| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,
,
*
The Hooper Foundation, San Francisco, CA 94143; and
Ontario Cancer Institute, and
Department of Immunology, University of Toronto, Toronto, Ontario, Canada
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In vivo, the heavy (H) chain is usually rearranged before any light chain (10). In the normal course of events, a D element is first joined to a JH element, and the resulting DJH segment is then joined to a VH element to form the VHDJH segment, which, if in frame, will encode the variable domain of an H chain (11). The germline gene segments are not used randomly (12, 13, 14, 15). At the mouse H chain locus, the most overused VH segment is V81x, the most D-proximal functional VH (13, 16, 17). We and others have demonstrated the overuse of this segment early in B cell development and the decline in its usage at later stages (15, 16, 18).
The rearrangement order may, in part, be a consequence of the rate at which the various segments rearrange. According to this model, the faster recombining segments would usually rearrange before the slower ones. Several studies in vivo and in vitro have investigated what factors affect this rate. The RSSs constitute one factor; the more closely an RSS sequence resembles the consensus (CACAGTG-spacer-ACAAAAACC), the more often it is recombined in extrachromosomal substrates (3, 19, 20, 21). The coding nucleotides adjacent to the RSSs constitute another factor; flanks of C or G are preferred over A or T (7, 22, 23, 24, 25). In addition to the sequences of the elements to be rearranged, their chromosomal location and accessibility affect the order of rearrangement (9, 26, 27, 28). These same factors will also affect the usage of individual V gene segments. In addition, it has been speculated that the extreme overuse of V81x in the primary repertoire is due its D-proximal location (13).
As soon as the primary repertoire has been generated, it begins to be
altered as a result of selection and somatic mutation. The decline in
usage of V81x seems to result from cellular selection, possibly
mediated by interaction of the heavy chain with 
5 surrogate light
chain (29, 53).
To understand the mechanisms for the initial overuse of the V81x gene in the absence of the cellular selection processes described above, we have constructed a series of microlocus plasmids, each of which contain germline coding sequences, RSS, and flanking sequences of a VH, D, and JH gene segment. With this plasmid configuration, we can compare, for example, the rearrangement of two VH gene segments to the same D and the same JH at the same distance away. Moreover, the recombination potential of the segments is assessed in the context of their natural flanking DNA sequences, but in the absence of any higher order chromatin structure, as well as in the absence of cellular selection. These plasmids lack VH promoters and, as no constant domain is included, they cannot produce functional µ polypeptide chains. Our aim is to uncouple the process of recombination from other cellular factors to investigate the extent to which that process itself can generate the repertoire in the absence of other cellular factors and processes that also affect it.
Here, we report the results of experiments with microlocus plasmids in which the VH is either V81x or VA1, with the other two elements always being DFL16.1 and JH1. VA1 is a member of the J558 family of VH gene segments, located at the 5' end of the VH locus. We show that when transfected into A-MuLV-transformed cell lines: 1) they give rise to most of the recombination products observed in intact animals, albeit at different relative frequencies; 2) the junctional sequences of these rearrangements closely resemble those that arise in intact animals, and in particular, the presence of Cys92 and the absence of DH reading frame 3—two characteristics that would be enforced by selection in intact animals—seem to be largely determined during the process of V(D)J recombination; and 3) V81x recombines much more frequently than VA1. This is a direct demonstration that many of the properties of VHD and DJH junctions as well as much of the differential usage of Ig gene segments can be attributed to the sequences of VH, D, and JH gene segments themselves in the context of local flanking sequences.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A-MuLV-transformed cell lines 204-1-8 and A3A-B6, both derived from adult bone marrow, were the kind gift of Naomi Rosenberg, Tufts University (30). Other A-MuLV lines were generated by us (31). Cell lines were grown in RPMI 1640 medium supplemented with 10% FCS (Life Technologies, Gaithersburg, MD), antibiotics (100 U/ml penicillin and 100/ml streptomycin sulfate), 50 µM ß-mercaptoethanol, 2 mM L-glutamine, and 24 mM NaHCO3 at 37°C and in the presence of 5% CO2. Cells are maintained at a density of 5 x 105 to 2 x 106 per ml.
