Download Cytomegalovirus replication in human retinal pigment

Cytomegalovirus Replication in Human Retinal Pigment
Epithelial Cells
Altered Expression of Viral Early Proteins
Barbara Detrick* Jean Rhame,* Yun Wang,~f Chandrasekharam N. Nagineni,~f
and John J Hooks f
Purpose. Cytomegalovirus (CMV) infections are frequent complications in patients who have
undergone kidney and bone marrow transplant and in patients with acquired immune deficiency syndrome. The mechanism by which CMV is activated and replicated within the retina
is unknown. The authors evaluated the ability of human CMV to initiate replication in human
retinal pigment epithelial (RPE) cells and compared this system with CMV replication in
human fibroblasts (HEL-299, MRC-5) and human amnion epithelial (WISH) cells.
Methods. Human RPE cells were obtained from donor eyes and propagated in vitro. Cells
were infected, and CMV replication was evaluated in three ways: the detection of viral antigen
by immunofluorescent, flow cytometry, and Western blot assays; the detection of virus-induced
cytopathic effect (cpe), and the detection of infectious virus.
Results. No evidence of viral replication in the epithelial (WISH) cells was found. Although
CMV does not usually replicate in vitro in epithelial cells, CMV replication was detected in RPE
cells. There are a number of distinct differences in CMV replication in RPE cells compared to
replication in human fibroblasts. Virus-induced cpe and the production of infectious virus by
RPE cells were delayed when compared to virus infection in either HEL or MRC 5 cells. At
a multiplicity of infection of 0.1 and 1, cpe and infectious virus yield reached maximum levels
at days 4 to 5 in fibroblasts and at days 19 to 46 in RPE cells, respectively. Nevertheless,
infectious virus produced by RPE cells (10b5 TCID50/0.1 ml) significantly surpassed levels
produced by HEL cells (1055 TCID50/0.1 ml). The permissive infection in RPE cells consisted
of a prolonged period (5 to 6 days) of virus production in the absence of cytopathology.
Virus protein expression evaluated by indirect immunofluorescence assays, Western blot analysis, and flow cytometry revealed a delay in viral protein expression in RPE cells compared to
viral protein expression infibroblasts.The pattern of viral protein evaluated by flow cytometry
was noticeably different in the two cell types. At the middle phase of CMV replication in RPE
cells, a low percentage of cells express immediate early (IE) protein at a time when a high
percentage of the cells express early (E) proteins. This IE-1 protein is a stable protein found
concurrently with E protein in fibroblasts. This difference in percentage of cells expressing
specific CMV proteins is transient, that is, it does not remain apparent at 100% cpe.
Conclusions. Retinal pigment epithelial cells appear to demonstrate a distinct pattern of CMV
infection. The low frequency of expression of IE viral protein in RPE cells, the subsequent
slow replication of CMV, and the altered expression of IE viral proteins may be critical
variables that impact on their relationship to viral persistence and activation within the retina.
Alterations in the IE gene product may indicate the existence of positive or negative nuclear
transcription factors within infected RPE cells. Invest Ophthalmol Vis Sci. 1996; 37:814-825.
From the * Department of Pathology, The George Washington University Medical
Center, Washington, DC, and t The Immunology & Virology Section, laboratory of
Immunology, National Eye Institute, National Institutes of Health, Bethesda,
Maryland.
Presented in part at the 1993 and 1994 annual meetings of the Association for
Research in Vision and Of/hllialmology, Sarasola, Florida.
Submitted for publication February 10, 1995; revised November 16, 1995; accepted
Deceml>er 19, 1995.
Proprietary interest category: N.
Reprint requests: Barbara Detrick, The George Washington University Medical
Center, Ross Hall, Room 502, 2300 Eye Street NW, Washington, DC 20037.
814
.Human CMV is a herpesvirus that is a major cause
of blindness in children born with congenital infections and in immunocompromised persons.1'2 During
the past decade, the number of persons with compromised immune systems is rapidly increasing because
of immunosuppressive therapy in cancer and transplantation and to complications of human immunodeficiency virus infection. Moreover, vision loss caused
Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
Copyright © Association for Research in Vision and Ophthalmology
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
CMV Replication in Human RPE Cells
by CMV retinitis is a rapidly expanding problem despite the use of newer treatment modalities.3 The natural course of CMV retinitis is incompletely understood. A basic understanding of CMV replication processes within selected retinal cells may provide a
rationale for alternative treatment modalities.
Cytomegalovirus is a DNA virus whose replication
is regulated sequentially and is dependent on viral
and cellular factors.4 In permissive infection, there are
three kinetic phases of gene expression: immediate
early (IE), early (E), and late (L) genes. 1-4~6 Immediately after viral infection, the CMV IE genes code for
IE-1 and IE-2 proteins. These genes depend on host
factors for their expression, which occur in the absence of viral DNA replication. E gene expression is
dependent on IE gene expression. E genes encode
for enzymes required for viral DNA synthesis. After
viral DNA synthesis is initiated, L genes are expressed
and encode for the major part of the virion. In nonpermissive infection, these kinetic phases of viral gene
expression are altered. Usually, no viral proteins are
expressed or only IE proteins are detected.7 Moreover,
the initial phase of IE gene activation is under the
control of the major IE promoter (MIEP), whose activation is dependent on host cell transcription factors.8"' IE-1 and IE-2 proteins themselves regulate homologous and heterologous promoters.10" The restricted nature of CMV replication can thus be
controlled by both viral and cellular factors.
Limited cellular distribution of virus in selected
tissue is a hallmark of CMV infection. Histopathologic
evaluation, viral isolation, and, more recently, in situ
hybridization have identified that CMV may be detected more frequently in selected tissue cell types.
