| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
*
Department of Microbiology and Immunology, The Medical School, Newcastle, United Kingdom;
Unité de Biologie Moléculaire du Gène, Institut National de la Santé et de la Recherche Médicale U277, Institut Pasteur, Paris, France;
Transplantation Biology Unit, Medical Research Council Clinical Sciences Centre, London, United Kingdom;
§
Department of Neurology, The Medical School, Newcastle, United Kingdom; and
¶
Institute for Cancer Studies, The Medical School, Sheffield, United Kingdom
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In humans, two sets of inhibitory receptor have been identified. One set belongs to the Ig superfamily and recognizes determinants found on classical class I molecules (5). The second set belongs to the C-type lectin family and comprises an invariant CD94 chain covalently linked to a member of the NKG2 family (6). CD94/NKG2 receptors were initially thought to recognize determinants shared by many classical class I molecules, but it has now been established that they recognize the nonclassical HLA-E molecule whose expression at the cell surface is largely dependent on its association with a peptide that is liberated by intracellular proteolytic digestion of the leader sequence of certain classical class I molecules (7, 8, 9).
Until recently the only class I receptors known to be expressed on murine NK cells were those belonging to the Ly49 family (10, 11). Those members of the Ly49 family that have been studied in detail have been found to recognize particular classical class I molecules and to be expressed on overlapping subpopulations of cells, raising important questions as to how a complex and diverse repertoire of functional self-tolerant NK cells is created during ontogeny. It is currently believed that during a critical stage in their development, NK cells activate Ly49 genes at random from the entire set of genes available on both homologous chromosomes. The resulting immature NK cell repertoire is then acted upon by a selection process that results in the survival only of those NK cells that express at least one inhibitory receptor for self class I molecules (12, 13, 14).
Understanding the nature of these events would be greatly facilitated if it were possible to study the development of NK cells in vitro from individual precursors. We have recently developed a system in which immature progenitor cells from day 14 fetal thymus (15) or liver (16) can develop into lines and clones of NK cells in vitro. The NK cells that arise in this system are phenotypically and functionally similar to adult splenic NK cells with two striking exceptions, namely that they are severely deficient in the expression of Ly49 molecules and show only limited ability to discriminate between class I-sufficient and class I-deficient targets (17). Thus, most fetal NK lines lyse syngeneic blast cells almost as well as blasts prepared from class I-deficient mice. Curiously though, all lines and clones of fetal NK cells that we examined gave strong lysis of a ß2-microglobulin (ß2m)3-deficient mutant of EL4 tumor cells and of a TAP2-deficient mutant of RBL5 tumor cells, but little or no lysis of the corresponding wild-type cells. These results strongly suggest that fetal NK cells express inhibitory class I receptors distinct from those previously characterized. One possibility would be that they express novel member(s) of the Ly49 family. In support of this we found that, although fetal NK cell lines generally lacked detectable levels of most Ly49 molecules, both at the cell surface and at the RNA level, they expressed high levels of mRNA for Ly49E (17). Alternatively, they may express inhibitory receptors homologous to those found in man, or ones of a novel nature.
The recent demonstration that adult mouse NK cells express inhibitory CD94/NKG2 receptors that recognize Qa1 (18, 19), the mouse homologue of HLA-E, led us to investigate the expression of receptors for Qa1 on fetal NK cells. Our studies reveal that, during their development in vitro, a proportion of C57BL/6 fetal NK cells do indeed acquire receptors for a soluble form of Qa1b bearing the Qdm peptide (hereinafter referred to as Qa1R), but not for soluble forms of syngeneic classical class I molecules. The acquisition of these Qa1R is under strict cytokine control, occurs in a stochastic manner, and correlates with the limited ability of fetal NK cells to distinguish between wild-type and class I-deficient blasts. However, these Qa1R do not account for the striking ability of fetal NK cells to discriminate between class I-sufficient and class I-deficient forms of certain tumor cells, suggesting the existence on NK cells of novel inhibitory receptors for structures of an unknown nature involving nonclassical class I molecules.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Culture reagents were purchased from Life Technologies (Paisley, U.K.) and Sigma (Poole, U.K.). Human rIL-2 was obtained from Cetus (Emeryville, CA). Mouse rIL-4 was obtained as the supernatant of X6310 cells transfected with the mouse IL-4 gene (20), kindly provided by Dr. F. Melchers (Basel Institute for Immunology, Basel, Switzerland).
Adult and fetal NK cells
Adult NK cells were prepared as described previously (17). Fetal NK cells were generated as described in detail elsewhere (15, 21). Briefly, thymocytes were prepared from day 14 embryos of timed-mated C57BL/6 mice (day of vaginal plug = day 0), and, unless stated otherwise, were cultured for 1–3 days in medium containing 10 U/ml IL-4 and 10 ng/ml PMA, then transferred to medium containing 104 U/ml IL-2.
