Two-Color CRISPR Imaging Reveals Dynamics of Herpes Simplex Virus 1 Replication Compartments and Virus-Host Interactions

ABSTRACT

Real-time imaging tools for single-virus tracking provide spatially resolved, quantitative measurements of viral replication and virus-host interactions. However, efficiently labeling both parental and progeny viruses in living host cells remains challenging. Here, we developed a novel strategy using the CRISPR-Tag system to detect herpes simplex virus 1 (HSV-1) DNA in host cells. We created recombinant HSV-1 harboring an ~600-bp CRISPR-Tag sequence which can be sufficiently recognized by dCas9-fluorescent protein (FP) fusion proteins. CRISPR-assisted single viral genome tracking (CASVIT) allows us to assess the temporal and spatial information of viral replication at the single-cell level. Combining the advantages of SunTag and tandem split green fluorescent protein (GFP) in amplifying fluorescent signals, dSaCas9-tdTomato_10x and dSpCas9-GFP_14x were constructed to enable efficient two-color CASVIT detection. Real-time two-color imaging indicates that replication compartments (RCs) frequently come into contact with each other but do not mix, suggesting that RC territory is highly stable. Last, two-color CASVIT enables simultaneous tracking of viral DNA and host chromatin, which reveals that a dramatic loss of telomeric and centromeric DNA occurs in host cells at the early stage of viral replication. Overall, our work has established a framework for developing CRISPR-Cas9-based imaging tools to study DNA viruses in living cells.

IMPORTANCE Herpes simplex virus 1 (HSV-1), a representative of the family Herpesviridae, is a ubiquitous pathogen that can establish lifelong infections and widely affects human health. Viral infection is a dynamic process that involves many steps and interactions with various cellular structures, including host chromatin. A common viral replication strategy is to form RCs that concentrate factors required for viral replication. Efficient strategies for imaging the dynamics of viral genomes, RC formation, and the interaction between the virus and host offer the opportunity to dissect the steps of the infection process and determine the mechanism underlying each step. We have developed an efficient two-color imaging system based on CRISPR-Cas9 technology to detect HSV-1 genomes quantitatively in living cells. Our results shed light on novel aspects of RC dynamics and virus-host interactions.

INTRODUCTION

Single-virus tracking in living cells offers the opportunity to monitor the virus journey in real time. This strategy allows us to directly probe dynamic interactions between viruses and cellular structures and capture quantitative information about infection kinetics (1–6). Furthermore, the combination of single-virus labeling methods and sensitive fluorescence microscopic techniques would facilitate mechanistic understanding of viral infection (7–10). Herpes simplex virus 1 (HSV-1) is a large (~150-kb) DNA virus that infects most of the human population. HSV-1 can productively infect a variety of epithelial cells and develop latent infections in neurons in vivo (11). HSV-1 infection of cells has long been used as a model to elucidate fundamental aspects of both virus infection and cellular biology (12–14).

The formation of membraneless nuclear assemblies termed replication compartments (RCs) occurs during productive infection of DNA viruses. These spatially defined compartments provide a dedicated environment where factors required for viral processes are enriched (15, 16) and the viral genome is transcribed and replicated (17–19). How invading viral genomes utilize host cell resources and coopt or antagonize cellular pathways to benefit viral replication remains poorly understood at the molecular level. Previous studies have shown that lytic infection with HSV-1 induces profound modification of host chromatins. One of the few examples is the displacement of host chromatin to the nuclear periphery to increase nuclear volume for viral DNA replication (20). Additionally, HSV-1 infection altered multiple aspects of host cell telomeres, likely promoting the formation of ICP8-associated prereplication foci (21). Similarly, the centromeric chromatin structure is also significantly remodeled by ICP0 during lytic HSV-1 infection (22–24). Dual-color DNA-fluorescence in situ hybridization (FISH) imaging indicates that latent HSV-1 genomes are nonrandomly associated with centromeres in neuron nuclei (25). However, it is not fully understood how centromeres are functionally involved in HSV-1 infection. Developing live-cell DNA labeling tools to visualize host chromatin and viral genomes simultaneously will provide a dynamic view of host-virus interactions in great detail.

Most virus genome-tracking methods are limited to the study of fixed cell samples, such as using FISH (7, 26, 27) or incorporation of nucleoside analogs (e.g., BrdU and EdU) into viral genomes (9, 15, 19, 28). ICP4 forms a novel DNA recognition complex to serve as a viral transcription factor (29, 30). ICP8 is a single-stranded DNA binding protein involved in HSV-1 DNA replication (31). Recombinant HSV-1 strains expressing ICP4 or ICP8 fused to fluorescent protein (FP) have been generated to visualize RCs in living cells (18, 32). However, ICP4-FP and ICP8-FP act only as RC protein markers. In contrast, incorporating binding sites for fluorescent DNA-binding proteins into the viral genome potentially enables real-time tracking of viral replication in living cells. ANCHOR DNA labeling technology has been developed for labeling human cytomegalovirus, which utilizes an ANCH DNA target sequence around 1 kb long that specifically binds dimers of fluorescent OR protein (33). However, whether OR proteins that undergo bidirectional spreading on adjacent DNA sequence may disturb viral functions needs further elucidation. The bacterial tetracycline operator/repressor (TetO/TetR) system has been adapted to detect HSV-1 replication compartments (34–36). However, the TetO array was inserted into an amplicon plasmid instead of the HSV-1 genome, and helper viruses were required to generate normal progeny viral particles, thus limiting its wide applications (4). The nuclease-dead Cas9 (dCas9) has been developed for labeling specific DNA sequences in living cells (37–39). CRISPR-Tag was created as a small DNA tag (~250 to 630 bp) which can be efficiently labeled by dCas9-GFP_14X, thus enabling visualization of a nonrepetitive locus (40).