For example, CMV is found in ductal epithelium of
salivary glands, in mucosal epithelium and submucosal
endothelial cells of the intestine, in bile duct epithelium of the liver, and in alveolar and bronchial epithelium of the lungs.1213 Within the retina, CMV also has
been observed within the retinal pigment epithelial
cells and within the neural retina.1415
The RPE cell is a potent regulatory cell within the
retina. It is known to participate in the transport of
vitamin A and nutrients to the neural retina, it phagocytizes the outer segments shed by photoreceptor
cells, it adsorbs light, and it provides adhesive properties for the retina."' This cell also appears to be particularly important in infectious and/or immunologic
conditions within the eye.17 First, intracellular infectious agents, including viruses and parasites, can infect
and replicate within this cell.18"20 Second, bacterial
products, such as lipopolysaccharides and cytokines,
can bind to and activate RPE cells.21'25 Third, interferon-y-activated RPE cells can express major histocompatibility class I and II molecules and the adhesion
molecule, ICAM-1. Furthermore, these activated RPE
cells can function as antigen-processing and antigen-
815
presenting cells within the retina.23 Finally, the infected or cytokine-activated cell can produce and release a variety of cytokines that may alter the ocular
microenvironment.2b~28 The effects of CMV infection
on the varied physiologic and immunologic actions of
RPE cells is unknown.
It is difficult to study CMV latency in humans.
Currently, cell culture models of CMV replication and
latency have been used to uncover clues about CMVhost interactions. Human fibroblasts often have
served as the traditional cell line; however, in vivo
these cells usually are not infected with CMV. Investigations using other cell models are needed to reflect
more accurately this in vivo process. Because CMV
replicates within human RPE cells, this pattern of virus
replication may reflect replication more accurately
within specialized cells and may provide novel ways to
evaluate factors that control CMV gene expression and
replication. In this article, we show that even though
CMV usually does not replicate in epithelial cells in
vitro, it can replicate within human RPE cells. However, there are a number of distinct differences in
virus replication in RPE cells compared to replication
in fibroblasts. Virus replication in RPE cells is atypically slow, particularly when compared to CMV replication in human fibroblasts. In addition, the permissive infection in RPE cells consists of a prolonged period (5 to 6 days) of virus production in the absence
of cytopathology. Characterization of viral protein expression indicates that there is an altered pattern of
CMV IE protein expression in RPE cells in comparison
to fibroblasts. These distinct changes may be important findings associated with CMV infection in resident ocular cells.
MATERIALS AND METHODS
Cells
Human RPE cells were obtained and propagated
as described previously.26 When subcultured cells
reached confluency, they typically were hexagonal and
formed monolayers with clear intercellular boundaries characteristic of epithelial cells. Two primary cell
lines of human RPE cells between passages 5 and 10
were used in this study. When different passages of
these cell lines were tested for the presence of cytokeratin, 100% of the cells reacted positively with monoclonal antibodies (mAb) to cytokeratin. This reactivity
supports the presence of epithelial cells. Immunoblotting analysis further confirmed the presence of cytokeratin and indicated that cytokeratin 18 (42 kDa) is
the predominant form. At the time of first passage,
RPE cells reacted positively with mAb developed
against RPE. However, continuous culture of these
cells results in the loss of reactivity with this RPE-specific mAb. As previously described, this reactivity is
completely lost after the first passage of the cells.26
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
816
Investigative Ophthalmology 8c Visual Science, April 1996, Vol. 37, No. 5
HEL-299 and MRC-5 are human fibroblast cell
lines obtained from American Type Culture Collection (CCL137, CCL171; ATCC, Rockville, MD), and
WISH cells are a human amnion epithelial cell line
obtained from ATCC (CCL25). All cells were maintained in minimum essential medium supplemented
with antibiotic-antimycotic (Xl) mixture (Gibco BRL,
Grand Island, NY) and 2% to 5% fetal bovine serum.
Human Research
The tenets of the Declaration of Helsinki were followed, and informed consent was obtained from all
participants. Approval was obtained from the Office
of Human Subjects Research.
Viruses
Cytomegalovirus, AD 169 strain, was used in this study.
For some studies, a clinical isolate of CMV was used.
The virus was isolated from peripheral blood lymphocytes of a patient who underwent bone marrow transplantation at The George Washington University Medical Center. The virus was passaged twice in human
fibroblasts. Virus stock was prepared by propagation
in HEL cells. Infected cultures were harvested by freezing and thawing one time, followed by centrifugation
for 20 minutes at 2000 rpm. Supernatant fluids were
used as virus inoculum. Virus infectivity assays were
performed on HEL cells propagated in 96-well microtiter plates. Infectivity was recorded as the induction
of cytopathic effect (cpe) by serial 10-fold dilutions of
the sample.
Antibodies
Murine monoclonal antibodies directed against human CMV immediate IE, E, and L antigen were obtained (Accurate Chemical 8c Scientific, Westbury,
NY). The MAS127 clone 13 is an immunoglobulin
(IgG) 1 antibody that reacts with IE nonstructural antigen of 72 kDa (IE-1). The MAS 550 clone 2A2 is an
IgGl antibody that reacts with "early" antigen present
16 hours after infection. The MAS 552 clone SL-20 is
an IgG3 antibody that reacts with "late" antigen present 72 hours after infection. Monoclonal antibodies
(IgGl and IgG3) were used as isotypic controls.
Indirect Immunofluorescent Assay
Retinal pigment epithelial, HEL, and WISH cells were
propagated on chamber slides. Cells were incubated
with CMV at an input multiplicity of 0.5 to 1 for 2
hours, washed three times in PBS, re-fed with maintenance media, and incubated at 37°C. Cells were then
incubated at 37°C and at various times after infection,
removed, and processed for immunofluorescent
assays. Slides containing cells were air dried and fixed
in acetone for 4 minutes and washed three times in
PBS. Primary antibodies were applied for 30 minutes
in a humidified chamber. Slides were washed three
times in PBS, followed by incubation with fluoresceinconjugated horse anti-mouse IgG for 30 minutes.
Slides were washed three times again, mounted, and
viewed under a fluorescent microscope.
For indirect immunofluorescence studies, slides
were reviewed in a coded fashion by three investigators. The following samples were evaluated in the immunofluorescent assay: uninfected cells incubated
with media, irrelevant mAb, or virus-specific antibodies; or infected cells incubated with media, irrelevant
mAb, and serial 2-fold dilutions of the virus-specific
antibodies. These cultures were then incubated with
the second fluorescein isothiocyanate conjugate-labeled anti-mouse Ig antibody. The irrelevant mAbs
were the same isotypes as the positive anti-virus antibodies and were used at the same concentration as
the sample tested.