Plasmid construction
Extrachromosomal recombination plasmids were constructed by
assembling germline Ig H chain gene sequences into pJC119
, a 6.2-kb
backbone plasmid containing all the sequences needed to replicate in
eukaryotic cells as well as to replicate and be selected
(ampicillin) in prokaryotic cells (19, 32). Plasmid pJC119
was derived from pJC119, which we previously used in extrachromosomal
V(D)J recombination assays by removing sequences containing the RSSs
(19). The H chain gene sequences were fragments isolated from other
plasmids, PCR-generated fragments, or double-stranded oligomers.
pV81x-D-J series.
These plasmids contain an 
1.4-kb BamHI fragment with
V81x, DFL16.1, and JH1 sequences. A 329-bp
PvuII-PstI fragment isolated from pV81x (a kind
gift of F. Alt, Harvard University) (13) includes V81x sequences
from the fourth codon through the third nucleotide downstream of the
RSS. An 860-bp BamHI fragment isolated from pN25 (a kind
gift of S. Tonegawa, Massachusetts Institute of Technology) includes
DFL16.1 sequences from 355 bp upstream of the 5' RSS through 426 bp
downstream of the 3' RSS (the D sequence is reported in 33 . A
120-bp XhoI-SalI PCR product includes
JH1 sequences from 20 bp upstream of the RSS through 10 bp
downstream of the coding sequence. It was generated from pGW3, which
contains all JH segments in an 3 kb
BamHI-EcoRI fragment by PCR methodology using
primers 5'J1CY and 3'J1CY. The three segments were assembled in
pBluescript (Stratagene, La Jolla, CA), and the 1.4-kb
BamHI fragment was then inserted into pJC119 in both
orientations. In all constructs, the 1.4-kb DNA sequence was verified
by sequencing at this stage.
We constructed two versions of the V81x-DFL16.1-JH1
microlocus differing only at the junction of V81x and DFL16.1. In one
version, the V81x and DFL16.1 fragments were juxtaposed; in the other
version, an additional 34 bp, containing PstI,
HindIII, ClaI, SalI, and
XhoI sites were inserted at the junction (pV81x-D-J Xho).
The two versions did not differ in recombination frequency. Each
version was cloned into the backbone vector in both orientations to
give plasmids pV81x-D-J+ and pV81x-D-J-, with
the + and - superscripts referring to the two orientations. The
constructs are shown schematically in Figure 1
.
|
20
XhoI-resistant plasmids and identified 3 that were sensitive
to SmaI; this means that the embedded heptamer TACTGTG
had been replaced with TCCCGGG. The plasmids were sequenced in the
relevant areas, and one was subcloned and used to generate
pEmBed--D-J series of plasmids, as described above for the
pV81x-D-J series.
pVA1-D-J series.
These plasmids differ from the pV81x-D-J series only in the
VH sequence. A 373-bp BamHI-SspI
fragment from pA1 (a kind gift of F. Alt, Harvard University; and
described in 11 includes VA1 sequences from the third nucleotide
of the first codon through 42 bp downstream of the RSS. It was cloned
into pV81x-D-J, where it replaced the V81x fragment. The resultant
plasmid, pVA1-D-J, contains a 1.5-kb
VHA1-DFL16.1-JH1 microlocus. The junction of
VA1 and DFL16.1 has 18 bp containing ClaI, SalI,
and XhoI sites. DFL16.1 and JH1 fragments are
directly juxtaposed to each other, exactly as in the V81x-D-J
microlocus. The VHA1-DFL16.1-JH1 microlocus was
also cloned in the other orientation with respect to the sequences of
the backbone plasmid and designated pVA1-D-J+ and
pVA1-D-J-, respectively. The constructs are shown
schematically in Figure 1
.
Recombination assay in vitro
The assay was performed essentially as described (34). Briefly, 1 µg of DNA was transfected into 2 x 107 A-MuLV cells in a DEAE-dextran solution. Transfectants were incubated in the presence of 1 mM caffeine (ICN Pharmaceuticals, Costa Mesa, CA). Cells were harvested after 48 h, and DNA plasmids were recovered by alkaline lysis. The recombinant plasmids were isolated by transformation. Typically, 104 to 105 transformants were recovered.
Analysis of recombination products
Colony blotting analysis of individual DNA clones.