In this study, we created recombinant HSV-1^CRISPR-Tag, which harbors the dCas9 binding sequence CRISPR-Tag. HSV-1^CRISPR-Tag infection in dCas9-FP-expressing cells enables visualization of parental and newly replicated viral genomes. This CRISPR-assisted single viral genome tracking (CASVIT) technology represents a promising new system for understanding the fundamental biology of virus infection and antiviral drug discovery.

RESULTS

Design and selection of CRISPR-Tags.

To label HSV-1 genomes in living cells, we hypothesized that if CRISPR-Tag could be successfully inserted into the HSV-1 genome, the recombinant viral DNA would be efficiently visualized by dCas9-FP. To test this hypothesis, we first designed five versions of CRISPR-Tags following the basic principle of CRISPR-Tag assembly revealed in our previous work (40). These CRISPR-Tags could be efficiently recognized by dCas9 from Streptococcus pyogenes (dSpCas9). Both CRISPR-TagV1 and -V2 include six repeats of four different targeting sequences, harboring 24 CRISPR targeting sites in total (Fig. 1A). All the other three CRISPR-Tags (V3 to V5) contain 12 copies of targeting sequence 1 (TS1), thus enabling dCas9-FP labeling using one single guide RNA (sgRNA) (sgTS1). To achieve optimal dCas9 binding, the spacing between two adjacent TS1 sites in V3 to V5 CRISPR-Tags is 37, 52, and 67 bp, respectively. We then performed CRISPR imaging of plasmids harboring individual CRISPR-Tag in clonal HeLa cells stably expressing dSpCas9-GFP_14X, which utilized the tandem split-green fluorescent protein (GFP) system to increase DNA detection sensitivity (40). Although DNA sequencing results revealed that CRISPR-TagV4 contains only 11 TS1 sites, we found that CRISPR-TagV4 achieved the most efficient labeling of plasmids by measuring signal-to-noise ratios (Fig. 1B and C). Thus, we selected CRISPR-TagV4 for viral DNA labeling.

FIG 1.

Design and selection of CRISPR-Tags for viral DNA labeling. (A) Schematic illustration of five CRISPR-Tag designs (V1 to V5). As shown in the diagram, commercially synthesized DNA fragments containing C. elegans genomic sequences (TS1 to TS8) were assembled in various combinations. CRISPR-TagV1 and CRISPR-TagV2 contain four different sgRNA recognizing sequences. CRISPR-TagV3, CRISPR-TagV4, and CRISPR-TagV5 contain the same sgRNA recognizing sequence (TS1) but harbor spacer sequences of different lengths. (B) Representative images showing CRISPR-Tag labeling by dSpCas9-GFP_14X in HeLa cells. These CRISPR-Tags were assembled in a common plasmid vector for CRISPR imaging. Nuclei are indicated with white dotted lines. Arrows indicate dCas9-GFP_14X spots. All images are maximum-intensity projections from z stacks. Bars, 5 μm. (C) Quantification of DNA labeling efficiency defined by quantifying SNR, which are averaged to the cell level. n ≥ 21 cells. Each dot represents a single cell. All values are means and SEM. One-way analysis of variance (ANOVA) with Tukey’s post hoc was used to test differences between groups. *, P < 0.1; ****, P < 0.0001.

Generation and characterization of recombinant HSV-1^{CRISPR-Tag(Sp)}.

Next, we generated recombinant HSV-1^{CRISPR-Tag(Sp)} and examined whether we could monitor HSV-1 DNA replication by CRISPR imaging in living cells (Fig. 2A). Using bacterial artificial chromosome (BAC) technology, we introduced the CRISPR-TagV4 sequence into the intergenic region between UL3 and UL4 (Fig. 2B). In both single-step and multiple-step analyses, HSV-1^{CRISPR-Tag(Sp)} replicated with kinetics similar to that of the parental virus, suggesting that the CRISPR-TagV4 insertion does not significantly affect infectivity (Fig. 2C; also, see Fig. S1 in the supplemental material). HSV-1^{CRISPR-Tag(Sp)} was then used to infect HeLa cells stably expressing dCas9-GFP_14X and transiently transfected with sgRNA-expressing plasmids (sgTS1 or sgGal4) at a multiplicity of infection (MOI) of 20. Notably, cells were transfected with sgRNA plasmids 24 h prior to the virus infection so that the sgRNA was highly expressed when viral DNA entered the nucleus. As expected, at 1 to 4 h postinfection (hpi), cells transfected with sgTS1 started to exhibit visible dCas9-GFP_14X spots, likely representing parental genomes. At a later stage (8 to 24 hpi), replication compartments (RCs) were formed (Fig. 2D). CRISPR imaging detected HSV-1^{CRISPR-Tag(Sp)} DNA with a signal-to-noise ratio of 25. HSV-1^{CRISPR-Tag(Sp)} genomes could not be labeled if sgGal4 (no targeting sites in the HSV-1 or human genome) was expressed instead of sgTS1 (Fig. 2E).

FIG 2.