Flow Cytometric Analysis
Uninfected and CMV-infected RPE and HEL cells
were removed from plastic dishes with trypsin-versine. Cell suspensions were fixed in 100% methanol
for 10 minutes at 4°C. After fixation, cells were washed
twice with PBS, and the cell count was adjusted to
10,000 cells///l. Primary incubations of 100 fj\ of cells,
with appropriate dilutions of isotypic controls and
monoclonal antibodies, were for 30 minutes at 4°C.
After washing with PBS, fluorescein isothiocyanate
goat F(ab)2 anti-mouse antibody was added, and the
cells were incubated for an additional 30 minutes at
4°C. Cells were again washed twice with PBS, then
fixed with 1% paraformaldehyde. All samples were
held at 4°C until analyzed.
Analysis was performed on a Coulter (Hiahleah,
FL) Profile II flow cytometer equipped with a 488
argon laser. Forward and side angle light scatter were
used for gating purposes, and 10,000 cells were collected. Isotypic controls, establishing the nonspecific
binding region, were followed by the associated monoclonal antibodies to determine percent positive fluorescence. Data presented are representative of at least
three separate experiments.
Western Blot Analysis
Uninfected and CMV-infected human RPE cells were
lysed by sonication in extraction buffer (50 raM Tris,
pH 8.5; 4 mM EDTA; 2 mM phenylmethylsulfonyl fluoride; 1% [wt/vol] deoxycholate; 1% [vol/vol] Triton
X-100). Cell lysates were clarified by centrifugation at
10,000 rpm for 10 minutes. Samples were adjusted
to a concentration of 20 /^g/40 [i\. The supernatant
fraction was subjected to sodium dodecyl sulfate-10%
polyacrylamide electrophoresis and transferred to nitrocellulose membranes. After a 6-hour incubation of
the membranes with either anti-CMV IE, anti-CMV E,
or anti-CMV L protein antibodies, they were washed
and further incubated for 2 hours with a 1:100 dilution
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
817
CMV Replication in Human RPE Cells
FIGURE 1. Cytomegalovirus-induced cytopathology in human fibroblast cells (HEL). (A) Normal, uninfected HEL cells. (B) Appearance of HEL cells 2 days after infection with CMV
at a multiplicity of infection of 0.1.
of goat anti-mouse IgG coupled to horseradish peroxidase. The color was developed with 4-chloro-l-naphthol and H^O-z as substrates.
Electron Microscopy
The cell culture medium was decanted, and the cells
were fixed in 2.5% glutaraldehyde in phosphate buffer
(pH 7.4) at 4°C. After 20 minutes, the cells were
scraped gently with a rubber policeman, poured into a
polypropylene test tube, and centrifuged into a pellet.
After a 2-hour incubation at 4°C, the fixative was removed, and the pellet was stored in 0.13 M sodium
phosphate buffer at 4°C. The cells were fixed in phosphate-buffered osmium tetroxide, dehydrated in alcohol, and embedded in Spurr's medium. Ultra-thin sections stained with uranyl acetate and lead citrate were
examined with a Zeiss (Jena, Germany) electron microscope.
RESULTS
Comparison of the Development of VirusInduced Cytopathology in Human Retinal
Pigment Epithelium, Fibroblasts, and
Epithelial Cells
Confluent monolayers of RPE, HEL-299, MRG-5, and
WISH cells were incubated with CMV at an input multiplicity of infection of 0.1 and 1. Typical CMV-cpe
consisting of ballooned swollen cells was noted in fibroblasts (HEL & MRC-5) within 24 hours after inoculation and encompassed 100% of the cells by days
4 to 5 (Fig. 1). In contrast, CMV did not alter the
appearance of WISH cells. Cytomegalovirus-induced
cpe was noted in RPE cells (Fig. 2); however, the timing was markedly different than that observed in the
fibroblast cell lines. Cytopathic effect wasfirstdetected
in small foci (<1% of the culture) at days 8 to 10,
FIGURE 2. Cytomegalovirus-induced cytopathology in human retinal pigment epithelial
(RPE) cells. (A) Normal, uninfected RPE cells. (B) Appearance of RPE cells 30 days after
infection with CMV at a multiplicity of infection of 0.1.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
818
20-
Days
FIGURE 3. Development of cytopathic effect (cpe) in cytomegalovirus (CMV)-infected cells. Human retinal pigment
epithelial cells and human fibroblasts (HEL, MRC-5) were
incubated with CMV at a multiplicity of infection of 1. After
a 2-hour incubation period, inoculum was removed, and
cells were washed three times and then re-fed maintenance
media. Cells were evaluated daily for the development of
cpe.
and 100% of the cells were not involved until approximately day 22 after inoculation (Fig. 3). These studies
indicate that CMV, which does not usually replicate
and induce cpe in vitro in epithelial cells, can do so
in human RPE cells. It is possible that a virus can
induce cytopathology in the absence of a productive
infection. To investigate this possibility, we next evaluated the cultures for the presence of infectious virus.
Replication of Cytomegalovirus in Human
Retinal Pigment Epithelial Cells
Monolayers of RPE, HEL, and MRC-5 cells were incubated with CMV at an input multiplicity of 1. After a 2hour virus adsorption period, cells were washed three
times, re-fed, and incubated at 37°C. At varying times,
supernatant fluids were removed and assayed for virus
infectivity. The remaining cells were resuspended in
the same volume of media and assayed for cell-associated infectious virus. The data presented in Figure 4
show the development of infectious virus in the supernatant fluid of CMV-infected cells. Production and
release of infectious virus into the supernatant fluid
occurs rapidly in human fibroblasts (HEL and MRC5 cells) (Fig. 4). At day 2, CMV was detected at a
concentration of 10' 5 TCID50/0.1 ml. Maximum virus
titers were detected at day 4 in both fibroblast cell
lines. In contrast, in RPE cells, CMV was detected at
day 7 and reached a maximum at day 19 (1065
TCID50/O.I ml).
