Analysis of herpes simplex virus type I nuclear particles by flow cytometry



Flow cytometry has been instrumental to characterize cell populations and examine their inner molecules and processes. In most instances, whole cells are analyzed, and hence, particle size is not an issue. Viruses are 2–3 orders of magnitude smaller than cells so flow cytometry has typically been used to study viral markers within whole infected cells. However, the ability to separate and purify viral particles representing different maturation stages within a viral life cycle would be a useful tool to analyze them in details and characterize the host proteins they associate with. Herpes simplex virus Type 1 is a 250 nm enveloped DNA virus that replicates in the nucleus where it assembles new viral particles called capsids. These capsids eventually travel to the cell surface and are modified along the way, producing several intermediate particles. In the nucleus, three types of stable nonenveloped 125 nm nuclear capsids exist that differ in protein composition and genome content. This includes so-called nuclear C-capsids that are the precursors of mature extracellular virions. We report that we can apply flow cytometry to sort these nuclear C-capsid intermediates by labeling the viral genome with Syto 13, a fluorescent marker that binds to nucleic acids. This is the first time flow cytometry has been used not only to detect but also to purify an intracellular viral maturation intermediate. This opens new research avenues in virology to study capsid assembly, maturation and egress, analyze mutant phenotypes, and define host factors associated with specific viral intermediates. © 2012 International Society for Advancement of Cytometry

Cells and their constituent molecules and processes have been studied by various techniques over the years, including biochemical and genetic approaches, diverse microscopy techniques [electron microscopy (EM), epifluorescence, confocal] and more recently by system biology tools such as DNA and RNA arrays, two-hybrid screens, and multiple “omics” approaches. One of the first widely available larger scale analysis tools has been flow cytometry and its derivative, flow cytometric sorting, both widely used since the early 1970s (1). Amazingly, refinements of these techniques now allow several markers to be followed in parallel, hence allowing complex processes to be finely monitored and new research avenues to be explored (2). One limitation of flow cytometry has been the size of objects under study. Typical experiments examine whole cells, which are up to tens of microns in diameters and are thus well above the recommended lower detection threshold of a half micron for most instruments. However, viruses are much smaller entities ranging in the tens (e.g., Adeno-associated virus, 20 nm icosahedral) to hundreds of nanometers (e.g., Poxvirus, 250 × 300-nm brick shape). Consequently, viral research using flow cytometry has often relied on viral markers within whole infected cells.

Thus far, a number of studies demonstrated the feasibility to use flow cytometry to analyze sub-micron particles (3–7). Bacteria, which are below the detection limit of common flow cytometry instruments, have indeed been detected with success when labeled with fluorescent dyes (8). Furthermore, early work by Shapiro and coworkers (9) indicated it might be possible to study by this technique the small T2 bacteriophage, a tubular virus of roughly 100 nm in diameter (10), and Reovirus, which has a 600–800 nm icosahedral capsid (11). More recently, the laboratories of Brussaard and Steen (12–15) extended these findings to several other viruses including herpes simplex virus Type 1 (HSV-1) and the related cytomegalovirus stained with SYBR green. However, this powerful technique has yet to be used to separate different viral particle populations. This would provide new means to learn about virus assembly, viral intermediates, and their interactions with host proteins.

HSV-1 is a human pathogen causing mild to severe conditions ranging from cold sores to blindness, encephalitis, and serious new born complications as well as being an aggravating factor for AIDS (16–18). It has a 152 kb double-stranded DNA genome within a 125 nm icosahedral capsid shell, itself wrapped by a complex layer of proteins termed the tegument and an external host-derived envelope, for a final size of roughly 250–300 nm (19). Following entry of the virus at the cell surface, the viral genome is ultimately delivered to the nucleus, where it replicates and leads to the assembly of new capsids. Four distinct nonenveloped capsid species are present in the nucleus. The first one is procapsids that are rarely detected because they are thermodynamically unstable (20). These procapsids are believed to give rise to three stable nuclear capsid populations. A-capsids appear to be capsids that fail to properly incorporate the viral genome (21). Similarly, B-capsids also lack viral DNA but, unlike A-capsids, are thought to never initiate the genome packaging step. They can be distinguished from the former capsids on the basis of their distinctive protein contents. In contrast to these two abortive particles, C-capsids incorporate the viral genome and ultimately form mature enveloped virions. According to common belief, only mature nuclear C-capsids travel across the two nuclear membranes by an envelopment/de-envelopment mechanism and are ultimately re-enveloped elsewhere in the cell, likely at the trans Golgi network (TGN) (22–25). These viral particles subsequently leave the TGN toward the cell surface by a constitutive pathway implicating the host protein kinase D and myosin Va (26–28). This maturation process from the nucleus to the cell surface is the subject of intense scrutiny as it is poorly understood, in part because of the several viral intermediates along the way. It would thus be most useful to individually isolate these intermediates to better characterize them.

