Conceived and designed the experiments: ELB. Performed the experiments: ELB PF PW S-BC. Analyzed the data: S-BC PF ELB. Contributed reagents/materials/analysis tools: PW ELB. Wrote the paper: S-BC ELB. Developed the synchronization protocol: ELB PF. Performed immunogold EM: PW ELB. Performed confocal imaging: ELB PF S-BC.
The authors have declared that no competing interests exist.
Herpes simplex type 1 (HSV1) replicates in epithelial cells and secondarily enters local sensory neuronal processes, traveling retrograde to the neuronal nucleus to enter latency. Upon reawakening newly synthesized viral particles travel anterograde back to the epithelial cells of the lip, causing the recurrent cold sore. HSV1 co-purifies with amyloid precursor protein (APP), a cellular transmembrane glycoprotein and receptor for anterograde transport machinery that when proteolyzed produces A-beta, the major component of senile plaques. Here we focus on transport inside epithelial cells of newly synthesized virus during its transit to the cell surface. We hypothesize that HSV1 recruits cellular APP during transport. We explore this with quantitative immuno-fluorescence, immuno-gold electron-microscopy and live cell confocal imaging. After synchronous infection most nascent VP26-GFP-labeled viral particles in the cytoplasm co-localize with APP (72.8+/−6.7%) and travel together with APP inside living cells (81.1+/−28.9%). This interaction has functional consequences: HSV1 infection decreases the average velocity of APP particles (from 1.1+/−0.2 to 0.3+/−0.1 µm/s) and results in APP mal-distribution in infected cells, while interplay with APP-particles increases the frequency (from 10% to 81% motile) and velocity (from 0.3+/−0.1 to 0.4+/−0.1 µm/s) of VP26-GFP transport. In cells infected with HSV1 lacking the viral Fc receptor, gE, an envelope glycoprotein also involved in viral axonal transport, APP-capsid interactions are preserved while the distribution and dynamics of dual-label particles differ from wild-type by both immuno-fluorescence and live imaging. Knock-down of APP with siRNA eliminates APP staining, confirming specificity. Our results indicate that most intracellular HSV1 particles undergo frequent dynamic interplay with APP in a manner that facilitates viral transport and interferes with normal APP transport and distribution. Such dynamic interactions between APP and HSV1 suggest a mechanistic basis for the observed clinical relationship between HSV1 seropositivity and risk of Alzheimer's disease.
Herpes simplex virus type I (HSV1), an alpha herpesvirus, is endemic in the general population, causing life-long latent infections in neurons. Like many other viruses, after assembly in the nucleus HSV1 nucleocapsids transport outwards through the cytoplasm towards the cell surface both in epithelial cells and in neurons
How HSV1 coordinates assembly with transport remains an important unresolved question. Such coordinated assembly may differ between epithelial cells and neurons, and between different types of alpha herpesviruses, which has led to some controversy. Recent evidence suggests that the swine alpha herpesvirus, pseudorabies virus (PRV), travels inside membranated vesicles within neurons
To coordinate envelopment with transport, the virus must take advantage of cellular synthetic and transport machinery. Such co-option of transport machinery may underlie HSV1 cellular pathology, injuring cells by interfering with this normal important cellular process. Exploiting green-fluorescent-protein (GFP)-labeled HSV1 as a tool to uncover cargo motor receptors led us to discover that the cellular transmembrane glycoprotein, amyloid precursor protein (APP), is a component of isolated HSV1 intracellular viral particles, with ∼1,000 or more copies on average per particle
While APP is physically associated with viral particles isolated from infected cell cytoplasm, it was not among the proteins identified by mass spectroscopy of extracellular infectious particles
We therefore set out to determine whether APP co-localizess with virus inside cells, and whether interplay between virus and APP-containing cellular membranes enhances transport of nascent virus from peri-nuclear region to the cell surface and/or affects APP transport dynamics. Two approaches were developed to assign viral particles as either out-going or in-coming: Synchronizing infection to a narrow time window and thus limiting in-coming viral particles during viral production; and live imaging. We used a GFP-tagged virus, VP26-GFP HSV1, to visualize viral particles as they form and travel. VP26-GFP labeling has proven to be a spectacular tool to image viral capsid movements