The data that describe the development of cpe
(Fig. 3) and the data that depict the development of
infectious virus (Fig. 4) are derived from the same
experiments. It is of interest to point out that high
levels of infectious virus are present in the supernatant
fluid of infected RPE cultures at a time (days 7 to 12)
when less than 1% of the cells demonstrate a cpe.
The studies shown in Figure 4 indicate that the
CMV replication cycle in RPE cells is delayed in comparison to the virus replication cycle in human fibroblasts. This was noted also at different multiplicities
of infection (MOI). At an MOI of 1, virus-infected
RPE cells yielded maximum virus titers (10(l-5) at day
19, whereas virus-infected HEL cells yielded maximum
virus titers (1050) at day 4. At a MOI of 0.1, virusinfected RPE cells yielded maximal virus titer (10b5)
at day 46, whereas virus-infected HEL cells yielded
maximum virus titer (1055) at day 5. These studies
indicate that CMV replication in human RPE cells is
delayed when compared to the time course of CMV
replication in fibroblast cells. Nevertheless, the
amount of infectious virus produced by RPE cells significantly exceeds the maximum levels attained in fibroblast (HEL or MRC-5) cultures. This was observed
when we compared peak levels of virus produced in
supernatant fluids or when we compared peak levels
of virus detected in a combination of supernatant fluids and cell-associated virus (cell-associated [c] and
supernatant fluid [s] virus in HEL cells versus RPE
cells, P = 0.001; c + s virus in MRC-5 cells versus RPE
cells, P < 0.0001; c + s virus HEL cells versus c + s
Days
FIGURE 4. Cytomegalovirus (CMV) replication in human fibroblasts (HEL, MRC-5) and retinal pigment epithelial cells.
Cells were incubated with CMV at an input multiplicity of
infection of 1. After a 2-hour incubation period, inoculum
was removed, and cells were washed three times and then
re-fed maintenance media. At varying times between days 1
and 22, supernatant fluids were removed and assayed for
infectious virus on HEL cells. Results are expressed as the
mean ± standard error of the mean for experiments performed in triplicate.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
CMV Replication in Human RPE Cells
819
virus MRC cells, P = 0.07; paired Kests). It should be
noted that when cell cultures were infected, all of the
cultures contained the same number of cells inoculated with the same amount of virus (MOI 1). However, optimal virus yields were obtained in 4 days in
fibroblast cultures and in 19 days in RPE cultures.
Therefore, the ability of CMV to reach higher maximal
virus yield in RPE cells than in fibroblasts may be
explained, in part, by the additional time and RPE
cell multiplication.
When a clinical isolate of CMV was used to infect
RPE and HEL cells, a similar pattern of virus-induced
cpe and replication was noted. The clinical isolate
induced cpe in fibroblasts within 2 days, whereas cpe
was not noted in RPE cells until 15 days. Passage of this
isolate in RPE cells did not alter the delayed pattern of
virus replication described above in this section.
Because herpesvirus replication is frequendy cell
associated, we compared the level of infectious virus
produced in the supernatant fluid with the level of
virus obtained from cell-associated samples. Cytomegalovirus-infected RPE, HEL, and MRC-5 cells assayed
in Figure 3 also were analyzed for cell-associated virus.
Data obtained for the RPE cells is shown in Figure 5.
The amount of infectious virus was consistently slightly
higher in the cell-associated samples. Similarly, the
virus yield in the cell-associated samples for HEL and
MRC-5 cells was higher than that observed in the supernatant fluids (data not shown).
Electron microscopic evaluation confirmed the
presence of CMV infection in RPE cells. Cytomegalovirus-infected RPE cells revealed typical herpesvirus particles within the nucleus and cytoplasm of the RPE
cells. Virus particles were observed budding from the
nuclear and cell membranes (data not shown).
Analysis of Cytomegalovirus Proteins by
Western Blot Assays
Uninfected RPE cells and RPE cells infected with CMV
at a MOI of 0.1 were harvested at days 10, 15, and 22
after inoculation and were analyzed by Western blot.
Monoclonal antibodies reactive to CMV proteins IE1, E, or L were used. As seen in Figure 6, reactivity
with IE-1, E, and L proteins was detected at days 15
and 22. A weak band of reactivity was observed at day
10 for the E protein. Infected RPE cells did not react
with an unrelated monoclonal antibody, OX-8. In addition, uninfected RPE cells did not react with the
monoclonal antibodies for CMV proteins IE-1, E, or
L. At this multiplicity of infection (0.1), the kinetics
of detectable virus by infectivity assays and detection
of viral proteins by Western blot assays were similar.
Analysis of the Sequential Expression of
Cytomegalovirus Proteins in Retinal Pigment
Epithelium, Fibroblasts, and Epithelial Cells by
Immunofluorescent Assays
Retinal pigment epithelial, HEL, and WISH cells propagated on chamber slides were incubated with CMV
Days
FIGURE 5. Cytomegalovirus (CMV) replication in human retinal pigment epithelial cells. Cells were incubated with CMV
at an input multiplicity of infection of 1. After a 2-hour
incubation period, inoculum was removed, and cells were
washed three times and then re-fed maintenance media. At
varying times between days 1 and 22, CMV-infected RPE
culture lysates or supernatants were assayed for the production of cell-associated or extracellular virus. Results are expressed as the mean ± standard error of the mean for two
separate experiments, each of which was performed in triplicate.
at an input multiplicity of 1. At varying days after inoculation (days 1 to 4), slides were fixed in acetone and
incubated with monoclonal antibodies reactive to
CMV IE-1, E, or L proteins and to an unrelated epitope, OX-8. Data obtained are summarized in Table
1. In CMV-infected HEL cells, the nuclear staining for
IE-1 protein was detected within 5 hours. This reactivity was observed at days 1 to 4 in more than 90% of
cells. Nuclear staining for E protein was first noted at
24 hours. This reactivity persisted and was observed
on days 2, 3, and 4 in more than 90% of the cells.
Diffuse cytoplasmic staining for L protein was first
noted at 48 hours.
