close

Вход

Забыли?

вход по аккаунту

?

In vitro evidence for a dual role of tumor necrosis factor- in human immunodeficiency virus type 1 encephalopathy.

код для вставкиСкачать
In Vitro Evidence for a Dual Role of Tumor
Necrosis Factor-cx in Human
Immunodeficiency Virus Type 1
Encephalopathy
Susan G. Wilt, PhD," Elizabeth Milward, PhD," Jia Min Zhou, MS," Kunihiko Nagasato, MD," Heather Patton,
BS," Ray Rusten," Diane E. Griffin, MD, PhD$ Michael OConnor, MD,t and Monique Dubois-Dalcq, MD"
Microglial cell activation, myelin alteration, and abundant tumor necrosis factor (TNF)-a message have been observed
in the brains of some human immunodeficiency virus type 1 (HIV-1)-infected and demented patients. We therefore
used cultures of purified human microglia and oligodendrocytes derived from adult human brain to examine the role
of TNF-a in HIV-1 encephalopathy. Human microglia synthesize TNF-a message and protein in vitro. When these
cells were infected with HIV-1 JrFL and maintained in the presence of TNF-a antibodies, soluble TNF-a receptors,
or the TNF-a inhibitor pentoxifylline, viral replication was delayed or strongly inhibited. Both human microglia and
oligodendrocytes express the two TNF receptors, TNF-R1, which has been implicated in cytotoxicity, and TNF-R2.
While TNF-a may enhance HIV-1 replication in an autocrine manner, it is not toxic for microglia. In contrast,
recombinant human TNF-a causes oligodendrocyte death in a dose-dependent manner. In situ detection of DNA
fragmentation in some cells indicated that oligodendrocyte death may occur by apoptosis. Addition of live microglia
or medium conditioned by these cells also resulted in 30 to 40% oligodendrocyte death, which was largely prevented
by TNF-a inhibitors. We propose that TNF-a plays a dual role in HIV-1 encephalopathy, enhancing viral replication
by activated microglia and damaging oligodendrocytes. Thus, TNF-a inhibitors may alleviate some of the neurological
manifestations of acquired immunodeficiency syndrome.
Wilt SG, Milward E, Zhou JM, Nagasato K, Patton H, Rusten R, Griffin DE, O'Connor M,
Dubois-Dalcq M. In vitro evidence for a dual role of tumor necrosis factor-a in human
immunodeficiency virus type 1 encephalopathy. Ann Neurol 1995;37:381-394
The most common neurological complication of acquired immunodeficiency syndrome (AIDS) is the
AIDS cognitive/motor complex, also called the AIDS
dementia complex C 1). Human immunodeficiency
virus type 1 (HIV-1) can spread to the central nervous
system (CNS) where it replicates mostly in cells of the
monocyte-macrophage lineage including microglia, the
resident macrophages of the brain 11-31. With in situ
hybridization and immunocytochemistry, productive
infection of cells of the monocyte-macrophage lineage
residing in the brain has been observed consistently
while viral transcripts and proteins have been identified
only rarely in neurons, oligodendrocytes, or astrocytes
13-51. Neuropathologists have observed a variety of
pathological changes ranging from microglial nodules,
gliosis, neuron loss, and myelin alterations to multinucleated giant cells. The terms used to describe the
neuropathology of HIV- 1-associated disease include
HIV-1 encephalitis, HIV leukoencephalopathy, vacuolar myelopathy or leukoencephalopathy, and diffuse
poliodystrophy C5-7).
How does the virus affect rhe brain and spinal cord
to cause such profound neurological dysfunction? The
virus may directly cause neuropathological changes by
replicating in microglia, inducing formation of multinucleated giant cells and microglial nodules in white matter, deep gray matter, and basal ganglia C3-41. Furthermore, nonviral factors induced by HIV-1 infection may
cause astrocytosis, myelin changes (coined "pallor"
based on reduced histological staining), blood-brain
barrier alterations, and neuronal death {l, 2, 61. In
HIV-l-induced leukoencephalopathy and myelopathy, myelin breakdown and/or vacuolation occur, suggesting dysfunction of oligodendrocytes, the cells that
synthesize and maintain CNS myelin { 3 , 4, 5 , 71. In
addition, one can detect widespread activation of mi-
From the *Laboratory of Viral and Molecular Pathogenesis, National
Institute of Neurological Disorders and Stroke, National Institutes
of Health, Bethesdat ?The Graduate Hospital, University of Permsylvmia, Philadelphia, PA; and $Johns Hopkins University School
of Medicine, Baltimore, MD.
Received Aug 18, 1994, and in revised form Nov l . Accepted Nov
3, 1994.
Address correspondence to Dr Dubois-Dalcq, Laboratory of Viral
and Molecular Pathogenesis, NINDS, NIH, Building 36, Room
5D04, 36 Convent Drive, MSC 4160, Bethesda, MD 20892-4160.
Copyright 0 1995 by the American Neurological Association 381
croglial cells that express major histocompatibility
complex (MHC) class I1 molecules and cytokines in
the CNS of AIDS patients { 3 , 8, 91.
One of these cytokines, tumor necrosis factor
(TNF)-a is of particular interest since TNF-a messenger RNA (mRNA) levels are significantly higher in the
subcortical white matter of demented AIDS patients
than in that of nondemented AIDS patients [lo}. As
TNF-a is known to enhance HIV-1 replication in
monocyte-macrophages [ll], it may also enhance
HIV-1 replication in activated microglia. In addition,
microglia-derived T N F a could cause alterations of the
white matter such as those observed in some AIDS
patients. In favor of this hypothesis are the observations that white matter alterations can be correlated
with the presence of activated microglial cells in some
AIDS patients 17,8}and that TNF-a, in a recombinant
form or on the surface of rat microglia, can be cytotoxic
for rat oligodendrocytes and disrupt myelin in vitro
[12-151.
To investigate the mechanisms of action of TNF-a
in HIV-1 infection of the human brain, we used an in
vitro system where microglia are purified from adult
human brain and, when cultured, become activated and
release TNF-a. We analyzed in this system how microglia-derived TNF-a may influence HIV- l replication in microglial cells and may trigger human oligodendrocyte death. HIV-1 replication in microglia was
studied in the presence of TNF-a inhibitors, which
strongly decreased HIV-1 production by these cells.
We also analyzed how TNF-a signaling in this system
can cause death of myelin-forming cells and examined
the role of activated microglia in these events. We
found that in vitro, TNF-a is a key mediator of HIV-1
replication in human microglia and death of human
oligodendrocytes in the vicinity of activated microglia.
Materials and Methods
Materials
Unless specifically noted in the text, chemicals and enzymes
were obtained from Sigma; media, from GIBCO BRL; restriction enzymes, from New England Biolabs; and conjugates and antisera, from Jackson Immunoresearch.
Tissues and Cells
PRIMARY HUMAN BRAIN CULTURES. These were established
from temporal lobe biopsy specimens resected from patients
with intractable epilepsy, by enzymatic dissociation and centrifugation through a 30% Percoll gradient as described {lb].
One minor modification was that Dulbecco’s modified Eagle
medium (DMEM) supplemented with 10% fetal bovine
serum (FBS) was substituted for phosphate-buffered saline
solution (PBS) in all centrifugations other than the Percoll
density gradient centrifugation and also in the initial overnight incubation. Surgically resected tissue was obtained from
patients enrolled in a hospital internal review boardapproved study. The most common histopathology found in
382
Annals of Neurology
Vol 37 No 3 March 1995
these patients is subpial cortical and subcortical white matter
gliosis as discussed previously {17). The tissues did not include the epileptic focus and were extensively cleaned from
contaminating blood before dissociation. This virtually eliminates blood macrophages from the cultures. These cultures
contained three major populations of cells-microglia,
astrocytes, and oligodendrocytes-which were subsequently
purified as described below.