We found that when confluent cell cultures are continuously exposed to virus, individual cells may acquire cytoplasmic viral particles from the media or through junctional contacts with other cells at any time during incubation. In such cultures individual viral particles in the cytoplasm of fixed cells cannot readily be identified as either incoming or outgoing. Since molecular motors mediating inward versus outward transport are different, the composition of incoming and outgoing viral particles must also differ. Without a reliable method to distinguish incoming from outgoing viral particles conclusions about the molecular composition of out-going particles is difficult. What is needed is a method to limit incoming particles. To do this we developed a natural yet rigorous infection protocol of sub-confluent cultures to limit in-coming virus within a short time period such that all infected cells were at a similar stage of infection. Since our goal was to image exiting viral particles, we did not aim for a uniform infection of all cells. We validated the protocol by counting VP26-GFP-labeled particles in the cytoplasm of infected cells at successive time points after various infection protocols (
(A) Diagram of the "in-coming/out-going" problem. Constitutive infections (left column), in which virus is added to the culture and allowed to remain throughout incubation, results in continuous entry of virus and mixing of in-coming and out-going viral particles in the cytoplasm. In synchronized infection (right column) viral exposure is limited by removal of the media, release of adherent virus from cell surface with acid-glycine and inactivation with human serum, resulting in few in-coming virus in the cytoplasm at later stages. Sub-confluent cultures are required to reduce viral transmission through cell-cell junctions. (B) An example of digital images used to quantify cellular locations of VP26-GFP particles. In this example, cells synchronously infected with VP26-GFP HSV1 (green) were fixed at 7 hr p.i., and counterstained with DAPI (blue). Images were collected by widefield fluorescence to detect VP26-GFP particles throughout the full thickness of the cells. Infected cell nuclei containing VP26-GFP appear turquoise at this exposure. Most of these infected cells display multiple viral particles in the cytoplasm. One cell (arrow) with no nuclear GFP fluorescence has three cytoplasmic GFP-particles (arrowheads) that therefore are in-coming virus. (C) Higher magnification of the boxed region in B in the GFP channel to show absence of diffuse nuclear fluorescence. Red arrows indicate two of the three GFP particles in this cell. See
While Campodelli-Fiume et al. reported that gD over-expression inhibits infection by measuring viral protein synthesis in cells expressing gD
In a careful, quantitative study of various protocols previously used to block continuous viral entry, we found that none eliminated incoming viral particles in infected cell cytoplasm, including viral expression of gD or the presence of inhibitory human serum in the media. Since our focus was to witness HSV1-APP interactions in out-going cytoplasmic particles through immuno-staining and dynamic imaging, we concluded that reliance on gD expression or human serum to block re-infection was inadequate. We used a combination approach to limit infection to a narrow time frame (1 hr) as detailed in
APP was visualized in mock-infected cells and in cells synchronously infected with VP26-GFP HSV1 by immuno-fluorescence (
(A) APP (red) co-localizes with TGN46 (green) in a compact tuft to one side of the nucleus (DAPI, blue) representing the trans-Golgi network in mock infected cells. (B) In cells infected with VP26-GFP HSV1 (green), APP (red) is distributed in particles throughout the cytoplasm at 7 hr p.i. Nuclei are stained with DAPI (blue). (C) Western blotting of uninfected (u) and infected (i) cells with anti-APP, anti-actin (loading control) and anti-VP5 (viral capsid) demonstrates significant increased amound of APP in infected cells. Actin bands remain similar, and VP5, as expected, is only detected in lanes loaded with infected cell lysate. Note no new APP bands are detected in the infected versus uninfected cell lysates. (D) Isolated virus separated on 7.5% SDS-PAGE and stained with amido black for protein and probed for APP by Western blotting with the same antibodies used for immuno-fluorescence. Note that only the 90–110 kD doublet representing APP is detected by anti-APP, with no additional viral protein bands detected. See also
HSV1 encodes an immunoglobulin Fc receptor, gE/gI, an envelope glycoprotein complex which binds the non-antigen-binding domain of immunoglobulin
To test whether the APP staining we observed was a result of specific antigen binding or an artifact of antibodies binding to the viral Fc receptor, gE/gI, we performed several control experiments: a) staining infected cells in parallel for APP and for a nuclear antigen, histone 3, using anti-histone rabbit antibodies of the same purity and at the same concentration as the anti-APP rabbit antibodies; b) staining cells infected with a virus deleted in gE; and, as shown later in this paper, c) knocking down APP with siRNA.