CTL clones
The Qdm-dependent Qa1-specific clone e1d (22) and the H-Y-specific clone CTL10 ((23); kindly provided by Dr. D. Roopenian) were maintained by periodic stimulation with irradiated spleen cells from male C57BL/6 mice in medium containing 100 U/ml IL-2. After 1 wk of growth, cells were transferred to medium containing 5 U/ml IL-2. Cells were used in experiments while growing in 100 U/ml IL-2.
Immunofluorescence and flow cytometry
Aliquots of 1–3 x 105 cells were
stained with combinations of FITC PK136 anti-NK1.1, PE 2C11
anti-CD3
, and PE or Red670 tetrameric class I molecules. Abs
were obtained from PharMingen (San Diego, CA) or Cambridge Bioscience
(Cambridge, U.K.). Soluble tetramers of Qa1b
(refolded with the Qdm peptide, AMAPRTLLL) and Kb
(construct kindly provided by Dr. J. D. Altman, Emory University,
Atlanta, GA; refolded with the OVA peptide, SIINFEKL) associated with
Red670 streptavidin (Life Technologies) and of Db
(refolded with the H-Y peptide, WMHHMLDI) associated with
PE-streptavidin were generated as described elsewhere
((18); D. Sourdive et al. in preparation). Staining was
analyzed on a FACScan (Becton Dickinson, San Jose, CA), using forward
and side scatter to gate on single viable cells. Compensation for
spectral overlap of dyes was set by running mixtures of unstained cells
and cells stained with each of the fluorochromes singly.
Qa1R+ and Qa1R- cells were
purified by sorting Qa1-tetramer stained fetal NK cells on a FACS
Vantage (Becton Dickinson).
RT-PCR
Total cytosolic RNA was extracted from 
1 x
106 cells using Trizol (Life Technologies). Ten
micrograms of DNase-treated RNA was reverse transcribed using an
oligo(dT)12–18 primer and a Superscript
preamplification kit (Life Technologies). PCR amplification was
performed on aliquots of cDNA using a hot start after an initial
denaturation at 95°C, followed by 35 cycles of 95°C for 1 min,
56°C for 1 min, 72°C for 50 s, with a final extension time of
7 min. The Ly49E primers have been described previously
(17). The forward and reverse primers for CD94 and NKG2A,
NKG2C, and NKG2D were based on sequences obtained by Vance et al.
(24) and Lohwasser et al. (25) and were as
follows: CD94f, TCTAGGATCACTCGGTGGAG; CD94r, AACCGTGGACCTTCCTTGTC;
NKG2Af, CTATGCAGAACTGAAGGTG; NKG2Cf, CTGCTTGGAACTGAACAGG; NKG2A/Cr,
CCAAGATTTTCTCTCCATAC; NKG2Df, GAGCAAATGCCATAATTACG; and NKG2Dr,
CATAGACAGCACAGGATCC. Products were analyzed on 3% NuSieve GTG agarose
(FMC BioProducts, Chicago, IL) containing ethidium bromide.
Cytotoxicity assays
These assays were performed in a standard manner by incubating
serial dilutions of effector cells with 5000
51Cr-labeled target cells for 4 h in
V-bottom microtest plates. The C57 lymphoma lines EL4 and RMA, together
with the ß2m-deficient mutant of EL4, C4.4
(26), and the TAP-deficient mutant of RMA, RMA/S
(3), were kindly provided by Drs. R. Glas and K.
Kärre (Karolinska Institute, Stockholm, Sweden). RMA/S cells
transfected with mouse CD1 cDNA (27), and the 1B1
anti-CD1 mAb (28) were kindly provided by Dr. M.
Kronenberg (Los Angeles, CA). The F12 cell line transfected with a
hybrid Qa1b/Db construct
has been described previously (22). During extended
cultivation, this cloned transfected L cell line spontaneously
generated a Qa1b/Db-ve revertant
subpopulation. Expressing and nonexpressing cells were separated on a
FACS Vantage following staining with HB27
anti-Db mAb (American Type Culture
Collection, Manassas, VA), and were returned to culture to give the
F12+ and F12- lines. T2
cells transfected with the Qa1b gene
(29), and control T2 cells were kindly provided by Dr. M.