HUMAN OLIGODENDROCYTES. Nonadherent cells were collected 24 hours after initial seeding, centrifuged, and plated
on poly-L-lysine (20 ng/ml)-coated dishes; they consist
mainly of oligodendrocytes as assayed by immunostaining for
the 0 4 antigen and galactocerebroside (GC) Eli). To enhance regeneration and survival of human oligodendrocytes
(HOs)in vitro, insulin-like growth factor 1 (R & D Systems),
basic fibroblast growth factor, and platelet-derived growth
factor AB (PDGF) (Upstate Biotechnology) at 50, 20, and
10 ng/ml, respectively, were added to defined medium as
described [li].Oligodendrocytes typically had regenerated
their processes by days 7 to 14. Growth factors were removed from defined medium 3 days prior to cytotoxicity
assays (see below).
Adherent cells in primary cultures
MICROGLIA CULTURES.
were fed with DMEM containing 5% FBS and 5% giant
cell tumor supernatant (GCT; AIDS Reagent and Reference
Program) 118) for 7 to 14 days. The cultures were then
placed on an orbital shaker (150 rpm) for 5 hours at room
temperature to dislodge astrocytes 1163. The remaining adherent cells were trypsinized and plated on poly-L-lysine (10
ng/ml)-coated dishes and consist of 97 to 99% microglia as
assayed by immunostaining for the low-density-lipoprotein
receptor (LDL-R) [18]. In some experiments, purified microglia were added after trypsinization to purified HOs at a
maximum ratio of 1 : 1. Alternatively, purified microglia
were maintained for 4 days in a nonadherent state by culturing in a 24-well Teflon plate (designed by D . Rausch, National Institute of Mental Health, and manufactured inhouse) before aliquots of live cells were added to purified
HO cultures.
Human fetal astrocytes (passage numbers
2-5, courtesy of D r Tornatore, National Institute of Neurological Diseases and Stroke) were grown in DMEM with
10% fetal calf serum (FCS) and seeded in 24-well plates at
1 to 2 x lo3cells/well. These cells, rather than adult human
astroctyes, were used because purified preparations of
astrocytes derived from adult human brain were difficult to
obtain in sufficient quantities, mostly due to their slow mitotic rate. The rat oligodendrocyte (RO) CG4 cell line was
grown and passaged 20 to 35 times as described El93 and
seeded in 24-well plates at 1 to 2 x lo3 cells/well or at 2 to
4 X lo4cells/35-mm dish. This cell line was used in quantitative assays requiring high numbers of cells impossible to generate from the small amount of human brain tissue obtained
from surgery. U138MG glioma, HeLa, and U937 promonocytic cell lines were obtained from American Type Culture
Collection (ATCC) and cultured in DMEM containing 10%
FCS.
OTHER CELLS.
of TNF-a Message and Protein
lmmunocytochemistry
Detection
IDENTIFICATION OF MICROGLIA A N D HUMAN OLIGODENDRO-
A partial cDNA clone for human
TNF-a containing 586 nucleotides (2807-3392) (gift of Dr
Peter Bressler, National Institute of Allergy and Infectious
Diseases) was cloned into a pBluescript KS( - ) vector (Stratagene). After linearization with Hind111 or BamHI, sense
and antisense riboprobes were synthesized using the T7 and
T 3 RNA polymerases, respectively, and labeled with digoxigenin (Genius 5 kit, Boehringer Mannheim). For in situ hybridization, cells were fixed in 4% PF for 20 minutes, treated
with O.OSn/O Triton X-100 for 3 minutes and 1 pg/ml of
proteinase K at 37°C for 15 minutes, and then postfixed with
4 % PF for 5 minutes and hybridized as described {22}, except that no chloroform treatment was performed and the
final wash was at 55°C.
Live microglia in cocultures with
HOs were labeled for LDL-R with 1,l’-dioctadecyl1-3,3,3’,3’-tetramethylindocarbocyanineperchlorate-conjugated LDL (DiI-LDL; Biomedical Technologies) and all cells
were stained for chromatin with Hoechst 33342 DNA binding dye (Molecular Probes). Cocultures were then fixed 20
minutes with 29%paraformaldehyde (PF) before staining with
mouse monoclonal anti-GC and goat anti-mouse IgG conjugated to fluorescein and mounting as described elsewhere
[l?]. Preparations were examined and photographed on a
Zeiss Axiophot microscope equipped with the appropriate
fluorescence filters.
CYTES IN COCULTURES.
CUltures were fixed for 40 minutes with 4% PF and permeabilized for 5 minutes with Triton X-100 (0.05%) before staining with mouse monoclonal anti-TNF receptor 1 (TNF-R1)
(Austral Biologicals, 4 pg/ml) or anti-CD14, a marker for
microglia (4 pg/ml) as control, followed by goat anti-mouse
IgG peroxidase con jugate and metal-enhanced diaminobenzidine (DAB; Pierce). Cultures were then dehydrated and a
coverslip applied using Permount before photomicrography
as above.
IN SITU HYBRIDIZATION.
STAINING OF TUMOR NECROSIS FACTOR RECEPTOR I .
R N A Isolation and Polymerase Chain Reaction
Cellular RNA from microglia, HOs, or U138MG glioma
(ATCC) was extracted with RNAzol B (BIOTECX Laboratories) and complementary DNA (cDNA) was generated by
M-MLV reverse transcriptase (GIBCO BIU) as per manufacturers’ instructions and used in polymerase chain reaction
(PCR) (95”C, 5 minutes; 30 cycles of 95”C, 1 minute; 60°C,
1 minute; 72”C, 1.5 minutes; 2.5 mM magnesium chloride,
200 pM each deoxynucleoside triphosphate (dNTP), 0.5 p M
each primer, 2 units of Taq/ 1 pg of cDNA) with the following primers (5’, 3’, respectively) specific for
1. TNF-a: CCTGTAGCCCACGTCGTAGC, ‘ITGACCTCAGCGCTGAGTTG (203. Detection of FCR products
was done by end-labeling the 5’ primer using yi2P ATP
(Amersham) and polynucleotide kinase (Stratagene) as
per the manufacturers’ instructions.
2. TNF-RI: TGGGC’ITCAGTCCCGTGCCC, CGCCTCCAGGTCGCCAGCAT. These primers (spanning nucleotides 1103-1481, G B accession no. X55313) bracket
most of the “death domain” (nucleotides 1232-1276 to
1493-1534) in the carboxy terminal region C21).
3. TNF-R2: CAC’ITGCCTGCCGATAAGGC, GCTACAGACGTTCACGATGC.
These primers span nucleotides 1020-1283 of G B accession
no. S63368.
Each primer set spans an intron; for TNF-R1 the intervening intron is small, resulting in a second band of 532 bp which
amplifies in the absence of reverse transcriptase treatment,
reflecting genomic DNA in some samples. Amplification of
glyceraldehyde 3-phosphate dehydrogenase mRNA (GAPDH)
was used when necessary to confirm the presence of RNA
in cell extracts (data not shown). The PAW1 18 plasmid (Stratagene) contains a segment of TNF-a cDNA.
IMMUNOPRECIPITATION. Biosynthetic labeling of cellular
proteins with sulfur 35 ( 3 5 S )cysteine (ICN Radiochemicals)
and immunoprecipitation with anti-TNF-a antibody (Genzyme) was performed as described (23) with extracts from
microglial and U937 cells. The immunoprecipitates were
electrophoresed through a 15n/O Laemmli polyacrylamide gel,
which was then fixed in acetic acid-methanol ( 5 : 95), dried
onto solid support, and exposed to x-ray film (Kodak
X-OMAT) for 14 days.
ENZYME-LINKED IMMUNOSORBENT ASSAY. TNF-a secreted
in serum-free medium conditioned by microglia for 6 or 24
hours was assayed using an enzyme-linked immunosorbent
assay (ELISA) kit (R & D Systems).
HIV Infection of Microglia Cultures and p24 Assay
HIV-1 JrFL (AIDS Reagent and Reference Program) inoculum containing either 6 or 20 ng of p24 antigen was adsorbed
onto 8 x lo4 microglial cells in 35-mm dishes for 6 hours
[24]. At the time of inoculation the microglia cultures were
95 to 98% pure, but 2 weeks later, some islands of fibroblasts
grew in these cultures. Supernatant samples were collected
at 3- to 4-day intervals and stored at - 70°C until p24 ELISA
( S R A Technology) was performed [24].