Design and validation of the CASVIT system. (A) Schematic illustration of CASVIT system for viral genome tracking in living cells. The model depicts the replication process of HSV-1 viral genomes during distinct stages of infection. The recombinant vDNA can be recognized by dSpCas9-GFP_14x, thus enabling the visualization of parental vDNA and RC formation. (B) Brief description of constructing the recombinant HSV-1 genome. CRISPR-TagV4 is inserted into the region between UL3 and UL4 in HSV-1 genome. (C) Time dependency of virus replication measured by plaque-forming assay. Recombinant HSV-1^CRISPR-Tag was compared to wild-type HSV-1. The single-step growth curves were generated by virus infection at an MOI of 5. n = 3 replicates. (D) Visualization of recombinant HSV-1^CRISPR-TagV4 in HeLa cells using CASVIT. sgGal4 was used as a negative control. sgTS1 can recognize CRISPR-TagV4. Arrows point to single viral particles or RCs. (E) Labeling efficiency of CASVIT was determined by the signal-to-noise ratio. Each dot represents a single viral particle (n = 100). The value is shown as mean and SEM. (F) (Left) Representative images showing the colabeling of RCs with dSpCas9-GFP_14x and ICP4 antibody. (Right) Line scan of the relative fluorescent signal indicated by the dotted lines in the left image. (G) Representative images showing recombinant HSV-1 replication (MOI = 20) in the presence or absence of ACV (500 μM). (H) Bar graph showing the proportion of cells with detectable RC formation. n ≥ 100 cells. All images are maximum-intensity projections from z stacks. Bars, 5 μm. Nuclei are indicated with white dotted lines.

The major HSV-1 transcriptional activator ICP4 is a viral DNA binding protein which is recruited into viral RCs containing large amounts of replicating viral DNA (41, 42). We performed ICP4 immunostaining to validate the specificity and efficiency of CRISPR labeling. We found that all dCas9-GFP_14X-labeled RCs contained abundant ICP4 proteins (Fig. 2F). Since small and large RCs represent early and late stages of RCs, respectively, we named them early and late RCs. Acyclovir (ACV) is used to treat HSV-1 infection; it acts by competitive inhibition of viral DNA polymerase (43). The addition of ACV to HSV-1^{CRISPR-Tag(Sp)}-infected cells resulted in the loss of RC formation, indicating impaired DNA replication (Fig. 2G and H). These results demonstrate that dCas9 fluorescent spots resulted from specific labeling of HSV-1 genomes. In addition to HeLa cells, we also monitored HSV-1^{CRISPR-Tag(Sp)} replication in two other cell types, including U2OS and Neuro-2a cells. CRISPR imaging could detect single viral genomes and RC formation in these cell types as well (Fig. S2). However, we mainly used HeLa cells as the host cells for this study.

Quantitative measurements of HSV-1 infection and genome replication in single cells.

To take advantage of our imaging system, we quantitated the number of parental genomes within individual cells by CRISPR imaging. HSV-1^{CRISPR-Tag(Sp)} was used to infect HeLa cells which stably expressed dCas9-GFP_14X and transiently expressed sgTS1. We noticed that the intensities and sizes of the individual dCas9-GFP_14X spots were always similar and did not change significantly as MOI increased (Fig. 3A to C). Thus, these spots are very likely to represent single genomes. Next, we sought to confirm whether CRISPR imaging has the sensitivity to detect the changes in the copy number of CRISPR-Tag(Sp)-containing plasmids in the cells. Our results revealed that the detected number of CRISPR-Tag plasmids increased with the number of plasmids transfected into the cells (Fig. S3). We then performed CRISPR imaging to test the effect of the virus dose on infection efficiency. HeLa cells were infected at different MOIs ranging from 5 to 80. However, as the MOI increased from low to high, the average number of dCas9-GFP_14X spots within a nucleus did not change significantly at 4 hpi (Fig. 3D). These quantifications indicate that only a limited number of HSV-1 parental genomes can enter the nucleus of HeLa cells, which is not consistent with the previous finding in RPE-1 cells (14, 44). We speculate that the inconsistency may be due to the differences in cell types, which can be further addressed in the future when side-by-side comparisons are carried out using the same viral DNA (vDNA) labeling method.

FIG 3.

CASVIT system facilitates live imaging of single viral DNA and HSV-1 replication. (A) Representative images showing parental HSV-1 genomes labeled by dCas9-GFP_14X at 4 h postinfection with HSV-1^CRISPR-TagV4 at different MOI. All images are maximum-intensity projections from z stacks. Bars, 5 μm. (B) Total fluorescence intensity of individual dCas9-GFP_14X spots. Each dot represents a single spot (n ≥ 50 spots). (C) Quantitative analysis of the size of individual dCas9-GFP_14X spots. Each dot represents a single spot (n ≥ 50). (D) Quantifications of vDNA labeling at 4 h after the infection of HSV-1^CRISPR-TagV4, by analyzing the number of dCas9-GFP_14X spots within one nucleus. Each dot represents one cell (n ≥ 37). All data are means and SEM. One-way ANOVA with Tukey’s post hoc test was used to test differences between groups. *, P < 0.1; NS, not significant. (E) Live-cell imaging snapshots of cells showing the replication process of HSV-1 viral DNA at MOI of 20. Real-time imaging was recorded in 30-min intervals. The cell in the first row was imaged between 4 to 12 hpi, while the other two cells from the second and third rows were imaged between 12 and 20 hpi. Nuclei are indicated with white dotted lines. Arrows point to the formation of RCs from single vDNA in cell 1. Purple dotted circles highlight two small RCs that moved and coalesced, while yellow dotted circles indicate multiple RCs that grew in situ and merged with neighboring RCs. All images are maximum-intensity projections from z stacks. Bars, 5 μm.