This study probes the possibility to sort intracellular viral intermediates by focusing on HSV-1 viral nuclear C-capsids. We now report that we can detect these 125 nm particles by flow cytometry and sort them according to their DNA content by labeling them with the capsid permeable and nucleophilic Syto 13 fluorescent stain. PCR and EM were consistent with the isolation of C-capsids with minimal contamination from A- and B-capsids. These findings now enable a detailed analysis of this important viral intermediate. Moreover, this approach should be applicable to other HSV-1 intermediates and even to other viruses to study different maturation stages in an effort to sort our way through their viral life cycles.


Cells and Viruses

Human HeLa S3 cells (ATCC # CCL-2.2) adapted to suspension culture were grown at 37°C in JMEM (Joklik's modified Eagle's medium; Sigma-Aldrich) supplemented with 5% fetal bovine serum, 0.1 mM nonessential amino acids, 100 U/mL penicillin, and 100 μg/mL streptomycin. HSV-1 F strain and the recombinant K26GFP KOS strain, which expresses green fluorescence protein (GFP) tagged VP26 capsid protein (29), were originally provided from Beate Sodeik (Hannover Medical School, Germany) and Prashant Desai (The Johns Hopkins University, USA), respectively. Both viruses were amplified on BHK and tittered on Vero cells as previously described (22).

Isolation of Total Nuclear Capsids

HeLa S3 cells were infected with wild type HSV-1 or K26GFP for 8 h at a multiplicity of infection of 5. The nuclei were then harvested by mechanical disruption of the cells and density centrifugation in the presence of a protease inhibition cocktail (8.25 μM de chymostatin, 1.05 μM leupeptin, 0.38 μM aprotinin, 0.73 μM pepstatin A; Sigma-Aldrich) as described before (30). These nuclei were subsequently topped up with 1% d'IGEPAL CA-630 (Sigma-Aldrich) and incubated for 30 min at 4°C. They were then treated with 500 U/mL DNase I (Roche) and 25 μg/mL of RNAse A for 10 min at 37°C to remove nonencapsidated nucleic acids. To isolate the nuclear viral capsids, the nuclear lysates were briefly sonicated and large debris removed by low speed centrifugation (5 min/4°C at 300g). The supernatant, containing the nuclear capsids, was filtered through a 0.45 μm filter and spun 1 h through a 35% (w/w) sucrose cushion at 100,000g at 4°C. The total capsid pellet was finally resuspended in 100 μL of MNT (30 mM MES, 100 mM NaCl, 20 mM Tris-HCl pH 7.4). The capsids were either frozen or analyzed immediately (see below).

Fluorescent Labeling of Viral Particles

Four microliters of a diluted fraction of DNAse/RNAse treated nuclear capsids were incubated for 30 min at 4°C with 1–5 μM Syto 11–14, 16, 21, 24, or 25 (Invitrogen). Alternatively, 5 μg/mL of Hoechst 33342 was used to label the capsid bound DNA. These labeled capsids were finally examined by fluorescence microscopy and by flow cytometry.

Efficient Detection of Intranuclear DNA

To determine the ability of Syto 13 to specifically label encapsidated viral genomes, 200 μL of wild type or K26GFP nuclear lysates (see above) were (A) treated for 10 min at 55°C with 1 mg/mL of proteinase K (recombinant, PCR grade, Roche) to disassemble the protein-based capsids and free their genome content and thus obtain the total DNA present in the nuclei; (B) first treated for 10 min at 37°C with 500 U/mL of DNAse I (Roche) to digest both cellular and nonencapsidated viral DNA then with proteinase K to only keep the capsid bound viral DNA; or (C) first treated with proteinase K subsequently inactivated for 10 min at 95°C (to avoid the degradation of the DNAse I) and finally incubated with DNAse I in the presence of 10 mM CaCl2 to protect the DNAse I from any residual proteinase K activity (31). In this last scenario, no DNA should remain because both encapsidated and nonencapsidated DNA should be digested. In all cases, the samples were processed as per manufacturer's instructions to isolate the non digested DNA with a GenElute Mammalian Genomic DNA Miniprep Kit (Sigma-Aldrich), which contains RNAse I. These samples were ultimately analyzed by quantitative PCR using GFP specific primers to detect the K26GFP viral genome (forward: 5′ACGTAAACGGCCACAAGTTC3′; reverse: 5′AAGTCGTGCTGCTTCATGTG3′) or GAPDH specific primers to detect cellular DNA (forward: 5′AGGGCCCTGACAACTCTTTT3′; reverse: 5′AGGGGTCTACATGGCAACTG3′). This was performed with a LightCycler 480 (Roche) using the PerfeCTa SYBR Green SuperMix (Quanta BIOSCIENCES). The data were finally analyzed using LightCycler® 480 software version 1.5 and expressed in relative genome copy. Two independent experiments, each done in duplicates, were pooled.