Soloski (Johns Hopkins, Baltimore, MD). Before cytotoxicity assays,
these cells were incubated overnight at 26°C in AIMV medium
containing 30 µM Qdm peptide. Chromium labeling and cytotoxicity
assays were also performed in the presence of 30 µM peptide. Blast
cells were prepared from frozen CD8-depleted spleen cells of C57BL/6
mice and mice homozygous for the ß2m knockout
mutation on a C57 background, kindly provided by Prof. E. Jenkinson
(Birmingham, U.K.). Thawed spleen cells were cultured for 2–3 days in
MEM containing 20 mM HEPES, 5 x 10-5 M
2-ME, 10% FBS, 2 µg/ml Con A, 50 µg/ml gentamicin, and 100
U/ml IL-2.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The left hand part of Fig. 1
shows
typical results obtained when polyclonal fetal and adult NK cell
preparations derived from C57BL/6 mice were stained with soluble
tetrameric forms of Kb, Db,
and Qa1b. As reported previously (18, 19), 50% of adult NK cells bore receptors that could be
detected with Qa1-Qdm tetramers. In most polyclonal fetal NK cell lines
that had been cultured for more than 7 days, 35% of cells bore Qa1R
of this type, the level of staining being similar to that on adult NK
cells. By contrast, although 40% of adult NK cells showed clear
staining with a Kb tetrameric construct, few if
any fetal NK cells stained. Neither adult nor fetal NK cells showed any
detectable staining with a Db tetrameric
construct that stained the cognate T cell clone CTL10. An exception to
the rule that fetal NK cell lines contained Qa1R+
cells was the long-term fetal NK cell line 1080.
|
The fetal NK cell lines studied here were derived by culturing day
14 fetal thymocytes in IL-4 + PMA then transferring them to high-dose
IL-2. The changes that occur in the expression of NK1.1 and Qa1R during
the early stages of this process are illustrated in Fig. 2 and summarized in Table I. Before culture, only about 0.7% of
fetal thymocytes had an NK phenotype (CD3-
NK1.1+), but interestingly 20% of these cells
expressed Qa1R. Following culture for 2 days in IL-4 + PMA, the
frequency of both CD3-
NK1.1+ cells and the proportion of these
expressing Qa1R did not change. Thus, only about 0.1% of the cells
present at this stage were CD3-
NK1.1+ Qa1R+ cells.
However, within 24 h of transfer to medium containing IL-2, 23%
of cells had acquired NK1.1 and 12% of these bore Qa1R. Thus, about
2.7% of the cells in culture were now CD3-
NK1.1+ Qa1R+ cells. Because
there is no noticeable cell death when cells are transferred to IL-2,
and 50–70% of input cells proliferate under these conditions
(15), the data strongly suggest that the dramatic increase
in the proportion of both CD3-
NK1.1+ cells and CD3-
NK1.1+ Qa1R+ cells that
occurs at this stage is due to the de novo expression of NK1.1 and Qa1R
on progenitor cells rather than the selective growth or survival of
preexisting cells of this type. This conclusion was directly confirmed
by single cell experiments (see below).
|
|
). Throughout the
culture period, the expression of Qa1R was largely restricted to cells
that expressed NK1.1. The addition of PMA accelerated the rate at which
cells acquired expression of NK1.1 and Qa1R. By contrast, the addition
of IL-4 had no effect on the acquisition of NK1.1 but completely
blocked the acquisition of Qa1R. Even more remarkably, when cells that
had been grown in IL-2-containing secondary cultures for 3 days were
transferred to medium containing IL-4, the expression of Qa1R was
virtually extinguished within 3 days. Kinetic analysis (data not shown)
revealed that, following transfer to medium containing IL-4, the
proportion of Qa1R+ cells declined 2- to 3-fold
each day. Cells that had been grown in secondary cultures containing
IL-2 + IL-4 for 3 days did not acquire Qa1R upon subsequent transfer to
medium lacking IL-4 (at least within 3 days). Thus, IL-4 exerts rapid,
dramatic, and largely irreversible effects on the expression of Qa1R on
developing NK cells. Single cell experiments demonstrate that developing NK cells acquire Qa1R in a stochastic manner
The finding that Qa1R are only expressed on a subpopulation of NK cells that develop in vitro has two possible explanations. The first is that Qa1R+ and Qa1R- NK cells arise from separate committed precursors that, respectively, switch on or do not switch on Qa1R expression during development. The second is that Qa1R+ and Qa1R- NK cells arise from a common precursor whose descendants randomly switch on Qa1R during a critical stage of development. To determine which of these possibilities was correct, IL-4/PMA-primed fetal thymocytes that consist of >99% NK1.1- Qa1R- cells were cloned in IL-2 at an average of 0.1 cell/well.
