Given the observation that PLE virions have smaller capsids than ICP1 virions (Figure 1A and B), we hypothesized that PLE encodes a single protein responsible for assembling these smaller capsids. The generation of small capsids would limit the amount of DNA that could be accommodated within the capsids and block the complete packaging of the larger ICP1 genome, reducing ICP1 plaquing. In line with this prediction, other satellites’ proteins that redirect capsid assembly are inhibitory to their helper phages for this reason (Shore et al., 1978; Damle et al., 2012; Carpena et al., 2016). Using this logic and PLE1 as the representative PLE, we identified one candidate PLE protein, TcaP (previously Orf17 (AGG09411.1)), as a putative capsid remodeling protein.

First, we used transmission electron microscopy (TEM) to evaluate the morphology of virions produced after a single round of ICP1 infection in a PLE(-) V. cholerae strain expressing tcaP from a plasmid (ptcaP). Lysates collected from the ptcaP host showed an abundance of small, ~50-nm-round particles reminiscent of PLE-sized procapsids (Figure 1C). Rarer particles in these samples resembled virions produced from a PLE(+) infection with small capsids and attached tails, though apparently lacking DNA (Figure 1—figure supplement 1A). In comparison, many virions produced from the empty vector control were ICP1-sized virions with large capsids, as expected (Figure 1D). Other particles observed in the control samples were full ICP1 capsids without tails, empty ICP1 procapsids, or empty expanded capsids (Figure 1—figure supplement 1B). Importantly, these data demonstrate that tcaP expression during ICP1 infection, in the absence of other PLE-encoded products, results in the production of small capsids.

In agreement with the expectation that small capsid formation would be inhibitory to ICP1, ptcaP inhibited ICP1 plaque formation by 100-fold (Figure 1E). Rare plaques isolated on the ptcaP strain were picked and serially propagated on the restrictive strain. Two escape phages were whole-genome sequenced, and each harbored a single nonsynonymous mutation in the gene encoding the coat protein, resulting in substitutions R223H or E234K (Supplementary file 1). Substitutions at unique but proximal positions in coat protein from independent escape phages suggest that TcaP interacts with the coat protein, an expected feature of TcaP’s predicted role as a capsid remodeling protein. Collectively, these genetic escapes, coupled with TcaP’s inhibition of ICP1 plaquing, which is accompanied by the production of small capsids, strongly support that TcaP is a capsid remodeling protein.

Expression of TcaP during ICP1 infection was sufficient to redirect capsid size, but other capsid remodeling satellites have been shown to use either a single protein (Shore et al., 1978; Carpena et al., 2016) or two proteins that act in concert to direct small capsid assembly (Damle et al., 2012). Virions resulting from ICP1 infection of wild type PLE and PLE∆tcaP were concentrated by centrifugation and their morphology was assessed by TEM. The wild type PLE control produced PLE-sized virions with small capsids as expected and no ICP1-sized virions were observed (Figure 2A). The PLE∆tcaP strain produced only virions with large, ICP1-sized capsids and no PLE-sized particles (Figure 2B), a phenotype that was rescued following the expression of tcaP in trans (Figure 2C). Importantly, tcaP is dispensable for PLE-mediated inhibition of ICP1 (Hays and Seed, 2020), indicating that the increase in capsid size in the absence of TcaP is specific to its role in capsid-size redirection and not a result of ICP1 escaping PLE. These data demonstrate that TcaP is necessary for PLE’s redirection of ICP1 coat protein into small capsids and that PLE does not encode a redundant mechanism for capsid remodeling.

Next, we measured the transduction of PLE’s genome following ICP1 infection in PLE∆tcaP hosts using a previously described assay using an antibiotic resistance marker in PLE (O’Hara et al., 2017). Interestingly, in the absence of tcaP, PLE transduction was tenfold less efficient than wild type (Figure 2D). The defect in the transduction of the PLE∆tcaP strain was largely restored with ptcaP (Figure 2D). Together with the morphological data of transducing particles, these data suggest that TcaP-mediated capsid remodeling facilitates more efficient horizontal spread of the PLE genome to recipient V. cholerae. PLE is the first satellite to show dependency on small capsids for efficient transduction.