Fluorescence Microscopy

Fluorescently labeled capsids were examined on an Axiophot microscope (Zeiss) equipped with a Retiga 1300 camera (Q imaging) or, for confocal microscopy, on a DM IRBE inverted microscope (Leica) coupled to a SP1 spectrometer and argon (488 nm), argon-krypton (568 nm), and helium-neon (647 nm) lasers. Epifluorescent images were analyzed with Northern Eclipse (Empix imaging) and Photoshop CS5 (Adobe), whereas confocal images were processed with the LCS Lite software and Photoshop CS5.

Flow Cytometry

Polystyrene beads from Estapor (100 and 800 nm) or Polysciences (200 nm) were diluted 1:1,000 in MNT and analyzed on a FACSAria IIu sorter (BD Biosciences) equipped with a 100 μm nozzle and 405, 488, and 633 nm lasers. The original concentrations of the samples were 6.36 × 1013 (100 nm), 5.68 × 1012 (200 nm), and 2.34 × 1011 (800 nm) beads/mL. A 100 μm nozzle, rather than a smaller one, was used to reduce the pressure and hence maximize the excitation time and signal strength. The fluorescently labeled viral capsid preparations were diluted 1:500 and analyzed on the above instrument. In that case, the concentration of the stocks is unknown and varied from preparation to preparation. For GFP and Syto labeled capsids, 502 nm longpass and 530/30 bandpass filters were used. Analysis and sorting were performed at low pressure (23 psi) and flow rate between 1 and 3 for a maximum of 3,000 events/sec to minimize coincidental events. In all cases, a minimal threshold of 200 for the SSC channel was use to remove some of the background signal. The data were initially acquired with the FACSDiva software (BD Biosciences) and then processed with FlowJo (TreeStar). After sorting, the diluted capsids were concentrated at 100,000g at 4°C for 1 h on a 35% sucrose cushion (w/w). The capsid pellet was then resuspended in MNT and frozen at −80°C.

The instrument was cleaned between every sample with BD FACS Clean (with 10% bleach) and back flushed with sheath. However, only one large sample was normally analyzed per day. When multiple samples were analyzed, we started with unlabeled capsids and finished with labeled ones. Each solution, including the control MNT buffer and sheath, was 0.22 μm filtered. At the end of the day, the instrument was cleaned with 1% Virkon (Antec International).

Electron Microscopy

The samples were processed for negative staining as previously described (32). Briefly, 5 μL of the sorted capsids were deposited on a copper grid previously coated with Formvar and carbon (Canemco & Marivac). Excess liquid was removed with a filter paper and the samples contrasted with 2% uranyl acetate (Canemco & Marivac). Following a quick wash in water, the grids were air dried and examined on a Philips 300 transmission electron microscope. Quantification of the capsids found in the samples was done from randomly selected fields from at least three independent experiments.


To independently evaluate the presence of viral DNA in the flow cytometry sorted samples (Fig. 5), they were analyzed by standard PCR using HSV-1 UL20 specific primers and limited cycling as before (29) (sense primer: 5′atgaccatgcgggatgaccttc3′; antisense primer: 5′ttagaacgcgacgggtgcattc3′). The amplified fragments were separated on a 1.5% agarose gel and revealed with ethidium bromide. For these experiments, the positive control consisted of viral DNA extracted from infected cells, whereas the negative control was DNA extracted from noninfected cells.


Detection of Small Particles by Flow Cytometry

To evaluate if flow cytometry may be a practical approach to sort viral capsids, we first tested commercially available microspheres of different sizes. This was particularly important given the 0.5 μm detection limit stated by the instrument manufacturer, a limit four times above the size of HSV-1 nuclear particles, which are 125 nm in diameter. We, therefore, analyzed by flow cytometry 100 nm, 200 nm, and 800 nm beads at a low flow rate for optimal detection. Figure 1 shows that the 800 nm beads (i.e., above the instrument threshold) were readily detected by both side and forward scattering. The smaller 100 and 200 nm beads differed substantially from the larger beads but not one from another. These smaller beads were also indistinguishable in the FSC channel from the buffer control, where 0.22 μm filtered MNT was passed over the instrument. However, one could clearly distinguish them in the SSC channel, which revealed both individual and aggregated beads. Together, this showed that particles as small as 100 nm could indeed be detected but not sorted from 200 nm particles based on their size. Interestingly, this implied that the 125 nm HSV-1 nuclear capsids should be detected by flow cytometry, not based of their size but rather the ability of such small particles to reflect light as measured by side scattering.