After electroporation of the DpnI DNA recovered from the
cell lines (described above), bacteria were incubated for 1 h in 1
ml SOC broth (SOB plus 20 mM glucose and 10 mM MgCl2
(35)) at 37°C with shaking (250 rpm) before being plated on LB
agar supplemented with 100 µg/ml ampicillin. After an overnight
incubation, colonies were picked and streaked onto fresh plates
overlaid with gridded nitrocellulose filters (70 colonies per
plate). Colony blotting of the bacteria was performed essentially as
described (35). From a random sample of recombinant colonies identified
by sequential oligomer hybridization (described below), DNA was
isolated and analyzed by restriction mapping and sequencing with a T7
sequencing kit and V1CY or polyoma primers. To enrich for certain
products to allow sequencing analysis, a selection step was sometimes
added; DNA recovered from transfection was treated with MscI
(to select for V-DJ recombinants) or XhoI (or
BglII, to select for VD-J recombinants). As plasmids
containing VDJ rearrangement products have neither restriction site,
they are present in both populations. This step was omitted in all
experiments used to determine frequencies of recombinants.
Southern blotting analysis of DNA populations. DNA recovered from transfections was treated with DpnI to digest and thus remove nonreplicated plasmid DNA. DpnI-treated DNA was transformed into ElectroMax DH10B-competent bacteria (Life Technologies) by electroporation with a GenePulser (Bio-Rad, Hercules, CA). In a typical experiment, one-half of the recovered DNA was treated with DpnI, and 1 to 10% of this treated DNA was used to transform 25 µl of competent bacteria. Transformants were amplified for 16 h in an additional 4 ml of LB containing 100 µg/ml ampicillin. Plasmids were recovered by alkaline lysis and digested with BamHI to release the insert from the vector. The resultant DNA fragments were fractionated by gel electrophoresis and analyzed on Southern blots with the oligomer probes described below. The microlocus and its various rearrangement products differed sufficiently in size for us to identify all recombinants generated by deletion with a single probe (usually 3'J1CY). Some products generated by inversion are the same size as the unrecombined plasmid, and these were identified with the VinvD and invDJ oligomers. The hybridization and washing conditions were according to the manufacturer’s suggestions (Hybond N; Amersham, Oakville, Ontario, Canada) and are described below. The bands were quantified on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) with ImageQuant 3.3 software.
Standardization of the assay conditions
We used two types of control experiments to ensure the quantitative accuracy of this assay:
1) Bacteria were transformed with 10 ng each of unrearranged
pV81x-D-J+ as well as rearranged pV81xD-J+ and
pV81x-DJ+ plasmids. Transformants were amplified as
described above, and plasmids were recovered, digested with
BamHI, electrophoresed, blotted on Hybond-N (Amersham)
membrane, and probed with 3'J1CY. As no further rearrangements were
detected from any of the three substrates (see Fig. 5A), we
conclude that these plasmids do not undergo detectable V(D)J
recombination-like changes in bacteria.
|
B). The 1.4-kb unrearranged pV-D-J
substrate was underamplified and the small pVDJ product was
overamplified, perhaps because of their different sizes. Nevertheless,
the ratio of VD-J to V-DJ in the output was very close to the 3:1 ratio
in the input. We conclude that our assay is valid for determining the
relative frequencies of VD-J and V-DJ recombinants. Oligodeoxynucleotide probes and PCR primers
Oligomers were purchased from Life Technologies and used as
described in Table I.
|
Genomic DNA was prepared from 20 A-MuLV day 17 fetal liver lines (31). All lines were RAG-positive with H chain loci in the DJ or VDJ configuration (36). Bone marrow and fetal liver DNA was prepared from C57BL/6 mice using standard methods. To detect VD and VinvD structures, the forward VHall primer or V81x primers were used with the Dgen or the invDgen primer. These primers and their characteristics have been described previously (15, 37, 38).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
1. The
microlocus substrate can undergo V(D)J-mediated recombination by
deletion or inversion to give rise to at least seven products. These
products are expected to arise, if at all, at very different
frequencies. Figure 2 shows the possible
rearrangement products containing coding joints, as well as the sizes
of these products after BamHI digestion of the plasmids,
which releases the microlocus. Other events giving rise to
hybrid junctions and open and shut coding/RSS junctions are also
possible. The microloci used in our study differed from those described
previously by other laboratories in that they lacked prokaryotic
transcription stop sequences and the chloramphenicol resistance gene in
the sequences flanking the gene segments. Thus, as there was no way to
select recombinants directly by antibiotic resistance, the frequency of
recombinants was determined by 1) colony blotting and manual
quantification of bacteria harboring plasmids containing the various
products classes and 2) PhosphorImager quantification of the relevant
bands identified by Southern blot analysis.
