ABSTRACT
Early/memory and plasma B-cell lines and fibroblasts were analyzed for their ability to use a 5
'
proximal (variant) versus a 3
'
distal (constant) poly(A) site, in the absence of a competing splice, from a
set of related constructs. The proximal:distal poly(A) site use (P:D ratio) of
the resulting cytoplasmic poly(A)
+
mRNA is a measure of poly(A) site strength. In this context the immunoglobulin
[gamma]
2b secretory-specific poly(A) site showed a P:D ratio of 1:1 in plasma cells, 0.43:1 in
early/memory B-cells and an intermediate value in fibroblasts. Meanwhile, a construct
with a proximal SV40 early-like poly(A) site produced mRNA with a P:D ratio of
&
50:1 in all cell types. Alterations in the region downstream of the proximal
poly(A) addition site and at the site itself resulted in changes in the P:D
ratio. However, these poly(A) sites, all with a P:D ratio of
<=
5:1, were used most efficiently in plasma cells. Constructs totally devoid of immunoglobulin sequences, but containing heterologous poly(A) sites producing mRNA with P:D ratios of
Figure 1.
<=
5:1, were also used more efficiently in plasma cells. We therefore conclude that
weak poly(A) sites, regardless of sequence composition, are used more efficiently in plasma cells than in the other cell types.
A single immunoglobulin (Ig) heavy chain gene can be alternatively expressed as mRNAs for a secretory-specific (sec) or a membrane-specific (mb) protein (
1
,
2
). Although both forms of mRNA are expressed at equal levels in early and memory
stages of B-cell development, there is 20-100 times more sec- than mb-specific mRNA produced in the plasma cell (
3
,
4
). There is >2 kb between the promoter proximal sec poly(A) site and the
downstream mb poly(A) site in the mouse Ig [gamma]2b gene. Splicing of the [gamma]2b encoded CH3 to M1 exons removes the sec poly(A) site from the transcript, thus allowing mb-specific mRNA to be expressed. If cleavage and
polyadenylation occur first at the secretory specific poly(A) site just
downstream of CH3, the CH3-M1 splice is prevented and only sec mRNA is produced. Transfection studies have shown that there is no change in splicing efficiency of a variety of constructs
between early B-cells and plasma cells (
5
-
8
).
In contrast, an increase in the polyadenylation efficiency
in vivo
of Ig sec poly(A) sites was seen in plasma cells versus early and memory B-cells for [mu] and [alpha] (
9
,
10
) and [gamma] sites (
11
). Studies of polyadenylation competent extracts prepared from mouse B-cell lines representing different stages of development have shown that there is as much as an 8-fold increase in the binding of the 64 kDa subunit of CstF and the 100 kDa subunit of CPSF to input polyadenylation substrates in the plasma cell extracts as compared with the early or memory B-cell extracts (
12
,
13
). The increased binding occurs without a commensurate increase in the amount of
either protein (
12
,
13
). An increase in the efficiency of the polyadenylation reaction in plasma cells could result in a large increase in the relative amount of the sec mRNA since use of the sec poly(A) site would
preclude splicing to the membrane exons. In addition, a factor was purified
from early/lymphoma B-cells which accelerates the decay of Ig sec polyadenylation complexes (
14
). Differential expression of the mouse [gamma] and [alpha] Ig heavy chain genes may be solely the result of regulation of polyadenylation
since transcription termination does not play a role in their expression as it does
in the [mu] Ig gene (
15
-
19
).
We previously studied the tissue specific polyadenylation of the Ig [gamma]2b gene by making deletions and substitutions at the sec poly(A) site in constructs stably transfected into cell lines showing processing patterns like either early or plasma B-cells (
11
,
20
). In constructs with different 5' proximal poly(A) sites in competition with a constant 3' distal site, we were able to measure the relative strengths of poly(A) sites by determining the proximal to distal site use
(P:D ratio). A strong site is defined here as one in which the P:D ratio
exceeds 30:1 while weak sites show a much lower value (~5:1 or less). By comparing the P:D ratio between different constructs in a given cell-type, the effect of nucleotide substitutions on the proximal poly(A) site use can be assessed. By comparing the
P:D ratio between cell-types for a given construct the effect of the presence or levels of
trans
acting factors which act on poly(A) sites in that cell can be assessed.