Capsid scaffolding proteins are characteristically responsible for directly promoting the assembly of coat proteins and controlling capsid size (Dokland, 1999). TcaP expression results in the formation of small capsids during ICP1 infection (Figure 1C), which suggests that it has scaffolding activity. However, it is possible that TcaP does not act directly on the coat proteins as a scaffold, but rather interferes with ICP1’s capsid morphogenesis pathway in some way that decreases capsid size. To directly address if TcaP is a scaffold, we set up a heterologous PLP assembly platform in E. coli similar to those previously described (Spilman et al., 2012; Wang et al., 2006; Cerritelli and Studier, 1996). Briefly, we co-expressed ICP1’s coat and putative scaffolds from either ICP1 or PLE and monitored the production of PLPs. By adding a C-terminal six-histidine (6xHis) tag to the coat protein, we were able to purify coat-containing complexes by affinity chromatography, examine their protein content by SDS-PAGE/Coomassie staining, and their morphology by TEM. First, as a control, coat::6xHis (referred to as ‘coat’ for simplicity) was expressed and purified. The coat proteins eluted as complexes, but they were irregular in size and shape, often forming spirals (Figure 3A1, Figure 3—figure supplement 1), as is expected for coat proteins in the absence of scaffolds (Dokland, 1999). As a second control, we confirmed that ICP1’s putative scaffold could assemble coat proteins into uniform, ICP1-sized PLPs (Figure 3A2, Figure 3—figure supplement 1). The presence of scaffold in these PLPs was supported by SDS-PAGE analysis in which we observed a band corresponding to the predicted size of the scaffold (39.3 kDa), indicating, as expected, that the scaffold co-eluted with coat-containing complexes (Figure 3A2). Interestingly, two smaller bands of approximately 23 and 17 kDa also appeared in these samples. The size of these bands is consistent with the cleavage of the scaffold resulting in two fragments. As scaffolds are generally not a part of the mature capsid, they are removed from the procapsid by self-cleavage (Robertson et al., 1997; Chang et al., 2008) or cleavage by a protease (Duda et al., 2013; Huet et al., 2016). Indeed, in the presence of protease inhibitors during purification, these bands were not observed, but the same-sized particles were produced, suggesting that scaffold cleavage is not a prerequisite for assembly (Figure 3—figure supplement 2A). Analysis of ICP1 PLPs by mass spectrometry (LC-MS/MS) further supported the presence of ICP1’s putative scaffold (Supplementary file 2). Together, these data confirm the predicted role of ICP1’s scaffold and demonstrate that PLPs can be assembled and purified using this platform.

Other known satellite systems have varying dependencies on the phage-encoded scaffold and/or additional satellite-encoded factors for the assembly of small procapsids (Damle et al., 2012; Chang et al., 2008; Marvik et al., 1994; Six, 1975). To test whether TcaP requires ICP1’s scaffold to produce small PLPs, we co-expressed coat with either just TcaP or with both TcaP and ICP1’s scaffold simultaneously. First, analysis of the stained SDS-PAGE gel following co-expression of coat and TcaP showed that TcaP bound to coat and was co-eluted (Figure 3A3). TEMs showed that TcaP was sufficient to assemble ICP1’s coat into PLPs that were homogenously PLE-sized (Figure 3A3, Figure 3—figure supplement 1). These data demonstrate that TcaP has scaffolding activity for ICP1’s coat protein and that it does not require additional factors to assemble small PLPs. Next, we constructed a plasmid expressing the scaffold and the coat as they occur natively in ICP1’s genome downstream of one of the promoters. Here, the gene gp123 is present between the scaffold and coat and predicted to encode for the decoration protein. The putative decoration protein was not expected to be incorporated into the PLPs as decoration proteins are typically added after genome packaging, and capsid expansion exposes their binding site, as has been shown for phage Lambda (Lander et al., 2012). In line with this, the stained gel of the resulting purified ICP1-sized PLPs indicated that while the ICP1 scaffold co-eluted with coat-containing complexes as expected, the decoration protein was produced but not incorporated into the assembled particles (Figure 3A4, Figure 3—figure supplement 1). Interestingly, when we co-expressed ICP1 coat, decoration, and scaffold along with TcaP, we did not observe any large particles, suggesting that TcaP is dominant over ICP1’s scaffold (Figure 3A5, Figure 3—figure supplement 1). SDS-PAGE analyses of particles produced in the presence of ICP1’s scaffold and TcaP confirmed the presence of TcaP as well as the phage-encoded scaffold and its cleavage products, suggesting that TcaP does not block the incorporation of ICP1’s scaffold into procapsids (Figure 3A5). This experiment was repeated in the presence of protease inhibitors which eliminated the ICP1 scaffold cleavage products but reproduced the small particle morphology (Figure 3—figure supplement 2B). Collectively, these data demonstrate scaffolding activity for both ICP1’s scaffold and PLE’s TcaP and show that TcaP can assemble small PLPs in the absence or presence of ICP1’s scaffold.