|
Figure 3 shows an example of the
colony blot analysis used to calculate the frequencies of the several
classes of recombinants. The colonies shown here were probed
sequentially with 5'DFL16.1 (Fig. 3A), 5'J1CY (Fig. 3B), and VinvD-J (Fig. 3C). As
illustrated, the first of these three oligomers is complementary to a
region between V and D, the second to a region between D and J, and the
third to a signal joint present only in the VinvD-J product. Thus, the
colonies negative for 5'DFL16.1 only or 5'J1CY only represent VD-J and
V-DJ recombinants, respectively, while those negative for both are VDJ
recombinants. The colony blot method is particularly useful for giving
the frequency of recombinants that arise by inversion. As no DNA is
lost during inversion, such recombinants can have the same size as
unrearranged plasmids and thus be indistinguishable by the Southern
blot detection method. In Table II, the
data generated in this experiment are tabulated. In the 264 plasmids
analyzed from transfections with pV81x-D-J+ plasmids (in
the positive orientation), there were 23 (8.7%) pVD-J+, 3
(1.1%) pV-DJ+, and 6 (2.2%) pVinvD-J+
plasmids. In the 312 plasmids analyzed from transfections with
pV81x-D-J- plasmids (in the negative orientation), we
found 23 (7.4%) pVD-J, 2 (0.64%) pV-DJ, and 21 (6.7%) pVinvD-J
plasmids. Pooling the data from both orientations, we found 13% V to D
joins, and 0.87% D to J joins. Although in other experiments we have
found 0.5 to 1% pVDJ plasmids with two joins, there were none in the
experiment shown in Table II. pV-invDJ plasmids were not found in any
experiment.
|
|
0.1%, the product of the
frequencies of plasmids with each single event. Sequences of the recombinants
The recombinants that arise by inversion and then deletion would
not be distinguishable from the single deletional VD-J and V-DJ
products by our methods. Thus, to verify and analyze recombinants, DNA
from plasmids identified by various methods were prepared and
sequenced. Figure 4, A–H,
shows sequences generated from several experiments. Of 198 sequences,
197 were the type designated by the colony blotting analysis, a result
that validates the methodology. Overall, the recombinants do not differ
qualitatively from endogenously generated Ig gene structures. Nine had
longer deletions at the joints than are generally observed in vivo
(<5%); there was one hybrid joint.
|
Homology joining. The 3' sequence of DFL16.1 (CTAC) is the same as the 5' sequence of JH1, a circumstance that permits joining by homology (8). In 21 of 67 pV-DJ sequences (31%), nucleotides at the joint could be assigned to either D or J, the usual criterion for joining by homology. This is similar to the frequency of DFL16.1 and JH1 joining by homology in intact animals (40, 41), further confirming the parallel between end processing in our system and intact mice.
Deletions at the VH border. Generally, the processing at the VD and DJ borders was similar to that found in vivo, including the almost complete absence of deletions into the embedded heptamer of the VH. The last fact has been taken as evidence of selection in vivo for the protein sequence encoded by the embedded heptamer, which is the site of an invariant Cys (42). Alternatively, we hypothesized that protein binding to, or structural features of, the embedded heptamer may prevent processing through it. We constructed vectors lacking the embedded heptamer to test the viability of this notion; experiments with the modified vector (pEmBed--D-J) are described below.
No reading frame 2 (RF2) in VDJ structures.
In the VDJ and DJ structures, it can be calculated what the
reading frame of DJ would have been if it were an endogenous gene
segment. Although about equal reading frame usage is found in the
DJH structures, there is no RF2 in the VDJ structures
(Table III). In one model to account for
this omission, DJ structures using RF2 would be inhibited from
secondarily rearranging to yield VDJ products. In an alternate model,
the VDJ recombinants would be derived mostly from the VD-J
intermediate. The latter model is tested below.
|
Is it possible that VD recombinants are more frequent in the
microlocus system because the secondary recombination of VD-J to VDJ
joining is disfavored? To investigate this possibility, we transfected
a pV81xD-J and pV81x-DJ clone (Figs. 4A, UP, and
5D, BX-16, respectively)
separately into A-MuLV line 1–8 and assayed the recombinants. As shown
in Figure 6, both recombinants underwent
second rounds of recombination. Thus, there is no sequence on pVD-J
plasmids preventing another recombination event, and therefore this
cannot be the reason for the accumulation of VD-J structures. Moreover,
the finding that pVD-J generates fewer pVDJ than does pV-DJ
demonstrates that in the absence of competition from each other for D
joining, there remains a preference for V81x over JH in the
recombination reaction.
