| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Promoter Regions of Multiple IgA Heavy Chain Genes1
Department of Microbiology and Immunology, Stritch School of Medicine, Loyola University Chicago, Maywood, IL 60153
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
4, C5, C6, C9, C10, C12, and C13), those
expressed at low levels (C1, C2, C7, and C11), and those
that are not expressed (C3 and C8). We tested whether the
differential in vivo expression is due to differential responses of the
I promoters to TGF-
stimulation. We stimulated the rabbit B cell
line 55D1 with TGF- and, using single-cell RT-PCR, found that
expression of germline (GL) transcripts of 3 and 8 could not be
induced. By luciferase reporter gene assay and EMSA we found that the
promoters of the unexpressed isotypes C3 and C8 are defective,
thereby explaining the absence of IgA3 and IgA8 in vivo. When comparing
the promoter activities of the other isotypes we found that the
activities did not reflect the degree of in vivo expression. Instead,
the promoters of the isotypes expressed at high or low levels promoted
expression of the luciferase gene to a similar degree, except for the
I4 promoter, which had much higher activity. Also the degree to
which TGF- induced GL expression of the various isotypes in 55D1 B
cells did not reflect in vivo expression. However, most of the
TGF--stimulated cells expressed GL mRNA of multiple isotypes; no
isotype was expressed preferentially. These results suggest that the
final switch to a single isotype is regulated in a step subsequent to
GL transcription, rather than by induction of GL transcripts by the
I promoter. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
genes (1).
The rabbit, a member of the lagomorph family, has 13 C genes,
whereas the human has two, and the mouse and bovine each have one C
gene. All 13 C genes of rabbit are potentially functional, as judged
by nucleotide sequence analysis (1) and in vitro
expression (2). RNase protection assay of mucosal tissues
of rabbit revealed that the 13 C genes are expressed at different
levels (3) and that two of the C genes, C3 and
C8, are not detectable in any tissue; C1, C2, C7, and
C11 are expressed at low levels, and C4, C5, C6, C9,
C10, C12, and C13 are expressed at high levels.
We showed previously that the C genes are distributed differently
among mucosal tissues (3). For example, a single C
gene, C4, is predominantly expressed in tissues of the upper
respiratory tract, particularly in lung, but is expressed unevenly in
the small intestine. Using IgA4-specific mAb we found that a high
percentage of IgA plasma cells in the duodenum and upper jejunum
produce IgA4, while in the ileum only a low percentage of the IgA
plasma cells produce IgA4 (4). These findings were
particularly interesting because this pattern of C4 distribution
resembles the distribution of IgA1 in humans (5), which is
predominant in the respiratory tract and the proximal end of the small
intestine, while IgA2 is predominant in the distal end of the small
intestine and in the large intestine. It is not known whether this
distribution of human IgA1 or rabbit IgA4 is a result of homing or
whether Ags and/or local environments direct switching to specific
isotypes. Resting splenic B lymphocytes (6, 7, 8), human B
cells from tonsil or blood (9, 10), and B cells from
nonmucosal tissues such as popliteal lymph nodes from rabbit
(4), can be induced in vitro to switch to IgA, indicating
that all B cells have the capacity to undergo isotype switch to IgA.
However, it is well established that switching to IgA in vivo occurs
mainly in mucosal tissues, particularly in Peyer’s patches
(PP),3 and not in systemic
lymphoid organs. Fagarasan et al. (11) recently identified
IgA switch recombination circle DNA in lamina propria lymphocytes,
suggesting that IgA+ cells in the gut lamina
propria are generated in situ. Thus, it is most likely that switching
to IgA is regulated in vivo by the local microenvironment and is not an
inherent characteristic of particular B cells.
The differential expression of IgA isotypes could be regulated at the
level of expression of germline (GL) mRNA. Class switch recombination
occurs in germinal centers after Ag stimulation. The first step leading
to isotype switch is the transcription of GL mRNA of that isotype to
which the cell will finally switch (12, 13, 14). This step is
regulated by ILs and other isotype-specific factors. Transcription of
GL mRNA in splenic B cells is specifically induced by TGF-
(6, 7, 15). The transcription of the GL mRNA is
initiated in the I region, which is controlled by a promoter of
approximately 150 nt (16, 17, 18). This promoter region is
highly conserved among human, mouse, and all rabbit I genes
(4). Several investigators have shown that the promoter,
when transfected into human or mouse B cell lines, drives expression of
a reporter gene (17, 18, 19, 20) and that the expression was
enhanced when the transfected cells were incubated with TGF-.
The I promoter region contains several recognition elements for
transcription factors, including a TGF- response element
(TGF-RE), a cAMP response element (CRE) site, and a site that binds
proteins of the ETS family. By nucleotide mutation and
deletions, several authors showed that for human (21) and
mouse (17, 22) B cell lines, all three transcription
factor binding sites, TGF-RE, CRE, and ETS are required for optimal
basal promoter activity, and they contribute to TGF- inducibility.
