Diverse eukaryotic viruses require dedicated genome replication compartments to concentrate host- and virusencoded replication machinery and shield their genomes from host defenses (133). These compartments can be assembled from host cell membranes or virally-encoded proteins, or formed via liquid-liquid phase separation. In contrast, no compartments are known to be required for genome replication of prokaryotic viruses. Bacteriophages of the recently described family Chimalliviridae are the only prokaryotic viruses known to assemble a compartment, the phage nucleus, within which their genomes replicate (436) (Fig. 1A). The phage nucleus protects the phage genomes from host defenses including DNA-targeting CRISPR/Cas systems and restriction enzymes, but whether it is intrinsically essential for genome replication has remained unclear (1, 739). The phage nucleus is enclosed by a continuous 2-D lattice composed primarily of the virallyencoded protein chimallin (ChmA) which gives Chimalliviridae its name ( 8, 10, 11 ). In a striking parallel to the subcellular organization orchestrated by the eukaryotic nucleus, this phage-generated organelle selectively imports DNA replication and transcription machinery and excludes metabolic enzymes and ribosomes in the bacterial cytoplasm. Here we applied a newly developed protein expression knockdown technology, CRISPRi-ART, to investigate whether the phage nucleus is essential for Chimallivirus genome replication.

CRISPRi-ART as a Genetics Tool for Nucleus-Forming Phages. While Chimalliviridae phages sequester their dsDNA genomes within the ChmA-based phage nucleus, rendering them inaccessible to traditional genome editing tools, viral mRNAs are exported to the cytoplasm to be translated by host ribosomes (436). Taking advantage of this vulnerability, we knocked down ChmA via CRISPRiART to determine how inhibiting phage nucleus assembly affects genome replication of E. coli phage Goslar. CRISPRiART (CRISPR interference by Antisense RNA Targeting) uses catalytically inactive Ruminococcus flavefaciens Cas13d (dRfxCas13d) to specifically suppress expression of proteins of interest (Fig. 1B) ( 12, 13 ). Guide-directed binding of targeted RNA triggers collateral RNAse trans-activity in wildtype Cas13, leading to death or dormancy of bacterial cells (14317). However, mutation of key active site residues in dRfxCas13d9s two HEPN (higher eukaryote and prokaryote nucleotide-binding) domains allows it to bind target RNA without inducing nuclease activity ( 12, 13, 18 ). Therefore, dRfxCas13d can be used to inhibit expression of a protein by binding and thereby occluding the translational start site of the corresponding mRNA ( 12, 13, 19 ). CRISPR i-ART has been shown to be highly effective as a genetics tool for diverse phages, including Goslar ( 12 ).

ChmA Knockdown Inhibits Goslar Reproduction. To characterize the effect of ChmA knockdown on Goslar reproduction, we confirmed the efficacy of two chmA transcript-targeting guides, ChmAg1 and ChmAg2 (Fig. 1C). We collected whole-cell lysates of Goslar-infected E. coli MC1000 strains coexpressing each guide with dRfxCas13d 90 minutes post infection (mpi) and determined ChmA expression levels via western blot (Fig. 1C, Fig. S1). Both guides dramatically reduced ChmA expression, but only ChmAg1 was used in our study as it consistently induced a slightly stronger knockdown.

Goslar9s efficiency of plaquing (EOP) was reduced ~1,000fold when infecting the host strain expressing dRfxCas13d and ChmAg1 (<knockdown strain=) compared to the host strain expressing dRfxCas13d and a non-targeting guide (<non-targeting strain=) (Fig. 1E-F), suggesting that ChmA is required for phage reproduction. Complementation by expression of ChmA from a chmA gene re-coded to avoid ChmAg1 targeting (chmA*, Fig. 1C) restored the EOP ~200fold (FL* for <full-length ChmA*=, Fig. 1G-I). This demonstrated that inhibition of phage reproduction was caused by lack of ChmA and not due to off-target effects of dRfxCas13d.