First, cells were stained in parallel for APP and for histones with rabbit primary antibodies and the same secondary antibody, all with similar purity and concentration, and imaged with the same camera settings. Viral particles were stained with anti-APP but not with anti-histone antibodies, which appropriately stained the nucleus predominantly (
Next, we infected cells with HSV1 deleted in the gE gene, gEnull. Parallel cultures were infected with the parental strain, and the gE deletion virus with the gE gene re-inserted (gE rescue)
A representative example of a cell infected with gE null virus stained for APP (red), gE (far red), and VP5 (green), the major capsid protein. (A) Low magnification wide-field image of the three channels, APP, VP5, and gE shown in color with merged channels (left), and then each individual channel, as labeled, of the same image. (B) Higher magnification of the boxed region shown in (A) and rendered in 3D. A Z-stack of 45 focal planes captured at 0.35 µm intervals was deconvolved using iterative processing and rendered in 3D. In four angles of rotation, co-localization of green VP5 capsids and red APP membranes throughout the cytoplasm of a cell infected with gEnull virus is shown. See
To characterize APP-capsid assemblies in more detail, we double-stained cells synchronously infected with VP26-GFP-HSV1 for APP and a viral envelope protein (either gE or gD) in cultures fixed at 7–9.5 hr p.i. (
Cells were synchronously infected with VP26-GFP HSV1 (green) and fixed at 7 hr p.i.. (A–D) Examples of infected cells stained for gE (blue) and APP (red). In (B) a high magnification of boxed region in (A) is shown. A white arrow indicates one of the particles displaying all three labels (gE, APP and VP26). Particles with APP only (white arrowhead) or gE and VP26 (pink arrowhead) are indicated. (F–H) Examples of infected cells stained for LAMP2 (blue) and APP (red). In (F) a high magnification of the boxed regions in (E) is shown. White arrows indicate APP and gE, pink arrowheads indicate LAMP2, and yellow arrowheads indicate APP alone. (C and G) Intensity profiles along a line (white) drawn across the merged image in (A) and (E). Arrows indicate the superposition of peaks for each channel. (D and H) Histograms showing the percentage of VP26-GFP particles in each category. VP26-GFP alone (D: 3.0±1.7% and H: 11.8±3.9%) or with APP (D: 2.0±1.5% and H: 61.3±14.7%); VP26-GFP with gE (D: 21.8±5.5%) or with LAMP2 (H: 4.6±1.6%); and VP26-GFP with both APP and gE (D: 72.8±6.7%), or both APP and LAMP2 (H: 22.3±8.7%). Experiments were performed in triplicate, and 2085 particles in 11 cells were counted in D and 1592 particles in 9 cells in H. See
A high proportion of VP26-GFP particles throughout the cytplasm co-localized with both APP and gE: 72.8±6.7% (
At higher magnification with split channels VP26-GFP particles were uniform in size and shape, whereas particles stained for APP or either of the viral glycoproteins, gD or gE, varied in shape and size (
Co-localization of viral particles with APP beyond the Golgi region was also biologically specific, since another organelle membrane protein, LAMP2, did not co-localize, except within the Golgi complex where membrane proteins are glycosylated (
First, to ensure that co-localization was not due to superimposition of separate particles from the full-thickness wide-field image, we captured images by confocal microscopy with 0.8 µm-thick optical sections (
(A) Example of a 0.8 µm optical section by confocal imaging of a cell infected with VP26-GFP HSV1 (green), fixed at 8.5 hr p.i., and stained for cellular APP (red) and viral glycoprotein, gE (blue). (B) Galleries of particles showing the co-localization of VP26-GFP with gE and APP. (C) Histogram showing the percentage of VP26-GFP particles in each category. VP26-GFP alone (3.2±2.0%), with APP (5.9±2.4%), with gE (12.4±5.6%) and with both APP and gE (78.4±5.7%). 569 particles in 5 cells were counted. (D) Thin section immunogold electron microscopy of HSV1 infected cells probed with anti-C-APP with protein-A linked 10 nm gold particles. Note single and multiple gold particles decorating membranes surrounding viral capsids in the cytoplasm. Bar = 100 nm. (E) Parallel sections from the same EM block treated with an irrelevant rabbit antibody of similar purity and dilution and probed with protein-A gold. Note the absence of gold labeling of viral particles. Also see
Co-localization of APP with viral particles was also detected at the ultrastructural level by immuno-gold thin section electron-microscopy (
Thus, membranes containing cellular APP are physically associated with membranated cytoplasmic HSV1 at the ultrastructural level.
Anti-APP stained centrally located viral particles in both wildtype and gEnull HSV1-infected cells which demonstrated that this staining pattern was not a consequence of antibodies binding non-specifically to the viral Fc receptor, gE (
To distinguish between these possibilities we knocked down APP expression using siRNA. If viral particles expressing gE do not stain for anti-APP after APP knock-down, this would demonstrate that anti-APP label is not due to the viral Fc receptor, and suggest that gE may mediate retention of APP-containing membranes to emerging viral particles during their maturation and transit to the cell periphery.
First, we confirmed knock-down by Western blotting in mock- and HSV1-infected cells, comparing no siRNA, non-silencing siRNA
HSV1 infected cells were transfected in parallel with either vehicle alone (None), non-silencing RNA (Ctrl) or siRNA against APP (APP). After 48 hr cells were scraped into lysis buffer, and loaded in parallel on a 10% gel for electrophoresis followed by transfer to nitrocellulose. The blot was divided in two horizontally, the top half probed for APP and the lower half for actin, a loading control. Non-silencing siRNA has little effect, while siRNA for APP decreases APP band intensity almost entirely, with no significant effect on actin.
The immunofluorescence pattern of APP staining of VP26-GFP HSV1 infected cells after transfection with non-silencing siRNA was similar to untreated cells (
(A) Representative example of a cell treated with non-silencing control siRNA, infected with VP26-GFP HSV1 (green), fixed at 7 hr p.i. and stained for APP (red), gE (blue). APP staining is bright and diffuse, and co-localizes with VP26-GFP viral particles (arrows). (B) Representative example of a cell treated in parallel to the cell shown in (A) but with siRNA for APP. Note the absence of most APP staining, while gE staining of viral particles remains strong.