Figure 1.

Flow cytometric sorting of fluorescent beads. Commercially available 100, 200, or 800 nm fluorescent beads were diluted 1:1,000 in MNT to reduce the number of events/sec and analyzed on a BD FACSAria sorter. The left panels denote the forward (FSC) and side (SSC) scattering, whereas the right panels show their forward scattering distribution. The smaller peaks for the 800 nm beads were attributed to bead aggregates. “Buffer” refers to the MNT used to dilute the beads. A minimal threshold of 200 for the SSC channel was applied to remove some of the background signal. No gates were applied so the dot plots represent 100% of the samples.

Detection of GFP Tagged HSV-1 Capsids

As for many other viruses, several components of the HSV-1 virions have been tagged with fluorescent moieties. Hence, genetically modified HSV-1 strains encode a labeled component of the capsid, the tegument or a glycoprotein present in the viral envelope. One such strain named K26GFP tags the essential HSV-1 VP26 viral capsid protein and is, thus, a useful marker to follow intracellular viral capsids at different stages of infection (22, 29). To directly assess the ability of the flow cytometer to detect HSV-1 capsids, we took advantage of this recombinant strain. Cells were infected with K26GFP or control nonfluorescent wild type virus and the nuclei harvested 8 h later. As detailed in “Materials and Methods” section, capsids were then isolated from these nuclei. As expected from the above results, both types of capsids were detectable by light scattering (Fig. 2, left panels). Excitingly, the K26GFP capsids gave a significant fluorescence signal, whereas the untagged wild type capsids had none (Fig. 2, middle and right panels). We, therefore, concluded that GFP tagged HSV-1 capsids can indeed be detected by flow cytometry by light scattering and fluorescence.

Figure 2.

Detection of GFP tagged capsids by flow cytometry. DNAse/RNAse treated nuclear capsids were prepared from wild type (i.e., nonfluorescent) infected cells or from cells infected with the K26GFP strain (i.e., fluorescent capsids). They were then diluted 500-fold in MNT to reduce the number of events/sec and examined by flow cytometry. A minimal threshold of 200 for the SSC channel was once again used to remove some of the background signal. The left panels show the dot plots of the scattering (SSC vs. FSC) that were used to define gates (see boxes). These gates contained the bulk of the capsids (>90%) and excluded the larger capsid aggregates. The samples were excited with a 488 nm laser coupled to an emission filter allowing the 515-545 nm wavelengths to go through. The middle and right panels show the fluorescence profiles of the gated material (dot plots and histograms, respectively).

Detection of Mature Nuclear C-Capsids by Labeling Their DNA Content

The ability to detect HSV-1 capsids enabled us to evaluate if it is possible to enrich the preparation in nuclear C-capsids, the precursor of mature virions. Because A- and B-capsids are devoid of viral genome and C-capsids are loaded with a complete viral genome, we tested various fluorescent DNA dyes to label the latter capsids. Initial staining of nuclear capsids with Hoechst or Dapi revealed spotty signals that partially overlapped with K26GFP by immunofluorescence but failed to give a signal by flow cytometry, presumably because the intensity of the dyes was too weak (data not show). Given that Syto 13, a RNA and DNA binding fluorescent stain, was recently used to successfully detect 0.1–1.0 μm microparticles (3), we probed a commercial sampler kit consisting of eight different Syto reagents (Invitrogen). Each of these membrane permeable cyanine derived stains has unique properties in terms of nucleic acid affinity and fluorescence energy with distinct but similar excitation/emission spectra. Collectively, they bind nucleic acids by multiple means, including intercalation, direct interaction with the nucleic acid backbone and reportedly by binding to the DNA groove (33). Most importantly, as free molecules they produce low background signals that are significantly enhanced once they bind to nucleic acid (quantum yield greater than 0.4). Enriched nuclear capsid preparations were, thus, individually labeled with each of the dyes as per manufacturer's instructions and first examined by immunofluorescence. While most gave no apparent specific signal over the background (i.e., Syto 12, 14, 16, 21, 24, 25), two proved promising and exhibited significant signals (Syto 11 and 13; data not shown). Moreover, this indicated that these two reagents could associate with viral DNA present in assembled capsids.