Having observed that TcaP is sufficient to make small PLPs in the heterologous assembly platform as well as when expressed from a plasmid during ICP1 infection, we next tested coat^R223H and coat^E234K, the previously identified in vivo genetic escapes of TcaP activity (Figure 1E), in the assembly platform. The coat^E234K variant was not assembly competent in our heterologous assembly platform (Figure 3—figure supplement 4), so we only continued with the coat^R223H variant. As expected, in the absence of any scaffolding proteins, coat^R223H formed similar spiral complexes as those observed with the wild type coat (Figure 3B1, Figure 3—figure supplement 1). These data demonstrate that the R223H substitution does not compromise the protein’s ability to bind to itself, nor does the substitution provide a means for the coat to form PLPs independent of a scaffold. Next, we expressed coat^R223H with TcaP, anticipating we would observe the formation of large ICP1-sized PLPs because this substitution was sufficient for ICP1 to avoid TcaP’s inhibitory activity (Figure 1E). Unexpectedly, TcaP still directed the assembly of small PLPs comprised of coat^R223H (Figure 3B3, Figure 3—figure supplement 1). However, TcaP was not robustly evident in the purified complexes as assessed by SDS-PAGE, and some particles were spirals, similar to those seen when the coat is expressed without a scaffold (Figure 3B3, Figure 3—figure supplement 1). These data suggest that TcaP’s scaffolding activity was partially compromised when the coat protein carried the R223H substitution. We hypothesized that with ICP1’s scaffold the coat^R223H proteins could be assembled into ICP1-sized PLPs. Leveraging the dual expression from a single promoter (Figure 3A4), we confirmed that the ICP1 scaffold alone directed the assembly of ICP1-sized PLPs with coat^R223H (Figure 3B4, Figure 3—figure supplement 1), but we observed a mix of PLE and ICP1-sized PLPs following addition of TcaP (Figure 3B5, Figure 3—figure supplement 1). We can attribute this phenotype to the substitution in the coat and not to the presence of protease inhibitors used during the purification as the addition of protease inhibitors during purification of the coat^{wild type} co-expressed with the ICP1 scaffold and TcaP resulted in only small PLPs (Figure 3—figure supplement 2B). These data suggest that the R223H substitution in ICP1’s coat can only partially escape TcaP-mediated small capsid formation, but in vivo this level of escape is sufficient to allow for approximately equal levels of viable progeny production as is seen in the absence of TcaP.

The data obtained from the assembly of PLPs in E. coli demonstrate that TcaP has scaffolding activity; however, these data do not distinguish between TcaP acting as an internal or an external scaffold. Therefore, we validated the protein content of PLE PLPs produced from co-expression of TcaP and coat by LC-MS/MS and subjected them to cryo-EM. As anticipated, LC-MS/MS confirmed PLE PLPs were comprised of coat and TcaP (Supplementary file 2). Representative micrographs from negatively stained samples (Figure 4A) and vitrified particles (Figure 4B) show a distinctly bumpy surface on the PLE PLPs, suggestive of external proteins. A total of 379,643 particles were used for icosahedral reconstruction of the PLE PLP particle, with a final resolution of 3.4 Å (Figure 4C, Figure 4—figure supplement 1). The external density of the PLPs corresponded to TcaP, demonstrating that it functions as an external scaffold. Internal density was assigned to coat proteins with a canonical HK97-like fold, which exist as pentamers and hexamers within the 48 nm, T = 4 icosahedral PLP. We used AlphaFold2 (Jumper et al., 2021) to predict the structures of the coat protein and TcaP, which were fitted into density from the cryo-EM and aided in modeling nearly all of the coat protein and approximately half of TcaP (residues 34–172) for which only a few side chains could be modeled. As expected, the A-domains of the coat proteins were centered in both pentamers and hexamers. TcaP dimers meet and form trimeric interactions at the threefold axes between hexamers. Broadly, the reconstruction of PLE PLPs shows TcaP functions as an external scaffold and provides context for how TcaP’s higher-order structure regulates the number of coat proteins that can be accommodated within the assembling PLP.