In CMV-infected RPE cells, IE-1 protein reactivity
was weak at day 1. At 48 hours, reactivity was observed
clearly. Although clear, bright nuclear fluorescent
staining for IE-1 protein could be detected in the RPE
cells, it was observed in less than 1% of the cell population. This is in sharp contrast to the staining pattern
observed in HEL cells, in which more than 90% of
the cells were reactive within 48 hours. E protein was
detected at 48 hours, and weak reactivity for L protein
was detected at 72 hours. In contrast to the reactivity
noted in HEL and RPE cells, CMV proteins IE-1, E,
and L were not detected in CMV-inoculated WISH
cells. The immunofluorescent staining data correlate
with the development of cpe and infectious virus
noted in HEL, RPE, and WISH cells.
Retinal pigment epithelial cells evaluated after day
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
820
TABLE l.
Comparison of the Expression of CMV Proteins on HEL, Human RPE,
and WISH Cells
IE Protein
Incubation
(days)
HEL
RPE
L Protein
E Protein
WISH
HEL
RPE
WISH
HEL
RPE
WISH
1
Reactivity
>90
0
Reactivity
>90
50
50
>90
>90
>60
100
100
100
Reactivity
Reactivity
<1
— = negative; ± weakly positive; + = positive; ++ = positive with increased intensity; % = percentage of cells reactive; HEL = human
fibroblast cell line; REP = retinal pigment epithelium; WISH = human amnion epithelial cell line; CMV = cytomegalovirus.
4 demonstrated an increase in the expression of IE,
E, and L proteins. When cytopathology was seen in
100% of the cells, 90% were positive for IE protein
and 100% were positive for E and L proteins.
Analysis of Cytomegalovirus Protein Expression
by Flow Cytometry
To quantitate accurately the percentage of cells staining and the intensity of the staining with anti-CMV
antibodies, these cultures were analyzed by flow cytometry. Seventy-two hours after CMV inoculation (MOI
of 0.1), HEL cells were harvested and evaluated by
flow cytometry. As is seen in Figure 7, 89.3% of the
cells were reactive for IE-1 protein, and 98.3% of the
cells were reactive for E protein. A similar pattern of
antigen expression has been described for IE-1 and E
proteins in CMV-infected fibroblast cells. This study
underscores the usefulness of flow cytometry for quantitating the number of cells expressing CMV proteins.
Four, 8, 26, and 30 days after CMV inoculation
(MOI of 0.1), RPE cells were harvested and evaluated
by flow cytometry (Fig. 8). Flow cytometric analysis
revealed no reactivity in CMV samples harvested 4 or
8 days after inoculation. In contrast, samples harvested
on day 26 were 19% reactive for IE-1, 84% reactive
for E, and 5% reactive for L proteins. In a separate
experiment, CMV protein expression was monitored
30 days after inoculation of RPE cells. Again, only
17.8% of the cells were reactive for IE-1, whereas
89.6% were reactive for E and 33.9% were reactive for
L proteins. Because of cell integrity requirements of
flow cytometry, it was not possible to evaluate the RPE
cells beyond this time point or when cpe was 100%.
Overall, these studies underscore a striking difference in the percentage of cells staining for the IE-1
protein. By flow cytometric analysis, we identified an
interesting pattern of viral gene expression in the mid-
dle of the CMV replication cycle. At this phase of
CMV replication in RPE cells, a low percentage of cells
express IE proteins at a time when a high percentage
of cells in the population express E proteins. This
CMV IE-1 protein is a stable protein found concurrently with E protein expression in HEL cells. This
difference in the percentage of cells expressing CMV
IE proteins in RPE cells is transient, that is, it does
not remain apparent at the time of 100% cpe (evaluation by immunofluorescent assays). This altered pattern of viral protein expression may be reminiscent of
recent studies29 identifying a transient expression of
CMV IE proteins in CMV-infected macrophages.
DISCUSSION
This study demonstrates that CMV can infect and replicate in human RPE cells. The virus induces a typical
CMV cytopathology in cultured RPE cells and HEL
cells, but not in another epithelial cell, WISH. Furthermore, this article describes an unusual pattern of viral
infection, that is, the virus replicates very slowly in RPE
cells. Although CMV induces cytopathology and cell
death in RPE cells, there is a prolonged period (days
6 to 12) of virus production in the absence of cytopathology. This pattern is strikingly different from CMV
infection in fibroblasts. Moreover, there appears to be
a unique pattern of viral proteins expressed at the
mid-point during CMV infection in RPE cells. This
shift is not seen in HEL cells. For example, at a time
when more than 89% of HEL-infected cells expressed
both IE and E proteins, RPE cells expressed a disproportionate level of these two proteins. Specifically,
RPE cells expressed 20% IE and 90% E proteins. Finally, of the three methods used to monitor this virus
infection, flow cytometric analysis allowed us to quantify the expression of CMV-induced proteins. In vivo
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
821
CMV Replication in Human RPE Cells
CMV infection frequently is associated with epithelial
cells, not fibroblasts. The kinetics of CMV replication
in RPE cells may not be the exception but rather the
typical mode of replication in specialized cells.
Cytomegalovirus infections are characterized by a
marked restriction in susceptible cells.1 In vitro studies
usually have relied on the efficient replication of CMV
in human fibroblast cells. However, a few reports have
identified low levels of CMV replication in cultured
o
in
>
>
CO
cj
106
80
-
49.5
-
1
CO
1
CM
CM
>
CO
T»
0)
1
o
o
C
>
>
>
C
'£
2
S
2
"c
o
o
o
>.
CO
•
in
>»
CO
•D
•
>
>
o
O
CM
CM
CO
•a
•
>
2
o
32.5 27.5 18.5 control
106
80
-
49.5
-
CMV - IE - protein
32.5 27.5 18.5 CMV - E - protein
CMV - L - protein
FIGURE 6. Western blot analysis of cytomegalovirus (CMV)
proteins in uninfected and CMV-infected retinal pigment
epithelial cells. Cells were infected with CMV at a multiplicity
of infection of 0.1. Uninfected and CMV-infected RPE cells
were harvested as described in Materials and Methods. After
sodium dodecyl sulfate-polyacrylamide gel electrophoresis,
the samples were transferred to nitrocellulose membranes
and incubated with monoclonal antibodies, which react with
immediate early (IE)-l, early (E), or late (L) viral proteins.