Poisson analysis indicated that >99% of the developing NK cell clones
arose from single progenitors. Because 30% of the cells that were
plated gave rise to clones, the vast majority of clones must therefore
have arisen from NK1.1-
Qa1R- progenitors. Twelve of 12 clones tested
showed uniform expression of NK1.1 on all cells. Of 34 clones examined
for expression of Qa1R (which included the 12 clones examined for
NK1.1), 31 expressed Qa1R. Importantly, every one of these 31 clones
displayed mosaic expression of Qa1R. The results for 10 randomly chosen
clones are shown in Fig. 3. The
percentage of Qa1R+ cells within these clones
ranged from 3% (J1.4) to >90% (J1.1). When heterogeneous clones were
restained after a further 2 wk of culture, the proportions of
Qa1R+ and Qa1R- cells were
very similar to what was originally found (data not shown). By
contrast, 4/6 long-established fetal NK clones that were examined
contained no detectable Qa1R+ cells, suggesting
that Qa1R may be lost during long-term culture. The results for two of
these, 773.2 and Cc5H10, are shown in Fig. 1, together with the results
for one long-term clone, I4, that contained >95%
Qa1R+ cells.
|
Compelling evidence has been provided that tetrameric molecules
formed from soluble forms of Qa1, or its human homologue HLA-E, bind to
CD94/NKG2 heterodimeric receptors on NK cells (7, 19). To
determine whether fetal NK cell lines and clones express CD94 and NKG2,
RT-PCR was performed on total cytosolic RNA extracted from appropriate
cells. Key results are shown in Fig. 4.
Of 13 polyclonal and monoclonal fetal NK lines examined, all expressed
mRNA for CD94. Polyclonal adult and fetal NK cell lines also expressed
mRNA for NKG2A, NKG2C, and NKG2D. The single exception was the
polyclonal line 1080, which lacked detectable mRNA for each of these
NKG2 species. In addition to the major bands, whose identity was
confirmed by sequencing, PCR amplification of NKG2 mRNA frequently
generated additional weak bands. In particular, the NKG2A primers
invariably gave rise to a subsidiary band 100 bp smaller than the
correct NKG2A amplimer. This band was close to the size expected for
NKG2B, a putative alternately spliced form of NKG2A. However,
sequencing revealed it to be derived from an NKG2A-like transcript that
lacked nt 188–289 of the published NKG2A sequence. This deletion
creates a potential in-frame translation product lacking the
transmembrane domain.
|
) expressed mRNA for both putative
inhibitory (NKG2A and NKG2D) and noninhibitory (NKG2C) NKG2 molecules.
Expression, and indeed, as far as could be judged, the level of
expression, bore no relationship to the level of expression of surface
Qa1R. In particular, it was quite clear that clones that contained few
(J1.4) or no (773.2 and Cc5H10) detectable Qa1R+
cells contained mRNA for CD94 and for NKG2A, NKG2C, and NKG2D, although
in the case of clone 773.2 the NKG2A and NKG2C signals were low. Each
of the polyclonal and monoclonal NK cell lines was also examined for
expression of mRNA encoding Ly49E, the only member of the Ly49 family
known to be expressed in these cells. All were strongly positive,
including the 1080 line that lacked expression of NKG2 molecules.
RT-PCR analysis was also performed on fresh day 14 fetal thymocytes.
Despite the very low abundance of fetal thymocytes with NK
characteristics, unexpectedly strong signals were found for CD94 and
NKG2A, with weaker but clear signals for NKG2C and NKG2D, and for Ly49E
(Fig. 4). Cytolytic activity of Qa1R+ and Qa1R- fetal NK cells
To determine the functionality of the Qa1R expressed on fetal NK
cells, the cytolytic specificity of the cells bearing them was
examined. Representative results of a large series of experiments are
shown in Fig. 5. As reported previously,
polyclonal fetal NK cells, in contrast to polyclonal adult splenic NK
cells, distinguished only weakly if at all between wild-type and
ß2m-deficient blast cells. By contrast,
purified Qa1R+, but not
Qa1R-, fetal NK cells preferentially lysed
ß2m-deficient blasts. The most likely
interpretation of this result is that many of the fetal NK cells that
bear Qa1R express inhibitory forms of these receptors that prevent the
cells that bear them from lysing Qa1+ blasts. As
shown by using the Qa1-specific T cell clone, e1d, most if not all of
the cells in the wild-type blast cell population did indeed express
Qa1, whereas ß2m-deficient blast cells lacked
Qa1 molecules that could be recognized by e1d. The fact that wild-type
blast cells were still killed to a significant extent by purified
Qa1R+ fetal NK cells could be because the
Qa1R+ cells were contaminated with some
Qa1R- cells, the Qa1R expressed by some
Qa1R+ cells are noninhibitory, or the inhibitory
signals delivered via Qa1R are not completely effective. Clearer
results were obtained using cloned fetal NK cells lines. The
predominantly Qa1R+ clone J1.1 discriminated very
clearly between wild-type and ß2m-deficient
blasts, whereas the clone J1.4, and also the polyclonal line 1080, that
were largely or completely devoid of Qa1R+ cells,
killed both wild-type and ß2m-deficient blasts
equally well. Strikingly different results were obtained when each of
these various effector cells was tested on wild-type EL4 cells and the
ß2m-deficient variant of EL4, C4.4. Not only
did adult NK cells and Qa1R+ fetal NK cells
preferentially lyse ß2m-deficient EL4 cells,
but so too did Qa1R- polyclonal fetal NK cells,
the Qa1R- clone J1.4, and all other fetal
Qa1R- clones tested. By contrast, the 1080 line
that lacked not only Qa1R+ cells but also NKG2A,
NKG2C, and NKG2D, killed wild-type EL4 cells more efficiently than
ß2m-deficient EL4 cells.