The results described above demonstrate a high frequency of interactions between intracellular APP and outgoing VP26-GFP-labeled viral particles. To discover whether interplay with APP makes a functional contribution to viral transport, we performed live confocal imaging of cells expressing mono-red-fluorescent-protein-labeled APP (APP-mRFP) infected with VP26-GFP-HSV1. Direct visualization of dynamic interactions does not require fixation and antibody staining, and also provides dynamic imaging of double-label viral particles. First, as a control, we confirmed that the mRFP signal represented APP by counter-staining fixed transfected/HSV1-infected cells with anti-APP antibodies. The majority (>95%) of mRFP particles also stained for APP by immunofluorescence (not shown), demonstrating that mRFP signal represents the APP-mRFP fusion protein.
VP26-GFP particles were co-localized with APP-mRFP and the two labels moved together through the cytoplasm (
(A) The first frame of a video sequence captured at 7–9 hr p.i. captured at 3-sec intervals for 900 sec (15 min). Many (64%) VP26-GFP particles (green) co-localize with APP-mRFP compartments (red) in this frame appearing bright yellow (arrows). VP26-GFP particles travel with APP-mRFP vesicles, and sometimes join and separate from APP. White lines show the boundaries between cells and position of the nuclei (N). Also see
The tracks followed by APP-VP26-GFP assemblies varied (
A functional link between APP-compartments and HSV1 became obvious when comparing the dynamics of VP26-GFP-particles with and without APP in cells expressing low levels of APP-mRFP. While velocities of VP26-GFP particles that moved were similar (
While VP26-GFP particles traveled more frequently when associated with APP-mRFP, the opposite was the case for APP-mRFP particles, for which virus association decreased both velocity and propensity to move. Small APP-mRFP particles in uninfected cells moved at 1.1+/−0.2 µm/sec instantaneous velocity, while APP-mRFP particles in infected cells with or without viral particles moved more slowly (0.3+/−0.1 µm/sec) (see
The velocity of APP/VP26 particles in the x-y plane suggests kinesin-driven transport, with the observed maximum velocity of 2.0 µm/s and average instantaneous velocity of 0.4±0.1 µm/s (n = 118 particles, from 20 cells in independent movies) (
When moving, instantaneous velocities of individual particles could be sustained (
(A) A double labeled GFP-mRFP particle moves away from the nucleus (arrow, from
Live imaging of cells double-transfected with VP26-GFP and APP-mRFP and then infected with the gEnull virus demonstrated the same pattern seen by immuno-fluorescence (
We also tested the specificity of APP-capsid interactions by probing for another Golgi-enriched cellular protein, TGN46
As for APP, TGN46 also moved from its normal peri-nuclear location to become distributed throughout the cytoplasm by 7 hr p.i. (
(A) Uninfected cells stained for TGN46 (red) show compact tufts on one side of the DAPI-stained nuclei (blue). (B) Cells synchronously infected with VP26-GFP HSV1 (green), fixed and stained for TGN46 (red) and DAPI for nuclei (blue). (C) Higher magnification of an example of cell infected with VP26-GFP-HSV1 (green) fixed at 7 h p.i. and stained for TGN46 (blue) and APP (red). (D) Co-localization in the peri-nuclear region shown at high magnification of boxed region "D" in cell shown in (C). Arrows indicate co-localization of VP26-GFP particles with APP in a TGN46-stained compartment. White arrowhead indicates a VP26-GFP particle apparently on the surface of a TGN46-stained vesicle. Yellow and cyan arrowheads indicate examples of single anti-APP-stained and both APP and TGN46 stained particles, respectively. (E) Loss of co-localization at the periphery is shown at high magnification of boxed region E at the periphery of the cell shown in (C). Pink and white arrows indicate co-localization of VP26-GFP HSV1 with APP alone at the periphery and in the cytoplasm close to the periphery, respectively. Yellow and cyan arrowheads indicate examples of single APP and both APP and TGN46 labels, respectively. (F) Linescan intensity profile of a region in the intermediate cytoplasm as seen in (C) shows both coincident (arrows) and non-coincident (arrowheads) peaks of TGN46 staining (blue line) with APP (red) and VP26-GFP particles (green). (G) Histogram of particles in the peri-nuclear region showing the percentage of VP26-GFP particles that co-localized with APP and TGN46. The majority of VP26-GFP particles co-localized with APP and TGN46 (51.8±9.5%) and fewer co-localized only with TGN46 without APP (11.6±3.1%). (H) Histogram of particles in the periphery showing the percentage of VP26-GFP particles that co-localized with APP and TGN46. None co-localized with TGN46 alone, although 25.5±14.2% were co-localized with both APP and TGN46. Note that many fewer particles co-localized with TGN46 in the periphery than in the peri-nuclear region, suggesting that membrane compartments co-localized with viral products retain the ability to sort their components. N = 10 cells, 3,782 particles from three experiments.