Zan et al. (22) found that anti-TGF- neutralized
the basal activity of the transfected promoters, suggesting that in B
cell lines, as in primary B cells, the promoter has no activity unless
activated by TGF-. TGF- signaling occurs through the Smad protein
family (23, 24), and Shi and Stavnezer (25)
identified several binding sites for Smad proteins and core binding
factors within the TGF-RE. The authors also found that in the mouse
cell line I 29, transfected with TGF-RE, core binding factor-3 is
synthesized in response to TGF- stimulation. Taken
together these studies suggest that TGF- initiates or enhances the
synthesis of transcription factors necessary for the expression of
GL mRNA, but additional transcription factors, the synthesis of
which may be independent of TGF-, are also required. Therefore,
since activation requires binding of several transcription factors to
the I promoters, one can predict that differences in the nucleotide
sequences of any single binding site can lead to differences in the
level of expression of GL mRNA, and this could explain the differences
in in vivo expression of the various IgA isotypes in rabbit.
We performed studies to investigate whether production of GL mRNA of a
given IgA isotype depends on the efficiency of its I promoter. We
first characterized the rabbit B cell line 55D1 and found that cells of
this line express sIgM and can be induced to express GL mRNA. We
used this cell line to compare the promoters of the various I genes
for their ability to support the expression of a reporter gene and to
bind nuclear proteins.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cells of the rabbit B cell line 55D1 were grown in RPMI with
15% FCS and maintained in log phase growth. To determine the level of
TGF--induced transcription of GL mRNA, the cells were incubated for
24 h in complete medium containing TGF- (human recombinant, 1
ng/ml; R&D Systems, Minneapolis, MN), 5% anti-Ig antiserum (rabbit
anti-VH allotypic antisera, anti-a1 and
anti-a3), and irradiated (5000 rad) murine CD40 ligand-transfected CHO
cells (CD40L-CHO) cells (a gift from M. Spriggs, Immunex, Seattle, WA)
at a 55D1 to CD40L-CHO ratio of 10:1.
Oligonucleotide primers
For PCR amplification of the 180-bp I promoters of the
different IgA isotypes we used as I-specific 5' sense primers,
20–23 nt with sequences starting at position -158. The I-specific
3' antisense primers were 20–23 nt with sequences from the 3' end
starting at position +21 (see Fig. 1
). For amplification of fragments
430–440 bp in size (positions approximately -415 to +21) we first
obtained sequences upstream of the TGF-RE (deposited in GenBank
database; for accession numbers, see Fig. 1) and designed sense primers
from the sequences around position -415 as follows: I3 and I4,
5'-CAATGGCTGTCCCCACCCTGAC; I8, 5'-TGCATAGAATAAGTCTTAAT; and
I9, 5'-TAATGGCTGTCTCCATCCTGAC. Truncated regions of the promoters
were amplified with 5' sense primers starting at various positions as
indicated in the text. Primers used to amplify oligomers for use in the
luciferase (LUC) assay were synthesized with cloning sites:
SacI for the sense primer and XhoI for the
antisense primer.
|
-C germline mRNA (GL mRNA) was
amplified using the sense primer 5'-AGGTGCAC/AAGCTC/TTC/GCAGGA located
in the I region and the antisense primer, either
5'-GGCCCGGGGGGAAGAAG located in exon 1 or
5'-CTCAGGGTGCAGGTGAGGCT located in exon 2. These three primers
recognize all IgA isotypes (4). For amplification of GL
mRNA of the different isotypes we used the I sense primer shown
above and gene-specific antisense primers, the sequences, locations,
and specificities of which have been published (4). The
positive control for all isotypes was cDNA from appendix. We also
tested for the expression of GL mRNA of the unexpressed isotypes
C3 and C8 in cDNA of activated 55D1 and PP cells, and found none.
The antisense primers of these two unexpressed isotypes, when tested on
cloned DNA in combination with a pan-specific exon 1 sense primer
(4) amplified the expected fragments of C3 or C8.
Oligonucleotides a–e for EMSA (40–45 nt; see Fig. 6) were obtained as
complimentary pairs with a 4-bp CTGA 5' overhang for end labeling. All
primers were synthesized by Integrated DNA Technology (Coralville,
IA).
|
The I promoters were amplified from previously subcloned
plasmids (4) or from C containing phage or cosmid DNA
(1). The isotype was confirmed by nucleotide sequence
analysis. The I promoters were cloned into the
SacI/XhoI sites of the PGL 3 vector containing
the SV40 enhancer (Promega, Madison, WI).