Protein molecular weight markers (in thousands) are shown
on the left.
epithelial cells, endothelial cells, and peripheral blood
mononuclear cell lines. In these cell types, the efficiency of CMV replication appeared to depend on the
maturity level and ploidy of the cell.30"*3 For example,
in the monocyte-macrophage cell system, CMV can
infect monocyte progenitors in bone marrow cells.
The CMV genome can exist within these cells for prolonged periods with litde or no viral gene expression.
When the cells reach a certain stage of differentiation,
the IE, E, and L genes are expressed sequentially.7tM
Alternatively, CMV replication in a differentiated
monocyte cell line, THP-1, is associated with changes
in viral enhancer-binding proteins and IE-1 gene expression, which occur with initiation of differentiation.32-35 The fact that CMV replicates in RPE cells
indicates that these cells are at a stage of differentiation to be permissive. The shift from a low level of
permissive cells early in the infection to 100% permissive cells may reflect alterations in cell cycle or level
of cellular differentiation. Preliminary studies in our
laboratory indicate that when RPE cells are maintained in serum-free conditions, the cells morphologically appear more differentiated. Cytomegalovirus infection of these cells occurs more rapidly than in RPE
cells maintained in serum (data not shown). Studies
are under way to evaluate the conditions that allow
these serum-free RPE cells to become more permissive
to CMV infection.
Of particular interest in this study was the observation that the slow replication of CMV in RPE cells was
associated with a low percentage of the cells displaying
viral proteins during the first 10 days after infection.
There are at least two possible explanations for the
low number of RPE cells expressing viral proteins, in
particular, the IE proteins. First, virus attachment and
penetration may be inhibited; second, IE gene expression may be altered by viral or RPE cellular transcription proteins, or both.1 The initiation of CMV infection is characterized by attachment of the virus to the
cell membrane and penetration by the virion into die
cell. Three viral proteins are associated with this cellular attachment.1 A One such protein, B2 microglobulin, is found on the virion surface, which may bind to
MHC proteins on the cell surface. In addition, viral
glycoprotein H(gH) and glycoprotein b (gB) bind to
90 kDa and 30 kDa cellular proteins, respectively. The
distribution or concentration of these cellular receptors on RPE cells may be insufficient to support CMV
infection. This may result in a low number of RPE
cells expressing IE protein. In fact, it has been reported that the MHC molecules are either not expressed or are expressed in low concentration on nonactivated RPE cells.36
Alternatively, the low number of CMV-infected
RPE cells during the first 10 days may be associated
with viral and/or cellular control of IE gene expression. It is known that IE gene expression is influenced
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
822
Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
control antibody
(1.4%)
No.
of
Cells
Log Fluorescence Intensity
FIGURE 7. Flow cytometric analysis of cytomegalovirus (CMV) proteins in HEL cells. Flow
cytometric analysis of CMV-infected HEL cells (3 days after inoculation), demonstrating
1.4% positivity for the isotypic control, 89.3% for the immediate early (IE) CMV protein,
and 99.3% positive for the early (E) CMV protein.
by cellular factors, such as NFkB.4810 12 IE genes are
controlled by the major IE promoter (MIEP) whose
activation depends on such host cell factors as cellular
transcription proteins. A number of investigators37
have shown that HCMV MIEP expression is different
in nonpermissive and permissive cells. Nonpermissive
cells may either lack positive transcriptional regulatory
factors or possess a repressor protein that acts on the
MIEP.38'39 On the other hand, permissive cells may
contain these positive transcriptional regulatory factors that support virus replication. Additional studies
are required to identify whether cellular regulatory
proteins are involved in the transient expression of IE
gene expression.
In addition to our description of alterations in the
early phase of CMV-infected RPE cells, flow cytometry
allowed us to identify a unique pattern of virus gene
expression in the middle of the CMV replication cycle.
This phase of CMV replication in RPE cells is characterized by a low percentage of cells expressing IE proteins at a time when a high percentage of the cells in
the population express E proteins. The IE-1, a 72-kDa
protein, is a stable protein found concurrently with E
protein expression in HEL cells. Whether viral factors,
such as IE-2, another immediate early protein, or cellular regulatory proteins are active in downregulating
IE protein expression in RPE cells remains to be determined.4041 Nevertheless, the low expression of IE-1
proteins late in RPE cell infection may be important in
viral persistence. Recendy, Fish and associates29 have
demonstrated that the growth kinetics of human CMV
are altered in monocyte-derived macrophages. They
described the productive nonlytic growth of CMV in
these cells, in which the replication cycle is delayed
No.
of
Cells
Log Fluorescence Intensity
FIGURE 8. Flow cytometric analysis of cytomegalovirus (CMV) proteins in retinal pigment
epithelial (RPE) cells. Flow cytometric analysis of CMV-infected RPE cells (30 days after
inoculation), demonstrating 1.6% positivity for the isotypic control, 17.8% for the immediate
early (IE) CMV protein, 89.6% for the early (E) CMV protein, and 33.9% for the late (L)
CMV protein.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
CMV Replication in Human RPE Cells
relative to replication of the virus in human fibroblasts. These investigators also showed that IE protein
expression appears early (day 3), disappears between
days 3 and 7, and reappears with L protein at day 7.
This pattern of viral protein expression appears reminiscent of CMV infection in RPE cells.
Two general tenets of viral persistence are downregulation of viral gene expression and failure of the
immune system to detect and clear virus-infected
cells.42 These principles are true for infections, including CMV infections. Our observation of a low expression of CMV IE proteins in RPE cells may contribute to
viral persistence within the retina by evading immune
elimination.