Cytokines, Inhibitors, and Antibodies
Recombinant human TNF-a (rhTNF-a; activity ranged from
0.5-1.0
lo8 units/mg of protein) and neutralizing rabbit
TNF-a antibody (added to HOs or microglia medium to
neutralize 700 to 4,000 units of T N F d m l ) were purchased
from Gentyme. Initially, normal rabbit IgG was used as a
control in one HIV-1 infection experiment, which showed
no effect of these IgG’s on viral production as measured by
p24 ELISA. As TNF-ol has a membrane-bound form, we also
used as control an antibody reacting with a component of the
microglia cell surface, CD14, the lipopolysaccharide (LPS)
receptor. The mouse monoclonal CD14 antibody (Sigma)
was dialyzed overnight against sterile PBS to remove azide
and the protein concentration (BCA kit, Pierce) adjusted to
match that of the TNF-a antibody added to the cultures.
Soluble TNF-R1 (sTNF-Rl), a recombinant chimeric protein
consisting of the extracellular domain of TNF-R1 fused to
the Fc immunoglobulin region [251, was used at concentrations ranging from 25 to 200 ng/ml. Pentoxifylline was used
at a concentration of 30 pglml.
+
Wilt et al: TNF in HIV-1 Encephalopathy
383
Cytotoxicity Assays
0.02% H20,, cells were treated with terminal transferase (10
HO cultures were exposed to increasing doses of rhTNF-a,
units/ml, 37°C for 1 hour in supplied buffer, Boehringer
Mannheim) and biotin-labeled deoxyuridine triphosphate
(dUTP; 1 pmol/ml, Boehringer Mannheim). Labeled nuclei
were detected by streptavidin conjugated to peroxidase and
DAB staining. Positive controls were obtained by treating
fixed cells with DNAse at 40 kg/ml for 10 minutes at 37°C.
Only cells with intense staining over the nuclei were counted
as positive and their ratio to the total number of cells in 20
fields was calculated.
microglia-conditioned medium (CM), or live microglia cells
in serum-free defined medium [17] for 20 to 24 hours. Control HOs survive well for at least 3 days in such medium.
After treatment the cultures were analyzed for HO viability
or signs of apoptosis.
CONDrrIoNED MEDIUM COLLECTION. Serum-free DMEM
was conditioned for 6 hours by purified microglia cultures
that had been previously maintained in medium containing
5% GCT alone, or treated for 24 hours with either LPS
(100 ng/ml) or pentoxifylline (30 Kglml) in GCT-containing
medium. CM of microglia or HeLa cells (used as a control)
were centrifuged for 10 minutes at 2,000 rpm, aliquoted,
and stored at -70°C until use after a single freeze-thaw.
Such CM were diluted 1 : 1 with 2 x HO serum-free defined medium [ 171 for testing on HO cultures.
VIABILITY ASSAY. After rhTNF-a or CM treatment, cultures were stained for l hour at 37°C with (3-[4,5dimethylthiazol-2-yl]-2,5-diphenyl tetrazolium bromide
which is reduced by active mitochondria
(MTT; Sigma) [26],
to a purple insoluble stain. In the case of cocultures, M I T
staining was performed after LDL-R staining of living cells
(see above). Eight randomly chosen fields were photographed
in each dish with a 10 x objective on a Nikon inverted
microscope. M’M-negative and -positive cells were counted
blindly in each set of photographs for each condition and the
ratio of unstained cells to total cells calculated. For the most
recent experiments, instead of counting cells on photographs,
cells were counted in each field after video capture of the
negatives and inversion to a positive “print” using the program NIH Image (public domain). Any cell containing less
than three spots of mitochondria1 staining was considered
negative; as M’M staining was often much fainter in experimental than in control HO cultures, unstained cells were
identified by the presence of a discernible cell body or processes. Such analysis yields a minimum estimate of cytotoxicity since the total number of H O s per dish was lower in all
experimental conditions, compared with control dishes, due
to detachment of dead cells from the dish. Variance analysis
of the M’M viability assay data was performed using SuperANOVA (Abacus) and applying the Scheffe’s S test criteria (99% confidence levels) to the mean averages. The mean
and standard errors were also calculated with SuperANOVA.
In the few cases where the sample size was small, standard
deviations (SDs) are shown. The p values are reported in the
text and figure comments.
TUNEL assays were performed on R O
cultures exposed to CM, live microglia, or increasing doses
of rhTNF-a for 24 hours as described 127). ROs survive well
in serum-free defined medium for 3 days {26}. HO cultures
derived from one brain biopsy were also stained by the
TUNEL method after rhTNF-a or CM treatment for 24
hours. ROs were fixed 10 minutes with acetic acid-ethanol
( 5 : 9 5 ) while HOs (which could not be preserved well
enough by this fixation) were fixed in PF and permeabilized
with Triton X-100 as described above (see Immunocytochemistry). After inactivation of endogenous peroxidase with
APOPTOSIS ASSAY.
384 Annals of Neurology
Vol 37
N o 3 March 1995
Results
Human Microglia Synthesize TNF-(Y
Message and Protein In Vitro
Microglial cells isolated from freshly resected adult
human brain tissue express many monocyte-macrophage markers but differ from peripheral blood monocytes by their elongated shape, increased adhesion and
survival, their nonproliferative properties, and their
high levels of MHC class I1 expression in vitro 116, 18,
28). When cultured in serum medium with or without
added growth factors, the majority of these cultured
microglial cells have properties of activated cells: They
are capable of phagocytosis of latex particles and of
antigen-antibody complexes bound to the Fc receptor;
they express class I1 molecules (HLA-DR) and this expression is further increased by interferon-?; they can
present antigen in the context of class I1 and stimulate
T-cell mitosis (discussed in {28]). This spontaneous activation of adherent human microglia in vitro may be
caused by myelin debris released during adult white
matter dissociation, as myelin vesicles have been shown
to activate microglia and to stimulate secretion of cytokines by these cells {29).
When RNA extracted from purified microglia cultures (3-5 days in vitro) was probed for the presence
of TNF-a mRNA using reverse-transcriptase PCR
(RT-PCR), a strong signal for TNF-a-specific sequences was detected (Fig 1A). Presence of TNF-aspecific transcripts was confirmed by in situ hybridization using a digoxigenin-labeled riboprobe that stained
approximately 75% of these microglial cells (Fig 1B).
Immunoprecipitation of extracts on human microglial
cells, and of the promonocytic cell line U937 as control, revealed both the 51-kD trimeric active form and
the 17-kD monomeric form of TNF-a protein {3032) (Fig 1C). Correspondingly, ELISA of medium of
purified microglia cultures detected up to 32 pg/ml of
TNF-a protein. Although human fetal microglia activated by LPS have been reported to release as much
as 1 ng of TNF-u/ml 1331, the values of up to 32 pg/
ml observed here for adult microglia are within the
range of those observed in the plasma and serum of
AIDS patients 181. Treatment of these already activated microglial cells with LPS or interferon-? increased membrane ruffling and caused some cell shape
changes but only a slight increase of TNF-a in the
~~~~
~
F i g 1 . Adult human microglia synthesize TNF-a message and
protein in vitro. (A) TNF-a-specific sequences detected by reverse-transcriptase polymerase chain reaction (RT-PCRI on total
RNA isolated from purzfied human microglia cultures at 5 days
in vitro. A 375-bp TNF-a RT-PCR product (arrowhead) is
detected in lane 2 (1.0 pg of R N A reverse transcribed).Lane 1
shows a 360-bp product amplzj2edfrom the PAW 118 plasmid
containing TNF-a gene sequences. The molecular weight (MW)
standard was @XIHinflII.Sequence specificity was confirmed
by restriction enzyme digestion (not shown) and identical PCR
products were observed with four different brain specimens. iB)
In situ hybridization on purihed human microglia cultuyes
using a digoxigenin-labeled riboprobe revealed TNF-a transcripts in microglial cells from seven different brain specimens.