55D1 cells were transiently transfected by electroporation; 2 x
107 log phase cells were suspended in 1 ml
protein free RPMI 1640 with 20 µg of the LUC reporter gene construct
and 10 µg of the Renilla LUC construct as an internal
control. The cells were incubated on ice for 5 min, pulsed once at 350
V and 800 µF, transferred into 10 ml complete medium, and incubated
for 24 h. For TGF- activation, the transfected cells were
suspended in medium containing 1 ng/ml TGF- and 5% anti-Ig
antiserum, plated in 24-well plates that were previously inoculated
with irradiated (5000 rad) CD40L-CHO cells at a concentration 10% that
of 55D1 cells, and incubated. After 24 h the cells were lysed, and
a LUC assay was performed as suggested by the manufacturer
(Promega)
Single-cell RT-PCR
Single 55D1 cells, unactivated or TGF--activated, were sorted
directly into lysis buffer by the FACStar Plus (BD Biosciences,
Mountain View, CA) in the FACS facility at Loyola University Chicago
(Maywood, IL). Dead cells were excluded with propidium iodide.
Single-cell RT-PCR was performed with two rounds of PCR using
seminested primers as previously described (4). Briefly,
total GL mRNA was amplified with a sense primer from the I region
in both rounds of PCR; the antisense primer was from exon 2 in the
first round PCR and from exon 1 in the second round. For
determination of GL mRNA of particular IgA isotypes the antisense
primers in the second round were isotype specific. Samples were
visualized on 5% polyacrylamide gels.
Preparation of nuclear extracts
Nuclear extracts were prepared from 55D1 cells or cells prepared from primary lymphoid tissues using a modified small-scale method described by Schreiber et al. (26). Briefly, 10–20 x 106 cells, washed once with complete medium and once with TBS (pH 7.9), were suspended in 400 µl buffer A (10 mM HEPES (pH 7.9), 10 mM KCl, 0.2 mM EDTA, and 1 mM DTT) and incubated for 15 min on ice. After 25 µl of 10% Nonidet P-40 was added, the samples were vortexed for 10 s, and nuclei were pelleted in a microcentrifuge for 30 s. The pellet was resuspended in 50–100 µl buffer C (20 mM HEPES (pH 7.9), 0.4 M NaCl, 1 mM EDTA, and 1 mM DTT), incubated for 30 min with frequent mixing, and then centrifuged for 20 min in a microcentrifuge at 12,000 rpm. All procedures were performed on ice. Both buffers contained protease inhibitor cocktail (P-8340; Sigma-Aldrich, St. Louis, MO) at the suggested concentration. The protein concentration of the supernatant was estimated using the Bradford assay (Bio-Rad, Hercules, CA). Samples were aliquoted and stored at -70°C.
EMSA
Double-stranded oligonucleotide probes were obtained by
annealing complimentary pairs of single-stranded probes. Equal
concentrations of the complimentary oligomers were mixed, incubated for
5 min at 56°C then for 30 min at room temperature and stored at
4°C. The double-stranded probes were end-labeled with
[-32P]dCTP (NEN, Boston, MA; 3000 Ci/mmol)
using the Klenow fragment of DNA polymerase I and purified on 5%
polyacrylamide gels. DNA binding reactions were performed in a volume
of 15 µl. The samples contained 3–5 µg nuclear extract, 2.5–3.5
µg poly(dI-dC), and 0.2–2 ng (20,000–100,000 cpm) probe. The final
concentrations of the binding buffer were 10 mM HEPES (pH 7.5), 50 mM
KCl, 1 mM DTT, 0.5 mM EDTA, 10% glycerol, and 0.05% Nonidet P40. The
reaction mixture was incubated for 20 min at room temperature. In
competitive binding assays, the competing probe was incubated with the
nuclear proteins for 20 min before addition of the labeled probe. The
samples were electrophoresed over 5% polyacrylamide gels (prerun for
30 min) at 100 V for 1.5–2.0 h at room temperature with 0.5x
Tris-borate buffer. Gels were soaked in water for 2 min, dried, and
placed on x-ray film.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The B cell line 55D1 was established from a transgenic rabbit
carrying the c-myc transgene with the rabbit 
-chain
enhancer (27). The cells have functional VDJ genes
rearranged on both IgH alleles, and they express IgM on the surface. We
have observed spontaneous switching to IgA in some of the 55D1 cells.
This observation led us to test whether these cells could be induced by
TGF- to express GL mRNA and could therefore be used to compare
the promoters of the various I genes.
We have established several clones of 55D1 cells, each expressing a
different IgA isotype on the surface, in addition to IgM. We found that
some clones that express only sIgM expressed GL mRNA. The
sIgM-expressing cells could be induced to transcribe more GL mRNA by
the addition of TGF-, anti-Ig antiserum, and CD40L-CHO cells,
i.e., the same factors that we used to induce primary B cells to
transcribe GL mRNA (4). To obtain information about the
expression of GL transcripts of the different isotypes we analyzed
cDNA of activated 55D1 cells by PCR using isotype-specific antisense
primers. We found that expression of all isotypes except 3 and 8
was induced in these activated cells (data not shown). No GL
transcripts for 3 and 8 were detected. Similar results were
obtained when we analyzed cDNA of activated primary PP cells.
