asymmetric unit ofgp13�C (A, B, C, and D in Fig. 4) while using the l odd reflections. Furtherdetails are given in SI Text and Table S1.

CryoEM Reconstruction. The production and purification of sus13(330) andsus13(342) particles has been described (21). Electron micrographs were re-corded at a magnification of 33,000 with a CM300 FEG microscope underlow-dose conditions. Both mutant virus reconstructions assumed no symmetryand were based on a procedure as described (4, 7, 34). The N- and C-terminaldomains of gp13 were initially fitted manually into the cryoEM density map byusing the program Chimera and then optimized with the program EMfit,assuming 2-fold symmetry. Further details are in SI Text and Table S2.

ACKNOWLEDGMENTS. We thank Kay Choi and Petr Leiman for many helpfuldiscussions; Sheryl Kelly and Cheryl Towell for the preparation of the manuscript;and Siyang Sun and the staff at the APS, BioCars, and GM/CA sectors for their helpin x-ray diffraction data collection. Those facilities are supported by the U.S.Department of Energy and/or the National Institutes of Health. This work wassupported by National Science Foundation Grant MCB-0443899 (to M.G.R.) andNational Institutes of Health Grant DE03606 (to D.L.A. and M.G.R.).

Fig. 6. Stereo diagrams showing two gp13 monomers (blue and red) fittedinto the 2-fold averaged cryoEM density of the emptied WT (green contours)and mutant full sus13(330) (black contours) cone densities. End-on (a) and side(b) views of a 40-Å-thick cross-section.

9556 � www.pnas.org�cgi�doi�10.1073�pnas.0803787105 Xiang et al.

1. Popham DL, Helin J, Costello CE, Setlow P (1996) Analysis of the peptidoglycanstructure of Bacillus subtilis endospores. J Bacteriol 178:6451–6458.

2. Labischinski H, Goodell EW, Goodell A, Hochberg ML (1991) Direct proof of a ‘‘more-than-single-layered’’ peptidoglycan architecture of Escherichia coli W7: A neutronsmall-angle scattering study. J Bacteriol 173:751–756.

3. Beveridge TJ, Graham LL (1991) Suface layers of bacteria. Microbiol Rev 55:684 –705.

4. Jiang W, et al. (2006) Structure of epsilon15 bacteriophage reveals genome organi-zation and DNA packaging/injection apparatus. Nature 439:612–616.

5. Lander GC, et al. (2006) The structure of an infectious P22 virion shows the signal forheadful DNA packaging. Science 312:1791–1795.

6. Leiman PG, Chipman PR, Kostyuchenko VA, Mesyanzhinov VV, Rossmann MG (2004)Three-dimensional rearrangement of proteins in the tail of bacteriophage T4 oninfection of its host. Cell 118:419–429.

7. Xiang Y, et al. (2006) Structural changes of bacteriophage �29 upon DNA packagingand release. EMBO J 25:5229–5239.

8. Moak M, Molineux IJ (2004) Peptidoglycan hydrolytic activities associated with bacte-riophage virions. Mol Microbiol 51:1169–1183.

9. Poranen MM, Daugelavicius R, Bamford DH (2002) Common principles in viral entry.Annu Rev Microbiol 56:521–538.

10. Koraimann G (2003) Lytic transglycosylases in macromolecular transport systems ofGram-negative bacteria. Cell Mol Life Sci 60:2371–2388.

11. Kanamaru S, et al. (2002) Structure of the cell-puncturing device of bacteriophage T4.Nature 415:553–557.

12. Kenny JG, McGrath S, Fitzgerald GF, van Sinderen D (2004) Bacteriophage Tuc2009encodes a tail-associated cell wall-degrading activity. J Bacteriol 186:3480–3491.

13. Anderson D, Reilly B (1993) in Bacillus subtilis and Other Gram-Positive Bacteria:Biochemistry, Physiology, and Molecular Genetics, eds Sonenshein AL, Hoch JA, LosickR (Am Soc Microbiol, Washington, DC), pp 859–867.