In sharp contrast, in the peripheral cytoplasm many VP26-GFP particles co-localized with APP but very few also had TGN46 staining (
At slightly later times (9 hr p.i.) the large triple-labeled clusters in the Golgi region seen at 7 hr p.i. were less pronounced, and some appeared to have drifted outwards to the intermediate cytoplasm. This change may reflect the fragmentation of the Golgi that occurs in HSV1-infected Vero cells
By synchronizing infection and performing live imaging of VP26-GFP-labeled capsids in cells expressing fluorescently labeled APP, we report results that provide new insights into the interaction between viral and host cell components during transit of virus to the surface. Functional relevance to the virus of interplay with APP, an identified motor receptor, is clear: association of viral capsids with membrane systems containing APP confers a propensity to move through the cytoplasm at fast-anterograde transport rates. Riding on or entering cellular membrane systems that normally traffic to the surface would thus promote efficient viral egress. By coordinating secondary envelopment with acquisition of the cellular vesicular transport machinery, the virus would be assured of reaching the cell surface in an infectious form.
In summary, the evidence includes: (1) VP26-GFP-labeled particles in the cytoplasm are frequently found together with both viral glycoproteins and cellular APP; (2) VP26-GFP-labeled particles travel together with APP for long, rapid trajectories, and GFP particles lacking APP move less often; (3) Co-localization with APP is specific, since GFP-labeled capsids and viral glycoproteins were less frequently found with other cellular organelle membrane proteins, LAMP2 and TGN46; (4) gEnull particles stain for APP, and siRNA knock-down of APP abolishes staining; and (5) VP26-GFP particles sustain co-localization with APP-mRFP throughout transport, while by immuno-fluorescence co-localization of VP26-GFP with TGN46 is lost. Thus sorting of cargo is preserved in HSV1 infected cells at least at these time points after infection, and VP26-GFP-labeled particles appear to interact with a select APP-containing Golgi-derived membrane compartment.
Interaction of viral particles with APP-containing membranes is not without functional consequence: APP-mRFP particles travel more slowly in infected than in un-infected cells, even without detectible viral cargo, and APP is mis-localized in HSV1-infected cells. Such mis-localization could contribute to increased APP proteolysis with HSV1 infection
Collectively, our results provide new information about dynamic interactions between nascent viral particles and cellular membranes, the molecular composition of virus during outbound transport, and suggest how secondary envelopment and transport to the surface may be coordinated. Such dynamic interactions between APP and HSV1 suggest a mechanistic basis for the observed clinical relationship between HSV1 seropositivity and risk of Alzheimer's disease
By synchronizing infection we minimized viral re-entry during replication, ensuring that at later time points during productive infections, >90% of cytoplasmic viral particles are progeny undergoing outbound transport. Other methods to synchronize infections include the use of mutant viruses, temperature shifts, and pharmacologic interference with synthetic pathways
Models for the location of envelopment and the molecular composition of viral particles transporting to the cell surface are quite varied and a topic of considerable debate. Two possibilities for capsid transport are proposed: independent or membrane-associated (see diagram,
A cartoon showing various types of interactions between cellular APP and VP26-GFP labeled viral particles documented here. (A) In the peri-nuclear region, VP26-GFP particles dance around and within large peri-nuclear compartments co-localizedd with viral envelope proteins, gE and gD, and cellular membrane proteins, LAMP2, TGN46 and APP. LAMP2 compartments separate from this apparent Golgi network early and rarely co-localize with viral components at the periphery. Some membrane systems with VP26-GFP also label for both APP and TGN46, primarily near the nucleus at the time points studied here. (B) TGN46 particles separate from VP26-GFP labeled viral components farther towards the periphery, while the APP particles remain with VP26-GFP particles and with viral envelope glycoproteins, gE and gD, en route towards the cell surface. (C) VP26-GFP particles may enter smaller post-Golgi APP-staining particles that undergo directed transport. (D) Some VP26-GFP particles remain separate from APP after leaving the nucleus. These may be inside unlabelled membrane systems or be free in the cytoplasm, some have the capacity to transport without APP. (E) VP26-GFP particles may ride on the cytoplasmic surface of APP-labeled membrane systems, come on or off these membranes, or bud into them. Any particular viral particle may employ all of these mechanisms during transit in the cytoplasm. In each case, we hypothesize that microtubule motors, such as kinesin, are recruited, possibly via APP or another cellular motor receptor.
In our videos it sometimes also appears that the VP26-GFP particle is riding on the cytoplasmic surface of cellular membranes that are undergoing transport. These ideas as well as other possible transport configurations are diagrammed in
Budding into dynamic vesicles may be critical for nascent viral particles to acquire anterograde motors, such as the kinesins, for transport to the surface. When co-localized with an APP membrane, VP26-GFP particles moved significantly more frequently and at higher velocities. Thus our live imaging of cellular APP and viral capsid reveals a functional link between them: APP-containing membrane alliance confers efficient motility to the virus.
Evidence that the other model of non-membrane-associated transport occurs is also presented here (see
Reorganization of microtubules in HSV1-infected cells must play a large role in these two mechanisms of viral transport. We show here that the microtubule organizing center (MTOC) is lost in infected Vero cells, and microtubules project from the full circumference of the nucleus as has been previous described
At later stages, the large triple-labeled clusters of capsid-envelope-APP, initially found in the peri-nuclear region, became less pronounced and some clusters appear to move outwards to the intermediate cytoplasm, probably a consequence of microtubule reorganization. Thus as productive infection proceeds, capsids must travel farther after exiting the nucleus to reach membrane compartments for envelopment. This observation indirectly supports a necessary role for membrane-free capsid transport from the nucleus to cortical membrane compartments for envelopment. Reorganization of the microtubules may allow retrograde motors to carry nascent viral capsid-tegument assemblies from the perinucleay area to the cortical Golgi at late time points of productive infections in epithelial cells.
