| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Centenary Institute of Cancer Medicine and Cell Biology, Royal Prince Alfred Hospital, Camperdown, Sydney, Australia
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
ßTCR+CD2+ T cells
scattered infrequently throughout the CNS parenchyme, 90% of which
were blast cells of donor origin. An unusual clustering of activated
microglia expressing strongly enhanced levels of CD11b/c and MHC class
II was a feature of the GvHD-affected CNS, and despite the paucity of T
lymphocytes present, activated microglial cell clusters were invariably
intimately associated with these T cells. Moreover, 70% of T cells in
the CNS were associated with single or clustered MHC class
II+ microglia, and interacting cells were predominantly
deep within the tissue parenchyme. Approximately 3.7% of the microglia
that were freshly isolated from the GvHD-affected CNS were cycling, and
proliferating cell nuclear Ag-positive microglia were detected in situ.
Microglia from GvHD-affected animals sorted to purity by flow cytometry
and cultured, extended long complex processes, exhibited spineous
processes, and were phagocytic and highly motile. These outcomes are
consistent with direct tissue macrophage-T cell interactions in situ
that lead to activation, proliferation, and expansion of the responding
tissue-resident cell. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Graft-vs-host disease (GvHD) is a systemic, multiorgan condition that follows the transfer of T lymphocytes into immunologically compromised (usually irradiated) recipients, in which major (16) or minor (17) histocompatibility differences exist between donor and host. Few T cells are required to elicit the condition (17), and therefore small numbers contaminating bone marrow or organ grafts are sufficient to cause GvHD in the host. Sporadic reports of the involvement of the CNS in GvHD have appeared. These include an early description of impairment of cerebellar growth in rats with GvHD (18) and CNS T cell infiltration, cerebral white matter degeneration or gliosis, and associated neurologic symptoms in two children with GvHD following allogeneic bone marrow transplantation (19). GvHD induction in rats that had recovered from a single episode of the normally monophasic form of EAE in this species was shown to precipitate a second episode of classical EAE disease signs (20). Finally, Hickey in 1987 (21) demonstrated that, like other tissues in GvHD (16, 22), MHC class II up-regulation also occurred in the CNS with MG thought to be the major cell type involved. Small numbers of T cells were the only obvious infiltrate to the CNS of GvHD-affected animals.
In view of the now-accepted notion that the CNS is normally subject to immunologic surveillance via non-antigen-dependent extravasation of activated T cells (23, 24), the description of a seemingly similar process in GvHD suggests that this disease may be a particularly valid model for exploring the processes involved in normal, noninflammatory T lymphocyte patrolling of the CNS. Furthermore, the interactions occurring between patrolling T lymphocytes and resident tissue cells are of particular interest in relation to up-regulation of molecules involved in Ag presentation, such as MHC class II. These interactions can probably best be investigated in an organ such as the CNS, in which resident lymphocyte numbers are normally minimal and constitutive MHC expression is generally low (reviewed in Refs. 1 and 23).
Here, we have exploited the GvHD model to explore the interactions occurring between CNS-infiltrating T cells and MG. The combination of phenotypic analysis of CNS-isolated MG and leukocytes by flow cytometry (12, 13) and in situ labeling techniques has enabled us to show unequivocally in the adult CNS in vivo that direct interaction between infiltrating semiallogeneic donor T lymphocyte blasts and tissue-resident MG is responsible for subsequent MG activation and proliferation, as well as MHC class II expression.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Eight- to ten-week old (Lewis x brown Norway)F1 (F1, MHC RT1l/n) rats were obtained from the Animal Resource Centre (Perth, Australia) and housed in the Centenary Institute animal facility under specific pathogen-free conditions. GvHD was induced in F1 animals by exposure to 2 Gy of gamma irradiation (Cs137) and injecting them i.v. with 1.5 to 2 x 108 viable Lewis strain splenocytes. Recipient animals were monitored for weight loss and other clinical signs and symptoms of GvHD, such as dermal erythema, coat condition, and general physical constitution.
Abs and reagents for flow cytometry
Mouse mAb specific for rat cell surface markers were MRC OX1
(anti-CD45 monomorphic (25)), MRC OX6 (anti-MHC class II, RT1B
(I-A) (26)), MRC OX18 (anti-MHC class I, monomorphic (27)), MRC
OX26 (anti-CD71, transferrin receptor (28)), MRC OX27 (anti-MHC
class I polymorphic (29)), MRC OX34 (anti-CD2 (29)), MRC OX42
(anti-CD11b and -CD11c (30, 31)), and W3/25 (anti-CD4 (29, 32)). Hybridomas of R73 (anti-ßTCR (33)), V65
(anti-
TCR (34)), and 10/78 (anti-NKR-P1A, (35)) were
kindly supplied by Prof. Thomas Hünig, Würzburg, Germany.