To determine the frequency at which GL mRNA was induced we performed
single-cell RT-PCR of uninduced and induced cells and determined
whether GL mRNA was present. We found that of 30 uninduced cells,
7% (two cells) expressed GL mRNA. We then compared 55D1 cells with
primary PP cells and found a similar result: 10% (5 of 48 cells)
uninduced PP cells expressed GL transcripts. However, if the cells
were induced with TGF-, anti-Ig antiserum, and CD40L-CHO cells,
the number of GL mRNA-containing 55D1 cells increased to 86% (12 of
14 cells). These results are similar to those published previously
(4) for activated primary B cells from PP where 78% (14
of 18 cells) expressed GL transcripts. In these earlier experiments
we also showed that activated single primary B cells can transcribe
GL mRNA of more than one IgA isotype (4). To determine
whether single cells from the 55D1 cell line also express multiple
GL mRNAs we analyzed those 55D1 cells that expressed GL mRNA with
PCR primers specific for 11 IgA isotypes. We found that single 55D1
cells, like PP cells, express GL mRNA of more than one IgA isotype
(Table I). One of the two GL
mRNA-expressing 55D1 cells from cultures that were not activated (Table I, expt. A), expressed four GL mRNAs and two of the five uninduced
GL-expressing PP cells expressed transcripts of more than one
isotype (Table I, expt. C). Of the 12 activated GL mRNA-expressing
55D1 cells, eight expressed GL mRNA of more than one isotype (Table I, expt. B): three cells expressed two, two cells expressed three, and
three cells expressed four different GL mRNAs. These results are
similar to those reported for activated PP cells (4) where
seven of seven GL-expressing cells expressed multiple isotypes.
Although no single isotype was preferentially expressed, we do not know
whether the different isotypes are expressed in similar concentrations
within individual cells. As expected, based on the results obtained
with cDNA of activated 55D1 cells, no GL mRNA of IgA3 or IgA8 was
detected in any of the single cells. Surprisingly, the number of cells
that expressed GL mRNA of IgA1, IgA2, IgA7, and IgA11 isotypes that are
expressed at low levels in vivo was similar to the number that
expressed GL mRNA of the isotypes expressed in vivo at higher levels.
None of the cells expressed mRNA of mature IgA (data not shown). We
conclude that the 55D1 cells respond to TGF- activation in a manner
similar to that of primary cells and can be used to study the I
promoters of the various isotypes.
|
genes
We have previously cloned and sequenced the promoter regions of
the I exons for each of 11 C genes and found that they are highly
similar to each other and to those of the human and mouse (Fig. 1
) (4). We have now added
the sequence of the I promoter of C10, and some changes have been
made in the sequences of I2, I4, I7, and I8. While the
nucleotide sequences among rabbit I promoters are not identical,
they all contain the transcription factor binding sites (TGF-RE,
CRE, and ETS sites) that have been described for the human and mouse
I promoter regions. We hypothesized that the few differences among
the I promoters could affect their efficiencies. This, in turn,
could explain why the IgA isotypes are expressed at different levels in
vivo and why IgA3 and IgA8 are not expressed in vivo or induced in 55D1
cells to transcribe GL mRNA in vitro. To test this hypothesis, we
cloned 180-bp fragments (positions -158 to +21) of the I promoters
into the PGL 3-SV40 enhancer vector with the LUC reporter gene and
transfected the rabbit B cell line 55D1. We found that nine of the 10
I promoters tested (all except I3) were able to promote
expression of the LUC gene, but the level of expression among them
varied over a wide range (Fig. 2A). Although the absolute
increase in expression varied between experiments, the relative
expression among the 10 I promoters remained the same. The promoter
of I4 was the most efficient and, in the experiment shown (Fig. 2A), was 170-fold higher than that of the vector without
promoter. The I8 promoter had approximately 50% the activity of
I4, and the activities of the other seven promoters were lower, but
relatively similar to each other, varying between 5 and 23% of the
I4 activity (Fig. 2A). The data show that the I
promoters differ in their ability to drive expression of the LUC
reporter gene.
|
promoter activities, as measured by LUC gene
expression, differed among the various I genes, the differences did
not reflect in vivo expression. Nilson and Sideras (18)
reported the presence of negative regulatory elements in the human I
promoter upstream of the TGF-RE, while Lin and Stavnezer
(17) reported upstream elements that increased TGF-
inducibility in the murine I promoter. Therefore, it was possible
that the discrepancy between in vitro promoter activity and in vivo
expression of the rabbit IgA isotypes could be due to upstream
regulatory elements that differentially affect the promoters in vivo.