Cell-mediated immune responses are important
in controlling CMV infections.43 In fact, in the murine
system cytotoxic T cells (CTLs), CD8+ T cells, are
instrumental in host defense against CMV infections.43
Adoptive transfer of CD8+ T cells prevents death in
the mouse CMV immunodeficiency model. The majority of the CTLs recognize immediate early products
and not late viral gene products. In fact, as many as
50% of the CTLs recognize IE-1 gene products. 44 ' 45
Additional studies with a vaccinia virus expressing
CMV IE-1 proteins demonstrated the importance of
these proteins in CMV-host cell interactions. 46 Inoculation of mice with the vaccinia virus expressing only
CMV IE-1 proteins protected mice from a lethal challenge of CMV. In humans, CTLs specific for CMV IE
proteins also have been described and shown to play
a critical role in controlling CMV infections.47"49 The
studies reported here show that during the first 10
days after infection, there is a low level of expression
of all viral proteins in CMV-infected RPE cells. When
virus replication increases, low levels of IE protein are
observed at a time when greater amounts of E and L
proteins are expressed. Because the majority of CTLs
generated in CMV infections can be directed against
IE proteins, this alteration in the eye may allow CMVinfected RPE cells to evade immune destruction.
In summary, CMV replication in RPE cells may provide clues to mechanisms involved in CMV latency and
activation within the retina. Alterations in the IE gene
product may indicate the existence of positive and negative nuclear regulatory factors within the infected RPE
cell. Additional factors, such as the availability of viral
receptors on RPE cells and the stage of cellular differentiation, may influence retinal CMV infections. Studies
are under way to evaluate some of these factors.
823
Cytomegalovirus Diseases. Becker Y, Darai G, Huang ES,
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
Key Words
cytomegalovirus,flowcytometry, immediate early genes, retinal pigment epithelium
References
1. Huang ES, Kowalik TF. The Pathogenirity ofHuman Cytomegalovirus: An Overview in Molecular Aspects of Human
15.
16.
eds. Heidelberg: Springer-Verlag; 1993:3-45.
Hirsch MS. Cytomegalovirus and its role in the pathogenesis of acquired immunodeficiency syndrome.
Transplant Proc. 23; 1991:118-121.
Maclean H, Dillon B. Cytomegalovirus retinitis: Diagnosis and treatment. MJSTD AJDS. 4; 1993:322-325.
Mocarski ES. Cytomegalovirus biology and replication. In: Roizman B, Whitley RJ, Lopez C, eds. The
Human Herpesviruses. New York: Raven Press; 1993:
173-226.
McDonough SH, Spector DH. Transcription in human fibroblasts permissively infected by human cytomegalovirus strain AD 169. Virology. 1983:125:31-46.
Wathen MW, Stinski MF. Temporal patterns of human
cytomegalovirus transcription: Mapping the viral
RNAs synthesized at immediate-early, early, and late
times after infection. / Virol. 1982;41:462-477.
Minton EJ, Tysoe C, Sinclair JH, Sissons JGP. Human
cytomegalovirus infection of the monocyte/macrophage lineage in bone marrow. / Virol. 1994; 68:40174021.
Sambucetti LC, Cherrington JM, Wilkinson GWG,
Mocarski ES. NF-kappa B activation of the cytomegalovirus enhancer is mediated by a viral trans-activator and by T cell stimulation. EMBOJ. 1989; 8:4251 4258.
Thomsen DR, Stenberg RM, Goins WF, Stinski MF.
Promoter regulatory region of the major immediate
early gene of human cytomegalovirus. Proc Natl Acad
Sci USA. 1984;81:665-663.
Lukac DM, ManuppelloJR, AlwineJC. Transcriptional
activation by the human cytomegalovirus immediateearly proteins: Requirements for simple promoter
structures and interactions with multiple components
of the transcription complex. J Virol. 1994; 68:51845193.
Kerry JA, Priddy MA, Stenberg RM. Identification of
sequence elements in the human cytomegalovirus
DNA polymerase gene promoter required for activation by viral gene products. J Virol. 1994; 68:41674176.
Mocarski ES, Abenes GB, Manning WC, Sambucetti
LC, Cherrington JM. Molecular genetic analysis of cytomegalovirus gene regulation in growth, persistence
and latency. Curr Top Microbiol Immunol. 1990; 154:4774.
Myerson D, Hackman RC, Nelson JA, Ward DC,
McDougall JK. Widespread presence of histologically
occult cytomegalovirus. Hum Pathol. 1984; 15:16451658.
Pepose JS, Newman C, Bach MC, et al. Pathologic
features of cytomegalovirus retinopathy after treatment with antiviral agent ganciclovir. Ophthalmology.
1987;94:414-424.
Egbert PR, Polland RB, Gallagher JG, Merigan TC.
Cytomegalovirus retinitis in immunosuppressed hosts:
II: Ocular manifestations. Ann Int Med. 1980; 93:664670.
Bok D. Retinal photoreceptor-pigment epithelium interactions. Invest Ophthalmol Vis Sci. 1985;26:16591694.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
824
Investigative Ophthalmology & Visual Science, April 1996, Vol. 37, No. 5
17. Detrick B, Hooks JJ. Autoimmune aspects of ocular
disease. In: Rose N, MacKay I, eds. The Autoimmune
Diseases. Vol. II. New York: Academic Press; 1992; 345361.
18. Miceli MV, Newsome DA, Novak LC, Beuerman RW.
Cytomegalovirus replication in culture human retinal pigment epithelial cells. Curr Eye Res. 1989;
8:835-839.
19. Pepose JS, Hilborne LH, Cancilla PA, Foos RY. Concurrent herpes simplex and cytomegalovirus retinitis and •jncephalitis in the acquired immune deficiency syndrome (AIDS). Ophthalmology. 1984;91:
1669-1677.
20. Wang Y, Detrick B, Hooks JJ. Coronavirus (JHM) replication within the retina: Analysis of cell tropism in
mouse retinal cell cultures. Virology. 1993; 193:124137.
21. Detrick B, Newsome D, Percopo C, Hooks JJ. Class II
antigen expression and gamma interferon modulation of monocytes and retinal pigment epithelial cells
from patients with retinitis pigmentosa. Clin Immunol
Immunopathol. 1985; 36:201-211.
22. Hooks JJ, Chan CC, Detrick G. Identification of the
lymphokines, interferon-gamma and interleukin-2, in
inflammatory eye diseases. Invest Ophthalmol Vis Sci.
1988; 29:1444-1451.
23. Percopo C, Hooks JJ, Shinohara T, Caspi R, Detrick
B. Cytokine-mediated activation of a neuronal retinal
resident cell provokes antigen presentation. J Immunol. 1990; 145:4101-4107.