We previously observed that deleting the C-terminal segments (CTS1 and CTS2) of Pseudomonas chlororaphis phage 201phi2-1 ChmA abolished self-assembly into oligomers in vitro ( 10 ). Therefore, to determine whether ChmA lattice assembly is required for Goslar reproduction, we investigated whether ChmA* lacking both C-terminal segments (&CTS*, Fig. 1G) could complement ChmA knockdown. Not only did ChmA&CTS* fail to complement ChmA knockdown, but its expression reduced Goslar EOP by an additional 1000-fold (Fig. 1H-I). Notably, ChmA&CTS* also acted as a dominant negative mutant, severely reducing Goslar EOP when expressed in the non-targeting strain and when expressed in the absence of the knockdown system (Fig. 1HI, Fig. S2). Previous fluorescence microscopy studies have shown that eGFP-tagged C-terminal segment ChmA truncations associate with the wildtype ChmA lattice produced by 201phi2-1 during infection, indicating that they co-assemble with the full-length protein ( 10 ). This suggests that ChmA&CTS*, untagged and expressed at a high concentration, likely poisons the lattice of phage-generated wildtype ChmA, preventing phage nucleus assembly.

Thus, two independent methods for inhibiting ChmA lattice assembly inhibit lytic phage growth. These results demonstrate that ChmA, and therefore the phage nucleus, is essential for Goslar9s life cycle.

The ChmA Lattice is Required for Phage DNA Replication. We used single-cell fluorescence microscopy to determine the stage of infection at which Goslar9s life cycle was inhibited by ChmA knockdown. At 90 mpi, infected no ntargeting cells were swollen, elongated and contained large phage nuclei visible by brightfield microscopy and DAPI staining (Fig. 2A, Fig. S3). The host chromosomes of infected cells were also partially degraded. This phenotype is characteristic of Goslar infections ( 9 ), indicat ing that nontargeting CRISPRi-ART did not disrupt normal infection progression. Infected ChmAg1 knockdown cells were also significantly swollen and elongated at 90 mpi but a lmost none of them had developed mature phage nuclei. Instead, infected cells contained small DAPI-stained puncta surrounded by a zone devoid of host chromosomal DNA (Fig. 2B, Fig. S3). These DAPI-stained puncta appeared similar to the initially injected phage genomes previously observed early in the Chimalliviridae life cycle ( 9, 20 ). Therefore, to determine whether the DAPI-stained puncta were Goslar genomes arrested at an early stage of infection, we investigated whether sfGFP-ChmA colocalized with them. sfGFP-ChmA can be expressed in the knockdown strain because the target sequence is not at the translational start site of the fusion protein ( 12, 13 ). We expressed sfGFPChmA at a low concentration from a second plasmid to avoid complementation of ChmA knockdown. In nontargeting cells, sfGFP-ChmA formed a punctum that colocalized with the DAPI-stained punctum of initially injected phage DNA early during infection (10 mpi) and integrated into the ChmA lattice of the mature phage nuclei surrounding the phage DNA at later time points (45 and 90 mpi) (Fig. 2D). This recapitulated the previously observed pattern of GFP-tagged ChmA localization over the course of Goslar infection ( 9 ). In knockdown cells, sfGFP-Chm A colocalized with the DAPI-stained puncta (>80% colocalization, n = 106, 122, 122) throughout infection (10, 45 and 90 mpi) ( Fig. 2E). sfGFP alone does not colocalize with the DAPI-stained puncta and is excluded from mature phage nuclei, remaining diffuse in the cytoplasm (Fig. S4). By quantifying the cross-sectional area of DAPI-stained phage DNA as well as phage DNA quantity in terms of total DAPI intensity, we found that ChmA depletion severely inhibited both phage nucleus expansion and viral DNA replication (Fig. 2F, H). In fact, the area and DAPI intensity of phage DNA in the knockdown strain were comparable to that of Goslar particles outside of cells in the same microscopy fields (Fig. 2C, G, I), suggesting that intracellular DAPI-stained puncta contain only a single, non-replicating Goslar genome (Fig. 2C). In the knockdown strain, phage DNA area was >0.1 µm2 in only 5.9% of infections (n = 171) compared to 0.0% for extracellular phage (n = 171, Fig. 2G) and 90.5% for non-targeting strain infections (n = 178, Fig. 2F). Similarly, the total DAPI intensity of phage DNA was >50 AU in 11.1% of infections in the knockdown strain (n = 171) versus 0.6% for extracellular phage (n = 171, Fig. 2I) and 97.8% for nontargeting strain infections (n = 178, Fig. 2H). Time-lapse brightfield microscopy revealed that arrested Goslar infections do not proceed even after 210 mpi (Fig. 3A). Unlike non-targeting strain infections (n = 248), for which the average time-to-lysis is ~160 minutes, infected knockdown strain cells remained intact (n = 271), often continuing to slowly grow in length (but not dividing or septating) up to at least 210 mpi without developing visible phage nuclei (Supplementary movies 1-2). Taken together, these data suggest that phage DNA is injected but fails to replicate during ChmA knockdown. We next investigated whether ChmA* complementation restored Goslar DNA replication. Introduction of the empty vector of our second plasmid for ChmA* (FL and &CTS) expression did not significantly alter phage nucleus size and DNA quantity in the non-targeting and knockdown strains (Fig. 3C). Complementation by untagged, full length, recoded ChmA (FL*, Fig. 1C and G) greatly restored phage nucleus expansion and DNA replication. Although frequently smaller than in the non-targeting strain (Fig. 3C), phage DNA area was >0.1 µm2 in 70.2% of infections when ChmA-FL* was expressed in the knockdown strain (ChmAg1, FL*; n = 181) compared to 8.6% without ChmAFL* (ChmAg1, EV; n = 198; Fig. 3D). Similarly, phage DNA DAPI intensity was >50 AU, indicating the presence of multiple genome copies (Fig. 2I), in 70.2% of infections when ChmA-FL* was expressed in the knockdown strain (ChmAg1, FL*; n = 181) compared to only 14.1% without ChmA-FL* (ChmAg1, EV; n = 198; Fig. 3H). In contrast, ChmA—CTS*, which acts as a dominant negative mutant in plaque assays (Fig.1F-I), failed to restore DNA replication during ChmA knockdown and actually inhibited phage nucleus growth and DNA replication in the non-targeting strain (NT, —CTS*). When ChmA&CTS* was expressed, phage DNA area was >0.1 µm2 in 9.1% of non-targetin g strain infections (NT, &CTS*; n = 197; Fig. 3E) and in 5.6% of knockdown strain infections (ChmAg1, &CTS*; n = 196; Fig. 3F). In keeping with previous results, phage DNA DAPI intensity was also reduced by this dominant negative mutant. ChmA&CTS* expression resulted in phage DNA DAPI intensity >50 AU in 11.2% of non-targeting strain infections (NT, &CTS*; n = 197; Fig. 3I) and 7.1% of knockdown strain infections (ChmAg1, &CTS*; n = 196 ; Fig. 3J). This independently confirms that ChmA and the phage nucleus compartment are essential for viral genome replication.