Our data show APP-containing membranes travel together with intracellular HSV1 particles, and contrast with a failure of mass spectroscopy to detect APP in preparations of extracellular HSV1 virions
Synthesizing our results with the work of others, we propose various models for coordination of capsid envelopment and transport to the cell surface after emergence from the nucleus (
Emerging
Once inside the neuron, HSV1 enters a latent stage that persists for the lifetime of the host. Decline in immunity associated with aging {Arlt, 2004 #292;Kurz, 2004 #291} increases chronic reactivation of herpesvirus
VP26-GFP HSV1 was a gift from Dr. Prashant Desai (Johns Hopkins University)
Virus was isolated as described
The number of viral genomes was determined in parallel by quantitative PCR both with and without DNAse treatment, using the gB primers as described
Quantification was determined by comparison of the DNA from our viral prep to known concentrations of genomes obtained commercially for clinical studies, HSV1 Type 1 DNA Macintyre strain (Advanced Biotechnologies Inc.) run in parallel. Viral preps and these standards with and without DNAse treatment are mixed with 12.5 µl SYBR Green Master Mix (Applied Biosystems Inc), 25 mM of each forward and reverse sense primer and 5 µl of diH2O. Primers were selected from the HSV1 glycoprotein B, an essential gene. The forward sense primer is 5′-CCA CGA GAC CGA CAT GGA GC-3′ and reverse primer 5′-GTG CTY GGT GTG CGA CCC CTC-3′
To determine plaque-forming units, viral preparations were titered in triplicate by infecting confluent Vero cells under synchronizing conditions. Vero cells were plated at 0.5×106 cells per 6-well plate and incubated overnight at 1.5×106 confluent cells per well after division. Gradient-purified virus was serially diluted in growth media (1/100 to 1/10−8 viral preparation/media). Each viral dilution (0.5 ml) was inoculated into three wells of confluent cells on ice. After an hour cultures were warmed to 37°C and incubated another hour and then rinsed in acid-glycine according to the synchronization protocol (below). To limit subsequent infection through the media, cultures were overlaid with 0.5% agarose in growth media plus 1% human serum (Gibco/invitrogen) after the acid-glycine rinse. Thus virus has 1 hr to enter cells. Cultures were incubated at 37°C for 2–3 days when plaques become distinctly visible.
A pMonoRed-APP695 plasmid encoding MonoRed (mRFP)-tagged 695-aa human APP (APP695) was generated from a backbone plasmid, pEGFP-N3/VP16
To synchronize viral infection, subconfluent cultures were first chilled on ice for 30 min-1 hr and then inoculated with VP26-GFP HSV1 virus, typically diluted to a viral concentration of 10 plaque-forming units per cell, i.e.10 MOI. We used subconfluent monolayers because our live video experiments had revealed viral particles passing through junctions between cells in confluent cultures (Bearer and Ferland, MS in preparation). Because our goal was to create culture conditions that limited in-coming virus without resorting to genetic manipulation of the virus or pharmacologic intervention, we used sub-confluent monolayers to reduce such junctional transfer of virus between cells, and short (1 hr) infection times. Even at 10 MOI with a 2.5 genome/PFU ratio, this did not result in infection of all cells. We accepted this lower infection efficiency in favor of better assurance that the possibility of in-coming virus at later time points was eliminated to the extent possible. Even though the virus was titered under synchronizing conditions, that titering requires confluent monolayers, where each virion has a better chance of encountering a cell. In subconfluent cultures, with cells not touching each other, virions also fall between cells necessitating higher viral MOIs for efficient infection rates.
After inoculation, cells were incubated on ice for another 30 min-1 hr to allow virus to adhere, and then warmed to 37°C to allow viral entry. After 1 hr infected cells were washed with acid glycine for 3 min at RT (0.14 M NaCl, 5 mM KCl, 1 mM MgCl2-6H2O, 0.7 mM CaCl-2 H2O, 0.1 M glycine, [pH 3])
Cells on coverslips were washed in serum-free media and fixed for 15 min in 4% paraformaldehyde in PBS. For microtubule staining, cells were fixed in −20°C methanol for 5 min. The following antibodies were used: rabbit anti-peptide antibody against APP C-terminus aa676–695 of APP695 (1∶2000, protein A purified IgG, 14 mg/ml Sigma A8717) or N-terminus aa46–60 of APP695 (1∶500, protein A purified IgG, 8.1 mg/ml, Sigma, A8967), amino acids); anti–APP-C aa673–695 of APP 695 (1∶1000, 0.25 mg/ml, affinity purified on the peptide, Zymed/Invitrogen 36–6900); mouse anti-VP5 antibody (IgG2b, 1∶1000, 3B6, EastCoast Bio); mouse anti-gE antibody (IgG2a at 1∶1000, East Coast Bio); mouse anti-gD antibody (IgG2a,1∶1000, East Coast Bio); mouse anti-LAMP2 (IgG1, 1∶200, Abcam); mouse anti-tubulin (IgG1, 1∶400, Sigma); sheep anti-TGN46 (1∶1000, AbD Serotec); rabbit anti-histone H3 (1∶200, protein A purified IgG, 0.9 mg/ml goat anti-rabbit IgG (Upstate Biologicals-Millipore); mouse anti-actin (1∶1000, Amersham/GE Healthcare) goat anti-rabbit Alexa-Fluor 555 (1∶2000, Molecular Probes); donkey anti-rabbit IgG Alexa-Fluor 555, donkey anti-sheep IgG Alexa-Fluor 647 and goat anti-mouse IgG Alexa-Fluor 647 (1∶2000, Molecular Probes); goat anti-mouse Cy5 (IgG2b, 1∶500, Jackson Labs); goat anti-mouse FITC (IgG2a, 1∶2000, Jackson Labs).