Further, the reconstruction reveals details of the residues that were substituted in phages that escape TcaP-mediated capsid remodeling (R223H and E234K). Both residues are found in the A-domain which is oriented near the center of the hexamer where TcaP binds (Figure 4D). The arginines at position 223 were clearly visible in the map. Due to the twofold symmetry in the hexamer, there are identical interactions between each half of the hexamer and the TcaP dimer. Importantly, there are three distinct interactions between TcaP and R223 (Figure 4E). First, R223 from coat_D is in proximity to TcaP’s residues 87–93, while coat_B has R223 positioned near TcaP’s residues 63–70, and R223 from coat_C lies close to the second TcaP subunit. In this third interaction, the negatively charged aspartic acid at position 218 from the coat coordinates with the positive charges from the arginine residues from coat (R223) and TcaP (R133) and creates a salt bridge with residues measuring less than 4 Å apart. The electrostatics of the complex, with the TcaP dimer oriented as in Figure 4D and F, show TcaP’s R133 fitting into a negatively charged pocket on the coat (Figure 4G). The side chain for the other residue substituted in the escape phage, E234, could not be confidently modeled, but its localization within the capsid is outside of the TcaP binding region. The exact nature of E234 is not yet clear, but perhaps it has a role in stabilizing the A-domain. Together, these data support the conclusion that ICP1 escapes TcaP’s scaffolding activity by altering the coat to affect TcaP’s binding site.

If ICP1 is under selective pressure to escape PLE-mediated capsid remodeling in nature, ICP1 isolates could be expected to harbor these coat substitutions. However, comparison of the coat alleles from 67 genetically distinct ICP1 isolates revealed no differences within the A-domain of the alleles. These data show that ICP1 does not naturally encode mutations in the A-domain that would escape TcaP-mediated capsid remodeling.

Having established that PLE1’s TcaP acts as an alternative external scaffold that directs the assembly of ICP1’s coat protein into small capsids, we assessed the conservation of tcaP across genetically distinct PLEs. While PLE1 is the most well-studied PLE, there are nine additional PLEs that have been discovered to date (Angermeyer et al., 2021). Every PLE encodes a tcaP allele in the same locus, which encodes for proteins sharing 62–100% amino acid identity with TcaP^PLE1 (Figure 5A, Figure 5—figure supplement 1A). Largely, the regions that contact coat proteins in the PLP are conserved (Figure 5—figure supplement 1C). The conservation of this protein in the 10 PLEs suggests that all PLEs share the TcaP-mediated capsid remodeling strategy.

Curiously, the most divergent tcaP allele, tcaP^PLE5, is 237 nucleotides (~26%) shorter than tcaP^PLE1 due to an apparent truncation of the N-terminal coding region. However, a nucleotide alignment of the tcaP^PLE1 sequence with the tcaP^PLE5 5′-UTR through the end of the coding sequence revealed two small deletions, one of which results in a frameshift and premature stop codon in the tcaP^PLE5 allele (Figure 5A and B). The alternative start site downstream of the deletions would restore the original reading frame for tcaP^PLE5 (Figure 5B). Given the conservation of full-length tcaP alleles in other PLEs, and the structural data highlighting interactions between the TcaP^PLE1 with ICP1’s coat proteins, we assessed if the truncated TcaP^PLE5 was functional to redirect ICP1 virion assembly.