Together, our results demonstrate a CRISPR-assisted single viral genome tracking (CASVIT) system which can be applied for specific labeling of HSV-1 genomes. Notably, we noticed that CASVIT has the sensitivity to detect nonreplicating vDNA at late stages of infection. By measuring the size of single dCas9-GFP spots (most likely representing single vDNA) at 4, 8, 12, 16, and 20 hpi, we found that the average size of viral genomes appearing at early times of infection (4 and 8 hpi) was significantly larger than that at late times of infection (12, 16, 20, and 24 hpi) (Fig. S4). A possible explanation is that some viral genomes become decompacted to facilitate gene expression and DNA replication at early stages. In contrast, nonreplicating vDNA at late stages of infection might be compacted into condensed heterochromatic structures. It has been reported that HSV-1 genomes become “chromatinized” upon entry into the nucleus and can undergo histone acetylation to overcome repressive chromatin marks, promoting viral gene expression in host cells (45–47). Thus, our observations based on viral genome imaging are consistent with this hypothesis.

Next, we investigated whether the CASVIT system could monitor the replication dynamics of viral genomes in real time. We observed that RCs had fully grown in HeLa cells at 20 hpi. Therefore, real-time imaging was performed between 4 and 20 hpi. At the beginning, multiple areas suggestive of replicative sites were formed around the initial dCas9-FP spots within the nucleus. These areas gradually increased in size, forming early pre-RCs (Fig. 3E). At the late stage, we observed how large RCs were formed in the nucleus. There seem to be two ways to develop large RCs: (i) small RCs gradually enlarged and moved within the nucleus, approaching and fusing with neighboring RCs (Video S1), and (ii) small RCs stayed in the same position, grew larger, and merged with nearby RCs. Similar observations have been reported in previous studies (18). However, it was unclear whether coalescent RCs undergo reorganization to exchange contents, such as progeny HSV-1 genomes. Therefore, we further developed a two-color CRISPR imaging technique to monitor the dynamics of coalescent RCs.

Two-color CASVIT reveals frequent RC interactions but stable RC territories.

To develop a two-color CASVIT, we utilized the two-color CRISPR imaging system which combines SpCas9 and a smaller Cas9 ortholog from Staphylococcus aureus (SaCas9) (48). The two orthologous CRISPR systems recognize different PAM sequences and provide an efficient tool for labeling multiple genomic loci in living cells (49). To optimize this two-color CRISPR imaging system, we combined the use of SunTag and tandem split-GFP system to improve the signal-to-noise ratio of DNA imaging. We first generated a clonal HeLa cell line by optimizing the coexpression of dSpCas9-GFP_14X and dSaCas9-(SunTag)-tdTomato_10x (Fig. S5). Similar to recombinant HSV-1^{CRISPR-Tag(Sp)}, we also created recombinant HSV-1^{CRISPR-Tag(Sa)}, which harbors 12 targeting sites (TS9) in its genome for dSaCas9 binding (Fig. 4A and B). Despite somewhat lower titers of HSV-1^{CRISPR-Tag(Sa)} in the initial eclipse period of the single-step growth curve, HSV-1^{CRISPR-Tag(Sa)} replicated with yields similar to those of the wild type in the synthetic period (Fig. 4C). Moreover, in multiple-step growth curve analysis, HSV-1^{CRISPR-Tag(Sa)} replicated with kinetics similar to that of the wild type (Fig. S1). dSaCas9 imaging revealed that HSV-1^{CRISPR-Tag(Sa)} underwent replication to form RCs, which was effectively inhibited following ACV treatment (Fig. 4D and E). These results suggest that the dSaCas9-tdTomato_10x system can efficiently monitor HSV-1 replication in living cells.

FIG 4.

Design of the dSaCas9-tdTomato_10X imaging system for imaging viral DNA replication. (A) Schematic illustration of CRISPR-Tag design to realize dSaCas9-mediated recombinant HSV-1 labeling. A commercially synthesized DNA fragment containing C. elegans genomic sequences (TS9) that can be recognized by dSaCas9 is inserted into the HSV-1 genome region between UL3 and UL4. (B) Representative images showing CRISPR-Tag labeling by dSaCas9-tdTomato_10X in HeLa cells. CRISPR-Tag(Sa) was cloned into a common plasmid vector for CRISPR imaging. Nuclei are indicated with white dotted lines. Arrows point to visible dSaCas9-tdTomato_10X spots. All images are maximum-intensity projections from z stacks. Bars, 5 μm. (C) Time dependency of virus replication measured by plaque-forming assay. Recombinant HSV-1^{CRISPR-Tag(Sa)} was compared to wild-type HSV-1. n = 3 replicates. These single-step growth curves were obtained together with those in Fig. 2C. Notably, Fig. 2C and panel C of this figure have the same wild-type growth curves. (D) Representative images of cells infected with HSV-1^{CRISPR-Tag(Sa)} (MOI = 8) and incubated in the presence or absence of ACV (500 μM). All nuclei are indicated with white dotted lines. All images are maximum-intensity projections from z stacks. Bars, 5 μm. (E) Bar graph showing the quantifications of RC formation for the conditions used for panel D. n ≥ 100 cells.