MRC OX21 (mouse anti-human C3bi and negative on rat cells (36)),
K1-21 (mouse anti-human free 
light chain, non-cross-reactive
with rat Ig (37)) and mouse anti-human CD4-PE (Serotec, Oxford,
U.K.) were used as negative controls. Mouse antiproliferating cell
nuclear Ag (PCNA, clone 19F4) was from Boehringer Mannheim Biochemica,
Mannheim, Germany. For direct staining, mAb were purified from ascites
or tissue culture supernatant, then conjugated to either FITC or biotin
as described (38). PE-conjugated OX1 (anti-CD45) or R73
(anti-ßTCR) were from PharMingen, San Diego, CA. For flow
cytometry in which mouse mAb tissue culture supernatants were used, mAb
were detected with rat-absorbed FITC-conjugated sheep
F(ab')2 anti-mouse Ig (Sigma Chemical, St. Louis, MO).
Biotinylated mAb were detected with either streptavidin-PE (Caltag, San
Francisco, CA; two-color flow cytometry) or
streptavidin-allophycocyanin (Molecular Probes, Eugene, OR; three-color
flow cytometry).
Cell isolation and flow cytometry
MG and other CNS-associated cell populations, including infiltrating leukocytes, were isolated from the thoroughly perfused rat CNS as detailed (12, 13) and immediately labeled with mAb plus fluorochromes. Two-color analysis was performed on a FACScan (Becton Dickinson, San Jose, CA) while three-color analysis or sorting was executed with a dual-laser FACStarPlus (Becton Dickinson) as described (13). Analysis of DNA content for assessment of cell cycle was achieved using the method of Telford (39, 40) with modifications. Briefly, isolated CNS cells were stained for CD45 or CD11b/c, then fixed in 1% paraformaldehyde (4°C, 1–2 h) followed by 70% ethanol (-20°C, 12–24 h). After washing in PBS, cells were resuspended in propidium iodide (PI) staining reagent (0.1% triton X-100, 0.1 mM EDTA, 50 µg/ml RNase (Boehringer Mannheim Biochemica)) and 50 µg/ml PI (Calbiochem, La Jolla CA) in PBS. Stained samples were analyzed by FACScan 0.5 to 1 h later.
Microglial cell culture
Highly purified MG from normal or GvHD-affected rat CNS were cultured on tissue culture grade plastic (Costar, Cambridge, MA) in RPMI 1640, 5% FCS. Cultures were maintained in a 5% CO2 atmosphere at 37°C. The phagocytic activity of activated MG was assessed by adding 1 µm latex FluoSpheres (Molecular Probes) to the 48-h cultures. Free beads were washed out with warm medium after 4 h, and cells were maintained as before.
Immunohistologic analysis
Single- and dual-color analysis for leukocyte cell surface markers
and vascular endothelium.
The spinal cords were removed from normal F1 and
GvHD-affected F1 rats, 5 mm blocks of tissue were snap
frozen in OCT compound, and 5-µm cryostat sections were cut and air
dried. Sections for single labeling were fixed in 100% ethanol
(-20°C, 30 min), then stained via the immunoperoxidase technique as
described previously (41). Standard dual-color immunohistochemical
labeling involved sequential addition of unconjugated mAb
(anti-ßTCR, CD2, NKR-P1A, transferrin receptor, at 20°C for
1 h), alkaline phosphatase-conjugated anti-mouse IgG (Sigma
Chemical, 20°C, 30 min), and the (blue) substrate
5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (8). This
was followed by addition of biotinylated anti-CD11b/c or MHC class
II mAb in the presence of 10% normal mouse serum (20°C, 1 h),
biotin/streptavidin-horseradish peroxidase complex (Dako, Carpinteria,
CA; 20°C, 30 min), and the (brown) substrate diaminobenzidine (Sigma
Chemical).
In vivo labeling of CNS vasculature. The luminal face of CNS vascular endothelial cells is constitutively positive for the transferrin receptor (28) as well as for MHC class I. Thus, to precisely localize the CNS vasculature, GvHD-affected rats were lightly anesthetized and injected i.v with a mixture of 0.5 ml ascites of MRC OX26 mAb (anti-rat transferrin receptor) and 2.0 ml ascites of MRC OX18 mAb (anti-rat MHC class I). Thirty minutes later, recipient rats were killed by CO2 overdose, perfused via the ascending aorta with PBS/0.2% (w/v) BSA/2 IU heparin-sodium (Fisons Pharmaceuticals, Rochester, NY), and CNS tissue was removed and frozen and sections cut therefrom. Bound mAb on sections was revealed by the addition of alkaline phosphatase-conjugated anti-mouse IgG and substrate, as described above.