It was also possible that the promoters responded differently to
TGF- activation. To test these possibilities we investigated whether
regulatory elements, positive or negative, upstream of the 180-bp
promoter were present. We cloned fragments of between 430 and 440 bp
(positions approximately -415 to +21) of I3, I4, I8, and
I9 into the LUC vector and assayed for LUC activity. We found that
the activities did not change significantly, implying that regulatory
elements are not present in this upstream region (data not shown). We
also investigated whether the promoter activities of transfected 55D1
cells would change in response to TGF- stimulation. We tested
whether those isotypes that are expressed in vivo at high levels but
have low LUC activity (for instance, I9 and I10) would attain
activity comparable to those of I4 and I8 as a result of TGF-
stimulation. We compared the activities of the promoters of I4,
I8, I9, and I10 of TGF--activated 55D1 cells. We found that
TGF- increased the promoter activity of each isotype (Fig. 2B). The increase was, in two experiments, 2- to 5-fold for
all isotypes. The activities of the I4 and I8 promoters remained
the highest even after TGF- activation. The I3 promoter remained
inactive (data not shown). The results show that with the exception of
I3 and I4, the activities of the I promoters, even if
TGF--activated, do not correspond to their in vivo expression.
Regulatory elements within I promoters
We next searched for nucleotides of the different regulatory
elements of the I promoters that could potentially explain
differential expression of the I genes in the LUC assay. We
truncated the I4 promoter and tested these shortened oligomers for
basal promoter activity (Fig. 3). We
constructed five truncated forms of the I4 promoter by shortening it
from the 5' end and tested these for promoter activities in the LUC
reporter gene assay. One oligomer of 150 bp (positions -133 to +21)
that starts at the 5' end of the TGF-RE (see Fig. 1) and an oligomer
of 128 bp (-106 to +21) in which the TGF-RE was deleted had
promoter activities that were only marginally reduced. Removal of the
TGF-RE reduced promoter activity by only 25% (Fig. 3). When the
TGF-RE and the CRE were both deleted (-98 to +21), the promoter
activity was reduced to 40% of its maximal value. A still shorter
oligomer (-66 to +21) in which the ETS site was also deleted had no
promoter activity (Fig. 3). These results indicate that the 30-bp 5' of
the TGF-RE and the TGF-RE itself have little effect on basal
promoter activity and also that the CRE site is not essential, even
though it increases promoter activity. The ETS site and/or sequences 5'
of it, however, are required for promoter activity. Similar results
have been reported for the human (18, 21) as well as the
mouse (17, 28) I promoters. To confirm the importance
of the ETS site we constructed an I4 oligomer (-98 to +21) in which
four nucleotides of the core element of the ETS site were mutated from
AAGG to CTCG. The activity of this construct was reduced to 15% of its
original value (Fig. 3), confirming the importance of the ETS site for
promoter activity.
|
3, I12, and I8 promoters,
three promoters that have widely different activities from that of
I4. We introduced the nucleotides that differed between the three
promoters and the I4 promoter in the region of the CRE and the ETS
sites (see Fig. 2) into the truncated I4 promoter (Fig. 4A). We cloned these mutated
I4 oligomers into the PGL 3 vector and assayed for LUC activity.
|
3 the region from position -106 to -66 differs from that of
I4 by only four nucleotides (positions -103, -95, -80, and -73;
see Fig. 1). We introduced all four nucleotide changes into the I4
promoter and found that the promoter activity decreased from 129 to 3
times above the control level (Fig. 4B). If only two
nucleotides, -103 within the CRE site and -95 just 3' of it, were
introduced, the activity was reduced to 14 times above the control
level (Fig. 4B). It appears therefore that the difference in
activity between the I4 and I3 promoters is not due to a single
nucleotide, but that at least two, and perhaps all four, of these
nucleotides in the region of the CRE and the ETS elements contribute to
promoter activity.
The I12 promoter differs from the I4 promoter in the CRE and 3'
of it by five nucleotides, which is probably the reason why I12 has
a lower promoter activity than I4. None of the nucleotide
differences is the same as those between the I3 and I4 promoters,
and in the ETS element, I12 has the same nucleotide sequence as
I4 (as do most I genes; see Fig. 1). When we mutated these five
nucleotides to those of I12, we found that the activity of I4 was
reduced from 129 to 42 times above the control level (Fig. 4B). The results show, as for the I3 promoter, that basal
promoter activity is largely dependent on the region containing the
CRE. The results also suggest that the activity is affected by several
nucleotides, but that some, like those by which I3 differs from
I4, are more critical than others.
The nucleotide sequence of the I8 promoter is identical with that of
I4 throughout the TGF-RE and the CRE, but there are four
nucleotide differences in the region 3' of the CRE and in the ETS site
at positions -94, -89, -78, and -73 (Fig. 4A). If the
three nucleotides at positions -94, -89, and -78 of the I8
promoter were introduced into the I4 promoter, the activity was
unchanged, but if, in addition, nucleotide -73 (part of the ETS site)
was introduced, the promoter activity dropped dramatically (Fig. 4C). These results show that nucleotide -73, located two
nucleotides 5' of the core element of the ETS site, is critical for
promoter activity. However, it is important to note that these changes
in activity only became apparent when the mutations were introduced
into the truncated promoters.