To start, we generated PLE particles from the wild type PLE5 strain carrying an empty vector and assessed their morphology by TEM. We found the PLE5 virions did not have small capsids, but rather they had large, ICP1-size capsids (Figure 5C), similar to virions from the PLE1∆tcaP strain (Figure 2B). Consistent with this, when we directly addressed the scaffolding activity of the truncated allele by expressing it in trans during ICP1 infection and assessing plaque formation, we found that unlike ptcaP^PLE1, which inhibited ICP1 plaquing (Figure 1E), ptcaP^PLE5 did not inhibit ICP1 plaque formation (Figure 5E), supporting the conclusion that TcaP^PLE5 is not a functional scaffold. The reconstruction of the PLE1 PLPs (Figure 4C) revealed residues 63–70 in TcaP^PLE1 that specifically contact coat_B. That region is notably lacking from the truncated TcaP (Figure 5—figure supplement 1C). Further, we predict that a TcaP protein missing 79 amino acids would not be long enough to form the scaffolding cage around an assembling T = 4 procapsid explaining the lack of scaffolding activity for this TcaP variant.

Next, we addressed the transduction efficiency of PLE5 since the efficiency of PLE1’s transduction decreased in the absence of TcaP (Figure 2D), demonstrating an advantage to packaging the PLE genome into smaller capsids. Previous work reported similar transduction efficiency for PLEs1-5 (O’Hara et al., 2017), and we recapitulated those results here, showing that PLE5 has only a subtle, approximately twofold decrease in transduction efficiency relative to PLE1 (Figure 5F). As this seemed in conflict with the PLE1∆tcaP data, where there was over a tenfold decrease in PLE1 transduction without TcaP (Figure 2D), we directly examined the effect of small capsid production on PLE5 transduction by expressing TcaP^PLE1 in trans. First, we confirmed that ptcaP^PLE1 in PLE5(+) V. cholerae during ICP1 infection directed the assembly of small capsids (Figure 5D), which supported our hypothesis that the unmodified capsids produced by PLE5 are the result of this PLE encoding a nonfunctional TcaP allele. Next, we found that PLE5 transduction increased with ptcaP^PLE1, exceeding the transduction efficiency of PLE5 and PLE1 (Figure 5F). This result suggests that the capacity for transduction is different for PLE1 and PLE5, but still nonetheless demonstrates that small capsids promote PLE transduction.

Since both PLE1 and PLE5 transduction efficiency was higher when small capsids were made, this raised the question of PLE5’s maintenance in the V. cholerae population while lacking a functional scaffold. A recent analysis of sequenced V. cholerae isolates outlined a pattern where a PLE variant emerges, often rises to dominance, then is replaced by a different variant (Angermeyer et al., 2021). This pattern revealed that after its dominance from before 1960 until its disappearance in 1991 (for which there are 21 V. cholerae isolates harboring identical PLE5 sequences), PLE5 re-emerged in 2016 and 2017. Genetic drift within a given PLE variant appears to be very rare: a particular PLE variant is typically 100% conserved at the nucleotide level across the entire mobile genetic element despite residing in different isolates of V. cholerae (O’Hara et al., 2017; McKitterick et al., 2019). However, on rare occasions, single-nucleotide polymorphisms between the same PLE variant are found in different V. cholerae isolates (Barth et al., 2021). Strikingly, the tcaP^PLE5 sequence from 2017 showed a 14-nucleotide insertion which restored a full-length tcaP allele (Figure 5G, Figure 5—figure supplement 1B). The re-establishment of the full-length tcaP allele in the contemporary PLE5, along with the preservation of full-length alleles in other PLEs, supports conservation of TcaP for its role in capsid remodeling to promote PLE transduction and inhibit ICP1.

V. cholerae and E. coli were propagated at 37°C on LB agar or in LB broth with aeration. Where needed, antibiotics were used at the following concentrations: streptomycin 100 µg/mL, kanamycin 75 µg/mL, spectinomycin 100 µg/mL, chloramphenicol 2.5 µg/mL (solid media) or 1.25 µg/mL (liquid media) for V. cholerae and 25 µg/mL for E. coli, and carbenicillin 50 µg/mL.