To evaluate if we specifically target viral capsids with the Syto dyes, we first treated whole nuclear lysates with DNAse I to digest both cellular and nonencapsidated viral DNA and then released the viral genome from the capsids with proteinase K. Under these conditions, only viral DNA within the capsids is expected (34). We then evaluated by qPCR the presence of viral and cellular DNA in the samples following the purification of total genomic and viral DNA with a commercial kit containing RNAse A (see “Materials and Methods” section). As control, we solely digested the samples with proteinase K to get the total amount of cellular or viral DNA present in the nuclear lysates. We additionally inverted the DNAse I/proteinase K steps by first treating the nuclear lysates with proteinase K then DNAse I to remove all DNA from the sample. As expected, strong cellular and viral signals were present when the nuclear lysates were treated with proteinase K only and no signal was found when first treated with proteinase K then DNAse I (Supporting Information Fig. 1). In contrast, when the samples were first digested with DNAse I then treated with proteinase K, 1–2% of the total viral DNA was detected within the capsids, whereas no cellular DNA was detected. This indicated that these conditions were appropriate to specifically label the encapsidated DNA, that only a small proportion of the viral DNA was encapsidated, and that no cellular DNA was detectable within the capsids. We concluded it was thus possible to label the encapsidated viral DNA with nucleic acid reagents under these conditions.

Because gating on the capsids was possible, we examined by flow cytometry the fluorescence levels of nuclear capsids labeled with either Syto 11 or 13. We, thus, isolated viral capsids from DNAse I/RNAse A infected nuclei and then stained the capsids. Naturally, proteinase K was absent in these experiments to preserve the integrity of the capsids. As controls, we included unstained wild type or K26GFP capsids along with wild type capsids stained with Syto 24, which labeled the capsids less efficient than Syto 11 or 13. As expected, the level of fluorescence for unlabeled wild type capsids was minimal [mean fluorescence intensity (MFI) of 13; Fig. 3]. The data confirmed a stronger signal for the tagged K26GFP capsids (MFI of 2,878), in agreement with our immunofluorescence observations. Interestingly, Syto 11- and Syto 13-stained capsids gave an even stronger fluorescent signal (MFI of 25,700 and 53,600, respectively). Syto 24-labeled capsids had an intermediate signal that was oddly bimodal (MFI of 11,400). Altogether, this meant that it is possible to stain the encapsidated viral genome in intact HSV-1 nuclear C- capsids and detect them by both light scattering and fluorescence. Given the best signal obtained with the Syto 13 dye, all remaining experiments were performed with this dye.

Figure 3.

Testing of Syto fluorescent nucleic acid stains. DNAse and RNAse treated nuclear capsids were isolated as before from wild type infected cells and stained with Syto 11, 13, or 24. As control, unstained wild type or K26GFP nuclear capsids were used. All capsids were diluted 500-fold and analyzed by flow cytometry. The samples were excited with a 488 nm laser coupled to an emission filter allowing the 515-545 nm wavelengths to go through. The left panels show the light scattering of the samples (SSC vs. FSC). Once again, a minimal threshold of 200 for the SSC channel was applied as before to remove some of the background signal. The boxes represent the gates (greater than 90% of the samples) used for analysis in the middle and right panels. These panels show the fluorescence profiles of the gated samples (dot plots and histograms, respectively). The signals for Syto 13 stained capsids (MFI of 53,600) was significantly above that for Syto 11 (MFI of 25,700), Syto24 (MFI of 11,400), or K26GFP (MFI of 2,878).

Sorting of Nuclear Capsids

Given the ability of flow cytometry to physically detect the viral particles and the possibility to specifically fluorescently stain the viral genome within intact capsids, we proceeded to sort by flow cytometry the HSV-1 nuclear capsids. Wild type nuclear capsids were isolated from infected cells as before and incubated with Syto 13 to label the encapsidated viral DNA. They were then sorted according to their fluorescence profiles, gating on the capsids by both side scattering and strength of Syto 13 emission. Interestingly, a reproducibly continuous fluorescence signal ranging from low to high was observed (Fig. 4). Moreover, a change of slope was always observed past the mid point in the fluorescence channel. We then arbitrarily defined three regions of interest, with fraction 1 having no fluorescence signal (no DNA content), fraction 2 with an intermediate signal and fraction 3 with the highest signal (high DNA content), with the change of slope as the boundary between fractions 2 and 3 (see Fig. 4). Fraction 1 was reproducibly detected and represented a small portion of all viral particles (1.86% in the experiment shown; 4.6% ± 0.9 on average from 9 independent flow cytometry analyses). Similarly, fractions 2 and 3 were also always detected and on average contained 57.1% ± 4.5 and 36.1% ± 5.9 of total capsids, respectively. The three fractions exhibited increased levels of fluorescence (fraction 1: MFI of −532; fraction 2: 9,970 and fraction 3: 70,719 in the experiment depicted in Fig. 4). In contrast, unlabeled wild type capsids were exclusively found in the first fraction (97.8% ± 0.8 of all capsids; average of 4 independent experiments) with an average fluorescence of 3 in the experiment shown in Figure 4. Thus, it seemed possible to distinguish different capsid populations in these nuclear preparations.