Dual-color staining for PCNA and microglia. Cryostat sections for intranuclear PCNA and cell surface CD11b/c were fixed in 1% paraformaldehyde (20°C, 2 min), then methanol (-20°C, 10 min) and finally permeabilized in 0.2% triton X-100 (20°C, 1 min). Sections were incubated sequentially with anti-PCNA mAb or the appropriate control mAb (20°C, 1 h), peroxidase conjugated anti-mouse IgG (Dako, Glostrup, Denmark; 20°C, 30 min), and metal-enhanced diaminobenzidine substrate (Pierce, Rockford, IL) for 15 min at 20°C. After washing, biotinylated anti-CD11b/c mAb was added to the sections (20°C, 1 h); then, biotin/streptavidin-alkaline phosphatase complex (Dako, 20°C, 30 min) and substrate were added.
Detection of apoptotic T cells in situ. Rats were killed and perfused first with 150 ml cold PBS-BSA-heparin, followed by 250 ml 4% (w/v) paraformaldehyde in PBS. Spinal cords were postfixed for 4 h in the same fixative, then through increasing concentrations of sucrose (10, 15, and 20%, w/v, in PBS), each for 12 h. Pieces of tissue were frozen and cryostat sections prepared as above. Localization of apoptotic T cells was achieved by dual staining for CD2 and TUNEL-positive cells (In Situ Cell Death Detection Kit, peroxidase, Boehringer Mannheim Biochemica); to localize microglia, serial sections were stained for MHC class II. Both the CD2 and MHC class II Ags were retained at high levels despite fixation. TUNEL-positive cells were revealed according to the manufacturer’s instructions using FITC-dUTP, terminal transferase, and peroxidase-conjugated anti-FITC Ab. Signal enhancement was achieved by adding biotinylated tyramine (TSA-indirect kit, NEN, Boston, MA), followed by biotin/streptavidin-horseradish peroxidase complex and the (red) substrate AEC (3-amino-9 ethylcarbazole, Sigma Chemical). There was no staining when terminal transferase was omitted. T cells were localized on the same sections by the subsequent addition of anti-CD2 mAb, alkaline phosphatase-conjugated anti-mouse IgG (Sigma Chemical, 20°C, 30 min), and (blue) substrate.
All sections were mounted permanently by adding CrystalMount (Biomeda, Foster City CA) and baking at 75°C for 10 min.
Photomicroscopy
Micrographs of immunohistology were generated on a Leitz DMRBE
microscope utilizing transmitted light interference optics (Figs. 3
and 6) or standard bright field optics (Fig. 4). Micrographs of cultured
cells (Fig. 7) were produced with a Leitz DMIL inverted microscope and
phase contrast optics.
|
|
|
|
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MG are parenchymal, radiation-resistant, long-lived cells that
remain of the host type in radiation bone marrow chimeras (13, 42) and
are distinct from other CNS macrophages that are radiation-sensitive,
predominantly associated with vessels, and present in a number of sites
in the CNS including the meninges and the choroid plexus (43). Included
in the latter broad definition are classic perivascular macrophages of
the CNS, present in the perivascular space of deep vessels and
sometimes called perivascular microglia or perivascular cells (reviewed
in 1 . Adult CNS MG can be distinguished by flow cytometry from
all other macrophages, since MG express lower levels of CD45 (12).
Thus, macrophage lineage cells recovered from the perfused normal CNS
(Fig. 1A, day 0) were divided
into parenchymal, fixed-tissue macrophages (the MG cell) exhibiting low
CD45 expression (population 1a) and other CNS macrophages that
expressed normal (high) CD45 levels (population 2). A small number of
other CD45high cells were also isolated (population 3),
50% of which were T cells (see below). Few MG express MHC class II
in normal (LEW x BN)F1 rats (Ref. 8 and Fig. 1B, arrow); most MHC class II-expressing cells lie within
the CD45high cell populations.