A Transcriptionally Active, Membrane-Bound Vesicle Precedes Phage Nucleus Assembly.

The initially injected phage DNA can be visualized as a DAPI-stained punctum in the host cell cytoplasm early in the Chimalliviridae life cycle ( 9, 21 ). In our fluorescence microscopy studies, Goslar9s DNA appeared arrested in a similarly condensed state during ChmA knockdown (Fig. 2 and 3). Using cryo-electron tomography (cryo-ET), we have recently observed vesicles containing DNA in the cytoplasm of E. coli and P. chlororaphis cells infected by Chimalliviridae phages Goslar and 201phi2-1 respectively ( 10 ). Notably, the composition of the interior of these vesicles was the same as the phage nucleus, but using subtomogram analysis we determined that they were surrounded by a lipid bilayer instead of a ChmA coat. We called these vesicles unidentified spherical bodies (USBs) and speculated that they may represent a pre-nuclear stage in the phage life cycle, providing protection for the initially injected viral genome before assembly of the phage nucleus ( 10 ). Based on these observations, we hypothesized that in the absence of ChmA, the injected Goslar genome was trapped within this vesicle. Using cryo-ET, we observed that at 90 mpi, 48.9% of knockdown cells (n = 47, Fig. 4B) contained vesicles with an average diameter of 194 ± 24 nm, s imilar to previously observed USBs enclosed by a lipid bilayer (n = 24, Fig. 4A, D and E). The lipid bilayer nature of these vesicles, which were similar in size to the previously observed USBs ( 10 ), was evident from direct visualization in the tomograms showing two leaflets of high density corresponding to the head groups, and an intermediate region of low density corresponding to the lipid tails, with an overall thickness of ~4 nm (Fig. 4E and Fig. S5). No cells had a ChmA phage nucleus, consistent with the lack of phage nuclei observed in cells from the same sample via fluorescence microscopy (Fig. 4B). Additionally, DAPIstained puncta of phage DNA were detected in cells from the same sample at a comparable frequency as the vesicles were observed by cryo-ET (38.6%, n = 491; Fig. 4B). Thus, before assembly of the ChmA phage nucleus, the injected Chimalliviridae genome is initially enclosed in a lipid bilayer, a structure we name the early phage infection (EPI) vesicle.