|
). Thus, the Qa1-transfected L cell line
F12+ was resistant to lysis by
Qa1R+ clones such as I4 and J1.1, compared with a
revertant Qa1- line derived from F12. That the
structures on target cells that inhibited lysis by
Qa1R+ fetal NK cells were most likely Qa1
molecules bearing the dominant Qdm peptide was strongly suggested by
the finding that the human T2 line transfected with Qa1 became
resistant to both purified polyclonal Qa1R+ fetal
NK cells and to Qa1R+ NK clones, but not to
Qa1R- fetal NK cells, when incubated with Qdm
peptide.
|
). Furthermore, CD1-transfected RMA/S
cells that vastly over-express CD1 show little or no resistance to
fetal NK cells (Fig. 7). Line 1080, which fails to distinguish between
wild-type and ß2m-deficient EL4 cells, is also
clearly unable to distinguish between wild-type and TAP-deficient RMA
cells.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
30%
developed into clones that displayed uniform expression of NK1.1 but,
in the vast majority of cases, mosaic expression of Qa1. It seems that
the probability of cells switching on expression of Qa1R is highest in
the first few days after transfer to IL-2, and declines substantially
thereafter. Thus, within 1 day of transfer to IL-2, the proportion of
NK cells that expresses Qa1R rises almost 30-fold from about 0.1% to
2.7%, but over the next 2 days it increases only 5-fold to about 12%,
and over the following 3 days by less that 3-fold to 33%, after which
there is no further increase no matter how long the cells are grown.
Furthermore, if after 8 days of culture in IL-2,
Qa1R- cells were purified to >99% purity by
cell sorting and returned to culture, only about 2% acquired Qa1R even
after 7 days of further culture (data not shown).
It therefore seems likely that the characteristic mosaic Qa1R
expression pattern of individual clones is established early in clonal
development. If we assume that the acquisition of Qa1R is largely
irreversible, then a clone that contains 50%
Qa1R+ cells most likely arose because one of the
two daughter cells of the first IL-2-driven division acquired Qa1R.
Similarly, a clone that contains about 25% Qa1R most likely arose
because one of the four grand-daughter cells switched on Qa1R. The
finding that about 12% of NK1.1+ cells acquire
Qa1R in the first 24 h (approximately two cell divisions)
indicates that the probability of NK cells switching on Qa1R is
initially about 0.1/division. If this model is correct it suggests that
NK clones that are uniformly Qa1R positive would be rare, as they would
only be likely to arise if both daughter cells of the first division
acquired Qa1R. This prediction is supported by the observation that
<5% of the clones examined appear even close to being uniformly
Qa1R+, the great majority having <50%
Qa1R+ cells. Coupled with a declining switching
rate, perhaps associated with a declining rate of cell division, this
model readily explains why no more than 35% of the cells in fetal
thymus-derived NK cell populations ever express Qa1R.
The issue of whether Qa1R expression can be switched off as well as on
is an important one. The fact that clones composed predominantly of
Qa1R+ cells remain predominantly
Qa1R+ for many weeks suggests that if cells can
switch off Qa1R they do so at a rate that is slow compared with the on
rate in early cultures. Indeed, the observation that fetal NK clones
show stable but distinctive patterns of Qa1R expression supports the
deterministic model described above, and argues strongly against an
alternative model in which the expression of Qa1R on 35% of
polyclonal fetal NK cells is due to a dynamic equilibrium being reached
between on and off rates. Whatever mechanisms are responsible for the
stochastic expression of Qa1R, it is possible that they influence the
expression of other molecules. In particular, several members of the
CD45, Ly6, and NKRP families have also been found to be expressed in a
mosaic manner on fetal NK clones (31). The rate of
switching appeared to be different for different molecules, being
particularly high for Ly6C and the NKRP molecules recognized by mAb
10A7. We have also found occasional expression of Ly49 molecules on
fetal NK clones, and in every case this has been of a mosaic nature
(Ref. 31 and our unpublished observations).