Figure 4.

Sorting of Syto 13 labeled capsids. DNAse and RNAse treated nuclear capsids were stained with Syto 13, diluted 500-fold and sorted by flow cytometry. As control, unstained nuclear capsids were used. Three gates were arbitrarily defined in the SSC/fluorescent channels. Fraction 1 was based on the control capsids, which were not fluorescent. The remainder of the signal was divided up into two fractions (2 and 3) with the change of slope in the SSC channel as the boundary between the two fractions. The mean fluorescence intensities were −532 (fraction 1), 9,970 (fraction 2), and 70,719 (fraction 3) in the experiments depicted here. The samples were excited with a 488 nm laser coupled to an emission filter allowing the 515-545 nm wavelengths to go through. A minimal threshold of 200 for the SSC channel was used to remove some of the background signal.

Characterization of the Viral Fractions Sorted by Flow Cytometry

To insure that the particles previously sorted were indeed of viral origin and examine how the three fractions correlate with A-, B-, and C-capsids, we performed a PCR amplification using HSV-1 specific primers, taking advantage of the fact that A- and B-capsids differ from C-capsids in their genome content. Note that the sorted capsids were pretreated with DNAse I as described in “Materials and Methods” section, so the PCR reaction only monitored encapsidated viral DNA. Although sorting was a slow process due to the low pressure and flow rate required to avoid coincidental events, we retrieved in the order of 1–10 μg of total proteins for A- and B-capsids for roughly 24 h of sorting. This was significantly higher for C-capsids, for which we obtained up to 30 μg. These differences in yields were not surprising because A-, B-, and C-capsids are not in equal abundance in the nucleus and even vary among cell types. We, therefore, normalized the PCR by total protein content, as determined by the commercial Bradford assay (Bio-Rad). As expected, a control HSV-1 total cell lysate was positive by PCR, whereas a cell lysate prepared from noninfected cells was negative (Fig. 5). The data further indicated that fraction 1 (no fluorescence by flow cytometry), did not contain any detectable viral genome, consistent with DNA free capsids. At the other end of the spectrum, particles from fraction 3, which were strongly stained by Syto 13, were clearly positive by PCR, thus confirming that it contained viral DNA as would be expected from C-capsids. Interestingly, fraction 2 contained a faint but detectable level of HSV-1 DNA in line with the weak Syto 13 staining pattern seen by flow cytometric sorting.

Figure 5.

PCR of flow cytometry sorted capsids. DNA I treated and flow cytometry sorted nuclear capsids were concentrated by high speed centrifugation and analyzed by standard PCR using primers specific for the HSV-1 UL20 gene. DNA purified from either mock or HSV-1 infected cell lysates was included as negative and positive controls respectively. Either 0.25 or 1.25 μg of total protein of each fraction was used for the PCR reactions. Following amplification, the samples were resolved by electrophoresis on an agarose gel stained with ethidium bromide. For clarity, the image of the stained gel was inverted to better show the signals. Note the positive signal for fractions 2 and 3. The size of the fragment is indicated to the left of the figure. The absence of signal for the 1.25 μg of fraction 2 is likely caused by the large volume of sample needed for the PCR reaction which may be inhibitory. In this particular experiment, 1.25 μg of protein corresponded to 5 μL, 31.5 μL, and 12.5 μL for fractions 1, 2, and 3, respectively. The PCR reaction is done in 50 μL, so fraction 2 constitutes more than 60% of the total volume.