It was previously unclear whether the EPI vesicles represented a stage of the viral life cycle or were failed or aborted infections ( 10 ). Surprisingly, we observed polysomes containing at least 10 ribosomes extending from the surface of EPI vesicles (Fig. 4C), demonstrating that these structures are highly transcriptionally active. Chimalliviridae phages use two multisubunit RNA polymerases to carry out transcription: a virion RNA polymerase (vRNAP) and non-virion RNA polymerase (nvRNAP). The vRNAP is packaged into the capsids and injected into the host cell with the viral DNA to transcribe early phage genes (22324). Based on the organization of polysomes extending from the EPI vesicle surface, it is likely that the vRNAP is injected with the genome into the EPI vesicle, where it transcribes mRNAs that are exported to the cytoplasm and translated. Indeed, Antonova et al. recently showed that vRNAP subunit gp180 of Chimallviridae phage PhiKZ colocalizes with the initially injected phage DNA via fluorescence microscopy ( 25 ). We used mass spectrometry to determine whether initial phage gene expression occurs in the absence of ChmA. During ChmA knockdown, many viral proteins were detected at 30 mpi (Fig. 5A, Table S1). Among the most abundant viral proteins detected, statistically significant reduction in expression was only observed for ChmA itself, further demonstrating that the EPI vesicle is transcriptionally active and is a bona fide stage in the Chimalliviridae life cycle.

The ChmA-enclosed phage nucleus selectively imports both host and phage proteins, concentrating DNA replication machinery within this compartment ( 6, 7, 26, 27 ). We hypothesized that the membrane-enclosed EPI vesicle lacks this import ability, thereby preventing accumulation of these key enzymes. To test this idea, we investigated the Figure 5 | ChmA knockdown effects on major processes of early infection. a, Average peak area of the 20 most abundant Goslar proteins detected by label free quantification mass spectrometry of Goslarinfected cells 30 mpi. Proteins encoded by genes in the Chimalliviridae core genome (8) are in bold. ** p < 0.01. b, Localization of GFP-tagged Goslar DNA polymerase (sfGFP-DNAPGos) in non-targeting (NT) and knockdown (ChmAg1) cells 90 mpi. Cell membranes were stained with FM4-64 (red) and DNA was stained with DAPI (cyan). White arrows indicate phage DNA position determined by DAPI signal. Scale bar = 1 µm. c, Schematic of early Chimalliviridae infection. i) Goslar injects its genome along with its virion RNA polymerase (vRNAP, pink) into the membrane-bound EPI vesicle. ii) Supported by nucleotide import into the EPI vesicle (light blue), early viral genes are transcribed by the vRNAP and the mRNAs are exported to the cytoplasm for translation by host ribosomes (yellow). iii) The viral genome is transferred from the EPI vesicle to the ChmA phage nucleus (dark blue). At this stage, the viral DNA polymerase (green) can be imported into the phage nucleus and DNA replication begins. localization of GFP-tagged Goslar DNA polymerase gp244 (sfGFP-DNAPGos) during ChmA knockdown. sfGFP-DNAPGos localized within the phage nucleus in our control strain, forming foci on the phage DNA. However, when ChmA was knocked down, sfGFP-DNAPGos remained diffuse throughout the host cell cytoplasm and failed to colocalize with the EPI vesicle (Fig. 5B). This suggests that the DNA replication machinery cannot access the injected phage genome in the EPI vesicles, potentially explaining why ChmA knockdown inhibits phage DNA replication while early phage gene expression, which is driven by the injected vRNAP, is unaffected.