|
A,
population 1b) and there was a very slight shift in the MHC class II
expression of MG (Fig. 1B, arrow), but no obvious increase
in cell recovery (Table I). Increased
forward and side scatter of the cells was further evidence of subtle MG
activation at this time (Fig. 1C). Population 3 (Fig. 1A), many of which were T cells (not shown), was
increased relative to other populations. By day 10 and 14, further CD45
up-regulation had occurred on MG (Fig. 1A, populations 1c
and 1d); all were MHC class II+ and had increased
considerably in volume (Fig. 1C). Almost six times
more MG were recovered than from the normal CNS (Table I). Population 3
cells increased also, peaking at day 10 (see below and Fig. 2 for further characterization). The
differences in MG MHC class II expression between day 10 and day 14
shown here were not consistent. Generally, MHC class II expression was
maximal by day 10 and reached a level, on average, twofold lower than
CD45high cells in the normal CNS (Fig. 1B, d0).
Note the loss of some population 2 cells (Fig. 1A)
and CD45highMHC class II+ cells (Fig. 1B) after day 7, possibly reflecting the radiation of
the F1 recipient preparatory to induction of GvHD.
|
|
The origin of the CD45highCD11b/c-ve
cells accumulating in the CNS of GvHD-affected rats (Fig. 1A, population 3) was determined by flow cytometry. The few
T cells present in the normal CNS (Fig. 2A, population 1) as
well as other CD45high leukocytes (Fig. 2A,
population 3), many of which were probably CD45high CNS
macrophages, were MRC OX27+ as expected. Consistent with
the low MHC expression of normal CNS glia, MG (Fig. 2A,
population 2) ranged from MRC OX27-negative to weakly positive. The
majority (90%) of T cells in the CNS of GvHD animals (Fig. 2B, population 1) were, in contrast, MRC
OX27-ve and thus derived from donor splenocytes. These
cells were blasts (Fig. 2B), while the few T cells
from the normal CNS were not (Fig. 2A). Previously
MHC class I-ve/low MG uniformly up-regulated MHC class I
and virtually all were MRC OX27+ (Fig. 2B,
population 2). Finally, the
CD45highßTCR-ve cells (Fig. 2B, population 3) also up-regulated MHC class I expression
relative to the normal CNS. It is assumed that at least a proportion of
these cells are CD45high CNS macrophages and so these too
may be activated in GvHD in a manner similar to MG. The majority of
population 3 expressed host MHC class I (MRC OX27+), while
a minor proportion (5–7%) of this population in GvHD-affected CNS
were MRC OX27-ve and thus derived from donor splenocytes.
While not further defined here, these cells could, in principle,
include splenic macrophages but appear not to be NK cells, based on
absence of NKR-P1A-positive cells in the GvHD-affected CNS (see
below).
Infiltrating T cells associate with activated microglial clusters within the CNS parenchyme
MG in the normal CNS were detected readily with anti-CD11b/c
mAb (Fig. 3A). The
cells were distributed throughout the white and gray matter, exhibiting
a typical ramified morphology with fine processes. Little MHC class II
was detectable (Fig. 3C) and few T cells (arrow, Fig. 3E), and CD4 expression was observed only
occasionally on rare T cells or at low level on ramified MG (not
shown). In similarly stained CNS from GvHD-affected rats, clusters of
strongly CD11b/c+ (Fig. 3B) and MHC class
II+ cells (Fig. 3D) were observed.
Additionally, most other MG not in clusters were activated as evidenced
by the reduction in number and length of cell processes (Fig. 3B) and extensive MHC class II expression throughout
the parenchyme (Fig. 3D). Apart from the clustered
MG, CD11b/c expression on the majority of cells in GvHD was no greater
than in the normal CNS, consistent with the flow cytometry data (Fig. 1A). The small numbers of ßTCR+ T
cells that were detected were dispersed rather than in obvious
perivascular accumulations (Fig. 3F), and the
majority of them were CD4+ (Fig. 3H). MG
cells expressing CD4 were now clearly visible (arrows, Fig. 3H).
To determine the relationship between infiltrating T cells and
activated MG, tissue sections from GvHD-affected CNS were double
labeled for T cells and MG, the latter by staining for CD11b/c (all MG)
or MHC class II (activated MG). Preliminary analysis of the best
reagents for T cell detection revealed ßTCR expression to be
variable and often weak, consistent with the activated nature of these
cells (Fig. 2B). CD2 expression in contrast was
stronger, although similar numbers of ßTCR+ and
CD2+ T cells were detected. The GvHD-affected CNS was
negative when stained with NKR-P1A (NK cell) or TCR-specific mAb.
Thus, only ßTCR+ T cells entered the CNS in GvHD, and
CD2 was, in this case, a useful marker for these cells.