To minimize non-specific binding of antibodies from rabbit and sheep to viral Fc receptor, gE/gI
Fluorescence images were captured either by a 63X/1.4 N.A. oil immersion Plan Apochrome objective on a Zeiss Axioscope Z1 using the MRM AxioCam and AxioVision 4.5 software, or on a Zeiss LSM 410 confocal laser scanning microscope equipped with a krypton-argon laser for excitation at 488, 568, and 647 nm running Phoenix, v,2,0,2524 software. (Microcosm, Inc.). Confocal images of fixed cells used a pinhole adjusted to a narrow optical section (0.8 µm). Capture time was set by imaging coverslips stained in parallel with secondary only, and using linear grayscale for each channel. Figures were created using Photoshop CS2 and 3 (Adobe). For widefield deconvolution a z-stack was collected at 0.375 nm stepsize with 45 focal planes using th 63x objective. A region of interest was selected and deconvolved using AxioVision 4.5 DCI program using the automated PSF and interative processing. Deconvolved stacks were projected into 3D and rotated using AxioVision 4D Rendering software (Zeiss.com).
After 9 hr, synchronously infected or mock-infected cell cultures (100 mm petri dish) were washed in warm serum-free media, and then scraped into lysis buffer (50 mM NaCl, 30 mM Tris-HCl (pH 7.2), 5 mM EDTA, 50 mM NaF, 2 mM sodium vandadate, 1 mM PMSF, 5 mM p-nitrophenylphosphate, 1% Nonidet-p-40), protein concentration measured by bicinchoninic acid kit (Sigma Aldritch), and volume adjusted to give equal protein concentrations across all samples which were then aliquoted into gel sample buffer and boiled for 5 min
Quantitative co-localization analysis was performed on raw data using AxioVision 4.5 software in two ways: (1) Scatter plot, a graphical display that compares total pixel coincidence between two channels across the entire image. The intensities of two channels are distributed along the x- and y-axis in a scatter plot. If intensity of images in each channel completely overlaps, then the plot displays a straight diagonal lin,e starting from the origin of the scatter plot. This computational approach provides an average pixel-coincidence between two channels of the same field. (2) Linescan showsw a graph of the pixel intensity in each channel versus its position along a straight line drawn across a merged image. A superimposition of peaks between channels indicates high intensity overlap per pixel along the line. Linescans detect overlap along a line in a region of interest, while scatterplots can measure the global degree of co-incident intensities across a whole image field.
The number of fluorescent particles was counted in randomly selected images of synchronously infected cells by an independent person. All results are presented as mean ± standard error of the mean (SEM). Histograms were made from spreadsheets of counts using Microsoft Office Excel.
Synchronously HSV1-infected Vero cell cultures were washed in warm serum-free media and fixed in 4% formaldehyde in 100 mM phosphate buffer overnight. Fixed cells were rinsed in PBS containing 0.15% glycine, and then scraped into PBS containing 5% bovine serum albumin using custom-made Teflon scrapers prepared from Teflon sheets. Scrapings were pelleted in an Eppendorf benchtop centrifuge, resuspended in warm PBS with 10% gelatin and 0.01% blue dextran (Sigma), re-pelleted, and cooled to 4°C to solidify the gelatin. The tips of the tubes containing a visibly blue pellet of cells were cut off, the cell pellets scooped out and post-fixed in buffered 4% formaldehyde for 30 min., cut into 0.1–2 mm cubes and soaked overnight in 2.3 M sucrose at 4°C. The next day, cell pellets were mounted onto metal specimen pins (Leica Microsystems Inc, Deerfield, IL), frozen in liquid nitrogen and placed in 100% methanol containing 1% uranyl acetate (SPI Inc) cooled to −80°C in an AFS2 Freeze Substitution Device (Leica Microsystems, Inc).