We then coinfected two-color HeLa cells with the two recombinant viruses, HSV-1^{CRISPR-Tag(Sp)} and HSV-1^{CRISPR-Tag(Sa)} (Fig. 5A and Fig. S6A). The cells were fixed and imaged at 12 hpi. We found that dSpCas9-GFP or dSaCas9-tdTomato spots were restricted to discrete foci, representing individual RCs (Fig. S6B), suggesting that each virus replicated within a distinct RC with its own territory within the nucleus. The lack of mixed RCs supports a general model in which each RC initiates from single viral genomes (27). Notably, two-color CASVIT detected frequent interactions between RCs. Following confocal imaging, 3D rendering of adjacent RCs, which were in close contact, revealed that they were most likely not overlapping or fusing (Fig. 5B). Surprisingly, at 12 hpi, the proportion of cells containing adjacent RCs was as high as 60% in both HeLa cells and U2OS cells (Fig. 5C). This was unlikely due to limited space, since ~73% of interacting RCs were still small (the total RC area only occupied <30% of nuclear space), suggesting a potential active mechanism for the RC interactions (Fig. S6C).

FIG 5.

Two-color CASVIT reveals frequent RC interactions and stable RC territories. (A) Model depicting the principle of two-color CASVIT for visualizing RC formation. Two independent CRISPR-Cas9 systems are used to image the replication of two types of recombinant HSV-1 viruses, respectively. Parental vDNA may replicate within discrete territories (#1 RC and #2 RC) or within the same area (#3 RC). (B) Representative 3D confocal and 3D rendering images to show RC interactions from two angles. Bars, 2 μm. (C and D) Statistics showing the relative positioning between two RCs in one nucleus of HeLa (C) or U2OS (D) cells at 12 hpi. Each dot represents an independent experiment (n ≥ 53 cells in each experiment). Data are means and SEM. (E) Live-cell imaging snapshots of RC dynamics within one cell. Real-time imaging was recorded at 30-min intervals between 12 and 22 hpi. Nuclei are indicated with white dotted lines. Arrows indicate individual RCs at the first time point. Magenta arrowheads point to separate RCs, while blue arrowheads highlight adjacent RCs. For two-color CRISPR imaging (B to E), HSV-1^{CRISPR-Tag(Sp)} was labeled by dSpCas9-GFP_14X (green), while HSV-1^{CRISPR-Tag(Sa)} was labeled by dSaCas9-tdTomato_10X (red or magenta). All images are maximum intensity projections from z stacks. Bars, 5 μm. For all two-color CASVIT, HSV-1^{CRISPR-Tag(Sp)} was infected at an MOI of 20 and HSV-1^{CRISPR-Tag(Sa)} was infected at an MOI of 8.

Next, we performed two-color imaging to track RC interactions in real time at the single-cell level. We observed three types of RC interactions: (i) two interacting RCs remained in contact for longer than 8 h as they grew in size; (ii) two spatially separated RCs migrated toward each other, eventually touching each other; and (iii) two closely contacting RCs separated at a later time (Fig. 5D and Video S2). Intriguingly, RCs tend not to mix with their neighboring RCs despite the close contact, suggesting that the territory of each RC is highly stable once formed in the nucleus. While the mechanisms of maintaining RC stability and frequent RC interactions need to be explored, we proposed models to show the formation of large RCs at the late stage of HSV-1 infection in HeLa cells based on our two-color CASVIT results (Fig. S7A and B; Video S3). Approximately 75% of RCs grow larger in situ, resulting in close contact with neighboring RCs, while the other ~25% of RCs grow in size and move into contact other RCs without mixing contents (Fig. S7C and D). Without two-color CASVIT imaging, these interacting RCs would be interpreted as one large RC resulting from the fusion of multiples RCs.

Visualization of host-virus interactions using two-color CRISPR imaging.

Previous studies have shown that the formation and growth of RCs require the remodeling of host chromatin to overcome the spatial constraints of the nucleus (5, 50). One of the findings is that HSV-1 infection reorganizes telomeres to promote viral genome replication (21). To test whether two-color CRISPR imaging could visualize host-virus interaction at the DNA level, we applied dSpCas9-GFP_14X to monitor HSV-1 replication. At the same time, dSaCas9-tdTomato_10x was utilized for tracking host chromatin (Fig. 6A). Telomeres were efficiently labeled by dSaCas9-tdTomato_10x in untreated cells. However, the number of telomeres labeled by dSaCas9 was reduced by about ~50% in the host cells in which early or late RCs were formed, indicating dramatic loss of telomeric DNA accompanied by viral replication. Unlike telomeres, MUC4 labeling was not affected by HSV-1 infection. Furthermore, we confirmed that dSaCas9 expression and its nuclear localization were unaffected by HSV-1 infection (Fig. 6B to E and Fig. S8).

FIG 6.

Visualization of host-virus interactions by two-color CRISPR imaging. (A) Schematic illustration of simultaneous imaging of HSV-1 genome and host chromatin. dSpCas9-GFP_14X is used for detecting vDNA, while dSaCas9-tdTomato_10X is used for labeling the genomic DNA of host cells. (B, D, and F) Representative images showing two-color labeling of HSV-1^{CRISPR-Tag(Sp)} DNA and MUC4 DNA (B), telomere DNA (D), and alpha satellite DNA (centromere) (F). All images are maximum-intensity projections from z stacks. Bars, 5 μm. (C, E, and G) Quantification of host chromatin labeling, including MUC4 loci (C), telomeres (E), and centromeres (G) without or with HSV-1 infection. The numbers and the signal-to-noise ratios of dSaCas9-tdTomato_10X spots were quantified to show the status of host chromatin at 22 hpi. Each dot represents the average value within one cell. n ≥ 12 cells. Data are means and SEM. One-way ANOVA with Tukey’s post hoc test was used to test differences between groups in panels C, E, and G. *, P < 0.1; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; NS, not significant. (H) Representative images showing simultaneous detection of centromeres (labeled by dSaCas9-tdTomato_10X) and ICP0 (stained by antibody) in HeLa cells without or with HSV-1 infection for 10 h. All images are from a single focal plane. Bars, 5 μm. (I) Colocalization analysis of dSaCas9-tdTomato_10X and ICP0 for the samples used for panel H. Each dot represents an independent experiment (n ≥ 20 cells in each experiment). Data are means and SEM. HSV-1^{CRISPR-Tag(Sp)} infections depicted here were performed at an MOI of 20.