Hybridization with the TNF-a sense probe did not produce
staining (not shown). 1Ci TNF-a monomer (17-kd) and trimer
15 1-kd)proteins (arrowheads)were detected after iiwnanoprecipitation of cell extracts from purijied microglia cultures (lane 2)
and from the U93 7 promonocyte line (lane 1 ). Both TNF-a proteins were detected in three immunoprecipitations,using microglial extract from two brain specimens.
supernatant and no changes in the levels of TNF-a
message (data not shown). As TNF-a activates NF-KB
which regulates expression of several genes including
the TNF-a gene itself [32), it is likely that TNF-a
released into the medium enhances its own expression
in a paracrineiautocrine manner.
TNF-a Inhibitors Decrease or Abolish HIV-1
Replication in Human Microglia
To determine whether endogenous microglia-derived
TNF-a enhances HIV-1 replication in these cells {ll,
341, purified microglia cultures (3-5 days in vitro)
were inoculated with the brain-derived HIV- I JrFl
isolate {24} and maintained in the presence or absence
of TNF-a inhibitors for 3 weeks. Both neutralizing
TNF-a antibodies and sTNF-R1 can inhibit binding
of TNF-a to its surface receptor in monocytoid cell
lines [11, 25, 341. in the absence of inhibitors, HlV-1
JrFL caused a productive infection of human microglial
cells with peak viral production at 10 to 14 days after
inoculation, when assayed by release of p24 antigen
124) (Fig 2). In contrast, in the presence of TNF-a
inhibitors, the rate of HIV-1 replication in human microglial cell cultures was very low for the first 2 weeks,
with levels of p24 antigen release 4 to 10 times lower
at 14 days than levels in untreated infected cultures
(see Fig 2). In addition, the peak of p24 antigen release
was 2 to 3 times lower and sometimes delayed for 3
days (see Fig 2). sTNF-R1 was as effective as TNF-a
antibodies in delaying viral production while an antibody to a surface protein expressed on cells of the
monocyte-macrophage lineage, CD14 {28}, produced
no significant decrease in the rate of HIV-1 replication
(see Fig 2). These TNF-a inhibitors also caused a decrease in virus-induced fusion 1261 and a threefold to
fivefold decrease in the number of cells positive for
p l 7 gag antigen by immunofluorescence. Moreover,
infected cultures maintained in the presence of TNF-a
antibodies had conserved 99 t 1.5% (SD) microglial
cells compared to uninfected cultures while infected
cultures had lost 29 & 1% (SD) of their microglial
cells after 3 weeks in vitro (n = 2; 2 brain specimens).
Thus TNF-a may act in an autocrine and/or paracrine
way to enhance HIV-1 expression and replication in
primary human microglia as it does in chronically infected U1 cells [ll, 343. During the third week of
infection, however, p24 release increased even in cultures with TNF-a inhibitors. As the HIV-1 promoter
(LTR) contains NF-KB-binding sites, inhibition of
TNF-CLinduction of this transcription activator may initially result in low viral expression 1351. However, slow
but progressive accumulation of the HIV- 1 tat protein,
which binds to the TAR region of the HIV-I LTR,
may trigger the virus replicative cycle by an NF-KBindependent mechanism at later times {35].
Wilt
et al:
TNF in HIV-1 Encephalopathy
385
75
4 mock
+--JrFl
~
-
+ TNF A 5
~
E
CT)
c=
m
50
j
i
JrFl+
tsTNFR
*.-
0 1 4 AB
1
25 j
1
Days post-infection
Fig 2. lnhibitorj of T N F - a decreased p24 antigen release by
HIV-1 JrFL-itzfected microglia cultures. Addition of antiTNF-a antibodies (7”-AB; at a roncentration neittralizing
-3.000 units TNI;-alml) or soluble TNF-R1 (sTNFR; 25
viglml- results were identical with 50 nglml) after imiral inoculation and ewvy 3 to 4 days for 3 u1eek.r delayed p24 antigen
release for weral days as compared t o HIV- 1 JrFL-infected
cultures without TNF ailtibodies or with C D 14 antibody
(CD14 AB) { 3 1 , 3.1}. A pretious experiment 136) demonstrated that tsiral production toas inhibited similarly uhether
T N F antibody was added prior to, roncurrent with, or immediately ajter viral inoculation. Pento?cifylline(PTX) (30 p g l m l )
added o~~ernight
prior t o lira1 inoculation (arrow) and readded
ei’ery 3 to 4 duys abolished p24 antigen release oixer a .$week PKriod. The gruph shows a representatiw experiment on cultures
derived from one brain specimen. Similar resultj. were obtained
with TNF antibody in cultures deriiJedfrom
brain specimens ai7d with CDI4 antibody, sTNF-R, or pentox-ifylline in
cultirres derizied from three brain specimem. Each p24 t’alue is
from a pool of three czhres.
We then tested a potent inhibitor of TNF-a synthesis, the methyl-xanthine derivative pentoxifylline, which
interferes with NF-KB activation and binding to its
D N A motif [36, 37). Pentoxifylline (30 mg/ml) itself
did not cause microglial cell loss as the ratio of LDL-Rpositive cells in treated versus untreated cultures was
1.02 after 2 weeks (duplicate dishes counted). RNA
from pentoxifylline-treated microglia did not show
TNF-a-specific sequences when assayed by RT-PCR
while signal for the constitutively expressed enzyme
GAPDH was readily detected (data not shown). Microglia cultures were treated with pentoxifylline 16 to
20 hours prior to adsorption with HIV-1 JrFL and continuously thereafter for 3 weeks. Such treatment usually abolished viral production completely over a 3week period (see Fig 2) as well as cell fusion [26) and
p l 7 gag antigen expression (not shown). In one experiment, release of p24 antigen did rise slightly above
basal levels at 14 days after infection; nevertheless, a
957h inhibition of p24 release by pentoxifylline treatment was observed at 21 days (7.0 nglml vs 175 ng/
ml in untreated cultures). A study of HIV-1 replication
in microglia exposed to increasing doses of pentoxifyl-
386 Annals of Neurology Vol 37 No 3 March 1995
line revealed that 30 mg/ml was sufficient to almost
abolish viral production by microglia in this system
(data not shown). Such a pronounced antiviral effect
may result not only from a block in TNF-a production
but also from inhibition of synthesis of other lymphokines as well as a direct effect on viral transcription
[37). Thus, of all inhibitors tested in our system, pentoxifylline had the longest lasting, most potent blocking effect on HIV-1 replication in human microglia
TNF Receptors Are Expressed by Human
Oligodendrocytes and M irrogfia
To analyze further the effects of TNF-a produced in
this in vitro system, we next examined the expression
of T N F receptors on both human microglia and HOs.
There are two known TNF-a receptors, a 55-kD
(TNF-R1) and a 75-kD (TNF-R2) protein with 27%,
homology in the cysteine-rich regions of their extracelMar domain but little intracellular domain homology
C21, 38). Only TNF-R1 contains an 80-amino acid
carboxy terminus domain responsible for signaling
cytotoxicity and termed the “death domain” {21, IS}.
We first searched for transcripts coding for these receptors in R N A extracted from both purified microglia
and purified HOs as well as from U138MG glioma
cells as control. These RNAs were analyzed by RTPCR using primers specific for TNF-R2 and for
TNF-R1 in the region of the death domain [21). A
band of the expected size for each receptor was detected in normal HOs and activated microglia as well
as in the glioma cells (Figs 3A, 3B).
Because TNF-R1 has been implicated in TNF-ainduced cytotoxicity t381, we used the immunoperoxidase technique to stain TNF-R1 protein on HOs and
compared with staining on microglia and astrocytes.
Interestingly, HOs showed the most intense staining
on their cell body and processes compared to other
cell types, suggesting a high density of receptors on
cultured H O s (Fig 3C). HOs did not stain for the
microglia-specific marker CD14 (Figs 3D, 3E). Human
microglia were also labeled by anti-TNF-R1 but less
intensely than HOs (data not shown).