In this study, we demonstrated that the phage nucleus is required for nucleus-forming phage DNA replication. Similar to replication compartments of eukaryotic viruses, which are believed to have evolved in response to host defenses, the phage nucleus may have originated as a protective mechanism against bacterial immune systems but is now intrinsically essential for the phage life cycle. We also demonstrated that infection is initiated through formation of a membrane-bound vesicle containing the injected genome, which we name the early phage infection (EPI) vesicle (Fig. 5C). EPI vesicles, which we hypothesize are formed from the host inner membrane during injection, are transcriptionally active and therefore must have yet-tobe identified mechanisms for importing nucleotides from and exporting mRNA to the cytoplasm. These early infection functions are likely mediated by Chimalliviridae proteins such as the vRNA polymerase that are packaged into phage capsids and injected into host cells ( 22, 23, 28, 29 ). After the early phage gene expression program is initiated, the phage genome becomes enclosed within the proteinaceous phage nucleus to carry out genome replication and proceed through the rest of its life cycle. Using membranes to decouple transcription from translation is a hallmark of eukaryotic cells but has never been observed before in prokaryotes. The discovery of this early membrane-bound intermediate also resolves the question of how Chimalliviridae protect their genomes from host DNAtargeting defense systems before the proteinaceous nuclear shell is synthesized. Thus, Chimalliviridae form sophisticated subcellular compartments of distinct compositions and functions that facilitate successive stages of the life cycle.

We thank the members of the Emerging Pathogens Initiative HHMI Consortium for their feedback and support. J.H. is an EMBO long-term postdoctoral Fellow. Z.K.R. is supported and E.G.A. was previously supported by an NIH PiBS training grant (T32 grant GM133351). J.A.D ., D.F.S. and E.V. are investigators of the Howard Hughes Medical Institute. We acknowledge the use of the UC San Diego cryo-EM facility, which was built and equipped with funds from UC San Diego and an initial gift from the Agouron Institute. The authors acknowledge funding from the Howard Hughes Medical Institute through the Emerging Pathogens Initiative grant (to E.V., J.P., K.D.C., and K.P.), the National Institutes of Health grants R01-GM129245 (to J.P. and E.V.), R35 GM144121 (to K.D.C.), MRI grant NSF DBI 1920374 (to E.V.) and R01-GM57045 (to K.P.), InCoGenTEC (to B.F.C. an d J.A.D.), m-CAFEs (B.A.A. and J.A.D.), and the Curci Foundation (B.F. C.). This paper was typeset with a bioRxiv template by @Chrelli: www.github.com/chrelli/bioRxivword-template

K.P. and J.P. have an equity interest in Linnaeus Bioscience Incorporated and receive income. The terms of this arrangement have been reviewed and approved by the University of California, San Diego, in accordance with its conflict-of-interest policies. The Regents of the University of California have patents issued and pending for CRISPR technologies on which J.A.D., B.F.C., B.A.A., D.F.S., E.C., J.P., K.P., E.G.A. and J.L. are inventors. J.A.D. is a cofounder of Caribou Biosciences, Editas Medicine, Scribe Therapeutics, Intellia Therapeutics, and Mammoth Biosciences. J.A.D. is a scientific advisory board member of Vertex, Caribou Biosciences, Intellia Therapeutics, Scribe Therapeutics, Mammoth Biosciences, Algen Biotechnologies, Felix Biosciences, The Column Group, and Inari. J.A.D. is Chief Science Advisor to Sixth Street, a Director at Johnson & Johnson, Alto s and Tempus, and has research projects sponsored by Apple Tree Partners and Roche. D.F.S. is a cofounder and scientific advisory board member of Scribe Therapeutics.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD045200 and 10.6019/PXD045200. A ny additional information required to reanalyze the data reported in this paper is available from the corresponding authors upon request.

Bacterial growth conditions.