A previous study demonstrated that the ICP0 protein of HSV-1 can induce the proteasomal degradation of centromeric proteins (CENPs) and the alteration of centromeric chromatin structure (24, 51). However, the change of centromeric chromatin has not been directly monitored at the single-cell level. We carried out CRISPR imaging to detect centromeric DNA (alpha satellite DNA repeats) and HSV-1 DNA simultaneously. The number of dSaCas9-FP spots representing centromeres was decreased by ~50%, and the CRISPR signal of the remaining spots became weaker in cells with detectable RCs (Fig. 6F and G). These changes were closely correlated with successful viral DNA replication, indicated by the formation of early or late RCs. Thus, the progress of viral genome replication indeed caused a dramatic loss of centromeric DNA. Notably, we noted that ICP0 was highly expressed and enriched on centromeric DNA at 10 hpi (Fig. 6H and I). Overall, our data further support the hypothesis that ICP0, an immediate early protein of HSV-1, affects the stability of CENPs while it resides on the centromere, thus leading to a breakdown of the centromere architecture (51). Hence, the two-color CRISPR imaging system will provide an opportunity to obtain spatiotemporal information for investigating host-virus interactions at the single-cell level.

DISCUSSION

In this study, we describe the application of CRISPR-Tag DNA tagging strategy to a DNA virus. We created CRISPR-Tag-modified viral DNA and quantitatively monitored HSV-1 DNA replication kinetics in living cells by CRISPR imaging. A major advantage of using CRISPR-Tag to construct recombinant viruses is that a large number of CRISPR-Tags can be designed to meet different requirements. The current CRISPR-Tag applied to construct recombinant viral DNA contains only 12 CRISPR target sites. To achieve higher detection sensitivity of the viral genome, additional copies of the target sequence can be assembled into CRISPR-Tag, as long as the larger DNA size of CRISPR-Tags does not perturb viral replication. Additionally, DNA sequences recognized by various CRISPR-Cas9 systems can be assembled for multicolor viral genome imaging. This flexible nature of the CRISPR-Tag design makes it a compelling DNA tag for a variety of applications.

The first sign of HSV-1^CRISPR-Tag infection in HeLa cells was the appearance of dCas9-GFP spots in the nucleus around 1 to 4 hpi, which represent incoming viral genomes after the virus loses its envelope and capsid. These individual spots then multiplied in small specific territories within the nucleus that represent replication sites. Following the initial replicative phase, a massive amplification step resulted in the formation of large RCs. In light of these observations, we concluded that our viral genome tracking system (CASVIT) is reliable and efficient for elucidating viral DNA replication. Furthermore, this method could be potentially applied to other DNA viruses.

Previous live-imaging studies illustrated that HSV-1 replication compartments could be formed by coalescence of small replication sites (18, 26, 52). However, two-color FISH imaging revealed that each HSV-1 RC was derived from a single incoming genome. Additionally, most viral intergenomic recombination events occur at the interface between mature RCs (26). Two-color DNA FISH assays also indicate that pseudorabies virus (PRV) RCs originate with a single viral genome (27). However, it was unclear if these interacting RCs were partially mixed or not. Our real-time two-color CRISPR imaging shows that RCs in close contact may separate again. We observed interacting RCs in up to 60% of cells at 12 hpi, but we never saw two types of progeny viral DNA (HSV-1^{CRISPR-Tag(Sp)} and HSV-1^{CRISPR-Tag(Sa)}) completely mixed in the same RC. Without two-color imaging, many interacting RCs would be identified as one large RC. Our dynamic imaging results suggest that RC territories are highly stable. Recombination is considered a major driving force in the evolution of most organisms because it accelerates adaptation (53). The high stability of RCs allows only limited recombination events between viral DNA or viral DNA and the human genome. This may partly contribute to the relative genetic stability of DNA viruses (54).

Our live-cell imaging results raise some intriguing questions about why and how RC territories remain stable over the course of HSV-1 replication. It would be interesting to determine the extent to which liquid phase separation and other mechanisms regulate the stability of RCs. Another question is why small RCs have a tendency to interact with each other. Actin and myosin have been reported to be at least partially involved in the movements of replicative sites and replication compartments (18). The movement leads to the bridging of transcriptionally active RCs with nuclear speckles to form structures that promote the export of late viral mRNAs. Further studies are needed to understand the biological significance of RC interactions.

Viral infection has a profound effect on the preexisting nuclear structures (55). However, the dynamic process remains poorly characterized. Telomere and centromeres can be visualized by their binding proteins, such as TRF1 and CHENPs. However, these proteins play roles in maintaining chromatin architecture and integrity (56–58). Therefore, their subcellular location and dynamics do not precisely account for chromatin states. Our two-color CRISPR imaging system provides spatially resolved, real-time measurements of both host chromatin organization and viral genome replication. The use of real-time imaging in living cells can better illuminate the dynamics of host-virus interactions, the sequence of events, and the consequences of chromatin destabilization. In addition, understanding how and why existing nuclear architecture must be modified to drive RC formation will undoubtedly benefit from real-time imaging analysis.