Activated MHC class II+ MG clusters (two or more MG) were
almost invariably associated with one or occasionally two T cells;
examples of this are given at low and high power in Figure 4, A–D. A single T cell is
centered in the MG cluster in A and revealed at higher power
in B. There were many isolated MHC class II+ MG
with no associated T cell. Similarly, while the association of MG
clusters and T cells was virtually absolute, T cells were not
necessarily always associated with MG (e.g., Fig. 4A), although the frequency of association was high.
Analysis of the relative frequency of association of an
ßTCR+ T cell or CD2+ T cell with MG in the
GvHD-affected CNS showed that 45% of all T cells were associated
with a single MG and 25% of T cells with an MG cluster, while only
30% of T cells were in complete isolation (Fig. 5). Significantly, comparable results
were obtained by analysis of T cell-MG association using CD11b/c or MHC
class II for MG detection, indicating that where there was close
association with a T cell and a MG; the MG was invariably activated. It
is also apparent that most of the observed T cell-MG interactions
involved donor T cell blasts semiallogeneic with the host, given that
the majority (>90%) of T cells within the GvHD-affected CNS were
donor (Lewis, RT1l) as defined by flow cytometry (Fig. 2B). It is possible that those 30% of T cells not
associated with MG (Fig. 4A and 5) were enriched for host
(F1) T cells. However, immunohistologic examination of
these T cells was not feasible in the current experimental system, as
the only marker distinguishing host and donor T cells was BN
(RT1n) MHC class I. The latter was also expressed by host
CNS tissue in which MHC class I was extensively expressed, particularly
on MG (Fig. 2B).
|
E, arrow). Colocalization of MG clusters,
vasculature, and T cells was enabled by staining of the vasculature in
vivo by i.v. injection of anti-transferrin and MHC class I mAb
before the animals were killed to obtain tissue for
immunohistochemistry. Sections were stained for T cells (CD2) and
CD11b/c, and mAb bound to the vasculature was additionally revealed
during the staining procedure. CD11b/c+ clusters were again
found to be dissociated from the vasculature, as were the T cells
associated with these MG clusters (Fig. 4F). Note
also that vascular staining was discrete, remaining essentially
luminal, indicating that injected mAb did not cross into the tissue, a
finding consistent with the maintenance of the blood-brain barrier in
GvHD (45). Microglia proliferate in GvHD
Whether increased MG recoveries (Table I) and MG clusters (Figs. 3 and 4) reflected cellular proliferation was determined. The majority of
cells isolated from the normal or GvHD-affected CNS were nondividing
cells in the G1 phase of cell cycle as expected. A
basal level of 2.5% of all cells (Fig. 6A) and 1% of the
MG (Fig. 6B) was proliferating in the normal CNS. The
latter figure is almost certainly an overestimate, representing the
background level of the assay. In the GvHD-affected CNS at day 10,
7.5% of total cells and 3.5% of the MG were proliferating, falling
thereafter to basal levels by day 14. Although the relative percentage
of the MG population dividing at any one time was low, the increase
above background was significant. Importantly, cell division was
detected by day 7, preceding increased cell recovery from the CNS, a
finding consistent with the possibility that cellular proliferation
contributed to enhanced recoveries. Labeling of isolated cells with
CD11b/c and PI revealed that, on average, 4.4% of
CD11b/c-ve cells at day 10 (some of which were T cells;
see Fig. 1A, population 3, and Fig. 2) contained cells
in S/G2/M (data not shown).
PCNA is an auxiliary protein for DNA polymerase-, forming complexes
with D-type cyclins, which are candidate G1 cyclins in
mammalian cells (46). PCNA is required both for DNA replication and DNA
repair and increases from basal levels during the cell cycle. In the
GvHD-affected rat CNS, MG (CD11b/c+) expressing nuclear
PCNA were detected at day 7 (Fig. 6C) and day 10 (not
shown). The majority of positive cells were found in clusters and only
occasionally as single cells. Most MG were PCNA-ve,
consistent with the DNA content data of Figure 6B. In the
normal CNS, occasional PCNA+ cells were detected in the
meninges, in the immediate perivascular region around deep vessels, and
in ependymal cells lining the spinal canal, but in contrast to GvHD,
never in the parenchyma.