V. cholerae carrying PLE1 marked with a kanamycin resistance cassette downstream of the last ORF (O’Hara et al., 2017) was made naturally competent and transformed with DNA fragments containing a spectinomycin resistance marker flanked by frt recombinase sites assembled to up and downstream regions of homology by splicing by overlap extension PCR as previously described (Dalia et al., 2014). Spectinomycin-resistant transformants were transformed with a plasmid carrying an FLP recombinase which was induced by 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) (Fischer BioReagents, 367-93-1) and 1.5 mM theophylline (Sigma, T1633-100G), allowing for the removal of the spectinomycin resistance cassette via recombination, resulting in an in-frame deletion. The strains were cured of the plasmid, and deletions were confirmed by PCR and Sanger sequencing.

Plasmid constructs were assembled using Gibson Assembly and transformed into E. coli BL21 or mated into V. cholerae via E. coli S17. A list of strains used in this study can be found in the Key resources table. Plasmids in V. cholerae have a P_tac promoter downstream of a theophylline sensitive riboswitch. A list of oligos used in this study can be found in Supplementary file 3.

Overnight cultures of V. cholerae were diluted to OD₆₀₀ = 0.05 and grown in LB (supplemented with antibiotics where appropriate) with aeration at 37°C either directly to OD₆₀₀ = 0.3 or to OD₆₀₀ = 0.2 then induced with 1 mM IPTG and 1.5 mM theophylline for 20 min to reach OD₆₀₀ = 0.3, then mixed with pre-diluted phage samples. Phage attachment was allowed for 7–10 min prior to plating in 0.5% molten top agar (supplemented with antibiotics and inducer where appropriate) followed by overnight incubation at 37°C. Resulting individual plaques were counted.

V. cholerae carrying a plasmid encoding tcaP was used in a plaque assay as described above. As a control, ICP1_2011_Dha_A was used to infect V. cholerae expressing an empty vector. Plaques that formed on the tcaP-expressing strain were picked and purified on that strain two more times by plaque assay. High-titer phage stocks were prepared by sodium chloride (1 mM) polyethylene glycol 8000 (10%) precipitation or by centrifugation (26,000 × g for 90 min) and stored in STE (5 mM Tris pH 8.0, 100 mM NaCl, 1 mM EDTA). Prior to collecting genomic DNA (gDNA), phage stocks were treated with DNase for 30 min at 37°C to remove non-encapsidated DNA, then the enzyme was heat inactivated. gDNA was collected using a QIAGEN DNeasy blood and tissue DNA purification kit (QIAGEN, 69506) according to the manufacturer’s protocols and genomic libraries were prepared for Illumina sequencing using the NEBNext Ultra II DNA Library preparation kit (NEB #E7645, E7103) as described in the manufacturer’s protocols. Using an Illumina HiSeq4000 (University of California, Berkeley QB3 Core Facility), samples were sequenced by paired-end (2 × 150 bp). The genomes were assembled using SPAdes and analyzed using BreSeq (v0.33).

50 mL cultures of V. cholerae strains carrying plasmids were grown and induced as described above. ICP1 was added at a multiplicity of infection (MOI) of 2.5 and cultures were incubated until lysis (30–90 min). 1 mL of lysate was used for transduction assays (see below). Remaining lysates were concentrated by centrifugation at 26,000 × g for 90 min, resuspended in Phage Buffer 2.0 (50 mM Tris–HCl, 100 mM NaCl, 10 mM MgSO₄, 1 mM CaCl₂) overnight, treated 1:1 with chloroform (Fisher Scientific, C606-1) for 15 min, and centrifuged at 5000 × g for 15 min. The aqueous layer was collected and 5 µL was applied to a grid for TEM.

PLE transduction assays were carried out as previously described (Netter et al., 2021; O’Hara et al., 2017). Briefly, 1 mL of lysates from strains carrying PLE marked with a kanamycin resistance cassette downstream of the last ORF were treated with 10 µL chloroform, which was removed, along with bacterial debris, by centrifugation at 5000 × g for 15 min. The supernatant was collected and mixed 10:100 with a saturated overnight culture of spectinomycin resistant recipient V. cholerae cells (∆lacZ::spec^R) supplemented with 10 mM MgSO₄ immediately prior to transduction. Recipients were incubated with lysates for 20 min at 37°C with shaking (220 rpm) and then serially tenfold diluted. The resulting dilutions were plated on LB agar plates supplemented with spectinomycin and kanamycin. A colony represents one PLE virion/transducing unit.