The embedded heptamer does not augment V to D rearrangements
The predominance of VD recombinants and the absence of deletions
into the embedded heptamer contained in the V gene segment led us to
consider whether the embedded heptamer sequence itself might influence
the outcome of recombination. The embedded heptamer resembles an RSS
heptamer and is found about seven nucleotides 5' of the VH
RSS in the opposite orientation (12, 43, 44). We reasoned that the
embedded heptamer on the microlocus substrate might recruit recombinase
components to the VH segment causing D to join to the
VH RSS more frequently than to J, which lacks an embedded
heptamer. Using site-directed mutagenesis, we altered TACTGTG (the
embedded heptamer) of the V81x-D-J plasmid to TCCCGGG, transfected it
(pEmBed--D-J) into an A-MuLV cell line, and assayed the
recombinants by Southern analysis. The construct lacking an embedded
heptamer also yielded more VD than DJ structures (3.4 and 0.63%,
respectively, in one representative experiment) with VD-J:V-DJ ratios
comparable to those generated from the microlocus with an intact
embedded heptamer sequence (data not shown). Moreover, the pattern of
deletions into V81x is similar to that found in the unmutated V gene
segment (Fig. 4H). Thus, the embedded heptamer sequence is
not a factor in the V81x microlocus constructs yielding more VD than DJ
recombinants, nor is it directly responsible for deletions stopping at
the location of the embedded heptamer.
Analysis of recombinants from pVA1-D-J
To examine the possibility that V81x is overused because of factors intrinsic to the gene segment itself, we constructed another set of microlocus vectors identical to the first series except that V81x was replaced by VA1, including its RSSs and flanking sequences. To control for the recombinatorial status of the A-MuLV lines, transfections were done in parallel with all test plasmids. As before, we used the colony blot method with probes to look for VD, DJ, or VinvD-J recombinants. Only one VinvD clone was found in 187 colonies recovered after transfection with pVA1-D-JH- plasmids. No recombinants were found in 226 pVA1-D-JH+ colonies screened. These levels of recombination were at least 10-fold below the levels of pV81x-D-J plasmids in parallel transfections.
To better analyze the recombinants for both pV81x-D-JH and
pVA1-D-JH, a quantitative Southern blot method was also used. Control
quantifications validating this method are shown in Figure 5B and Materials and Methods. Figure
7 shows a result from an analysis of the
recombination products of substrates pV81x-D-J+,
pV81x-D-J-, pVA1-D-J+, and
pVA1-D-J-. Substrates with V81x yield more recombinants
than VA1 plasmids. pV81x-D-J+, and pV81x-D-J-
yielded recombinants in which V to D rearrangements were 14- to 17-fold
more frequent than D to J rearrangements. pVA1-D-J+ and
pVA1-D-J- yielded recombinants in which V to D
rearrangements were 2- to 5-fold less frequent than D to J. As the
orientation of the microlocus in the plasmids did not change the
qualitative differences, there seem to be no contextual effect of the
backbone vector on rearrangement of the gene segments (Fig. 7). Table IV summarizes the data from
this experiment. In all experiments, there are less VD than
DJ rearrangements on VA1-D-J plasmids and more VD than DJ
rearrangements on V81x-D-J plasmids. In four independent transfections,
VD:DJ ratios range from 5.8 to 17 for V81x-D-J and from 0.19
to 0.51 for VA1-D-J. Given that the frequency of DFL16.1 to
JH1 recombination is the same in both plasmids, we
calculate that the rate of V to D recombination is about 30 times
higher for V81x than for VA1.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
VH segments are not used randomly in B cell
development, and V81x is a particularly crass example of that fact (12, 13, 15, 45, 46, 47, 48, 49, 50). V81x is overutilized in the earliest detectable
rearrangements in fetal liver, where >25% of all VDJ rearrangements
use V81x; its usage declines during the course of development, being
<5% in adult spleen. Several hypotheses have been put forward to
explain the decline: 1) other VH regions might become open
during development and thereby become favored for rearrangement; 2)
cells utilizing V81x might fail to progress past the pre-B stage
because they usually fail to associate with 5 and thereby fail to
receive a signal required for differentiation (51, 52, 53); or 3) cells
utilizing other VH elements might be selected (15).