Our previous studies showed that the relatively weak Ig [gamma]2b secretory polyadenylation site was more efficiently used in plasma
cells than in earlier B-cell stages. In contrast, we observed no B-cell specific regulation of the SV40 early or the [gamma]2a membrane poly(A) sites in linked poly(A) site constructs (
20
). Since the strong SV40 early and the weak, regulated Ig sec poly(A) regions in
these constructs differ by only a small number of nucleotides in the region
downstream of the proximal poly(A) addition site and at the site itself, we
made substitution and alterations in these regions. These alterations result in
up to 7-fold changes in the P:D ratio in a given cell type. However, these poly(A)
sites, all with P:D ratios of <= 5:1, are used most efficiently in plasma cells. We made other constructs
totally devoid of Ig sequences and show that weak [alpha]2-globin and viral poly(A) sites are also used more efficiently in
plasma cells. We conclude that weak sites (P:D ratio <= 5:1) regardless of sequence are regulated in a B-cell stage-specific manner. This set of constructs was also analyzed in mouse L
cells, a fibroblast line, in which the splicing reaction was shown to dominate
over polyadenylation at the Ig [gamma] sec poly(A) site in intact Ig constructs with splicing and
polyadenylation in competition (
6
). We find that weak poly(A) sites are generally less efficient in L cells than
in plasma cells.
pSm series.
The 1285 nt
Sst
I-
Sma
I fragment from the mouse immunoglobulin [gamma]2b CH3-M1 intron was cloned into the
Sst
I and
Sma
I site of pGEM4 (Promega). An 830 nt fragment, containing the sec poly(A) site
and downstream region, was removed by
Kpn
I digestion creating pSq[Delta]Kpn the forerunner of pSm0. The 351 nt region downstream of the sec
poly(A) signal is identical to that found in the [Delta]Kpn constructs from Kobrin
et al.
(
20
) and which, when part of the entire [gamma]2b gene, directs virtually no polyadenylation on its own. Subsequent
constructs were made by inserting double stranded oligonucleotides with
Asp
718 ends into the
Asp
718 (
Kpn
I isochizomer) site within the
SstI
-
Sma
I fragment of pSq[Delta]Kpn. The double-stranded oligonucleotide was 20 nt for pSm0* and 31 nt for all the
other pSm1 or 2 constructs. Following insertion of annealed complementary oligonucleotides, the constructs were sequenced using the Genesis 2000 system (Dupont) to confirm
orientation and integrity. The sequences of the top strands of the inserted
oligos are shown in Figure
1
; the bottom strands are not indicated but they all extend 4 nt to the right (5') side with 5'-GTAC-3'. The 455 nt
SstI
-
Sma
I (lacking the
Kpn
I-
Kpn
I interval) fragment, but now containing the oligonucleotide, was removed from
the pGEM4 constructs by digestion with
Eco
RI and
Sma
I. The fragment ends were made even by a Klenow DNA polymerase fill-in reaction and inserted into the
Hpa
I site of the pSV2gpt Eh5a vector, which has the Ig enhancer added to pSV2gpt;
addition of the enhancer increases the transfection efficiency of the plasmid
in lymphocytes (
11
,
21
). This created the pSm series of plasmids which were stably transfected into
the eucaryotic cells; the distance between the inserted poly(A) site and that
of the downstream vector poly(A) site, the SV40 early site, in these constructs
is 452 nt.
The mouse plasmacytoma cell line J558L is representative of a late stage or
plasma B-cell line (
24
). The mouse cell line A20.2J subclone #1 was used because it produces an endogenous [gamma]2a message at 2:1 sec:mb levels (
20
). A20 is representative of a memory or lymphoma B-cell, an earlier step in development than the plasma cell. Lipofectant
reagent (BRL) was used to introduce pSV2gpt derived plasmid DNAs into mammalian
cells following the manufacturer's recommendations. Positive transformants arose after 2-4 weeks following the mycophenolic selection schema previously described (
20
,
24
). Mouse L cell transfectants were selected using the conditions described for
J558L cells (
20
). Clones from at least three and generally four independent transfectants of each cell
type were grown to mass culture under conditions of drug selection and cytoplasmic poly(A) containing RNA was isolated using oligo (dT)-cellulose (Collaborative Research). Each tranfectant was analyzed in the T2 ribonuclease protection assay a minimum of three times. Because of the previously observed pleomorphic
nature of the line, levels of endogenous [gamma]2a mRNA in the individual A20 transfectants were monitored by Northern
analysis and those rare clones exhibiting >3:1 sec:mb mRNA ratios were
eliminated from further study. Removal of the mycophenolic selection from the
clones for >5 days had no effect on the poly(A) site use (data not shown).