Thus far, the results were most consistent with the enrichment of C-viral particles in the third fraction and the presence of DNA free capsids in fractions 1 and 2. To insure this was the case, we next examined the sorted fractions by EM negative staining. This technique is widely used to formally identify viral particles and distinguish different HSV-1 viral intermediates. C-capsids have a dark center by this technique, whereas DNA free A- and B-capsids are whitish and grayish, respectively. We, thus, contrasted the samples with uranyl acetate and visually inspected them (Fig. 6). Although somewhat dilute, even after concentration of the samples by high speed centrifugation, capsids were indeed detected in each of the fractions. Although some heterogeneity was present, capsids in fraction 1 nearly always contained capsids with a whitish empty core reminiscent of A-capsids (21), whereas fraction 3 mostly contained strongly stained electron dense capsid cores typical of C-capsids (30). Finally, fraction 2 was the most heterogeneous with capsids with cores of intermediate densities between A- and C- capsids similar to B-capsids reported elsewhere (21, 35).

Figure 6.

EM of flow cytometry sorted capsids. HSV-1 viral capsids were harvested from nuclei as above and flow cytometry sorted into three different fractions based on their DNA content (none, low, and high). Following their concentration by high speed centrifugation and resuspension in MNT, the capsids were examined by negative staining. Typical capsids are shown for each of the three fractions. Fraction 1 (top panels) contained mostly empty A-capsids, whereas fractions 2 (medium panels) and 3 (lower panels) were enriched for B- and C-capsids, respectively.

Quantification of the capsids by EM mirrored the flow cytometry results. Hence, the distribution of nuclear capsids before flow cytometric sorting was 13.7% ± 5.7 (A-capsids), 58.0% ± 2.6 (B-capsids), and 28.0% ± 7.0 (C-capsids) as determined by negative staining (n = 387). Meanwhile, as mentioned above, 4.6% ± 0.9 of the capsids were sorted to fraction 1, whereas 57.1% ± 4.5 and 36.1% ± 5.9 were in fractions 2 and 3, respectively. Unfortunately, it was not possible to evaluate the purity of fractions 1 and 2 by EM due to their low capsid content. However, EM quantification revealed that fraction 3 was composed of 90.7% ± 5.7 of C-capsids and only minor amounts of contaminating A- and B-capsids (6.3% ± 6.8 and 3% ± 3.5, respectively; n = 167), representing ∼3-fold enrichment of the C-capsids by flow cytometry. Taken together, the data hint that fractions 1 and 2 could tentatively be A- and B-capsids, respectively, and that fraction 3 was most likely C-capsids. Thus, flow cytometric sorting constitutes a valid and novel method in the current toolbox to purify HSV-1 nuclear C-capsids, which are the precursors of mature extracellular virions.


Viruses are small entities that are below the recommended size fractionation limit of common flow cytometry instruments. This study reveals that while particles in the 100–200 nm range cannot be resolved one from another, they are nonetheless detectable (Fig. 1). A slow flow rate is likely an important factor to prevent coincidental events and favor single detection of such small particles. Furthermore, incorporation of a fluorescent tag in the capsid or staining of the viral genome with a fluorescent reagent is also sufficient to detect them. Why some dyes worked better than others is unclear at the moment. It is likely an issue of specific binding sites, affinity, and emission energy. It may also be that some dyes better traverse the protein shell of the capsids. Interestingly, labeling of the HSV-1 DNA with the permeable nuclei acid Syto 13 stain proved more efficient to detect these particles than tagging one of the capsid proteins with GFP (Figs. 2 and 3). This is somewhat surprising given the high copy number of VP26 in the capsids, which is estimated at 952 copies (36). However, it is not known how many copies of Syto 13 bind to the very large HSV-1 genome and what the relative fluorescence intensity of this stain is compared with GFP. Importantly, this signal was not due to contaminating cellular nucleic acids or free viral DNA because the capsids were pretreated with RNAse A and DNAse I to remove all the nonencapsidated nuclei acids. Altogether, this means that it is possible to detect viral capsids by both light scattering and fluorescence and that labeling of the viral genome with Syto 13 is even better than when the minor capsid component VP26 is GFP tagged.

Light scattering and fluorescence signals provided a mean to sort the nuclear HSV-1 viral capsids. Although populations were initially arbitrarily defined (Fig. 4), an analysis of these fractions by PCR (Fig. 5) and EM (Fig. 6) strongly suggested that fraction 3 is enriched in C-nuclear capsids, whereas fractions 1 and 2 were suggestive of A- and B- nuclear capsids, respectively. Hence, fraction 1 was devoid of any detectable viral DNA and had an empty core by negative staining, which is reminiscent of nuclear A-capsids as shown by others (21). Fraction 2 was somewhat problematic to define, because it was minimally positive for the viral genome and had an intermediate core density by negative staining. Although their appearance by EM hinted at nuclear B-capsids, the low DNA signals observed by flow cytometry and PCR were not consistent with that conclusion given these capsids are normally considered DNA free (21, 35). The most likely explanation is that some contaminating C-capsids were present in this fraction and accounted for the weak DNA signal. Technically, one alternative explanation would be if we detected a partial incorporation of DNA in B-capsids or if they contained DNA in a form that binds less efficiently the Syto dye. However, these possibilities require further validation before they can be considered. To formally define fractions 1 and 2, a detailed biochemical characterization will be needed. However, this may be complicated by cross contamination because it is not be possible to distinguish by Western blotting between reduced but biologically relevant levels of particular proteins in one capsid type versus a limited detection due to low contamination by other capsid types. This is exemplified by sucrose sedimentation studies of HSV-1 nuclear capsids that show the enrichment of viral scaffold proteins in B-capsids but also trace amounts of these molecules in other fractions (e.g., (37–40)).