Although the rise and decline of V81x containing variable domain genes in fetal and adult mouse B cell development has been studied in great detail and has led to valuable insights into B cell selection, the reasons for its initial overuse and subsequent decline remain elusive. At least two hypotheses have been put forward to explain the initial overutilization. 1) V81x is the most 3' VH functional gene segment so far identified; i.e., the one closest to the stretch of D gene segments (13). The "DNA-tracking" hypothesis postulates that the V(D)J recombinase travels along the double helix from D to V, and so the first V encountered would have the best chance of being used (models discussed in 1 . However, even without tracking, which is diffusion in one dimension, it is not unreasonable that formation of the entire recombinase-RSS complex involves a reaction rate limited by diffusion in three dimensions and thereby dependent on the distance between the gene segments to be recombined. 2) Using simple extrachromosomal recombination substrates, we and others have previously demonstrated the existence of cis-acting elements within or flanking V, D, and J gene segments that might determine their utilization in the primary repertoire (3, 7, 19, 23, 24, 54, 55). Additionally, the finding that the V81x coding and RSS sequence is the same in five strains of mice adds evidence to the importance of the sequences themselves (our unpublished data). The present study is an attempt to critically test whether such cis-acting elements can quantitatively explain the overutilization of V81x.
We have examined VH usage in a unique way, in a microlocus containing entire V, D, and J coding sequences and RSSs embedded in their natural flanking sequences. The latter fact is of particular significance since the same sequences naturally flanking DFL16.1 used in our microloci have been shown previously to influence the choice of VH or JH RSS for recombination to this D segment (24). Thus, our system combines most of the known cis elements recognized to be important in recombination between VH, DH, and JH segments, while still isolating the process of V(D)J recombination to exclude cellular selection at the level of encoded protein and complex chromatin structure. VA1 was chosen as a control VH gene segment because its usage is not out of the ordinary in either fetal or adult tissue, yet it is highly transcribed in the germline configuration, a trait that most likely indicates that it is a target for V(D)J recombination (56, 57, 58, 59, 60). On the other hand, V81x does not have detectable transcription in the germline state.
In the experiments reported here, pV81x-D-J generates up to 15% recombinants in a pre-B cell line, while pVA1-D-J yields <1%. We estimated the relative recombination potential of V81x and VA1 in several ways, which yielded ratios of 15- to 30-fold in favor of V81x. Since the only difference between these plasmids is the V gene segment, we conclude that in the absence of chromosomal context and in the absence of cellular selection, sequences in or around these V elements themselves are sufficient for marked differential usage. If there are 200 VH elements and if V81x is used in 25% of the rearrangements in mouse fetal liver, then V81x would be utilized 67 times as frequently as a "typical" VH segment. Thus, even if VA1 recombines two to four times as frequently as a "typical" VH, we have quantitatively explained the overuse of V81x.
It is implied that the differential usage of V81x and VA1 segments could be due to their RSS sequences or coding flank sequences. The V81x and VA1 coding flanks are CA and GA, respectively. However, it is the latter, the CA of the V81x coding flank, that is more competent in V(D)J recombination (roughly twice as competent) (25). Moreover, the consensus VH7183 family RSS, of which V81x is a member, is severalfold more recombinogenic than the consensus VHJ558 family RSS, of which VA1 is a member (19). Thus, we believe a fruitful focus will be the RSS and such studies are underway in our laboratory.
Other transacting factors not tested with our microlocus system, including those inducing accessibility, doubtlessly play a role in the overuse of V81x in vivo. In the section below on VD joints and ordered rearrangement, we discuss a simple model in which the rate of recombination is determined by the distance between the segments to be rearranged as well as by the local sequences.
Sequences of joints in pV81x-D-J are normal
Sequences of both VD and DJ junctions in the microlocus plasmids seem to be similar to those found in vivo (15, 49, 61). This finding confirms that junctional diversity is established during the actual V(D)J recombination process. Remarkably, deletions into the V gene segment stopped in the same region as found in expressed Ig genes; i.e., deletions rarely extended past Cys92, the amino acid residue required for intrachain disulfide bonding. This result indicates that somatic selection against cells with "bad" H chain proteins is not primarily responsible for enforcing the presence of Cys92. Instead, selection in an evolutionary time frame has resulted in a gene rearrangement process that restricts the length of deletions. Moreover, the presence or absence of the embedded heptamer sequence itself does not affect the length of deletions into V.