For the preparation of grids, 5 µL samples were incubated on a copper mesh grid (Formvar/Carbon 300, Electron Microscopy Sciences) for 60 s, wicked, immediately washed with sterile ddH₂O for 15 s, wicked, immediately stained with 1% uranyl acetate (Electron Microscopy Sciences, 22400-1) for 30 s, wicked and allowed to dry completely. Micrographs were collected with a FEI Tecnai-12 electron microscope operating at 120 kV.

Using Gibson Cloning, the coat gene fused to 6xHis was inserted into the pETDUET vector or the untagged coat was cloned into pCDFDuet. Either the ICP1 or PLE scaffold proteins were similarly inserted via Gibson Assembly downstream of a second T7 promoter. In other variations of this plasmid, the genes were cloned with the intergenic sequences found in ICP1_2011_Dha_A. The pETDUET constructs were transformed into E. coli BL21 and grown on LB agar or in LB broth supplemented with carbenicillin (50 µg/mL). To reduce aggregation of PLE PLPs for cryo-EM, a pETDUET construct encoding gp122::6xhis and tcaP was co-transformed with a pCDFDuet construct expressing gp122 (untagged), grown on LB agar or in LB broth supplemented with carbenicillin (50 µg/mL) and streptomycin (100 µg/mL). For protein production, overnight cultures were diluted 1:100 in 0.5–1 L, grown to OD₆₀₀ = 0.2–0.4, induced with 1 mM IPTG for 3–5 hr, and collected by centrifugation at 4000 × g for 20 min at 4°C. Pellets were then resuspended in 10% of the volume of Phage Purification Buffer (Phage Buffer 2.0 supplemented with 20 mM imidazole, 1 mM BME, and protease and phosphatase inhibitors [Cat# A32961/A32965, Pierce]), and frozen at –80°C. For protein purification, frozen samples were thawed, sonicated, and centrifuged at 12,000 × g for 60–90 min to remove membranes and debris. The particles were then pelleted at 26,000 × g for 90 min. The pellet was nutated overnight in Phage Purification Buffer, and the nutate was loaded onto a column packed with HisPur nickel-nitrilotriacetic acid (Ni-NTA) resin (Cat# PI88222, Thermo Scientific) pre-equilibrated with ice-cold Phage Purification Buffer. After two passes over the column, the flow-through was collected and the column was washed with wash buffer (Phage Buffer 2.0 supplemented with 50 mM imidazole, 1 mM BME, and 10% glycerol). Proteins were then eluted with one column volume of Elution Buffers 1–3 (Phage Buffer 2.0 supplemented with 1 mM BME and 10% glycerol and 150, 250, or 350 mM imidazole) and a final six column volume elution with Elution Buffer 4 (Phage Buffer 2.0 supplemented with 1 mM BME and 10% glycerol and 500 mM imidazole). For all experiments with coat^R223H and the noted replicates of wild type coat, decoration, scaffold, and TcaP, protease and phosphatase inhibitors were added to all purification and elution buffers. Aliquots of the fractions were either boiled in Laemmli buffer and assessed by SDS-PAGE stained with Coomassie or applied to a grid and imaged by TEM.

Cesium chloride-based purification following affinity purification was carried out as follows. PLE PLP (coat::6xHis +TcaP) samples from the final elutions from affinity chromatography were concentrated on a 100K MWCO Amicon ultra filter (Cat# UFC810024, Millipore) to a final volume of ~1 mL. 1 mL steps of 1.6 g/cm³, 1.5 g/cm³, 1.4 g/cm³, and 1.3 g/cm³ CsCl were made in Phage Buffer 2.0 (50 mM Tris–HCl, 100 mM NaCl, 10 mM MgSO₄, 1 mM CaCl₂ pH 8.0), filter sterilized, and layered in ultracentrifuge tubes (thinwalled WX 5 mL, Cat# 1131, Thermo Scientific) then topped with 500 µL of PLE PLP sample. Samples were spun for 2 hr at 36,000 rpm (~110,000 × g) at 18°C (AH-650 swinging bucket rotor Thermo Scientific), then 250 µL fractions were manually collected. The protein content of fractions was assessed by SDS-PAGE/Coomassie. Fractions containing the most pure coat::6xHis and TcaP complexes were pooled, dialyzed against Phage Buffer 2.0 in 20k MWCO Slide-A-Lyzer dialysis cassettes (Cat# 66003, Thermo Scientific), concentrated on Amicon ultra filters (Cat# UFC810024, Millipore), and prepared for cryo-EM.