In summary, our approach raises exciting possibilities for studying the dynamics of virus replication and host-virus interactions at the DNA level. This method is simple to operate and introduces little perturbation of the virus genome. CASVIT can also be used to tag other DNA viruses. Viral DNA imaging tools are crucial for elucidating the role of host factors in virus latency, activation, and replication at the single-cell level. Notably, the two-color CASVIT permits precise analysis of the dynamics of RC formation and movements. In addition to gaining new insight into the fundamental biology of DNA viruses, CASVIT technology is particularly suited to rapid detection of the effects of compounds on viral replication, thus serving as a drug screening reporter system for viral replication inhibitors. Taken together, our observations indicate that CASVIT technology is a promising tool for both basic research and biotechnology applications.

MATERIALS AND METHODS

Cell culture.

HeLa cells and Neuro-2a cells were grown in Dulbecco's modified Eagle medium (DMEM) with high glucose (Gibco) in 10% fetal bovine serum (FBS) (HyClone) and 1% penicillin-streptomycin (Gibco). U2OS cells were grown in McCoy’s 5A (Procell) supplemented with 10% FBS and 1% penicillin and streptomycin. RPE-1 cells were cultured in DMEM/F-12 (HyClone) supplemented with 10% FBS and 1% penicillin and streptomycin. All cells were cultured at 37°C and 5% CO₂ in a humidified incubator. Regular testing for mycoplasma was performed for cultured cells.

Plasmid construction.

The construction of dSpCas9-GFP11_14X and GFP_1-10 plasmids was described in our previous work (40). We utilized the SunTag system to amplify the DNA labeling signal via dSaCas9. DNA fragments of scFV and tdTomato were assembled into the lentiviral vector phage-ubc (Addgene no. 40649) using a NEBuilder HiFi DNA assembly cloning kit (New England Biolabs). The fragments of dSaCas9, 10XGCN4, and P2A-blasticidin (BSD) were amplified and cloned into the lentiviral vector (Addgene no. 40649) to generate the plasmid pUbc-dSaCas9-10XGCN4-P2A-BSD using a DNA assembly cloning kit. To assemble CRISPR-Tags, DNA fragments (one repeat unit) containing a series of Caenorhabditis elegans genomic sequences (GNNNNNNNNNNNNNNNNNNNNGG) that can be recognized by the CRISPR-Cas9 system were commercially synthesized. All the sgRNAs used in this study were constructed by modifying our previous sgRNA vector (Addgene no. 164043). The sgRNA spacer determining the target sequence can be changed to recognize a new site by the PCR-based QuikChange cloning method. sgRNAs are listed in Table S1. To assemble a CRISPR-Tag with a desired number of targeting sites, the synthesized repeat unit containing 2 or 4 CRISPR targeting sites was amplified and assembled to obtain six repeats in a plasmid using the Golden Gate cloning method as we described previously (40). The sequences of CRISPR-Tags used to construct recombinant HSV-1^CRISPR-Tag are shown in Tables S2 and S3. To prepare donor DNA sequences for constructing recombinant HSV-1^CRISPR-Tag, the CRISPR-Tag sequence and the I-SceI-Kan^r (kanamycin resistance gene) fragment were amplified, respectively, and ligated into a plasmid vector, resulting in the CRISPR-Tag-(I-SceI)-Kan^r plasmid.

Construction of HSV-1^CRISPR-Tag.

The HSV-1 wild-type (WT) BAC based on strain KOS was described previously (59, 60). Based on WT BAC, we constructed HSV-1^mCherry with mCherry inserted between US9 and US10 of the virus following the procedure used to construct HSV-1^GFP that we have described (61). The CRISPR-Tag^Sp-(I-SceI)-Kan^r fragment containing arms homologous to an intergenic region between UL3 and UL4 was amplified to construct an HSV-1^{CRISPR-Tag(Sp)} BAC on the basis of the HSV-1^mCherry BAC using a previously described method (62). The HSV-1^{CRISPR-Tag(Sa)} BAC was constructed by modifying the HSV-1 WT BAC. BAC colonies were screened by PCR and verified by sequencing of the PCR products and restriction fragment length polymorphism. The resulting BACs were transfected into Vero cells to produce viruses, which were plaque purified once before propagation to make virus stocks. Viral titers were determined by plaque assays in Vero cells.

Virus growth curves.

To generate single-step growth curves, after infection with wild-type or recombinant virus (MOI = 5), HeLa cells were incubated at 4°C for 1 h with gentle rocking every 15 min to allow attachment of virus to cells. After 1 h of incubation, the medium containing the virus was removed. Cells were then washed three times with cold phosphate-buffered saline (PBS) and replaced with prewarmed fresh medium, which is considered the 0-hpi time point. Cells and the medium were harvested at various time points (0, 4, 8, 12, 18, and 24 h). Samples were frozen (at −80°C) and thawed three times. Virus titration was finally determined by plaque assays in Vero cells. Multiple-step growth curves were generated by virus infection at MOI of 0.2. Only the time points at 1, 12, 24, and 48 hpi were analyzed to compare the virus replication between wild-type and recombinant viruses.

Lentivirus production and generation of clonal cell lines.