Apoptotic T cells in the GvHD-affected CNS
Apoptotic T cell death is a feature of the inflamed CNS, particularly in experimental autoimmune encephalomyelitis. Myelin Ag-reactive CD4+ T cells are more frequently involved than those T cell reactive with non-CNS Ags, indicating that Ag recognition within the CNS may be important in the apoptotic process (reviewed in 47 . To examine the possibility that T cells in the GvHD-affected CNS similarly die by apoptosis and, if so, whether this is associated with interaction with MHC class II+ MG, cryostat sections were dual labeled for CD2 and TUNEL, with serial sections stained for MHC class II to identify activated MG.
Of the T cells identified in the GvHD-affected CNS by CD2 staining
(200 T cells assessed on five separate sections taken on day 8–11),
only very few (<5%) were also TUNEL-positive, indicating either that
few were dying or that apoptotic cells were removed (engulfed) rapidly
and thus not detectable. Moreover, of the few CD2+
TUNEL+ cells observed, no consistent association with MHC
class II+ MG stained on serial sections was observed—thus,
providing an inconclusive result. Some CD2-negative, TUNEL-positive
nuclei were also observed, although the nature of these cells was not
determined (data not shown).
In vivo-activated microglia are adherent, highly processed, phagocytic, and motile in vitro
The growth characteristics of MG in vitro following in situ
activation in GvHD were determined. Only a minority (usually <20%) of
highly purified MG sorted from the normal CNS adhered to plastic
culture wells, extending small, usually bipolar, blunt processes (Fig. 7A). These processes
generally occurred within 12 h and were maximal within 24 to
36 h. Fewer (5–10%) of the MG elaborated complex processes
compared with their in vivo counterparts. After 36 to 48 h, most
cells had retracted processes and were nonadherent but remained viable.
The addition of various growth and differentiation factors
(granulocyte-macrophage CSF and macrophage CSF, for example) had no
obvious effect on cell morphology or growth characteristics.
The majority (50–70%) of GvHD-activated MG, in contrast, adhered
rapidly to plastic (Fig. 7B) and extended processes
that became more complex with time (Fig. 7C). By day
5 in vitro, many cells were motile and exhibited classical unipolar
migration, with lamellipodia and a trailing tail of cytoplasm (see
examples in Fig. 7, D and E). Cells were
observed passing over neighboring cells. The presence of fine spines on
the MG—already apparent at 12 h and seen clearly at 48 h at
high power (day 5; Fig. 7, D and E)—was a
feature of these cells. Activated MG remained adherent and viable in
vitro for extended periods (>3 wk). When tested at 48 h,
GvHD-activated MG were highly phagocytic (Fig. 7, F and
G). There was some relationship between the number of
beads internalized and cell morphology, with increased bead load (Fig. 7G) resulting in adoption of a more rounded, less
processed "ameboid" or reactive form (48).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Here, a GvHD model was employed to assess MG-T cell interactions in an
environment more typical of that occurring during normal immunologic
surveillance of the CNS by activated T cells. A number of aspects of
this process are illustrated. First, the capacity of T cells not
specifically reactive with CNS Ag (in this case, semiallogeneic T
cells) to cross the blood-brain barrier and move into the tissue (23, 24) is clearly demonstrated (
Figs. 2–4). The studied T cells were
blasts (Fig. 2) and accounted for 90% of T lymphocytes in the
GvHD-affected CNS. Significantly, few other inflammatory leukocytes
(CD45high), monocytes in particular but also NK cells
and T cells, accompanied the effector ß T cells into the
CNS (Fig. 1), and injected mAb did not leak into the CNS parenchyme
(Fig. 4F). Collectively, these results are consistent
with an earlier study (45) showing that the integrity of the
blood-brain barrier is maintained in GvHD and also arguing against the
possible involvement of systemic soluble factors, such as cytokines
produced during GvHD, in the initial MG activation process. Second, the
possibility that MG in situ can be activated directly by infiltrating T
cells receives support here, and the proliferative potential of MG
after T cell interaction is demonstrated. Within 7 to 10 days after
transfer of semiallogeneic splenocytes, single, activated microglia or
small clusters of MG were apparent, and almost invariably, these were
associated with a single T cell or occasionally a second T cell (Fig. 4). By day 10, virtually all MG were MHC class II+ (Figs. 1 and 3), with MG clusters still apparent (Figs. 3 and 4), presumably the
product of an initial T cell-MG interaction that was shown to occur at
high frequency (Fig. 5). Importantly, the parenchymal distribution of
activated MG, particularly MG clusters and their association with T
cells (Fig. 4), is highly suggestive of an MG activation process
initiated by infiltrating T cells. The converse possibility, that MG
preactivated by systemic factors attracted the T cells, seems unlikely,
as localization of activated MG and associated T cells predominantly
around vessels rather than throughout the CNS parenchyme would be
expected in this case.