Poly(A) site use was assayed as described previously (
25
) by hybridizing the cytoplasmic poly(A)+ RNAs from the transfectants with
32
P-U-labeled antisense probes generated from the pGEM clones containing
the appropriate fragments. The hybridization mix was subjected to T2 ribonuclease digestion. The protected fragments were resolved on denaturing urea:polyacrylamide gels. Cells which were not transfected with the constructs show no protected fragments (Fig.
7
and data not shown). Labeled molecular weight markers used as standards
included 1 kb DNA ladder (BRL) and Marker V (Boehringer Mannheim) labeled with
[
32
P]ATP and kinase. Gels were exposed to X-ray film with an intensifier screen at -70oC for between 24 h and 5 days for the photographs.
Quantitation was done on the Molecular Dynamics Phosphoimager with exposures in
the linear response range of the instrument. A correction factor for the number
of U residues in the sequence was applied to the promoter proximal to distal
ratio. Results were analyzed using the Instat (GraphPAD Software, Inc.) Statistical Package and Microsoft Excel to determine standard deviations and standard errors of the mean. For each
construct the J558L data were compared with the A20.2J data, using the Students
T test, to exclude the null hypothesis,
P
= <0.05 and passed that test in all cases.
We have used an approach which assesses both the strength and the contribution to B-lineage regulated expression of sequences around the secretory specific
poly(A) site of the Ig [gamma]2b heavy chain gene by allowing it to control polyadenylation of a non-immunoglobulin gene, bacterial guanosyl phosphotransferase (gpt). This gpt gene, present in the pSV2gpt based vector, confers
mycophenolic acid resistance to stably transfected cells and is under the
control of SV40 early promoter and polyadenylation sequences (
24
). The unique
Hpa
I site found 83 bp upstream of the vector SV40 early poly(A) site of pSV2 Eh5a
was used to insert a series of mutations of the Ig [gamma]2b secretory poly(A) site, making the pSm constructs. All splice sites
are missing from this interval. The inserted sequences are shown in Figure
1
; otherwise these constructs are identical to each other. The region downstream
of the proximal poly(A) site is identical to that found in the [Delta]Kpn constructs (
20
) and which, when part of the entire [gamma]2b gene, directs virtually no polyadenylation on its own.
Constructs were transfected into either the A20.2J line, a Balb/c mouse tumor
line representative of an early or memory B-cell line, J558L, a Balb/c mouse tumor line representative of the latest
stage in B-cell development, the plasma cell, or mouse L cells, a fibroblast-like line representative of the non-Ig regulated state. We analyzed poly(A)+ cytoplasmic RNA from
at least three, and generally four, independent transfectants of each cell type
that were stably transfected with the gpt constructs. Each transfectant was
analyzed by a T2 nuclease protection assay a minimum of three times. To
determine how often each of the two possible poly(A) sites had been used in
mRNA processing, we set up the nuclease protection assay illustrated in Figure
2
. Transfected cell mRNAs from pSm1(Ig sec) or pSm2(SVE) constructs which have
used the promoter distal, vector poly(A) site, will protect a fragment from T2
ribonuclease digestion of 576 nt. Transfected cell mRNAs which have used the
promoter proximal, or oligonucleotide modified, poly(A) site will protect a
fragment of 477 nt. Several representative nuclease digests are shown in Figure
2
; pSm2(SVE), which contains the sequence from the SV40 early poly(A) cleavage
site and downstream region, shows exclusive use of the promoter proximal
poly(A) site versus distal site (P:D ratio &50:1), regardless of which cell was transfected. We call this site strong
because of the high P:D ratio and unregulated since it shows no cell type
specificity, a conclusion consistent with the observation we made previously
with a similar construct (
11
). In construct pSm1(Ig sec) the oligomer has the [gamma]2b secretory-specific poly(A) site and its downstream sequence. As shown in Figure
2