In mice, VDJH structures rarely express DH in
RF2 (41, 62). There is good evidence that this is because cells with
DJH structures in RF2 are selectively deleted. Be that as
it may, RF2 was not utilized in VDJH recombinants generated
by our microlocus plasmids (Table III). Many studies have determined
that DJH reading frame usage in V81xDJH joints
and the productivity of the sequences vary between fetal and adult B
cell development, presumably due to the absence and presence,
respectively, of an N addition (29, 53, 63, 64). Our system, however,
confers N/P addition to the same approximate degree in all classes of
recombinants and shows that at least part of the apparent selection
against RF2 seems to be built into the recombination process itself.
Interestingly, the bias against RF2 was not evident in VDH
or DJ H structures in the microlocus plasmids, only in the
double recombinant, and may occur in part because of preferred joining
by homology. As there are many fewer VDJH than
VDH or DJH structures, more work is clearly
needed to clarify this issue.
Recently, it has been proposed that V81x encoded H chains with N
addition do not pair with surrogate light chain 5, whereas those
without N addition are able to do so (51, 53). Furthermore, it has been
proposed that V81x sequences without N addition (such as those arising
during fetal B cell development in which TdT is absent) are inherently
more productive than other VH regions (64). We did not
observe any pattern of joint sequences (deletion/addition) that would
indicate an inherent propensity of the V81x segment for a unique
pattern of N addition. As, indeed, coding end sequences have been found
to affect such joint processing (65), it would be interesting to build
a series of other VH microloci plasmids to pursue this
puzzle.
VinvD joints are common in microlocus rearrangements
Not only did pV81x-D-J generate more deletional VD joints than DJ, V81x surprisingly joined to DFL16.1 almost as frequently by inversion as by deletion. The 3' RSSs of D segments are considered stronger than the 5' RSSs (4, 38, 54). Indeed, the majority of murine D segments have close to consensus 3' RSS heptamers and nonamers, but nonconsensus 5' RSSs. Using RSS extrachromosomal plasmids, Teale and coworkers found that the 3' was better than the 5' RSS of a D segment RSS for joining to JH as well as to VH (54). This effect would not be seen in intact animals, because D to J joining generally occurs first, and the 3' RSS of the D segment would be deleted in the process. Thus, we effectively demonstrate that there is no inherent bias in the V(D)J recombination process against the generation of VinvD-J structures. One reason that such structures are not found in vivo might be the deletion of the 3' D RSS during the initial DJ H rearrangement. VanDyk et al. have previously shown that in the presence of the natural DH flanking sequences, the 5' RSS of DH is preferred over the 3' RSS for joining to VH (24). One possible explanation for the high frequency of VinvD-J as compared with direct VD-J products in our microlocus system would be the presence of authentic VH sequences and the strength of the VH RSS.
VD joints and ordered rearrangement
Given the well-established order of H chain rearrangements, VD
joints have rarely been found in intact animals (66, 67). Although we
were unable to find such joints by PCR, others have recently described
direct, inversional, and hybrid VD joints in vivo (24). With the
commonly accepted notion that locus accessibility is the cause of
ordered rearrangement, it was not clear whether VD joints could serve
as intermediates in the production of successful coding VDJ joints. In
addition, with the whole bone marrow PCR approach to isolating such
rare joints, it is possible that these joints are rare mistakes in the
recombination process; it is not even clear what cells the VD joints
are found in. It was plausible that the sequences that flank D regions
on the 5' and 3' sides play a role in inhibiting V to D or VD to VDJ
joints. In such a model, a VD joint represents an aberrant event,
possibly due to a failure in controlling locus accessibility and the
fact that cells harboring such rearrangements are prevented from
differentiating further. Our present results demonstrate that both VD-J
and V-DJ plasmids are able to recombine again to form full rearranged
VDJ segments. In particular, the 400 bp downstream of DFL16.1, a
sequence stretch thought to be important in ordered H chain
rearrangement (24), cannot prevent a secondary rearrangement of a VD-J
structure.
It was also considered possible that VD joints are not easily detectable in intact animals because they rapidly recombine again. In our plasmids, however, VD-J structures generated fewer VDJs than did V-DJ structures.
Why are VD joints so common in microlocus plasmids and so rare in intact animals?