14. Salas M, Mellado RP, Vinuela E, Sogo JM (1978) Characterization of a protein covalentlylinked to the 5� termini of the DNA of Bacillus subtilis phage �29. J Mol Biol 119:269–291.

15. Morais MC, et al. (2005) Conservation of the capsid structure in tailed dsDNA bacte-riophages: The pseudoatomic structure of �29. Mol Cell 18:149–159.

16. Simpson AA, et al. (2000) Structure of the bacteriophage �29 DNA packaging motor.Nature 408:745–750.

17. Tao Y, et al. (1998) Assembly of a tailed bacterial virus and its genome release studiedin three dimensions. Cell 95:431–437.

18. Kamtekar S, et al. (2006) The �29 DNA polymerase: Protein-primer structure suggestsa model for the initiation to elongation transition. EMBO J 25:1335–1343.

19. Garcıa JA, Carrascosa JL, Salas M (1983) Assembly of the tail protein of the Bacillussubtilis phage �29. Virology 125:18–30.

20. Hagen EW, Reilly BE, Tosi ME, Anderson DL (1976) Analysis of gene function ofbacteriophage �29 of Bacillus subtilis: Identification of cistrons essential for viralassembly. J Virol 19:501–517.

21. Cohen DN, Erickson SE, Xiang Y, Rossmann MG, Anderson DL (2008) Multifunctionalroles of a bacteriophage �29 morphogenetic factor in assembly and infection. J MolBiol 378:804–817.

22. Holm L, Sander C (1998) Touring protein fold space with Dali/FSSP. Nucleic Acids Res26:316–319.

23. Firczuk M, Mucha A, Bochtler M (2005) Crystal structures of active LytM. J Mol Biol354:578–590.

24. Hooper NM (1994) Families of zinc metalloproteases. FEBS Lett 354:1–6.25. Blake CCF, et al. (1965) Structure of hen egg-white lysozyme: A three-dimensional

Fourier synthesis at 2-Å resolution. Nature 206:757–761.26. Phillips DC (1966) The three-dimensional structure of an enzyme molecule. Sci Am

215:78–90.27. Chipman DM, Sharon N (1969) Mechanism of lysozyme action. Science 165:454–465.28. Sun DP, Liao DI, Remington S (1989) Electrostatic fields in the active sites of lysozymes.

Proc Natl Acad Sci USA 86:5361–5365.29. Weaver LH, Matthews BW (1987) Structure of bacteriophage T4 lysozyme refined at

1.7-Å resolution. J Mol Biol 193:189–199.30. Evrard C, Fastrez J, Declercq JP (1998) Crystal structure of the lysozyme from bacterio-

phage lambda and its relationship with V and C-type lysozymes. J Mol Biol 276:151–164.

31. Weaver LH, Grutter MG, Matthews BW (1995) The refined structures of goose lysozymeand its complex with a bound trisaccharide show that the ‘‘Goose-type’’ lysozymes lacka catalytic aspartate residue. J Mol Biol 245:54–68.

32. Odintsov SG, Sabala I, Marcyjaniak M, Bochtler M (2004) Latent LytM at 1.3-Å resolu-tion. J Mol Biol 335:775–785.

33. Meroueh SO, et al. (2006) Three-dimensional structure of the bacterial cell wallpeptidoglycan. Proc Natl Acad Sci USA 103:4404–4409.

34. Morais MC, et al. (2001) Cryoelectron microscopy image reconstruction of symmetrymismatches in bacteriophage �29. J Struct Biol 135:38–46.

35. Walsh MA, et al. (1998) Refinement of triclinic hen egg-white lysozyme at atomicresolution. Acta Crystallogr D 54:522–546.

Xiang et al. PNAS � July 15, 2008 � vol. 105 � no. 28 � 9557

BIO

CHEM

ISTR

Y