To produce lentivirus, HEK293T cells were seeded into T25 flasks on day 1. The next day, cells were transiently transfected with 3,000 ng lentivirus constructs (dCas9-GFP11_14X, GFP_1-10, scFV-tdTomato, or dSaCas9-GCN4_10X-P2A-BSD), 2,800 ng pCMV-dR8.91, and 340 ng PMD2.G, using FuGENE (Promega) following the manufacture's recommended protocol. After 60 h of transfection, virus supernatant was harvested and mixed with PEG 6000 (Solarbio) in a ratio of 2:1. The virus was then concentrated after 12 h storage at 4°C. Finally, the virus precipitation was resuspended in PBS and stored at −80°C. HeLa, U2OS, and Neuro-2a cells were transduced with dCas9-GFP11_14X and GFP1-10 for HSV-1 DNA labeling. Clonal cell lines that expressed optimal level of dCas9-GFP_14X were isolated for efficient DNA labeling. To achieve two-color CRISPR imaging in HeLa cells, a dCas9-GFP stable cell line was further transduced with dSaCas9-GCN4_10X-P2A-BSD and scFV-tdTomato lentiviruses. BSD (2 μg/mL) was applied for positive-cell screening. Two-color clonal cell lines were also isolated to achieve the best signal-to-noise ratio (SNR) of DNA labeling.

Selection of CRISPR-Tags for viral DNA labeling.

Clonal HeLa cells expressing particular reporters (dSpCas9-GFP_14X or dSaCas9-tdTomato_10X) were grown in 8-well chambered cover glasses and transiently transfected with 50 ng CRISPR-Tag-containing plasmids and 500 ng sgRNA expression vectors. Twenty-four hours after transfection, fluorescent images were recorded using wide-field microscopy and analyzed to calculate the SNR of CRISPR labeling.

DNA labeling by dCas9-FP.

Clonal HeLa, U2OS, or Neuro-2a cells expressing particular dCas9-FP reporters were seeded in 8-well chambered cover glass. The next day, cells (50% to 70% density) were transiently transfected with 500 ng sgRNA expression vectors (targeting viral DNA or/and host chromatin). Twenty-four hours after transfection, cells were infected with WT or recombinant HSV-1 at a required MOI. To achieve two-color CASVIT, cells were infected with HSV-1^{CRISPR-Tag(Sp)} and HSV-1^{CRISPR-Tag(Sa)} at MOIs of 20 and 8, respectively. After infection for a specific time, cells were directly used for live-cell imaging or fixed with 4% paraformaldehyde before imaging. To label CRISPR-Tag containing plasmids, CRISPR-Tag plasmids (different doses), sgRNA (targeting TS1 sites in CRISPR-Tag; 500 ng), and a cotransfection marker (100 ng) were transfected into dCas9-GFP_14X cells seeded in 8-well chambered cover glasses. Imaging was performed at 24 h after transfection.

Immunofluorescence staining.

Samples were fixed, washed, blocked in blocking solution (3% bovine serum albumin [BSA] and 0.5% Triton X-100 in PBS) for 10 min at room temperature (RT), incubated with the ICP0 primary antibody (Santa Cruz, sc-53070; 1:2,000) in blocking buffer for 4 h at RT, rinsed three times with PBS, incubated with Alexa Fluor 647-conjugated secondary antibody (Abcam, ab150115; 1:1,000) for 45 min at RT, and finally washed three times before imaging.

Wide-field microscopy.

All wide-field microscopy images were obtained with a Nikon Ti2-E fluorescence microscope equipped with a 100× 1.45 numerical aperture (NA) PlanApo oil immersion objective, an LED source (SPECTRA 4), an sCMOS camera (ZYLA 4.2MP Plus), a Perfect Focus unit (Nikon), and a motorized stage (Nikon) with a stage incubator (Tokai Hit, STRF-WELSX-SET). Cells were grown in 8-well chambered cover glass for cell imaging. The quantifications in Fig. 2D and G, 3A, 4B, and 6B, D, and F and in Fig. S2B, S4A, S6B, and S8A were analyzed based on wide-field images.

Confocal microscopy.

The two-color images in Fig. 5B were recorded with an Airyscan detector on a Zeiss LSM880 microscope. All other confocal images were acquired on Olympus spinning-disk confocal system SpinSR, equipped with Yokogawa CSU-W1 scanner a 60× 1.49 NA oil Apochromat objective, an sCMOS camera (Prime 95B), 405/488/561/640 nm lasers (OBIS), and a piezo stage (ASI) with a stage incubator (Tokai Hit). To perform live-cell imaging, cells were maintained at 37°C and 5% CO₂ in a humidified chamber. Cells for confocal imaging were plated in 8-well chambered cover glasses. The images in Fig. 2F, 3E, 5E, and 6H and Fig. S7A were acquired on this confocal microscope.

Data analysis.

All the fluorescence imaging data were analyzed by ImageJ to calculate the area of RCs, the mean intensity, total intensity, SNR, and the area of fluorescent spots. GraphPad Prism (version 5; GraphPad Software, La Jolla, CA, USA [https://www.graphpad.com]) was used to calculate the mean values, P value, and the standard error of the mean (SEM) for the statistical analysis. A line scan was performed using the Analyze/Plot Profile function, a plugin for ImageJ. The parameters were then analyzed in Excel and plotted in GraphPad Prism. Acquired confocal z stacks were reconstructed into three-dimensional (3D) images using Imaris 9.3.1 software (Bitplane AG, Zurich, Switzerland). Signal-to-noise ratio was calculated as the ratio of the intensity of a fluorescent signal and the power of background noise, as follows: (maximum intensity of fluorescent spot − mean intensity of background signal)/SD of background signal, where SD is standard deviation.

Data availability.

Imaging data have been deposited in the Mendeley Database with doi:10.17632/t2why6p3mg.1. All other data are available from the corresponding authors upon request.