Binding of nuclear proteins to I promoters
The data obtained by the LUC assay with the truncated promoters
suggested that the I8 promoter is not active in vivo because of one
nucleotide difference at position -73 in the ETS site, even though the
full-length I8 promoter is highly active in the LUC assay. At this
position the nucleotide of the I3 and I8 promoters is T, whereas
it is C for all other I promoters (the I3 promoter has additional
differences that affect promoter activity), indicating that a correct
nucleotide at this position may be critical. To obtain additional
evidence for the importance of the ETS site we performed EMSA. To first
test whether the 55D1 cells contain the same nuclear proteins as
primary cells we compared the gel shifts resulting from binding of the
180-bp promoters of highly expressed genes, C4 and C9, and
unexpressed genes, C3 and C8 (Fig. 5A) to nuclear proteins from
55D1 cells and from cells of various lymphoid tissues (popliteal lymph
nodes, tonsil, and sacculus rotundus). We found that the gel shifts of
nuclear extracts from primary cells are similar to those from 55D1
cells for all four isotypes. However, the shifted bands produced by the
I3 and I8 promoters were less intense than those produced by the
I4 and I9 promoters, indicating that fewer proteins are bound to
the I3 and I8 than to the I4 and I9 promoters (Fig. 5A). The gel shift from 55D1 nuclear proteins did not change
appreciably if the proteins were prepared from cells activated by
TGF-, CD40L, and anti-Ig Ab (Fig. 5B). These results
suggest that the promoters of I3 and I8 may not be able to
support transcription of GL mRNA because one or more essential
transcription factors cannot bind to the DNA or binds with lower
affinity. These data also show that nuclear proteins of the 55D1 cell
line and of primary cells bind to the I promoters in a similar
manner.
|
8 promoter bind fewer nuclear proteins than the equivalent regions
of the I4 promoter, we designed five short overlapping fragments,
oligomers a–e, (Fig. 6A), and
compared their abilities to compete with binding of proteins to the
full-length 180-bp I4 promoter. When we determined the competition
of binding by oligomers a–e of the I4 promoter, we found that only
oligomer c, containing the ETS-binding site, competed for nuclear
proteins (Fig. 6B). This competition suggests that the same
proteins that bind to the 180-bp promoter also bind to the shorter
I4c oligomer. To test this idea we compared the binding of nuclear
proteins to the 180-bp I4 promoter with that to the I4c oligomer
directly (Fig. 7A). We found
that although both oligomers formed two major protein complexes, one of
the complexes of the 180-bp oligomer was larger than those formed by
the I4c oligomer (Fig. 7A). We conclude that this larger
complex contains more protein than the complexes formed by the I4c
oligomer. The formation of both protein complexes is reduced in the
presence of an excess of I4c, as shown by the competition
experiments presented in Fig. 6B. The other oligomers,
I4a, I4b, I4d, and I4e, competed little or not at all with
binding of proteins to the full-length 180-bp I4 promoter (Fig. 6B). The competition of binding of the full-length I4
promoter by c oligomers of other isotypes (I2c, I7c, I9c,
I10c, and I12c) showed a degree of competition similar to that of
I4c (data not shown). In contrast, when we compared the c oligomer
of I8 to that of I4 for the ability to compete with the
full-length I4 promoter, we found that the I8c oligomer competed
less efficiently (Fig. 6B). These results suggest that the
reduced binding of nuclear proteins to the c oligomer region of the
I8 promoter may be responsible for the lack of C8 expression
in vivo.
|
4c and
I8c to nuclear proteins of 55D1 cells. We found that I8c indeed
binds fewer proteins than I4c (Fig. 7B). We also compared
binding of these two c oligomers with nuclear extracts of cells from PP
and tonsil and found that proteins from primary lymphocytes also bind
better to the I4c than to I8c (data not shown). To determine
which nucleotides of the c probe are responsible for binding the
nuclear proteins we performed EMSA with mutated probes. The I8c
probe differs from the I4c probe by two nucleotides at positions
-73 and -78, two and seven nucleotides 5' of the core element, GGAA,
of the ETS site, respectively (see Fig. 1). To determine whether either
of these two nucleotides in I8 is responsible for the reduced
binding, we constructed I4c probes in which either of these two
nucleotides was replaced with that of I8. We found that by changing
nucleotide -73 the I4c probe bound less protein, while changing
nucleotide -78 had little or no effect on the binding (Fig. 7B). The results show that changing one nucleotide 2 bp 5'
of the core element of ETS (-73) reduces binding of nuclear proteins
to the I promoter. We conclude that the ETS site is important for
binding of nuclear proteins to the I promoter. Taking these data
together with the results obtained by LUC gene expression, we suggest
that binding of factors to the ETS site is necessary for full promoter
activity. Further, we suggest that IgA8 is not expressed in vivo
because the I8 promoter has a defective binding site that does not
support transcription of GL mRNA |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, which is
secreted by various cell types in response to Ag stimulation. TGF-
induces the transcription of an GL mRNA from the I exon,
resulting in the first step leading to isotype switch. Rabbits have 13
IgA genes (1), and each of these genes has its own I
exon preceded by an I promoter (4). The degree of
expression of these genes in vivo varies over a wide range, and two of
the 13 IgA isotypes, C3 and C8, are not expressed in vivo
(3). We tested whether the differences in in vivo
expression are due to differences in the I promoters. We found that
in the rabbit B cell line 55D1, TGF- induces transcription of GL
mRNA of all isotypes except I3 and I8, suggesting that the
endogenous promoters of these two isotypes cannot be activated in
response to TGF- stimulation. The I3 promoter, if transfected
with the LUC reporter gene into 55D1 cells, does not promote expression
of the LUC gene, even when the transfected cells are stimulated by
TGF-, indicating that the I3 promoter is defective.