Thus, the experimental outcomes are consistent with the genesis of a MG response that emanates from localized interactions to encompass the entire organ. The eventual MHC class II expression by all MG in the GvHD-affected CNS clearly cannot be a result of direct T cell activation, and other signals subsequent to initial T cell involvement must account for the spreading MG response. Soluble factors within the CNS, such as neurotrophins within the nerve growth factor gene family, may be relevant in this context (49) .
Substantially increased numbers of MG were recovered from the
GvHD-affected CNS (Table I). There are three potential mechanisms that
may account for this. 1) "Precursor" cells, derived from blood,
could have entered the CNS along with activated T cells and adopted a
MG phenotype. This is unlikely in view of the demonstration elsewhere
in long-established radiation bone marrow-chimeric animals (13, 42, 43)
that most MG remain of the host type and the evidence here (Fig. 2B, population 2) that in GvHD, all MG expressed MHC class I
of the host type. In contrast, a proportion of
ßTCR-ve CD45high cells in the
GvHD-affected CNS (Fig. 2B, population 3) were negative for
the host MHC class I marker and thus derived from donor splenocytes. 2)
Activated MG may more readily be extracted due to changes in cellular
architecture and interaction with other cells in the local
microenvironment. As noted, MG in the GvHD-affected CNS tended to have
slightly shorter processes, adopting a more ameboid form (Fig. 3B). Thus, this possibility cannot be excluded and
may have contributed to enhanced cell yields to some extent. 3)
Finally, upon T cell-induced activation, MG not only up-regulated
surface molecules (MHC, CD45, CD4) and increased in size (Fig. 1C), but also proliferated.
There exists good evidence that MG activated by nerve transection and
toxin-induced neuronal death (3), ischemia (50), or IFN- injection
(51) divide in vivo, expressing cyclin D1 during the onset of
proliferation (50). However, the normal presence of inflammatory
macrophages in most T cell-mediated immunopathologies of the CNS (52),
as well as the difficulties involved in distinguishing between these
cells and MG, has resulted in a lack of consensus regarding
immunologically driven MG proliferation in the adult CNS. Here, two
independent cell cycle analyses were employed to demonstrate that MG
proliferation may indeed follow T cell infiltration in GvHD.
Furthermore, MG proliferation seems to be triggered by direct
interaction with extravasating T cells. While the proportion of MG in
the cell cycle was low (Fig. 6B), it is accepted that
both PI and PCNA analyses provide only a snapshot of the process and,
for example, while only one PCNA-positive cell was detected in a
cluster (Fig. 6C), it seems likely that other cells
within the cluster had divided previously. The existence of
proliferating cells within MG clusters also provides evidence that such
clustering could have followed clonal expansion of MG, rather than
movement of MG toward a given T cell, and formation of a cellular
aggregate. It is tempting to speculate that the MG clusters so typical
of Alzheimer’s disease (4) also result from similar localized MG
proliferation, although widespread astrocytic proliferation has also
been observed in the Alzheimer-diseased brain (53).
The best studied MG preparations morphologically are those derived from
neonatal rodent brain (54, 55) rather than from the adult CNS. The
purification of these cells from mixed brain cell culture depends on
differential adherence properties and requires extended culture times.
The use of these cells is justified on the basis of morphologic
evidence that hemopoietically derived MG precursors seed into the CNS
during gestation and develop over some weeks after birth, from what
appears to be a normal tissue macrophage, to become ramified, fully
differentiated MG (56). Sorting MG from the GvHD-affected CNS and
culturing them for only short periods produced MG that were adherent,
highly processed, motile, and phagocytic (Fig. 7), all properties
described for "normal," nonactivated neonatal MG in vitro (55, 57).
In the inflamed (EAE-affected) adult CNS as well, phagocytic MG have
been observed (58). Notably, normal adult CNS MG (Fig. 7A) exhibited few of these properties; so in this
regard, neonatal CNS-derived MG appear to represent an already
activated morphologic phenotype that is atypical of resting adult
microglia. As a note of caution, Giulian and colleagues (59) have
demonstrated that FCS is inhibitory in MG cultures, at least for MG
derived from neonatal CNS. This could account for the incapacity of
sorted, normal CNS adult MG to adhere and become processed in vitro in
the present study (Fig. 7A). However, once activated
in vivo during GvHD, MG readily adhered to plastic, processes were
formed, and cells became motile, despite the presence of FCS (Fig. 7, B–G).
A notable feature of sorted MG in culture was the projection of spiny
processes (Fig. 7). This appears to be a feature unique to MG, not
observed with any other type of tissue-derived macrophage (59).