Recombinant TNF-a and TNF-a Deritied from
M icroglia Can Cause Oligodendroryte Death
Detection of TNF-R1 death domain sequences and
TNF-R1 protein in HOs suggested that TNF-a binding to this receptor could trigger cytotoxic signaling in
HO. We therefore examined the cytotoxic effects of
increasing doses of rhTNF-a for HOs as studied previously in ROs [12, 14, 15, 26). rhTNF-a at 1,600 units/
ml caused cytotoxicity in approximately 30% of HOs
after 24 hours as measured by the MTT viability assay
(Figs 4A-4C) and the percentage of M’IT-negative
cells was found to be dose dependent (see Fig 4A).
Moreover, cytotoxicity for HOs could be neutralized
B
A
Human Microglia
Human Oligodendrocytes
Fig 3. Human oligodendrocytes (HO) express TNF-Rl message
and protein. (A,B) Reverse-transcriptase polymerase chain reaction (RT-PCRI demonstration of sequences speczjk for TNF-Rl
(lanes 2-5: top arrow-genomic DNA; middle arrowcDNA) and TNF-R2 (lanes 7, 8: bottom arrow) in R N A extracted from purified HOs (A) and microglia (B) (lanes 2, 3, 7)
or U138MG glioma cells (A and B, lanes 4, 5 , 8) afer incubation with (lanes 3, 3 , 7 , 8) or without (lanes 2, 4) reverse transcriptase 18). Lanes 1 and 6 are water controls. (Cj Immunoperoxidzse staining for TNF-RI protein of HO cultares revealed
intense staining throughout the cell body and processes. (D, E)
CD14 antibody, which labels microglia, did not prodace any
staining (0)although HOs were readily identified by phase
microscopy in the same field (E). (C, D, and E: x 125.)
by TNF-a antibodies (see Fig 4A). Treatment of HOs
f o r 3 days with 800 units of TNF-a also caused cytotoxicity in a significant proportion of HOs compared
to untreated controls (33.4
8.6% vs 7.3
4.4%;
n = 3, p = 0.007). Retraction and/or degeneration
of HO processes was often the first indication of
rhTNF-a action 139, 401, a process that could result
in demyelination in vivo even without cell death.
We then investigated whether TNF-a derived from
*
*
activated human microglial cells could cause changes
in morphology or death of HOs. Cytotoxic activity for
HOs was detected in CM of purified microglia after
24 hours and was equivalent to activity of 800 to 1,600
units of rhTNF-a (40-80 ng/ml, 3.6-7.2 nM; see Fig
4D). An equivalent degree of cytotoxicity was observed in microglial CM irrespective o f whether or not
microglia were activated by LPS (see Fig 4D). Since
such CM may contain as little as 32 pg/ml of TNF-a
(see above), biological activity from microglia CM appears far greater than that of equivalent amounts of
rhTNF-a, Nevertheless, the cytotoxic effect was entirely neutralized by TNF-a antibodies, was reduced
by a pretreatment of microglial cells with pentoxifylline
before collecting CM, and was not observed to any
significant level with CM of HeLa cells (see Fig 4D),
indicating that microglia-derived TNF-a was mostly responsible for this cytotoxic activity. Moreover, such
human microglia CM did not significantly affect viability of human fetal astrocytes (data not shown). Since
adult microglia and astrocytes in our cultures are simi-
Wilt et al: TNF in HIV-1 Encephalopathy
387
A
8
F
5
-TNFAB
i TNF AB
30
*O
ae 10
0
0
400
TNF
'P
E
*
800
CI
1600
Ulml
-TNFAB
iTNF AB
40
30
*O
10
0
pglia CM
Fig 4. Human oligodendrocytes (HO) are killed by recombinant
human TNF-a (rhTNF-a) and microglia-conditioned medium
(CM). (A)rhTNF-a is cytotoxicfor HOs in a 24-hour assay
in a dose-dependent manner. Values shown for HOs represent
the mean average of pooled data from four experiments (duplicates) on cultures of four brain specimens (n = 8). The stundavd errov of the mean is indicated. Addition of neutralizing
TNF antibody (TNF AB) (suficient to neutralize 4,000
unitslml) reduced the cytotoxicity induced by 400 unitslml or
800 unitslml of rhTNF-a to control levels (p = 0.99, n = 6
in three experimentsl. (B, C ) Representative photographs of
MTT-stained HO cultuves in the absence or presence, respective,$. of I .600 unitslml of rhTNF-a. There are fmer cells in
the rhTNF-a-treated cultures. (Magnijhtion x 60.1 (D)
CM from microglia is cytotoxic for HOs within 24 hours. The
proportion of dead cells was not enhanced by lipopolysaccbarzde
(LPS) treatment, was comparable to that seen with maximal
doses of rbTNF-a (A),and was neutralized by TNF antibody
(all p = 0.001, n = 6). Similarly, CM from cultures treated
with pentoxifilline ( E X )for 24 hours demonstrated signifcantly less cytotoxicity than CM from untreated microglia (p =
0.001. n = 6). The effect of pentoxzfilline on TNF-a synthesis
is reversible. accounting for the slight toxicity observed since pentoxifilline was withdrawn from the cultures 6 hours before CM
u~ascollected. CM from HeLa cell cultures iwhich release < 1
pglml of TNF-a) caused no signifcant cytotoxicity when compared to microglia CM ip = 0.99, n = 4). The graph shows
the merage of the mean, with standard error, of data from three
experiments using three brain specimens ( n = 6), except where
there are bars with an asterisk, indicating n = 4 in two experiments. M T T = 3 4 4 2 -dimethylthiazol-2-yl)-2.5-diphenyl tetrazolium bromide.
388 Annals of Neurology
Vol 37 N o 3 March 1995
larly unaffected by CM, the oligodendrocyte response
to microglia CM appears unique among glial cell types.
As activated microglia may come in contact with oligodendrocytes in vivo, we added live, purified microglial cells to pure HO cultures at a ratio of 0.1 to 1
microglia to 1 HO. In these experiments, coculture
with microglial cells also resulted in alteration of processes and loss of mitochondrial activity in HOs, especially at the ratio of 1 microglia to 1 HO (Figs 5 , 6).
Indeed, an even greater proportion of HOs (up to
40%) was affected than when microglia CM alone was
added (see Fig 5). Again, TNF-a antibodies and
sTNF-R1 neutralized the bulk of this cytotoxicity while
the control CD14 antibody did not (see Fig 5). Interactions between activated microglia and HOs were analyzed in cocultures triple-stained for LDL-R (microglia
in red), for G C (HOs in green), and for nuclei with
Hoechst (in blue) (see Figures 6A-6D). When microglial cells closely interacted with groups of HOs, they
lost their elongated shape, becoming ruffled and larger
(compare Figs 6B and 6C). As shown (see Fig bB),
HOs in the vicinity of these enlarged microglial cells
often showed fragmented processes (and decreased
M?T staining, not shown), whiIe HO processes were
preserved in cultures without microglia (see Fig 6A)
or in cocultures treated with TNF-a antibodies (see
Fig 6C). In contrast, cocultures treated with CD14 antibody, which is not predicted to block cytotoxicity, also
I-TNFAB
a
501
t
CD14 AB
ItTNFAB
40
0
d,
aJ
c
I=
I
*
30
*O
10
n
0
0.1
0.2
1
pglia: HO
Fig 5 . Activated microglia are toxicfar human oligodendrocytes
(HO) and induce alterations in HO shape and processes. In a
24-hr assay, coculture of microglia at ratios ranging from 0.1 to
1 microglia t o I HO signzjicantly increased (n = 6, p =
0.005) the observed proportion of dead cells when compared to
control HO cultures without microglia (see graph; 0 on x axi.4.