In contrast to the I3 promoter, we found that the promoter of I8
could drive expression of the LUC reporter gene in 55D1 cells, and that
the activity increased in response to TGF- stimulation. These in
vitro results appear contradictory to the finding that no GL-I8
transcripts are found either in vivo or in TGF--activated 55D1
cells. However, by EMSA, neither the full-length I8 promoter nor a
short oligomer containing the ETS site bound transcription factors as
effectively as promoters of expressed genes, indicating that in vivo
the I8 promoter may not bind transcription factors efficiently. The
reduced binding of transcription factors by the I8 promoter is due
to a single nucleotide change at position -73, adjacent to the ETS
site. ETS elements are present in many promoters and enhancers, and as
many as 17 mammalian ETS proteins have been described
(29). By using mutated promoters in reporter gene assays,
the ETS site has been shown to be essential for promoter activity of
human and murine I genes (21, 28, 30). It remains
puzzling that the rabbit I8 promoter, which also has a mutation in
the ETS site, appears to be highly active in promoting expression of
the LUC gene even though by EMSA it displays reduced binding of factors
to the ETS site. We used the LUC reporter gene assay, assuming that the
regulatory elements would function similarly in vivo and in vitro. It
may well be, however, that in vivo the I8 promoter functions
differently than in the LUC reporter assay because of the chromosomal
context of the promoter. In the context of chromatin, the promoter may
not be accessible to transcription factors or cis elements
required for activation of the promoter. It may also be that the
requirements for transcription factor binding are different for the
transfected and endogenous promoters, as discussed below.
ETS factors are known to interact with other transcription factors and
influence their binding to DNA. Xie et al. (30) suggested
that interaction of ETS and avian myeloblastosis proteins is important
in regulation of the human I1 germline promoter. Because
transcription factors, such as ETS factors, associate not only with
DNA, but also with each other, and affect each other’s binding to DNA,
we speculate that for the full-length I8 promoter, binding of
transcription factor to sites other than the ETS site causes the ETS
site also to be occupied. That the ETS proteins may interact with other
factors that also bind to the I promoter is suggested by competition
experiments in EMSA. We showed that an excess of the 45-bp oligomer
from the I4 promoter containing the ETS site prevented binding of
all or most other nuclear proteins to the I4 promoter. It is
possible that all shifted bands are due to binding of ETS proteins
only. However, it is also possible that the ETS proteins bind not only
to DNA, but to other transcription factors, and influence, positively
or negatively, their binding to cis elements of DNA. We
suggest that the apparent contradiction between in vitro and in vivo
activity of the I promoter results from binding of a "wrong"
transcription factor that may occur with low affinity and be sufficient
to support LUC gene expression in vitro, but not sufficient to activate
the endogenous promoter. We conclude that both IgA3 and IgA8 isotypes
are not expressed in vivo because I3 and I8 have faulty or
inefficient promoters that cannot be induced to express GL mRNA.
Differential expression of IgA isotypes is regulated subsequent to
expression of GL transcripts. We set out to determine whether the cause
of differential expression of the different IgA isotypes lies in
differences in I promoter activities. We found that the activities
of the promoters of the various IgA isotypes, as measured by the
expression of the LUC reporter gene, vary because of nucleotide
differences between them, mainly in the CRE and ETS sites. The I4
promoter has the highest activity, reflecting the high degree of in
vivo expression of C4. The promoters of I3 and I8 each differ
from I4 by four nucleotides in the region of the CRE and ETS sites,
and we showed that these differences probably explain the lack of
expression of IgA3 and IgA8 in vivo. For other isotypes, however, the
promoter activity does not reflect the in vivo expression. For
instance, C9, C10, C12, and C13 are expressed at a level
similar to I4 in vivo; however, their activity, as measured by LUC
gene expression, is much lower. Other discrepancies between in vitro
and in vivo expressions are that the I1, I2, and I11 promoters
have the same in vitro activity as the I9, I10, I12, and
I13 promoters, while their in vivo expression is much lower.