and summarized in Figure
3
, in J558L transfectants of pSm1 the promoter proximal, Ig sec, poly(A) site is
used about as often as the promoter distal, vector, poly(A) site in the
production of poly(A)+ cytoplasmic RNAs (1:1 ratio). Therefore, the Ig sec
poly(A) site of pSm1 is very weak relative to the SV40 early-like poly(A) site in pSm2. However, in A20.2J cells transfected with
pSm1(Ig sec) the proximal, inserted poly(A) site is even less efficient than in
the J558L cells (0.43:1 proximal to distal ratio). The regulation index (1:1
for J558L divided by 0.43:1 for A20) indicates a 2.6-fold more frequent use of the promoter proximal site in the plasma cell
tumor than in the early, memory B-cell tumor. The L cell value falls between these two. Since the distance
between the poly(A) sites was kept small (452 nt), the regulation index is not as large as had been previously seen with other constructs or the intact Ig gene (
11
,
20
,
21
). The small distance minimizes the possibility of transcription termination which generally occurs >1
kb downstream of a poly(A) site in this system (
15
,
16
).
In this study, poly(A) site strength or weakness is assessed by comparing the
amount of use of the promoter proximal versus promoter distal poly(A) sites,
the P:D ratio, in the absence of a competing splicing reaction by using
mutations in the region around the mouse Ig [gamma]2b sec poly(A) site. To determine the extent of B-lineage regulation, constructs were stably transfected into mouse
cell lines representing different B-cell stages or into fibroblasts (L cells). In pSm2, a [gamma]2b sec poly(A) site was modified, just at the cleavage and
downstream region, to look similar to that part of the SV40 early site; when
this site was in the promoter proximal location it was used predominantly, with
a P:D ratio &50:1, and was therefore a `strong' site in our assay. The SV40 early-like poly(A) site of pSm2 was used to the same same extent regardless of
cell type transfected. All the other modified [gamma]2b sec proximal poly(A) sites were used to a higher degree in the plasma
cell (J558L) than in the early, memory or lymphoma cell (A20); plasma cells
were more efficient at polyadenylation than L cells for all but one of these
constructs. This increased efficiency in plasma cells was also seen in
constructs totally lacking Ig sequence but instead containing ~480 nt of the human [alpha]2-globin poly(A) site or a fragment of ~600 nt with a minor SV40 poly(A) sequence repeated four
times. We therefore conclude that
cis
acting sequences are not responsible for the observed increase in proximal
poly(A) site use in plasma cells. Instead, it is the weakness of a site which
allows it to be acted upon
in vivo
by the increased efficiency of the polyadenylation machinery in plasma cells.
Having a very strong site (like that in pSm2) proximal to the promoter, allows
that site to be used to such a great extent relative to the downstream site
that an increase in efficiency may not be observable; thus, regulation can be
seen only on weaker sites. In contrast,
in vitro
, with sites that are not tandemly linked, we can see an increase in the UV-cross linking of constitutive polyadenylation factors from plasma cell extracts to both weak Ig secretory poly(A) sites
and a variety of strong sites (
12
,
13
). This indicates that all sites are enhanced in plasma cells but weak sites
which are promoter proximal are the ones which benefit most from this type of control mechanism
in vivo
. Taken together, the evidence indicates that
trans
acting factors change during differentiation to plasma cells and Ig secretory
mRNA expression; this transition may either involve alterations to generic
polyadenylation factors or induction of plasma cell specific factors to increase polyadenylation efficiency. The amount of 64 kDa CstF and 100 kDa CPSF essential
polyadenylation factors do not change between the cells (
12
,
13
, and data not shown). Differences in growth rates cannot account for the
differences we see since all three cell types grow with approximately the same
doubling time and all transfected cells were harvested in the exponential phase
of growth.