Table S2 lists the bacterial strains used in this study. All strains were generated from lab strain Escherichia coli MC1000, which was originally derived from MG1655 ( 30 ). Strains were grown on Luria -Bertani (LB) plates containing 10 g Bacto-Tryptone, 5 g NaCl, 5 g Bacto-yeast extract and 16 g agar per liter ddH2O with appropriate antibiotics to maintain plasmids. Liquid cultures were generated by inoculating LB with single colonies. p15A-CmR backbone plasmids (see table S2) were main tained using 30 µg/mL chloramphenicol (Sigma-Aldrich) in plates, im aging pads and liquid cultures. dRfxCas13d expression was induced by aTc (anhydrotetracycline, Cayman Chemical Company) as indicated. pDW20 6 plasmids (see table S2) were similarly maintained using 100 µg/mL a mpicillin (Cayman Chemical Company) and protein expression was induce d by IPTG (isopropyl ß-D-1-thiogalactopyranoside, Teknova) as indica ted. See additional details on induction conditions in specific procedure methods below. Streak plates were incubated overnight at 37°C.

Bacteriophage preparation.

We originally acquired phage Goslar from Johannes Wittman at the DSMZ ( 9 ). Goslar lysates were amplified as follows. 10 uL of high titer lysate, 500 uL of E. coli MC1000 overnight liquid culture and 4.5 mL of LB top agar (0.35%) were mixed and plated on 3-5 LB plates, the n incubated overnight at 37 °C. The next day, 5 mL of chilled LB broth were added to each plate and incubated ~5 hours at room temperature. All liquid was then drawn off of the plates into a single tube and centrifuged at 3220 x g for 10 minutes. The supernatant was transferred to a new tube and cleared with chloroform (3 drops chloroform/5 mL lysate). After 10 minutes incubating at room temperature with periodic mixing by inversion, the aqueous phase of the lysate was filtered through a 0.45 µm Corning membrane filter by syringe into a sterile conical tube and stored at 4°C.

Plasmid construction, spacer design and bacterial transformation. Plasmids and crRNA spacer sequences used are listed in table S2 and S3. dRfxCas13d and guides were co-expressed from a p15A-CmR vector, while sfGFP, sfGFP fusions, ChmA* and ChmA*&CTS were expressed from vector pDSW206 ( 9 ). The initial p15A-CmR backbone dRfxCas1 3d entry vector and non-targeting guide plasmids were generously provided by the Doudna and Savage labs ( 13 ). All other plasmids wer e designed manually in Benchling, then synthesized by GenScript and delivered lyophilized. We manually designed the two chmA transcript targeting crRNA spacers to be 31 bp long and reverse complementary to the 59 end of the Goslar chmA ORF, following the same design conventions described in Adler et al. ( 12 ) ( Fig. 1C). Goslar9s chmA ORF ends 92 bp upstream of the fol lowing non-virion RNA polymerase subunit (nvRNAP ³9 1) gen e, which has its own RBS (aggaga) 6 bp upstream of its ATG site, strongl y indicating that these genes are translated independently. Our non-targeting control was a previously validated crRNA targeting the T4 phage major capsid protein ( 17 ).

GenScript plasmids were hydrated at 200 µg/mL and diluted to 20 µg/mL in dH2O. 1 µL diluted plasmid was added to 30-50 µL of electroporation competent E. coli, which were prepared by washing in 10% glycerol, and electroporated with 2.5 kV. After transformants recovered in 1 mL SOC at 37°C for 60 minutes, they were spread on LB plates with appropriate antibiotics and incubated overnight at 37°C.

Western blot.