The present experiments were not designed to address this issue. Indeed, they could not have been, for we were quite surprised to find so many VDH joints. We think, however, that the most likely explanation lies in the closeness of VH to D in the microlocus plasmids as compared with the normal locus in intact animals.
Thus, as often happens in immunology, an argument about the validity of two proposed mechanism to explain the same facts ends up stimulating work suggesting that both mechanisms are at work. Higher order chromatin structure resulting from proximity to recombination enhancers as well as sequence are both likely to contribute to the overuse of V81x. One model is that the ordering of rearrangement at the H chain locus is predominantly due to the relative proximity of the D stretch to the JH stretch and the relative remoteness of the D stretch from the VH stretch. Interestingly, DFL16.1, which is an overutilized DH segment, is located at the 5' end of the DH locus. This usage pattern, in concert with our findings, indicates that although proximity is important in determining gene usage, there are clearly other factors including, as shown here, the intrinsic gene segment sequences. Looking toward clarifying the relative importance of these factors, it would be interesting to knock out a large part of the "inert" DNA between the D stretch and the V stretch. We predict that ordered rearrangement would be disturbed. Furthermore, although our studies did not include T cells, there is no reason to expect that TCR repertoire formation would not be similarly affected by such cis elements, as described in this report.
|
|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 C.C.K.Y. and M.L. contributed equally to this work.
3 Address correspondence and reprint requests to Dr. Gillian E. Wu, Ontario Cancer Institute, Room 8-113, 610 University Avenue, Toronto, Ontario, Canada M5G 2M9.
4 Abbreviations used in this paper: RSS, recombination signal sequence; H chain, heavy chain; RF2, reading frame 2.
Received for publication February 27, 1998. Accepted for publication May 27, 1998.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
light-chain recombination signal seguences mediate recombination more frequently than those of light chain. Proc. Natl. Acad. Sci. USA 88:10721. 5. Ann. NY Acad. Sci. 764:39.[Medline]
segments correlates with ongoing T-cell receptor -chain rearrangement. Mol. Cell. Biol. 12:1480.L chain gene locus in V(D)J recombinase-deficient progenitor B cells. Int. Immunol. 7:1915. genes in BLIN-1 human pre-B cell line correlates with germline J-C and V transcription. J. Exp. Med. 173:639.This article has been cited by other articles:
|
|
 |
|
  M. Larijani, A. Zaheen, D. Frieder, Y. Wang, G. E. Wu, W. Edelmann, and A. Martin Lack of MSH2 involvement differentiates V(D)J recombination from other non-homologous end joining events Nucleic Acids Res., November 27, 2005; 33(21): 6733 - 6742. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. S. Williams, A. Martinez, A. Montalbano, A. Tang, A. Mauhar, K. M. Ogwaro, D. Merz, C. Chevillard, R. Riblet, and A. J. Feeney Unequal VH Gene Rearrangement Frequency Within the Large VH7183 Gene Family Is Not Due to Recombination Signal Sequence Variation, and Mapping of the Genes Shows a Bias of Rearrangement Based on Chromosomal Location J. Immunol., July 1, 2001; 167(1): 257 - 263. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Y. Mortuza, I. M. Moreira, M. Papaioannou, P. Gameiro, L. A. Coyle, C. S. Gricks, P. Amlot, H. G. Prentice, A. Madrigal, A. V. Hoffbrand, et al. Immunoglobulin heavy-chain gene rearrangement in adult acute lymphoblastic leukemia reveals preferential usage of JH-proximal variable gene segments Blood, May 1, 2001; 97(9): 2716 - 2726. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Li, R. E. Hammer, J. B. George-Raizen, K. C. Meyers, and W. T. Garrard High-Level Rearrangement and Transcription of Yeast Artificial Chromosome-Based Mouse Ig{kappa} Transgenes Containing Distal Regions of the Contig J. Immunol., January 15, 2000; 164(2): 812 - 824. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. C. Nicholson, X. Zou, A. V. Popov, G. P. Cook, E. M. Corps, S. Humphries, C. Ayling, B. Goyenechea, J. Xian, M. J. Taussig, et al. Antibody Repertoires of Four- and Five-Feature Translocus Mice Carrying Human Immunoglobulin Heavy Chain and {kappa} and {lambda} Light Chain Yeast Artificial Chromosomes J. Immunol., December 15, 1999; 163(12): 6898 - 6906. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