The distance between the poly(A) sites was kept to a minimum in all our
constructs (452, 481 and 612 nt), so that transcription termination, which
generally occurs >1 kb downstream of most poly(A) sites, would not be a
confounding variable in the poly(A) site choice assay. With the short distance
between sites the difference in the proximal to distal ratios in J558L versus
A20 is not as large as has been previously seen with other constructs or the
intact Ig gene. Even a small change in polyadenylation efficiency can result in a large increase in the use of the secretory specific
site in the intact Ig gene because of the large (>2 kb) distance between the poly(A) sites, which would allow polyadenylation factors to work on
the first (sec) site before the second (mb) site is even transcribed by the slowly advancing RNA Pol II.
Using constructs with competing splice versus polyadenylation sites, it has been
shown that non-Ig RNA can be alternatively processed in a B-lineage regulated manner (
5
). Our results agree with those and extend the findings by showing that
regulation is seen at non-Ig poly(A) sites alone, in the absence of a competing splice site, and by
showing that sequences at the cleavage site and downstream region, while influencing poly(A) site strength, all show B-lineage regulated expression.
When Brown and Morrison looked at mouse L cells transfected with the intact and two different mutant Ig [gamma]2b genes, with splicing and polyadenylation sites in competition, the constructs
produced heavy chain mRNAs with patterns, on the aggregate, of promoter
proximal (sec) poly(A) site use intermediate between that seen in the A20 and
J558L cell transfectants (
6
). With our constructs the L cells show use of the promoter proximal weak poly(A) sites at a level
intermediate between that in the A20 and J558L transfectants, suggesting that the relative weakness of the [gamma]2b sec poly(A) in L versus plasma cells caused the sec site to be passed
over in the L cells in the Brown and Morrison constructs.
The constructs pSm1,2 and pSm2,1 are at least as regulated as pSm1(Ig sec) but
are much stronger in comparison. The increase in site strength relative to
pSm1(Ig sec), in this case, is due to the inclusion of SV40-like sequences both at the cleavage site and in the downstream region. The
effect of having both SVE elements in pSm2 (SVE-like) is more than additive, indicating a cooperativity between the cleavage site and the downstream element. Meanwhile, the constructs pSm1,3', pSm1,5', and pSm1,5'3' have substitutions in the Ig secretory specific downstream
consensus which make them more similar to the stronger SV40 early downstream
region of pSm2. However, pSm1,3' is unexpectedly weaker than pSm1(Ig sec), even while extending the run
of GT sequences. The mutation does interrupt a pyrimidine run, which, coupled
to the TGGTT core element, may be important for efficient poly(A) site use. The
construct pSm1,5'3' has a downstream region that is very similar to pSm1,2 except for
three out of the last four nucleotides. The weak use of the proximal poly(A)
site in pSm1,5'3' may reflect the complex nature of regulation at this site and of
the
trans
acting factors which could bind; the sequences from +17 to +21 downstream of
the poly(A) cleavage site must play a role in these interactions.
An intact Ig [gamma] or [mu] gene has a secretory terminal exon with a 3' splice site, 360 nt of coding sequence, a 5' splice site, ~100 nt of 3' UTR and a poly(A) site. A weak poly(A)
site at the end of the terminal secretory exon is needed to maintain regulated processing of that exon (
11
,
16
,
21
). While constitutive splicing seems unaffected between B-cell stages (
5
-
8
), the interplay of splicing and the poly(A) site during exon scanning must be
altered if poly(A) site efficiencies change. Understanding of the interaction
between the 5' splice site and the polyadenylation reaction is just beginning to emerge
(
28
-
30
). Continued study of the Ig gene and its regulation may help us understand
regulated polyadenylation, exon definition and the interplay of both.
We thank Makiko Hartman for superb technical assistance. This work was supported by National Institutes of Health Grant GM-50145 to C.M., a grant to the Pittsburgh Supercomputer Center (RR06009), and
fellowship support to S.M. from the Richard S. Caliguiri Amyloidosis Foundation
and the Keck Foundation.
*To whom correspondence should be addressed. Tel: +1 412 648 9098; Fax: +1 412
624 1401; Email: chris@hoffman.mgen.pitt.edu
+
Present address: Oak Ridge National Laboratory, PO Box 2008, Oak Ridge, TN
37831, USA
REFERENCES
Return