Further, when we compared the frequency of TGF--induced expression
of GL mRNA for the various isotypes in single 55D1 cells, the results
did not show the differences that we observe in in vivo expression.
Rather, with the exception of I3 and I8, all isotypes can be
induced to express GL mRNA at a similar frequency. Other explanations
could be that the 3' enhancers differentially activate the various
I promoters (31). However, since the different rabbit
IGL mRNAs of 55D1 cells are induced by TGF- at similar
frequencies, we think that such enhancers must have similar effects on
all I promoters.
Recently, Volgina et al. (32) reported that the rabbit IgH
locus has a single 3' (hs1, 2) enhancer, and while this element
significantly enhanced the activity of a VH
promoter, it had little effect on an I promoter. We suggest that
neither the strength of the I promoters nor the 3' (hs1, 2)
enhancer is responsible for the differential expression of IgA isotypes
in vivo. Instead, we suggest that the observation that activated single
cells express GL mRNA of more than one isotype is key to an explanation
for the differential expression of the IgA isotypes. Although we
cannot rule out the possibility that the intracellular
concentration of the GL mRNAs, which may be different for the different
genes, is important for the final switch recombination, we do not
think this is likely in view of the fact that no single isotype is
preferentially expressed as GL mRNA. We believe that the expression of
GL mRNA, while essential, is not sufficient to determine to which
isotype the cell will finally switch. Although Revy et al.
(33) have shown that cytidine deaminase is required for
isotype switch recombination, the mechanism of this recombination
remains to be elucidated. One can postulate that of the proteins that
are involved in the actual switch, one or more may be isotype specific
and may therefore play a role in selecting a single isotype to undergo
switch recombination. We propose that the final switch is regulated at
a step subsequent to transcription of GL mRNA and may involve specific
cytokines or other factors that are present in the environment of a
given tissue at a given time.
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Katherine L. Knight, Department of Microbiology and Immunology, Stritch School of Medicine, Loyola University Chicago, 2160 South First Avenue, Maywood, IL 60153. E-mail address: kknight{at}lumc.edu
3 Abbreviations used in this paper: PP, Peyer’s patch; CD40L, CD40 ligand; CRE, cAMP response element; GL, germline; LUC, luciferase; TGF-RE, TGF- response element.
Received for publication June 12, 2001. Accepted for publication January 23, 2002.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
genes. EMBO J. 8:4041.[Medline]
genes as chimeric rabbit-mouse IgA antibodies. Proc. Natl. Acad. Sci. USA 86:7561. and IL-2. J. Immunol. 162:5380. specifically enhances IgA production by lipopolysaccharide-stimulated murine B lymphocytes. J. Exp. Med. 170:1039. and IL-2 enhancement of IgA expression in lipopolysaccharide-stimulated B cell cultures. J. Immunol. 144:952.[Abstract]
1 increases IgA isotype switching at the clonal level. J. Immunol. 145:3773.[Abstract]
1 and C2 germ-line and mature mRNA transcripts in human peripheral blood B cells. J. Immunol. 153:1466.[Abstract]
-induced isotype switching (polymerase chain reaction). Proc. Natl. Acad. Sci. USA 87:3962.1-induced human immunoglobulin C1 and C2 germ-line transcripts. Int. Immunol. 3:1107. constant region gene by an ATF element and by novel transforming growth factor-1-responsive elements. J. Immunol. 149:2914.[Abstract]
1 and I2 germline promoter elements: proximal positive and distal negative elements may regulate the tissue specific expression of C1 and C2 germline transcripts. Int. Immunol. 5:271. germline transcripts in the B lymphoma I.29µ: transforming growth factor- activates transcription of the unrearranged C gene. J. Immunol. 147:4374.[Abstract]
constant region gene. Curr. Top. Microbiol. Immunol. 182:157.[Medline]
1 region contains a TGF-1 responsive enhancer and a putative recombination hotspot. Int. Immunol. 7:1191.: evidence for TGF- but not IL-10-dependent direct Sµ
S and sequential SµS
, SS DNA recombination. J. Immunol. 161:5217.. EMBO J. 18:1280.[Medline]
signaling. Genes Dev. 14:627.3 (AML2) is induced by TGF-1 to bind and activate the mouse germline Ig promoter. J. Immunol. 161:6751.-myc transgenic rabbits. Leukemia 8:2144.[Medline]
B and nocodazole in regulating induction of transcription of mouse germline Ig RNA by transforming growth factor-1. Int. Immunol. 13:733.
) with
the activities of TGF-
). The fold
increase refers to activity relative to the PGL3 vector alone. All
values are normalized to the activity of the cotransfected
Renilla vector. The promoter of I