For 200 µL of OD600 0.2 cells were spread on 1% agarose, 25% LB, 6 cm diameter plates containing 30 ug/mL chloramphenicol and 5 nM aTc. After incubating at 37°C for 2 hours inside a humidor, cells were infected with 100 µL ~1010 PFU/mL Goslar lysate and incubated again at 37°C. 90 mpi, cells were collected by the addition of 1 mL of 25% LB to each pad and gentle scraping with the bottom of an Eppendorf tube followed by aspiration. After checking the OD600 via nanodrop, ~4.0 x 108 cells were spun down for 1 minute at 211 x g and resuspended in 100 µL 2x SDS loading buffer (0.1 M Tris HCl, 0.004% bromophenol blue, 4% SDS, 20% glycerol, pH 6.8 plus 5% BME). Samples were boiled for 4 minutes at 95°C and vortexed for 30 seconds. After being diluted 1:5 in the same loading buffer, 10 µL of each sample were loaded into a Novex 4-20% Tris-Glycine Gel and run at 150 V. Protein was transferred to PVDF (Pall Life Sciences) and blocked by incubation at room temperature with StartingBlock Blocking Buffer (Thermo Scientific). The membrane w as cut below the 130 kDa molecular weight marker band to separate the regions containing ChmA and RpoB (loading control). The membrane secti ons were then incubated overnight at 4°C with the corresponding antibody: anti-ChmA (dilution = 1:500; custom GenScript polyclonal rabbit antibody), HRPconjugated mouse anti-RpoB (dilution = 1:5,000; Biolegend Cat# 663907, RRID:AB_2629625). RpoB is the E. coli RNA polymerase ³ subunit. The membranes were washed three times for 15 minutes with TBS-T and the ChmA-containing membrane was incubated with HRP-conjugated goat anti-rabbit IgG (H + L) secondary antibody (dilutio n = 1:10,000; Thermo Fisher Scientific Cat# 65-6120, RRID:AB_2533967) fo r 1.5 hrs at room temperature and washed as before with TBS-T. The membranes were treated with ECL western blotting detection reagent (Cytiva Amersham) and imaged using a ChemiDoc MP Imaging System (Bio-Rad). The amido black imaging setting was used to image the molecular weight marker ladder and chemiluminescence setting for visualizing the target protein bands. Figure panels were generated in Adobe Photoshop (21.2.0) and Adobe Illustrator (24.2). The western blot in Fig. 1C was run in biological and technical triplicate.

Antibody generation.

For the expression and purification of Goslar ChmA protein, the gene was cloned into UC Berkeley Macrolab vectors 2-BT (addgene #29666) for Nterminal TEV protease-cleavable His6-tag constructs. For protein expression, the construct was transformed into E. coli Rosetta2 pLysS (EMD Millipore), and the transformants were obtaine d on the nutrient agar plates containing carbenicillin and chloramphenicol antibiotics. A single transformant colony was inoculated into LB medium, and the culture was allowed to grow overnight in a 37°C incubator shaker. The following morning, the overnight culture (5 mL) was used to i noculate 1 L 2XYT medium and allowed to grow until the growth reached an OD600 of 0.6-0.8. This was followed by induction with 0.33 mM IPTG. T he induced cultures were incubated overnight at 20°C for the further expression of ChmA protein. Subsequently, the cells were harvested by centrifugation, and the bacterial pellets were resuspended in ice-cold resuspension buffer containing 50 mM Tris, pH 7.5, 300 mM NaCl, 10 mM imidazole, 10% glycerol, and 2 mM ³-mercaptoethanol. Resuspended cells were subjected to lysis using a sonicator, and the lysate was clarified using centrifugation. The protein was purified using Ni2+ affinity chromatography. Eluted protein was concentrated, and buffer exchanged to remove imidazole, and the N-terminal His6-tag was cleaved using TEV protease. The tag-cleaved protein was further subjected to purification by size exclusion chromatography using a Superdex 200 Increase 10/300 GL column (Cytiva) in a buffer containing 20 mM Tris, pH 7.4, 250 mM NaCl, and 2 mM ³-mercaptoethanol. Finally, the quality of purified protein was assessed by running an SDS-PAGE, and the protein aliquots (~2 mg/mL concentration) were flash-frozen using liquid nitrogen and stored at -80°C until shipment for antibody generation. GenScript generated polyclonal ³ChmA rabbit antibodies via their Polyexpress method, which were stored at -20°C upon being received. xxx Plaque Assay.