The proportion of HO cell death reached 40% at the highest ratio of microglia to HO tested ( I :I ) and was significantly reduced by addition of TNF antibody (TNF AB) (at a concentration neutralizing 4,000 unitslml of recombinant human
TNF-a; p = 0.01, n = 6) or soluble TNF receptor I
(sTNFR1) (p = 0.01, n = 4) but not by CD14 antibody
(CD14 AB) (n = 4). The cytotoxicity of the lowest microglia
to HO ratio tested (0.1 :1) was entirely neutralized by TNF
antibody (p = 0.001, n = 6).
showed disruption of HO processes, as expected (see
Fig 6D). Microglia in the vicinity of dying oligodendrocytes often acquired G C staining, suggesting active
phagocytosis of HO membrane (see Figs 6B, 6D). This
may coincide with increased activation, as occurs with
myelin phagocytosis by these cells 1291. An increased
local concentration of TNF-a and/or membrane-bound
TNF-a on the surface of human microglia may cause
HO alterations in shape, depolarized membrane potential and mitochondrial activity, and even possibly,
apoptotic death [14, 27, 31, 32, 38).
T o determine whether oligodendrocyte death induced by TNF-a can indeed occur by an apoptotic
process accompanied by DNA fragmentation, we used
the TUNEL method to detect nicks in D N A in situ in
the nuclei {271. In each cell type, the ability of TUNEL
to detect breaks in chromatin D N A was first tested by
treating the cells with DNAse, which caused nuclear
staining in 100% of the cells. The effects of TNF-a
on nuclear DNA was then tested on an RO line [19]
in order to obtain cells in sufficient numbers for quantitative analysis. This cell line previously was shown to
exhibit responses to TNF-a similar to those reported
here for HOs 115, 26). rhTNF-cr caused apoptosis of
ROs, with a dose response (not shown) closely resembling those previously obtained in HOs and ROs with
the MTT viability assay (see Fig 5 and 126)).Apoptotic
nuclei were also detected after treatment with CM or
coculture with live microglia for 24 hours in 21 & 2%
and 24 k 295, respectively, of R O cells (Figs 7A, 7C);
these values are closely similar to those observed after
treatment of ROs with 800 units of rhTNF-a/ml (23
k 1.3%) and significantly higher than was seen in untreated ROs (7 t 2%) (n = 3, p = 0.01) (Fig 7B).
Apoptosis was inhibited by TNF-a antibody, which reduced the percentage of labeled nuclei in ROs to 10
? 1% (n = 3, p = 0.01). In preliminary experiments
on HO cultures derived from one brain specimen,
apoptotic nuclei were also observed after treatment
with 800 unitdm1 of TNF-cu or microglia CM while
they were rare in the presence of TNF-a antibodies
(Figs 7D-7F). These observations suggest that at least
a portion of those cultured rat and human oligodendrocytes killed by TNF-a die as a result of an apoptotic
mechanism.
Discussion
The present study demonstrated that human brain microglia synthesized TNF-a in vitro and that TNF-a
inhibitors such as anti-TNF-a antibodies, soluble T N F
receptor, and pentoxifylline delayed and/or strongly
inhibited HIV-1 replication in these cells. This suggests that TNF-a acts in an autocrine/paracrine fashion
in this system to stimulate the NF-KB-mediated activation of the promoters of HIV-1 and the TNF-a gene
itself [ l l , 32). Moreover, we showed that death of
HOs can be induced in vitro by rhTNF-a in a dosedependent manner and that, likewise, activated human
microglia can mediate death of HOs by contact or
through their CM. These microglial cytotoxic effects
were also largely prevented by TNF-a inhibitors.
Taken together, our in vitro observations suggest that
microglia-derived TNF-a may play a central role in
the pathogenesis of HIV- 1 encephalopathy, enhancing
viral replication and causing oligodendrocyte damage
~411.
The observations that both TNF-R1 and -R2 are
expressed by human microglia and oligodendrocytes in
vitro was predictable as many cell types express both
receptors 1321. Most of the known TNF-a responses
are probably mediated by the activation of TNF-R1,
as suggested by recent studies in TNF-R1 knockout
mice [42). TNF-R1 may be the principal receptor signaling cell death 121, 42). The intracellular domain
of TNF-R1 has a unique “death” sequence coding for
80-amino acid residues (differing from the TNF-R2
intracellular domain) with homology to a domain of
the Fas antigen which can mediate activation-driven
T-cell suicide 121, 43). It appears however that Fas or
TNF-R1 receptor signaling can lead to either apoptosis
or proliferation, depending on the cell type, the stage
of differentiation, and/or the presence of other stimuli
143). TNF-R2 is particularly abundant on myeloid and
Wilt et al: T N F in HIV-1 Encephalopathy
389
Fig 6. Triple immunofluorescence staining of cocultures of micvoglia and human oligodendrocytes (HOs)at 24 hours. Microglia are stained red for low-density-lipoprotein receptors; HO,
green for galactocerebroside and nuclei, blue (Hoechst).(A) Control H O cultures without microglia showed highb arborized processes and GC staining on cell surfaces. (B)Addition of microglia (red) to HO results in dramatic deterioration of HO
processes. Note the hrge multipolar microglia in the center of the
field. (Ci In cocultures with miwoglia (arrows) in the presence
of TNF antibody, complex HO processes were preserved. Microglia remained elongated and smaller than observed in (B).
ID) Addition ofCDl4 antibody did not inhibit disrnption
of HO processes in response t o microglia (arrow); compare with
(A).( X 220 before 5% reduction.)
390 Annals of Neurology Vol 37 No 3 March 1995
Fig 7 . Detection of apoptotic nuclei in rat (ROs) and human
oligodendroqtes (HOs) at 24 hours. RO cultures were stained
by the TUNEL technique afer treatment with microgliaconditioned medium (CM) (A) or after coculture with live
human microglia (C, open arrowheads) at a ratio of 1 microglia to 1 RO. More apoptotic nuclei (black arrows) were observed
as compared to control RO cultures (B) (see numbers in text).
HO cultures treated with 800 units of recombinant human
TNF-a (0)or microglia CM (F) showed20.1% and 15.1%
apoptotic nuclei, respectively, while less than 1% stained nuclei
were detected in HO cultures treated with mzcroglia CM with
TNF antibody (E). A total of 450 to 550 cells were counted in
each condition. In HO, there appears to be some d;ffusion of the
stain from the nucleus t o the cytoplasm but this staining pattern was also observed in DNAse-treated cells. (Magnification
X 153.)
lymphoid cells where it is involved in cell proliferation
but it may also mediate cytotoxic effects through a
different mechanism from TNF-R1 {32, 44). TNF-R1
can mediate apoptosis in U937 and HL-60 cells
through the ceramide signaling pathway {44, 451. In
these cells, rapid hydrolysis of sphingomyelin to ceramide occurs through the action of a membraneassociated neutral sphingomyelinase and apoptosis can
be induced by ceramide analogs. Of particular relevance to microglia is that TNF-R1 appears to mediate
NF-KBactivation through an acidic sphingomyelinase
in the cytoplasm 1441. Clearly TNF-R1 does not signal
Wilt et al: TNF in HIV-1 Encephalopathy
391
death in microglia. At the present time, we can only
hypothesize that the same TNF-R1 may trigger different downstream signaling pathways in each cell type in
our system, leading to viral and host gene expression
in microglia as opposed to cytotoxicity in some oligodendrocytes. Elucidation of mechanisms of TNF-R1
signaling and the role of TNF-R2 in these CNS cells
awaits studies in single and double knockout mice for
each T N F receptor f421. In addition, the regulation of
both T N F receptors in microglia and oligodendrocytes
during demyelinating processes should be examined in
appropriate animal models.