24. Liversidge HFS, Thomson AW, Forrester JV. Lymphokine-induced MHC class II antigen expression of cultured retinal pigment epithelial cells and the influence of cyclosporine A. Immunology. 1988;63:313-317.
25. Baudouin C, Fred-Reygrobellet D, Jambou P, Gastaud P, Lapalus P. HLA-DR and DQ expression on
human retinal pigment epithelial cells in vitro. Graefe's
Arch ClinExp Ophthalmol. 1989; 228:86-88.
26. Nagineni CN, Detrick B, Hooks JJ. Synergistic effects
of gamma interferon on inflammatory mediators that
induce interleukin-6 gene expression and secretion
by human retinal pigment epithelial cells. Clin Diag
Lab Immunol. 1994; 1:569-577.
27. Elner SG, Elner VM, Pavilack MA, et al. Modulation
and function of intercellular adhesion molecule-1
(CD54) on human retinal pigment epithelial cells.
Lab Invest. 1992;66:200-211.
28. Elner VM, Scales S, Elner SG, Danforth J, Kunkel SL,
Strieter RM. Interleukin-6 gene expression and secretion by cytokine stimulated human retinal pigment
epithelial cells. Exp Eye Res. 1992;54:361-368.
29. Fish KN, Depto BS, Moses AV, Britt W, Nelson JA.
Growth kinetics of human cytomegalovirus are altered
in monocyte-derived macrophages. J Virol. 1995;
69:3737-3743.
30. Dutko FJ, Oldstone MBA. Cytomegalovirus causes a
latent infection in undifferentiated cells and is activated by induction of cell differentiation. J Exp Med.
1981;154:1636-1651.
31. Gonczol E, Andrews PW, Plotkin SA. Cytomegalovirus
replicates in differentiated but not in undifferentiated
32.
33.
34.
35.
36.
37.
human embryonal carcinoma cells. Science. 1984;
224:159-161.
Weinshenker BG, Wilton S, Rice GPA. Phorbol esterinduced differentiation permits productive human cytomegalovirus infection in a monocytic cell line. J Immunol. 1988; 140:1625-1631.
Tanaka J, Sadanari H, Sato H, Fukuda S. Sodium butyrate-inducible replication of human cytomegalovirus
in a human epithelial cell line. Virology. 1991; 185:271280.
Maciejenski JP, Bruening EE, Donahue RE, Mocarski
ES, Young NS, St. Jeor SC. Infection of hematopoietic
cells by human cytomegalovirus. Blood. 1992; 80:170178.
Sinclair JH, Baillie J, Bryant LA, Taylor-Wiedeman
JA, Sissons JG. Repression of human cytomegalovirus
major immediate early gene expression in a monocytic cell line. / Gen Virol. 1992;73:433-435.
Detrick B, Rodrigues M, Chan CC, Tso MOM, Hooks
JJ. Expression of HLA-DR antigen on retinal pigment
epithelial cells in retinitis pigmentosa. Am] Ophthalmol. 1986; 101:584-590.
Nelson JA, Fish K, Ibanez C, et al. Dependence of
cytomegalovirus replication on monocyte differentiation. In: Michelson S, Plotkin SA, eds. Multidisciplinary
Approach to Understanding CMVDisease. Amsterdam: El-
sevier; 1993:77-86.
38. Shelbourn SL, Kothari SK, Sissons JPG, Sinclair JH.
Repression of human cytomegalovirus gene expression associated with a novel immediate-early regulatory region binding factor. Nucl Acids Res. 1989;
17:9165-9171.
39. Ghazal P, DeMattei C, Giulietti E, Kliewer SA, Umesono K, Evans RM. Retinoic acid receptors initiate induction of the cytomegalovirus enhancer in embryonic cells. PNAS USA. 1992; 89:7630-7634.
40. Wolff D, Sinzger C, Drescher P, Jahn G, Plachter B.
Reduced levels of IE2 gene expression and shutdown
of early and late viral genes during latent infection of
the glioblastoma cell line U138-MG with selectable
recombinants of human cytomegalovirus. Virology.
1994; 204:101-113.
41. Arlt H, Lang D, Gebert S, Stamminger T. Identification of binding sites for the 86-kilodalton IE2 protein
of human cytomegalovirus within an IE2-responsive
viral early promoter. / Virol. 1994;68:4117-4125.
42. Oldstone MBA. Molecular anatomy of viral persistence. / Virol. 1991;65:6381-6386.
43. Koszinski UH, Del Val M, Reddehase MJ. Cellular and
molecular basis of the protective immune response to
cytomegalovirus infection. Curr Top Microbiol Immunol.
1990; 154:189-220.
44. Reddehase MJ, Mutter W, Munch K, Buhring HJ,
Koszinowski UH. CD8-positive T lymphocytes specific for murine cytomegalovirus immediate-early
antigens mediate protective immunity. / Virol.
1987;61:3102-3108.
45. Koszinski UH, Reddehase MJ, Keil GM, et al. Molecular analysis of herpesviral gene products recognized
by protective cytolytic T lymphocytes. Immunol Lett.
1987; 16:185-192.
46. Jonjic S, Del Val M, Keil GM, Reddehase MJ, Koszi-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017
CMV Replication in Human RPE Cells
nowski UH. A nonstructural viral protein expressed
by a recombinant vaccinia virus protects against lethal cytomegalovirus infection. J Virol. 1988;62:
1653-1658.
47. Rodgers B, Borysiewicz L, Mundin J, Graham S, Sissons P. Immunoaffinity purification of a 72K early
antigen of human cytomegalovirus: analysis of humoral and cell-mediated immunity to the purified
polypeptide. / Gen Virol. 1987; 68:2371-2378.
825
48. Borysiewicz LK, Hickling JK, Graham S, et al. Human
cytomegalovirus specific cytotoxic T cells. / Exp Med.
1988; 168:919-931.
49. Reusser P, Riddell SR, Meyers JD, Greenberg PD. Cytotoxic T lymphocyte response to cytomegalovirus after
human allogeneic bone marrow transplantation:
Pattern of recovery and correlation with cytomegalovirus infection and disease. Blood. 1991; 78:
1373-1380.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933416/ on 06/19/2017