The mosaic expression of Qa1R within NK cell clones implies that the
individual cells that constitute a clone vary in their expression of
one or more of the components required for the synthesis or expression
of Qa1R. All available evidence indicates that CD94/NKG2 heterodimeric
complexes are the principal if not sole receptors for Qa1 and its human
homologue HLA-E (7, 8, 9, 19). The finding that some
established NK clones lack any detectable surface Qa1R yet contain mRNA
for both CD94 and NKG2 molecules indicates that Qa1R expression may be
controlled, at least in part, at the posttranscriptional level. This
contrasts with the situation for Ly49, where in both adult
(32) and fetal NK clones (17) expression
appears to be controlled predominantly or exclusively at the
transcriptional level. However, it is in agreement with a recent report
that some 
T cells contain intracellular but not surface CD94
complexes (33), and suggests that in some cases surface
expression of CD94/NKG2 complexes may be controlled at the level of
assembly and transport to the cell surface. It has been established
that for CD94/NKG2C complexes, association with DAP12 is required for
surface expression (34), but whether additional components
are required for surface expression of CD94/NKG2A complexes is less
clear. Other explanations for the finding that some clones
contain mRNA for CD94 and NKG2 yet lack Qa1R would be that the CD94 and
NKG2 messages are being expressed in a mutually exclusive manner among
the individual cells of such clones, or that alternatively spliced
forms of CD94 and NKG2 molecules (see below) interfere with the
expression of Qa1R.
RT-PCR analysis revealed that all of the fetal NK clones examined contained mRNA for the inhibitory NKG2A molecule, the noninhibitory NKG2C molecule, and for the distantly related inhibitory NKG2D molecule. Again, further study will be needed to establish whether these different molecules can be simultaneously expressed by individual cells within clones. In humans, it appears that within the mature NK cell population there are cells that express either NKG2A or NKG2C or both (35, 36). Surprisingly, in both adult and fetal mouse NK cells we were unable to detect expression of NKG2B, a putative alternately spliced form of NKG2A that lacks the stalk region in the extracellular domain. Instead, all clones and lines examined, including adult NK cells, were found to contain what appears to be another alternately spliced form of NKG2A that lacks the transmembrane domain. Based on current knowledge of the translocation of type II proteins, it would seem that a molecule of this type could not leave the cytosol and hence would be nonfunctional. Curiously, though, we have found that adult mouse NK cells contain substantial quantities of alternatively spliced mRNA that would encode Ly49 molecules lacking transmembrane domains (S. Shrestha et al., unpublished data), suggesting that the production of molecules of this type is of some significance.
Another very unexpected finding was that Qa1R expression was strongly regulated by IL-4. When present during the secondary phase of culture in IL-2, IL-4 completely inhibited the acquisition of Qa1R by developing NK cells yet had no effect on the acquisition of NK1.1. Although the growth of NK cells is eventually inhibited by IL-4 (15), the presence of IL-4 during the first few days of culture in IL-2 has no obvious toxic or growth inhibitory effects. It therefore seems likely that IL-4 acts by preventing the de novo expression of Qa1R. Cells that have been exposed to IL-4 during the first 3 days of culture in IL-2 show no tendency to acquire Qa1R when transferred to IL-4-free medium indicating that the effect of IL-4 is largely irreversible, and providing further support for the view that the switching on of Qa1R expression is restricted to a critical time-limited period early in NK development. Remarkably, the effect of IL-4 itself is not confined to this period since, even after 6 days of culture in IL-2, the addition of IL-4 could rapidly reverse the expression of Qa1R (data not shown). Paradoxically, although developing NK cells eventually become susceptible to growth inhibitory effects of high dose IL-4, the long-term growth of fetal NK cells is enhanced by low dose IL-4 (21). It is possible that our use of low doses of IL-4 for this purpose explains why most of our long-term lines and clones lack Qa1R.