Size-exclusion purifications were carried out as follows. Lysates were prepared as described above and particles were pelleted and nutated in Phage Buffer 2.1 (same as Phage Buffer but with pH 7.4). The nutate was then spun at 12,000 × g for 10 min to remove aggregates then applied to an AKTA HiPrep 16/60 Sephacryl S-500 HR (Cat# 28935606, Cytiva) column and passed through at a flow rate of 0.5 mL/min. Protein content from elution peaks was assessed by SDS-PAGE/Coomassie, and samples containing PLE PLPs (elutions at 26–32 mL) were concentrated by centrifugation at 26,000 × g for 90 min, resuspended in ~100–500 µL Phage Buffer 2.0, and prepared for cryo-EM.

Cryo-EM samples were prepared by applying 5 μL aliquots of purified PLE PLPs to R2/2 Quantifoil grids and R2/2 Quantifoil grids coated with 2 nm ultrathin Carbon (QUANTIFOIL) that had been glow discharged for 45 s in a Pelco Easiglow glow discharging unit. The samples were plunge frozen in liquid ethane using a Vitrobot Mark IV operated at 4°C and 100% humidity, with a blot force of 1 and 5 s of blotting time per grid. The grids were screened at RTSF Cryo-EM facility using a Talos Arctica equipped with a Falcon 3EC direct electron detector and Cryo-EM data were collected at Purdue Cryo-EM facility using a Titan Krios equipped with a K3 direct electron detector, and operating at 300 keV with a post-column GIF (20 eV slit width) under low-dose conditions. Micrographs were collected at 64,000× nominal magnification (0.664 Å/pixel) by recording 40 frames over 3.1 s for a total dose of 36.41 e^-/Ų.

Data processing for icosahedral reconstruction of PLE procapsids was carried out using cryoSPARC v4.0.3. Briefly, the dose-fractionated movies were subjected to motion correction using patch motion correction with 2× binning, and CTF estimation of the resulting images was done using the patch CTF estimation jobs. The particles were picked using the Template picker job with templates generated from blob picker. Particles were then extracted and subjected to 2D classification. Cryo-EM image processing statistics, along with data collection parameters, are listed in Figure 4—source data 1. A total of 790,843 particles were used for 3D refinement with C1 symmetry, with a model generated from previous low-resolution data serving as the initial model. Refined particles were then subjected to a round of 3D classification in RELION 4.0 (Kimanius et al., 2021; Asarnow et al., 2019). A total of 379,643 particles were then imported into cryoSPARC (Punjani et al., 2017) and used for 3D refinement with icosahedral symmetry. The overall resolution was estimated using Postprocess job with a spherical mask in RELION 4.0 based on the gold-standard Fourier shell correlation (FSC) = 0.143 criterion (Kimanius et al., 2021; delaRosa-Trevín et al., 2016). The final map was sharpened with a B-factor of –130. The final maps were deposited into EMDB (accession number EMD-29675).

AlphaFold2 (Jumper et al., 2021) was used to generate homology models for all modeled protein chains. Initial models were then docked into EM maps for further refinement. For TcaP, only residues 34–172 were modeled. Refinement was carried out using Phenix (Liebschner et al., 2019) and model adjustments were carried out in COOT (Emsley et al., 2010). Model parameters were monitored using Molprobity in Phenix, and the values are listed in Figure 4—source data 2.

PLE and ICP1 PLPs subjected to mass spectrometry were further concentrated by trichloroacetic acid (TCA) precipitation, washed (10 mM HCl 90% acetone) three times, and air dried. The pellets were then resuspended in 100 mM Tris pH 8.5, 8 M urea, digested by trypsin, and analyzed by LC-MS/MS (Thermo LTQ XL linear ion trap mass spectrometer at the Vincent J. Coates Proteomics/Mass Spectrometry Laboratory at UC Berkeley). A sequence database containing E. coli proteins as well as ICP1 coat, scaffold, and decoration, as well as PLE TcaP, were used to compare the masses recorded and identify proteins in the sample.