On the basis of our observations on HIV-1-infected
human microglia cultures maintained in the presence
of TNF-a inhibitors, we propose the following sequence of events in HIV-1-infected patients. When
virus invades the brain, only a small number of resident
microglial cells get infected. If these cells are not activated and do not synthesize TNF-a, HIV-1 will only
replicate to low levels, a situation that can be observed
in vitro in human fetal microglia [46, 471 and in
HIV- 1-infected microglia treated with pentoxifylline
(this study). With time, HIV-1 tat protein may reach
levels high enough to turn on viral replication by an
NF-KB-independent mechanism. This would allow
the infection to spread slowly until widespread activation of microglial cells in the brain is triggered by one
or several factors @-lo}. The gp120 glycoprotein
could be one such factor since transgenic mice expressing this protein in astrocytes showed signs of microglial
activation and clustering reminiscent of microglial cell
nodules seen in HIV-1 encephalitis [48}.Activated microglia would then synthesize TNF-a, which in turn
would enhance viral replication f 3 , 8, 10, 11, 341. The
resulting increase in viral load in the brain may contribute to progression toward dementia. Therefore, an inhibitor such as pentoxifylline, which interferes with
TNF-a synthesis and HIV-LTR activation, could slow
this progression, possibly by inhibiting viral expression
in microglia. Indeed, pentoxifylline attenuates histological signs of experimental allergic encephalitis in mice,
suggesting that this drug can modulate CNS disease
1491. Other inhibitors such as anti-TNF-a antibodies
may not be able to cross the blood-brain barrier or
may have a more transient antiviral effect as they do
in vitro.
Apoptosis has been associated with programmed cell
death, a selective process of physiological cell deletion
during development and adulthood that is accompanied by endogenous endonuclease activity and chromatin cleavage (discussed in [27]). The observation that
cell death, possibly by apoptosis, can occur in some
oligodendrocytes in the presence of TNF-a raises the
question of the role of TNF-a in demyelinating processes in the intact CNS. Myelin-forming cells may be
exposed chronically to TNF-a bound to or derived
392 Annals of Neurology Vol 37 No 3 March 1995
from activated microglia in vivo. This could result in
the striking demyelination that can be observed in
HIV-1 vacuolar myelopathy. Such events may occur
not only in the HIV-1-infected CNS but also in multiple sclerosis (MS), an autoimmune demyelinating disease. Studies on MS patients have shown that in some
cases, increased TNF-a levels in the cerebrospinal fluid
correlate with disease activity {so], that blood mononuclear cells make significantly more TNF-a mRNA
in relapsing than stable states [5l}, and that TNF-a is
found on the edge of demyelinating lesions f52, 531.
In this regard, it is interesting that low doses of
rhTNF-a can be cytotoxic for cultured ROs after 3
days 1261. Thus, chronic exposure of oligodendrocytes
even to low levels of TNF-a derived from microglia
may possibly result in demyelination in vivo. Immune
cells infiltrating the CNS may also affect oligodendrocytes as activated T4 cells can induce oligodendrocyte
lysis in a non-MHC-restricted manner in vitro 1541.
Clearly a variety of factors may determine the degree,
extent, and irreversibility of oligodendrocyte damage
in the diseased CNS tissue.
In addition, TNF-a can induce nitric oxide synthase
in rat microglial cells, giving rise to nitric oxide (NO),
which can mediate microglial cell toxicity [55]. Thus,
inhibition of binding of TNF-a could block not only
cytotoxicity in HOs but also production of NO by
activated microglia. However, preliminary experiments
have shown that adult human microglia produced no
significant amounts of NO in vitro (J. Snell, S. G. Wilt,
and C. A. Colton, unpublished observations, 1994).
O n the other hand, several protective mechanisms may
counteract the TNF-a cytotoxic effects and prevent
more widespread HO alteration and death [321. While
TNF-a is a powerful inducer of reactive superoxides, it
can also induce the manganese-dependent superoxide
dismutase which confers resistance to TNF-a, indicating that mitochrondrially generated superoxide radicals
are a key component of TNF-mediated cell killing
1561. Such protective mechanisms may occur in the
majority of oligodendrocytes which do appear resistant
to TNF-(r. cytotoxic effects in vitro even after 3 days.
Whether these protective mechanisms operate in vivo
remains to be demonstrated.
In summary, microglia-derived TNF-a enhances
HIV-1 replication in human microglial cells and mediates cytotoxicity toward human CNS myelin-forming
cells. These events can be delayed or even abolished
by TNF-a inhibitors. We predict from these in vitro
observations that a powerful inhibitor of TNF-a and
HIV- 1 transcription such as pentoxifylline may well be
effective in preventing or slowing certain neurological
manifestations of AIDS. Similarly, inhibitors of TNF-a
may be beneficial in demyelinating diseases such as
MS. Clearly these proposals need to be investigated in
appropriate animal models.
D. E. Griffin was supported by National Institutes of Health (NIH)
grant NS 26643 and K. Nagasato by a fellowship of the Human
Frontier Science Program Organization.
We thank Zack Howard (Dyncorp) for soluble TNF-R1 receptor,
Carlo Tornatore for human fetal astrocytes, Steve Wesselingh for
TNF-a primers, Pierre Henkart for advice on apoptosis, and several
N I H colleagues, especially Vittorio Gallo, Rhonda Voskuhl,
Mohammad Hajihosseini, and Ellen Meier, for critical reading of the
manuscript.
References
1. Spencer DC, Price RW. Human immunodeficiency virus and
the central nervous system. Annu Rev Microbiol 1993;46:655-
693
2. Epstein LG, Gendelman HE. Human immunodeficiency virus
type 1 infection of the nervous system: pathogenic mechanisms.
Ann Neurol 1993;33:429-436
3. Dickson DW, Lee SC, Hatch W, et al. Macrophages and microglia in HIV-related central nervous system neuropathology. In:
Price RW, Perry SW, eds. HIV, AIDS and the brain. New
York: Raven, 1994:99-118
4. Budka H , Wiley CA, Kleihues P, et al. HIV-associated disease
of the nervous system: review of nomenclature and proposal for
neuropathology-based terminology. Brain Pathol 1991;l:143152
5. Weiser B, Peress N, LaNeve D, Eilbot DJ. HIV-I expression
in the CNS correlates directly with the extent of disease. Proc
Natl Acad Sci USA 1990;87:3997-4001
6. Power C, Kong P-A, Crawford TO, et al. Cerebral white matter
changes in acquired immunodeficiency syndrome dementia: alterations of the blood-brain barrier. Ann Neurol 1993;34:339350
7. Esiri MM, Morris CS, Millard PR. Fate of oligodendrocytes in
HIV-1 infection. Curr Sci 1991;5:1081-1088
8. Tyor WR, Glass JD, Griffin JW, et al. Cytokine expression in
the brain during the acquired immunodeficiency syndrome. Ann
Neurol 1992;31:349-360
9. Tyor WR, Glass JD, Baumrind N, et al. Cytokine expression
of macrophages in HIV-I-associated vacuolar myelopathy. Neurology 1993;43:1002-1009
10. Wesselingh SL, Power C, Glass JD, et al. Intracerebral cytokine
messenger RNA expression in acquired immunodeficiency syndrome dementia. Ann Neurol 1993;33:576-582
11. Poli G, Kinter A, Justement JE, et al. Tumor necrosis factor a
functions in an autocrine manner in the induction of human
immunodeficiency virus expression. Proc Natl Acad Sci USA
1990;87:782-785
12. Robbins D, Shirazi Y, Drysdale B-E, et al. Production of cytotoxic factors for oligodendrocytes by stimulated astrocytes.