The findings reported here concerning the expression and regulation of class I receptors on fetal NK cells go a considerable way toward explaining the lytic specificity of these cells. We previously found that, in contrast to adult NK cells, most preparations of fetal thymus-derived NK cells lysed wild-type blast cells almost as well as class I-deficient blast cells (17). The present finding, that only a minority of the fetal NK cells that develop under the culture conditions used expresses Qa1R, implies that it was these cells that were responsible for the limited discrimination observed, and that the majority of fetal NK cells, due to their lack of receptors for both Qa1-Qdm and classical class I molecules, were unable to discriminate between wild-type and class I-deficient blasts. This hypothesis was directly confirmed by testing purified Qa1R- and Qa1R+ fetal NK cells and clones. Although we cannot formally exclude the possibility that Qa1R merely act as markers for other class I receptors, strong supporting evidence that Qa1R are directly involved was provided by the observations that 1) L cells expressing a Qa1/Db hybrid molecule, and 2) T2-Qa1 cells incubated with Qdm peptide, were selectively resistant to Qa1R+ but not Qa1R- fetal cells. This finding also implies that, as is the case for adult NK cells (19), most Qa1R+ fetal NK cells receive net inhibitory signals following interaction with ligand-expressing cells. Given the stochastic nature of Qa1R expression on NK cells, we suspect that the fetal NK line 1270 that was found to discriminate efficiently between wild-type and class I-deficient blasts in our earlier study (17) contained an unusually large proportion of Qa1R+ cells. Likewise, reports that other populations of Ly49-deficient NK cells can discriminate relatively efficiently between wild-type and class I-deficient blasts (37, 38) may be explained by the different culture conditions or sources of cells used resulting in a higher proportion of the cells acquiring Qa1R.
What is clearly not explained by the class I receptor data we have obtained is the striking ability of fetal NK cells to discriminate between class I-deficient and wild-type EL4 and RMA cells. The fact that these are both tumor cells and are syngeneic to the effector cells is intriguing, but the observation that fetal NK cells efficiently kill a wide variety of other tumor cells, including syngeneic C1498 cells (31), and that both fetal and adult NK cells derived from allogeneic mice can discriminate between wild-type and class I-deficient forms of these cells (unpublished data), argues against either property being relevant. A rational explanation of the results would be that wild-type EL4 and RMA cells express one or more nonclassical class I molecules that are ß2m- and TAP-dependent and are not widely expressed on other tumor cells or blast cells. A potential candidate for such a molecule is CD1, which, although not generally TAP-dependent (39), can associate with peptides (40, 41), and has been reported to protect cells from NK lysis (30). However, the finding that both RMA and RMA/S express similar levels of CD1, and that RMA/S cells overexpressing CD1 do not become resistant to lysis by fetal NK cells, rules out a role for CD1 unless the structure recognized by NK cells on EL4/RMA cells is CD1 loaded with TAP-dependent peptides. In our hands, adult NK cells could also not distinguish between untransfected and CD1-transfected RMA/S cells. The reason for the difference between our results and those of others (30) is currently unclear.
If the resistance of EL4 and RMA cells to lysis by NK cells is due to their expression of a class I molecule, most or all fetal NK cells, and probably also adult NK cells, must express an inhibitory receptor for this molecule. The finding of a variant fetal NK cell line, 1080, that can no longer discriminate between wild-type and class I-deficient EL4/RMA cells, and the discovery that this line has lost expression of at least three members of the NKG2 family, provides strong support for the hypothesis that NK cells express novel class I receptors, and suggests that these may involve members of the NKG2 family. Given the current evidence that several members of this family recognize Qa1/HLA-E (7, 19), an intriguing possibility is that the protective element on EL4/RMA cells is a form of Qa1 that associates with non-Qdm peptides (42) or with other proteins (43, 44). It also strongly suggests that the defect in fetal NK line 1080 resides in a previously unsuspected factor that is selectively required for the transcription of NKG2 genes. This factor clearly does not affect transcription of the closely related CD94 molecule, implying that CD94 molecules may have a biological role that is independent of their association with members of the NKG2 family.
Finally, we reported previously that, although fetal NK cells lack expression of most Ly49 molecules, they express high levels of mRNA for the putatively inhibitory Ly49 molecule, Ly49E (17). We report here that Ly49E mRNA is expressed in 11/11 fetal NK clones examined, a pattern quite distinct from that found for other Ly49 molecules (32, 45). Whether it is expressed in all individual NK cells present within a clone and whether Ly49E protein is expressed on the cell surface requires further investigation. If these assumptions are correct, the presence of Ly49E mRNA in Qa1R- NK cells that readily lyse normal blasts would indicate that Ly49E is not an inhibitory receptor for any of the class I molecules expressed on blast cells of C57 mice. Similarly, the presence of Ly49E mRNA in the 1080 line would indicate that Ly49E does not recognize the putative protective class I molecule expressed on EL4/RMA cells unless it forms a heterodimeric structure with an NKG2 molecule. The possibility that the natural ligand for Ly49E may not be a class I molecule should therefore be considered seriously.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Colin Brooks, Department of Microbiology and Immunology, The Medical School, Newcastle, NE2 4HH, U.K. E-mail address:
3 Abbreviations used in this paper: ß2m, ß2-microglobulin; Qa1R, receptors for soluble Qa1b-Qdm complexes.
Received for publication March 31, 1999. Accepted for publication July 9, 1999.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
9 V2 T cells but not on CD94- thymic or mature T cell clones. Eur. J. Immunol. 28:3399.[Medline]
This article has been cited by other articles:
|
|
 |
|
  F. Gays, J. G. Aust, D. M. Reid, J. Falconer, N. Toyama-Sorimachi, P. R. Taylor, and C. G. Brooks Ly49B Is Expressed on Multiple Subpopulations of Myeloid Cells J. Immunol., November 1, 2006; 177(9): 5840 - 5851. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Nguyen, N. Dhedin, J.-P. Vernant, M. Kuentz, A. A. Jijakli, N. Rouas-Freiss, E. D. Carosella, A. Boudifa, P. Debre, and V. Vieillard NK-cell reconstitution after haploidentical hematopoietic stem-cell transplantations: immaturity of NK cells and inhibitory effect of NKG2A override GvL effect Blood, May 15, 2005; 105(10): 4135 - 4142. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. T. Wilhelm, J.-R. Landry, F. Takei, and D. L. Mager Transcriptional Control of Murine CD94 Gene: Differential Usage of Dual Promoters by Lymphoid Cell Types J. Immunol., October 15, 2003; 171(8): 4219 - 4226. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Toomey, F. Gays, D. Foster, and C. G. Brooks Cytokine requirements for the growth and development of mouse NK cells in vitro J. Leukoc. Biol., August 1, 2003; 74(2): 233 - 242. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Stevenaert, K. Van Beneden, A. De Creus, V. Debacker, J. Plum, and G. Leclercq Ly49E expression points toward overlapping, but distinct, natural killer (NK) cell differentiation kinetics and potential of fetal versus adult lymphoid progenitors J. Leukoc. Biol., June 1, 2003; 73(6): 731 - 738. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. H. Lian, M. Maeda, S. Lohwasser, M. Delcommenne, T. Nakano, R. E. Vance, D. H. Raulet, and F. Takei Orderly and Nonstochastic Acquisition of CD94/NKG2 Receptors by Developing NK Cells Derived from Embryonic Stem Cells In Vitro J. Immunol., May 15, 2002; 168(10): 4980 - 4987. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Van Beneden, A. De Creus, F. Stevenaert, V. Debacker, J. Plum, and G. Leclercq Expression of Inhibitory Receptors Ly49E and CD94/NKG2 on Fetal Thymic and Adult Epidermal TCR V{gamma}3 Lymphocytes J. Immunol., April 1, 2002; 168(7): 3295 - 3302. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Van Beneden, F. Stevenaert, A. De Creus, V. Debacker, J. De Boever, J. Plum, and G. Leclercq Expression of Ly49E and CD94/NKG2 on Fetal and Adult NK Cells J. Immunol., April 1, 2001; 166(7): 4302 - 4311. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Toyama-Sorimachi, Y. Taguchi, H. Yagita, F. Kitamura, A. Kawasaki, S. Koyasu, and H. Karasuyama Mouse CD94 Participates in Qa-1-Mediated Self Recognition by NK Cells and Delivers Inhibitory Signals Independent of Ly-49 J. Immunol., March 15, 2001; 166(6): 3771 - 3779. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Gays, K. P. Fraser, J. A. Toomey, A. G. Diamond, M. M. Millrain, P. J. Dyson, and C. G. Brooks Functional Analysis of the Molecular Factors Controlling Qa1-Mediated Protection of Target Cells from NK Lysis J. Immunol., February 1, 2001; 166(3): 1601 - 1610. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Hui Jia, Z. Kurepa, A. Bai, and J. Forman Comparative Ability of Qdm/Qa-1b, Kb, and Db to Protect Class Ilow Cells from NK-Mediated Lysis In Vivo J. Immunol., December 1, 2000; 165(11): 6142 - 6147. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Gays, M. Unnikrishnan, S. Shrestha, K. P. Fraser, A. R. Brown, C. M. G. Tristram, Z. M. A. Chrzanowska-Lightowlers, and C. G. Brooks The Mouse Tumor Cell Lines EL4 and RMA Display Mosaic Expression of NK-Related and Certain Other Surface Molecules and Appear to Have a Common Origin J. Immunol., May 15, 2000; 164(10): 5094 - 5102. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