All plaque assays were performed in biological triplicate. 1 mL of overnight culture (no preinduction) was mixed with 9 mL of 0. 35% LB top agar containing antibiotics and inducing agents and poured onto 25 mL LB plates containing the required antibiotics. As the plates did not contain inducing agents, aTc and IPTG were added to the top agar at initial concentrations to account for diffusion into the total final plate volume. For all plaque assays involving dRfxCas13d-expressing strains (Fig. 1E-F, H-I), aTc = 5 nM. For all plaque assays involving ChmA*- or ChmA*&CTSexpressing strains (Fig. 1H-I and Fig. S2), IPTG = 2 mM. Plates were spotted with a 10-fold dilution series of Goslar from 101-108 (prepared in LB) in technical triplicate with 3 µL/spot and were dried in a biosafety cabinet with lids removed. Once the spots dried (6-10 minutes), the plates were incubated at 37°C for 15-18 hours. Plates were photographed using an iPhone 13 camera and efficiency of plaquing was recorded. Viral reproduction inhibition via ChmA knockdown or ChmA*&CTS expression frequently precluded countable plaque formation and instead produced faint zones of clearing at higher dilutions. In these cases, the most concentrated dilution at which no clearing was observed was treated as 1 PFU. Figure panels were generated using Adobe Photoshop (21.2.0) and Adobe Illustrator (24.2). Data analysis was perform ed in Microsoft Excel (version 16.75.2) and GraphPad Prism (version 10.0. 0 (131)) and graphs were generated in Graphpad Prism.

Live single cell static fluorescence microscopy and time-lapse microscopy.

Fluorescence microscopy and time-lapse microscopy experiments were performed in at least biological duplicate. Host cells were suspended in 25% LB at OD600 0.4 as measured by nanodrop. 5 uL were spotted and spread on the surface of 1% agarose, 25% LB imaging pads containing required antibiotics and inducing agents on single-well concavity glass slides, then incubated for 1.5 to 2 hours at 37°C without coverslips in a humidor. 10 µL of ~1010 PFU/mL Goslar lysate were spotted and spread onto the imaging pads and the pads were incubated further at 37°C until the desired infection time point. For all microscopy experiments, aTc in imaging pads = 5 nM. Plasmid leakiness in the absence of IPTG was sufficient to express sfGFP-ChmA and sfGFP at appropriately low levels for related experiments (Fig. 2D-E and Fig. S4). To induce ChmA* and ChmA&CTS* expression (Fig. 3B-J), 8 uL of 2 mM IPTG was spread on the pad surface 20 minutes before Goslar infection.

All live cell microscopy was performed on a DeltaVision Elite Deconvolution microscope (Applied Precision, Issaquah, WA, USA). Directly before imaging, a glass coverslip was placed on top of the agarose pad. For static fluorescence microscopy, imaging pads were stained with 8 µL of dye mix (25 µg/mL DAPI, 3.75 µg/mL FM4-64) at room temperature 5 minutes before imaging. DAPI: Thermo Fisher Scientific Cat# D1306, FM4-64: Thermo Fisher Scientific Cat# T13320. Cells were imaged with 820 slices in the Z-axis at 0.2 µm increments. For time-lapses, no fluorescent stains were used and cells were imaged in 12 or 20 slices in the Z-axis at 5 minute intervals with Ultimate Focus mode enabled. Slides were kept at 37°C within the environmental control unit enclosing the microscope stage. Images were deconvolved in DeltaVision SoftWoRx (version 6.5.2). Figure panels were created in Adobe Photoshop (21.2.0) and Adobe Illustrator (24.2).

Quantification and analysis of fluorescence and time-lapse microscopy data.

All quantification of fluorescence microscopy and time-lapse microscopy data was performed on non-deconvolved DeltaVision image files (opened as TIFFs) in FIJI version 2.3.0/1.53q. All images w ere automatically scaled appropriately depending on the individual DeltaVision Elite Deconvolution microscope used for data collection. All cells that were in focus and fully within a field were analyzed for each image. To obtain a representative sample, n > 150 when possible. The cross-sectional area and DAPI intensity of phage DNA (intra- or extracellular) were measure d using the FIJI freehand selection tool at the point in the z-series in which the measured object was in focus. If a cell was infected multiple times (indicated by multiple distinct foci of phage DNA or mature phage nuclei), the phage DNA with the greatest cross-sectional area was measured. Data was analyzed and figures were generated using Microsoft Excel (version 16.75.2) and Python 3.

Percentage of phage nuclei that colocalize with sfGFP-ChmA during ChmA knockdown was determined via manual inspection of GFP and DAPI channels. Data analysis was performed in Microsoft Excel (version 16.75.2).

For time-lapse microscopy, each cell visibly swollen or containing a phage nucleus was tracked over 210 minutes. The data was analyzed and final figure generation was performed in GraphPad Prism (version 10.0.0 (131)).