J Immunol 1987;139:1398-2597
13. Selmaj K, Raine C. Tumor necrosis factor mediates myelin and
oligodendrocyte damage in vitro. Ann Neurol 1988;23:339346
14. Zajicek JP, Wing M, Scolding NJ, Compston AS. Interactions
between oligodendrocytes and microglia: a major role for complement and tumor necrosis factor in oligodendrocyte adherence and killing. Brain 1992;115:1611-1631
15. Louis J-C, Magal E, Takayama S, Varon S. CNTF protection of
oligodendrocytes against natural tumor necrosis factor induced
death. Science 1993;259:689-692
16. Yong VW, Antel JP. Culture of glial cells from human brain
biopsies. In: Fedoroff S, Richardson A, eds. Protocols for neural
cell culture. New York: Humana, 1992: 81-95
17. Gogate N, Verma L, Zhou JM, et al. Plasticity in the adult
human oligodendrocyte lineage. J Neurosci 1994;14:45714587
18. Watkins BA, Dorn H H , Kelly WB, et al. Specific tropism of
HIV-1 for microglial cells in primary human brain cultures. Science 1990;249:549-553
19. Louis JC, Magal D, Muir D, et al. CG4, a new bipotential glial
cell line from the rat brain, is capable if differentiating in vitro
into either mature oligodendrocytes or type-2 astrocytes. J Neurosci Res 1992;31:193-197
20. Wesselingh S, Gough NM, Finlay-Jones JJ, McDonald PJ. Detection of cytokine mRNA in astrocyte cultures using the polymerase chain reaction. Lymphokine Res 1990;9:177-185
21. Tartaglia LA, Ayres TM, Wong GHW, Goeddel DV. A novel
domain within the 55 kd TNF receptor signals cell death. Cell
1993;74:845-853
22. Hodgkinson CA, Moore K, Nakayama A, et al. Mutations at
the mouse microphthalmia locus are associated with defects encoding a novel basic helix-loop-helix zipper protein. Cell 1993;
741395-404
23. Bonifacino JS, Springer TA. Analysis of proteins. In: Ausaubel
FM, Brent R, Kingston RE, et al, eds. Current protocols in
molecular biology. New York: Greene Publishing Associates
Wiley-Interscience, 1991:10.16-10.18
24. Sharpless NE, OBrien WA, Verdin E, et al. Human immunodeficiency virus type 1 tropism for brain microglial cells is determined by a region of the env glycoprotein that also controls
macrophage tropism. J Virol 1992;66:2588-2593
25. Howard OMZ, Clouse KA, Smith C, Goodwin RG. Soluble
tumor necrosis factor receptor: inhibition of human immunodeficiency virus activation. Prod Natl Acad Sci USA 1993;90:
2335-2339
26. Wilt SG, Zhou JM, Wesselingh S, et al. Tumor necrosis factor
alpha derived from human microglia enhances HIV-1 replication and is toxic for rat oligodendrocytes in vitro. In: Major EO,
ed. Technical advances in AIDS research in the human nervous
system. New York: Plenum, 1994 (in press)
27. Yashuda T, Grinspan J, Stern J, et al. Apoptosis occurs in the
oligodendroglial lineage and is prevented by basic fibroblast
growth factor. J Neurosci Res 1994 (in press)
28. Williams K, Bar-Or A, Ulvestad E, et al. Biology of adult human
microglia in culture: comparisons with peripheral blood monocytes and astrocytes. J Neuropathol Exp Neurol 1992;51:538549
29. Williams K, Ulvestad E, Wage A, et al. Activation of adult
human derived microglia by myelin phagocytosis in vitro. J Neurosci 1994;38:433-443
30. Smith R, Baglioni C. The active form of tumor necrosis factor
is a trimer. J Biol Chem 1987;262:6951-6954
3 1. Perez C, Albert I, DeFay K, et al. A nonsecretable cell surface
mutant of tumor necrosis factor (TNF) kills by cell-to-cell contact. Cell 1990;63:251-258
32. Vassalli P. The pathophysiology of tumor necrosis factor. Annu
Rev Immunol 1992;10:411-452
33. Lee SC, Liu W, Dickson DW, et al. Cytokine production by
human fetal microglia and astrocytes: differential induction by
lipopolysaccharide and IL-1 beta. J Immunol 1993;150:26592667
34. Peterson PK, Gekker G, Hu S, et al. Microglial cell upregulation of HIV-1 expression in the chronically infected promonocytic cell line U1: the role of tumor necrosis factor-a. J Neuroimmunol 1992;4 1:81-88
35. Popik W, Pitha PM. Role of tumor necrosis factor alpha in
activation and replication of the tat-defective human immunodeficiency virus type 1. J Virol 1993;67:1094-1099
36. Chao C, H u S, Close K, et al. Cytokine release from microglia:
differential inhibition by pentoxifylline and dexamethasone. J
Infect Dis 1992;166:847-853
Wilt e t al: TNF in HIV-1 Encephalopathy
393
37. Biswas DK, Ahlers CM, Dezube BJ, Pardee AB. Cooperative
inhibition of NF-kappa B and tat-induced superactivation of
human immunodeficiency virus type 1 long terminal repeat.
Proc Natl Acad Sci USA 1993;90:11044-11048
38. Tartaglia LA, Rothe M, Hu Y-F, Goeddel D. Tumor necrosis
factor’s cytotoxic activity is signaled by the p55 TNF receptor.
Cell 1993;7 3:213-216
39. Soliven B, Szuchet S, Nelson DJ. Tumor necrosis factor inhibits
K + current expression in cultured oligodendrocytes. J Memhr
Biol 1991;124:127- 137
40. McLarnon JG, Michikawa M, Kim SU. Effects of rumor necrosis
factor on inward potassium current and cell morphology in cultured human oligodendrocytes. Glia 1993;9:120-126
41. Benveniste EN. Cytokine circuits in brain: implication for AIDS
dementia complex. In: Price RW, Perry SW, eds. HIV, AIDS,
and the brain. New York: Raven, 1994:71-88
42. Rothe J, Lesslauer W, Lotscher H, et al. Mice lacking the tumour
necrosis factor receptor 1 are resistant to TNF-mediated toxicity
but highly susceptible to infection by Lirterza monocytogenes.
Nature 1993; 364:798-802
43. Smith CA, Farrah T, Goodwin RG. The TNF receptor superfamily of cellular and viral proteins: activation, costimulation,
and death. Cell 1994;76:959-962
44. Heller RA, Kronke M. TNF receptor-mediated signaling pathways. J Cell Biol 1994;126:5-9
45. Kolesnick RN, Golde DW. The sphingomyelin pathway in TNF
and 1L-1 signalling. Cell 1994;77:325-328
46. Peudenier SC, Hery C, Montagnier L, Tardieu M. Human microglial cells: characterization in cerebral tissue and in primary
culture, and study of their susceptibility to HIV-1 infection.
Ann Neurol 1991;29:152- 161
394
Annals of Neurology
Vol 37
No 3 March 1995
47. Sebire G, Emilie D, Wallon C, Hery C. In vitro production of
IL-6, IL-beta and tumor necrosis factor alpha by human embryonic microglid and neural cells. J Immunol 1993;150:15171523
48. Toggas SM, Masliah E, Rockenstein EM, et al. Central nervous
system damage produced by expression of the HIV-1 coat protein a 1 2 0 in transgenic mice. Nature 1994;367:188-193
49. Nataf S, Louboutin JP, Chabannes D, Fieve JR. Pentoxifylline
inhibits experimental allergic encephalomyelitis. Acta Neurol
Scand 1993;88:97-99
50. Sharief MK, Hentges R. Association between tumor necrosis
factor-alpha and disease progression in patients with multiple
sclerosis (comment). N Engl J Med 1992;326:272-273
51. Rieckmann P, Albrecht BA, Kitze B, et al. Cytokine mRNA
levels in mononuclear blood cells from patients with multiple
sclerosis. Neurology 1994;44:1523-1526
52. Hofman M, Hinton DR, Johnson K, Merrill JE. Tumor necrosis
factor identified in multiple sclerosis brain. J Exp Med 1989;170:
607-6 12
53. Seimaj K, Raine CS, Canella B, Brosnan CF. Identification of
lymphotoxin and tumor necrosis factor in multiple sclerosis lesions. J Clin Invest 1991;87:949-954
54. Ante1 JP, Williams K, Blain M, et al. Oligodendrocyte lysis by
CD4+ T cells independent of tumor necrosis factor. Ann Neurol 1994;.35:341-348
55. Merrill JE, Ignarro LJ, Sherman MP, et al. Microglial cell cytotoxicity for oligodendrocytes is mediated through nitric oxide.
J Immunol 1993;15 1:2132-2141
56. Wong GHW, Elwell JH, Oberley LW, Goeddel DV. Manganous superoxide dismutase is essential for cellular resistance to
cytotoxicity of tumor necrosis factor. Cell 1989;58:923-931
Документ
Категория
Без категории
Просмотров
8
Размер файла
1 579 Кб
Теги
necrosis, factors, immunodeficiency, encephalopathy, dual, evidence, role, virus, typed, human, vitro, tumors
1/--страниц
Пожаловаться на содержимое документа