In contrast to first two fractions, the data strongly suggest that fraction 3 is C-capsids because PCR specifically detected the viral genome in those particles, that this viral genome is normally DNAse I resistant but DNAse I sensitive when pretreated with proteinase K (i.e., that genome is present within a protein shell), and the particles are of the right size. At times, one can also see the angular form of the icosahedral capsids on some images. Finally, these particles also exhibited a classical dense core by negative staining typical of C-capsids (30). EM analysis of this fraction revealed it was significantly enriched for that capsid type (90.7% ± 5.7 were indeed C-capsids) with less than 10% contamination by A- or B-capsids. This is slightly better than the reported 25% contamination by the classical 20–50% sucrose gradient purification (40). While neither approach yields perfectly pure samples, flow cytometric sorting thus constitutes a novel purification step that does not supplant classical sedimentation protocols but rather complements them when strongly enriched nuclear capsids are desired.

One particularly puzzling observation was the reproducible change of slope between fractions 2 and 3 in the SSC channel (Fig. 4). At this point, we do not know what is causing such change. Given that side scattering is influenced by surface granularity as opposition to particle size, it might be that C-capsids differ in that respect from the other two capsid populations but at this point it is difficult to ascertain the nature of this difference. Such slope change could theoretically be imparted by different DNA conformations, as mentioned above. Additional work is needed to clarify this point.

Flow cytometric analysis and sorting of cells has been instrumental to study various important cellular processes such as B and T cell biology, hematopoiesis, and stem cell differentiation to name a few. However, host-pathogen interactions involving viruses has often been limited to follow viral markers in infected cells due to the small size of these pathogens. This study shows that one can use flow cytometry not only to detect but also to sort nuclear HSV-1 intermediates (Table 1), despite an inappropriate size resolution at that range. This hints that the sorting limit of flow cytometry instruments is in essence lower than the half micron stated by the manufacturers when viral particles are coupled to a DNA labeling dye or a GFP tagged virion component. This opens up exciting new avenues to study the life cycle of viruses and their interaction with their hosts. For instance, it should now be possible to incorporate flow cytometric sorting as a purification step in combination with classical sedimentation techniques to enrich specific viral intermediates and further analyze them in isolation from other viral intermediates. This may be particularly useful to study capsid maturation and egress, the complex coating of the capsids with the tegument, analyze mutants or define the interactions of specific viral intermediates with host proteins such as motor proteins, kinases, cytoskeletal components, or cellular proteins that might be incorporated in the viral particles. Moreover, this opens new research avenues to study different maturation stages of other viruses.

Table 1. HSV-1 capsid types analyzed
 Stability at 4°CDNA contentCapsid shapeFinal outcomeComment
Properties Nuclear capsids     
 ProcapsidsUnstableNoneRoundA, B and C capsidsNot Analyzed
 C-capsidsStableViral genomeIcosahedralMature virus 
  Detection by flow cytometry   
 Side scatteringFluorescencePCREMInterpretation
  • The four nuclear capsid types found in HSV-1 infected cells are listed along with their known stability at 4°C and DNA content in the top portion of table. The bottom half of the table shows the results of this study.

  • Procapsids were not analyzed in this study since the samples were isolated at 4°C.

  • a

    Biochemical analysis required (see text).

  • b

    Likely due to some contamination by C-capsids.

This study Fraction     
 1YESNo fluorescenceNo SignalLight corePossible A-capsidsa
 2YESLow fluorescencebLow signalbVariablePossible B-capsidsa
 3YESHigh fluorescenceStrong signalDense dark coreC-capsids


The authors thank Danièle Gagné of the IRIC's flow cytometry platform for her excellent technical skills, expertise, and advice. They also thank Kerstin Radtke, Danièle Gagné, and Daniel Henaff for critical reading of the manuscript.