The AFS2 was programmed to warm to −60°C over 16 hr after which time the methanol/uranyl acetate was replaced with pre-cooled absolute ethanol. The specimen pins were removed from the solvent leaving the cell pellets to fall to the bottom of the tubes and the cell pellets were washed six times in fresh, cold ethanol over 6 hr. The cell pellets were then gradually warmed to −50°C and infiltrated with increasing amounts of Lowicryl HM20 resin (Electron Microscopy Sciences, Hatfield, PA) dissolved in ethanol and left overnight in a 1∶3 mixture of ethanol and resin. The next day, the cells were soaked in 3 changes of 100% Lowicryl resin and then placed in gelatin embedding capsules for polymerization. The cell pellets were covered with fresh 100% resin, labeled and polymerized under UV light at −50°C. After 5 days, the specimen blocks were gradually warmed to RT overnight.
Polymerized blocks, which were pink in color, were removed from the embedding capsules and placed overnight under UV illumination (in a laminar flow cabinet) at RT to allow unpolymerized volatile resin components to escape. Cells were sectioned with an Ultracut S ultramicrotome (Leica Microsystems Inc) equipped with a diamond knife (Diatome USA Inc). Sections were mounted on Formvar/carbon-coated metal specimen grids. After blocking with 20 µg/ml human IgG in PBS, sections were fixed with 1% gluteraldehdye for 10 min, quenched in 1% glycine in PBS for 10 min, blocked in PBS containing 10% BSA and 0.5% fish skin gelatin (blocker; Sigma), and then labeled with rabbit anti-cAPP antibodies (Sigma) or with irrelevant rabbit antibodies (rabbit anti-mouse IgG, Invitrogen) diluted 1∶20 in blocker, followed by 10 nm protein A-gold also diluted in blocker (University of Utrecht, The Netherlands). Omitting the human IgG block had no effect on the level of labeling. After immunogold staining, sections were contrasted with uranyl acetate and lead citrate
Three formulations of siRNA were purchased from Qiagen (Valencia, CA, USA): non-silencing siRNA sequence, shown by Basic Local Alignment Search Tool (BLAST) search not to share sequence homology with any known human mRNA; 2) specific siRNA against an unrelated gene product, MAPK-1, (Qiagen SI00300755); and specific siRNA against the APP target sequence
Vero cells were grown on chambered glass coverslips (Lab-Tek, Nalge Nunc International) and transfected with the pMonoRed-APP695 plasmid, or with both pMonoRed-APP695 and pK26GFP for cells infected with the gEnull, gE wildtype (NS) or gE rescue virus, using 1 µg plasmid DNA/chamber containing 40–80,000 cells using lipofectamine2000/Optimem standard protocols (Invitrogen). After 48 hr incubation to allow for fusion protein expression, cells were synchronously infected with VP26-GFP HSV1 at 1–10 PFU/cell, and either fixed or imaged at 5–10 hr p.i.. Only cells with low levels of mRFP expression as determined by weak fluorescent signal, no aggregation of label in nuclear envelope, and clearly visible small, rapidly motile mRFP particles, as previously described for YFP-APP transfections
Dynamic interactions between APP-mRFP particles and VP26-GFP-labeled capsids or viral particles were recorded under a 63X/1.4 N.A. oil objective, and time-lapse image sequences were collected at 3-second intervals simultaneously in FITC- and Cy3-fluorescent channels as well as a transmitted light channel, with 3 photo multiplier tubes using a Zeiss 410 confocal laser scanning microscope as described above. During observation, cells were maintained at 37°C through a temperature-controlled sample chamber and ring (Tempcontrol 37-2 digital, Carl Zeiss, Inc.). Movies from resulting time-lapse series were produced using NIH ImageJ (NIH).
Cells with low to moderate expression of APP-mRFP were selected for long-term imaging. Expression levels were determined by the intensity of the image after scanning at a fixed pinhole (40), gain (Detector gain: 80–120, Amplifier gain: 1–6) and Detector offset (3990). Our rationale was to avoid high levels of expression that could alter normal APP distribution. The large pinhole size gave an optical section of 12 µm that captured the full cell thickness. Most particled moved perpendicular to the optical section and remained within the filed of view. A few particles released to the apical (dorsal) cell surface remain visible until they separate from the cell and move out into the media.
Instantaneous velocity was measured from confocal sequences of synchronously infected cells using MetaMorph (Molecular Devices, Inc, Sunnyvale CA, USA). Videos of cells with active movements were selected for analysis. All mobile VP26-GFP and/or APP particles were measured inside each cell selected. Briefly, the distance moved between two consecutive frames was measured by marking the center of a fluorescent particle in each consecutive frame with a mouse-driven cursor. The distance in pixels was converted to real measurements in microns based on calibration (typically, 1 pixel represented 0.14 mm). To evaluate the accuracy of cursor marking and stage drift
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(PDF)
(MOV)
(MOV)
(MOV)
(MOV)
(MOV)
(MOV)
We are indebted to many former and current Brown University undergraduates for their participation in many aspects of this paper: Samuel Uhl, Aleksey Novikov, Arkady Rasin, Kristen Sylvester, Alison Glasgow and Crystal Yu, as well as graduate student Anda Chirila. We also wish to thank Bryan Kinney, Kelly Cleveland and Kathleen Kilpatrick for their technical contributions and excellent comments.