GvHD-activated MG in culture retained a ramified, processed, and
spinelike profile unless involved in phagocytosis (Fig. 7, F
and G) when spines were retained but a more rounded
or ameboid form was adopted.
Many tissue-resident cells (60) can be induced to express MHC class II
when affected by inflammatory or patrolling T lymphocytes, although
their role in promoting T cell clonal expansion in vivo is unclear. It
is apparent, for example, that there is minimal autoreactive CD4 T cell
proliferation in the CNS in EAE, despite the presence of a network of
MHC class II+ MG (61), and that Ag-reactive T cells (those
responding to myelin basic protein) preferentially die by apoptosis in
EAE (47, 62). The demonstration here of a direct, immunologically
mediated process of MG activation and MHC class II up-regulation,
coupled with the ability to obtain these cells at high purity from the
adult CNS, provides an ideal vehicle for analysis of the role of
resident MG in regulation of T cell responses within the CNS. Thus, we
demonstrated recently that interaction in vitro between GvHD-activated
MHC class II+ adult CNS MG and myelin basic
protein-reactive CD4+ T cells in the presence of Ag induced
only partial activation of T cells, resulting in an increase in T cell
volume but no proliferation, the secretion of TNF and IFN- but not
IL-2, and T cell death by apoptosis (63). Other CNS-derived
macrophages, in contrast, supported T cell proliferation and survival.
These observations may explain, at least in part, the incapacity of
CD4+ T cells to proliferate within the CNS and their
propensity to die there.
In the present study, localized MG activation and proliferation
followed stimulation by infiltrating T cells. However, the possibility
arose, in view of our in vitro data, that there may be a reciprocal but
negative signal delivered back to the T cell, that results in apoptotic
death. Attempts to observe this in vivo were unsuccessful (see
Results); one reason for this may be that activated T
cells in GvHD do not die in situ within the CNS as they do in EAE and
MS, for example. Nevertheless, the significance of this observation
should not be overinterpreted in view of the infrequency of detectable
TUNEL-positive T cells (which may be explainable by the rapid removal
of apoptotic T cells directly by interacting MG), coupled with the
relatively small number of T cells in the GvHD-affected CNS that can be
enumerated using histologic analysis. Alternative techniques, using
flow cytometric analysis of T cells derived from the GvHD-affected CNS,
are being established to determine the incidence of T cell apoptosis in
the CNS in GvHD. It will also be possible in these studies to ascertain
the activation state of T cells within the GvHD-affected CNS, an issue
of particular interest in view of the close and high frequency
association of T cells with MG. For example, a predominance of only
partially activated T cells expressing enhanced CD25 and MRC OX40, with
production of some cytokines (IFN- and TNF) but not others (IL-2),
would provide an invaluable in situ correlate for the consequences of T
cell-MG interactions described to date only in vitro (63).
Furthermore, the observations on MG activation reported here are currently being translated to an in vitro model in which the outcome of interactions between normal vs in vivo-activated MG and activated allogeneic or protein-Ag-specific (e.g., myelin protein) T cells can be assessed more precisely. This will enable a direct examination of the signals necessary for MG activation and proliferation following MG-T cell interaction.
|
Footnotes |
|---|
2 J.D.S. and A.L.F. made an equal contribution to this study.
3 Address correspondence and reprint requests to Dr. Jonathon D. Sedgwick, Centenary Institute of Cancer Medicine and Cell Biology, Building 93, Royal Prince Alfred Hospital, Missenden Road, Camperdown, Sydney NSW 2042, Australia. E-mail address:
4 Abbreviations used in this paper: CNS, central nervous system; MG, microglial cell; EAE, experimental autoimmune encephalomyelitis; GvHD, graft vs host disease; F1, (Lewis x brown Norway)F1 rat strain; PCNA, proliferating cell nuclear Ag; PI, propidium iodide; PE, phycoerythrin; TUNEL, terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end labeled.
Received for publication September 3, 1997. Accepted for publication February 3, 1998.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-treated microglial cells. Eur. J. Immunol. 17:1271.[Medline]
, T cells in sections costained for
CD11b/c+ MG; 
, T cells in sections costained for MHC
class II. A, T cells in isolation with no association with
CD11b/c+ MG or an MHC class II+ cell.
B, T cell associated with a single CD11b/c+ MG
or single MHC class II cell with MG morphology. C, T cell
associated with clustered (two or more) CD11b/c+ MG or MHC
class II+ cells with MG morphology.