90 233 Rearrangement of Immunoglobulin Genes

wea25324_ch23_732-758.indd Page 740
740
12/21/10
1:53 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 23 / Transposition
P elements are now commonly used as mutagenic agents
in genetic experiments with Drosophila. One advantage of
this approach is that the mutations are easy to locate; we
just look for the P element and it leads us to the interrupted
gene. Molecular biologists also use P elements to transform
flies—that is, to carry manipulated genes into flies.
Antigen-binding sites
S
SUMMARY The P-M system of hybrid dysgenesis in
Drosophila is caused by the conjunction of two
factors: (1) a transposable element (P) contributed by
the male, and (2) M cytoplasm contributed by the
female, which allows transposition of the P element.
Hybrid offspring of P males and M females therefore
suffer multiple transpositions of the P element. This
causes damaging chromosomal mutations that render
the hybrids sterile. On the other hand, P females
contain a suppressor of transposition (a group of
piRNAs targeting the P element), so offspring of
either P or M males and P females are fertile. P elements have practical value as mutagenic and transforming agents in genetic experiments with Drosophila.
23.3 Rearrangement of
Immunoglobulin Genes
Rearrangements of the mammalian genes in B cells that
produce antibodies, or immunoglobulins, and in T cells that
produce T-cell receptors, use a process that closely resembles transposition. Even the recombinases involved in
antibody and T-cell receptor gene rearrangements resemble
transposases. Because of these similarities, we include these
rearrangements in this chapter.
As mentioned in Chapter 3, an antibody is composed of
four polypeptides: two heavy chains and two light chains.
(Similarly, T-cell receptors contain one large b-chain and
one smaller a-chain.) Figure 23.11 illustrates an antibody
schematically and shows the sites that combine with an
invading antigen. These sites, called variable regions, vary
from one antibody to the next and give these proteins their
specificities; the rest of the protein (the constant region)
does not vary from one antibody to another within an
antibody class, though some variation occurs between the
few classes of antibodies. Any given immune cell can make
antibody with only one kind of specificity. Remarkably
enough, humans have immune cells capable of producing
antibodies to react with virtually any foreign substance
we would ever encounter. That means we can make many
millions of different antibodies.
Does this imply that we have millions of different antibody genes? That is an untenable hypothesis; it would
place an impossible burden on our genomes to carry all the
S
S
SS
S
Figure 23.11 Structure of an antibody. The antibody is composed
of two light chains (blue) bound through disulfide bridges to two heavy
chains (pink), which are themselves held together by a disulfide
bridge. The antigen-binding sites are at the amino termini of the
protein chains, where the variable regions lie.
necessary genes. So how do we solve the antibody diversity
problem? As unlikely as it may seem, a maturing B cell, a
cell that is destined to make an antibody, rearranges its
genome to bring together separate parts of its antibody
genes. The machinery that puts together the gene selects
these parts at random from heterogeneous groups of parts,
rather like ordering from a luncheon menu (“Choose one
from column A and one from column B”). This arrangement greatly increases the variability of the genes. For instance, if 41 possibilities are present in “column A” and 5 in
“column B,” the total number of combinations of A 1 B is
41 3 5 or 205. Thus, from 46 gene fragments, we can assemble 205 genes. And this is just for one of the antibody
polypeptides. If a similar situation exists for the other, the
total number of antibodies will be the product of the numbers of the two polypeptides. This description, though correct in principle, is actually an oversimplification of the
situation in the antibody genes; as we will see, they have
somewhat more complex mechanisms for introducing diversity, which lead to an even greater number of possible
antibody products.
Studies on mammalian antibodies have revealed two
families of antibody light chains called kappa (k) and
lambda (l). Figure 23.12 illustrates the arrangement of the
gene parts for a human k light chain. “Column A” of this
“menu” contains 41 variable region parts (V); “Column B”
contains 5 joining region parts (J). The J segments actually
encode the last 12 amino acids of the variable region, but
they are located far away from the rest of the V region and
close to a single constant region part. This is the situation
in the germ cells, before the antibody-producing cells differentiate and before rearrangement brings the two unlinked regions together. The rearrangement and expression
events are depicted in Figure 23.12.
wea25324_ch23_732-758.indd Page 741
12/21/10
1:53 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
23.3 Rearrangement of Immunoglobulin Genes
741
(a) κ light chain coding regions
5J
41 V
C
(b) Rearrangement
J1 J3 J5
Germ-line DNA
V1
V2
V3
V4
C
Recombination
J2 J4
B-cell DNA
C
V3 J2J3 J4 J5
V2
Transcription
J2 J3 J4 J5
RNA transcript
C
RNA splicing
V3
Messenger RNA
V3 J 2
C
Translation
Protein
V
C
Figure 23.12 Rearrangement of an antibody light chain gene.
(a) The human k-antibody light chain is encoded in 41 variable gene
segments (V; light green), five joining segments (J; red), and one
constant segment (C; blue). (b) During maturation of an antibodyproducing cell, a DNA segment is deleted, bringing a V segment
(V3, in this case) together with a J segment (J2 in this case). The gene
can now be transcribed to produce the mRNA precursor shown
here, with extra J segments and intervening sequences. The material
between J2 and C is then spliced out, yielding the mature mRNA,
which is translated to the antibody protein shown at the bottom. The
J segment of the mRNA is translated into part of the variable region of
the antibody.
First, a recombination event brings one of the V regions
together with one of the J regions. In this case, V3 and J2
fuse together, but it could just as easily have been V1 and J4;
the selection is random. After the two parts of the gene assemble, transcription occurs, starting at the beginning of
V3 and continuing until the end of C. Next, the splicing
machinery joins the J2 region of the transcript to C, removing the extra J regions and the intervening sequence between the J regions and C. It is important to remember that
the rearrangement step takes place at the DNA level, but
this splicing step occurs at the RNA level by mechanisms
we studied in Chapter 14. The messenger RNA thus assembled moves into the cytoplasm to be translated into an
antibody light chain with a variable region (encoded in
both V and J) and a constant region (encoded in C).
Why does transcription begin at the beginning of V3
and not farther upstream? The answer seems to be that an
enhancer in the intron between the J regions and the C region activates the promoter closest to it: the V3 promoter in
this case. This also provides a convenient way of activating
the gene after it rearranges; only then is the enhancer close
enough to turn on the promoter.
The rearrangement of the heavy chain gene is even
more complex, because there is an extra set of gene parts in
between the V’s and J’s. These gene fragments are called D,
for “diversity,” and they represent a third column on our
menu. Figure 23.13 shows that the heavy chain is assembled from 48 V regions, 23 D regions, and 6 J regions. On
this basis alone, the cell can put together 48 3 23 3 6, or
6624 different heavy chain genes. Furthermore, 6624 different heavy chains combined with 205 k light chains and
170 l light chains yield almost 2.5 million different antibodies or, strictly speaking, 2.5 million different combinations of variable regions.
But there are even more sources of diversity. The first
derives from the fact that the mechanism joining V, D, and
J segments, which we call V(D)J joining, is not precise. It
can add or delete bases on either side of the joining site.
Heavy chain coding regions
48 V
23 D
6J
C
Figure 23.13 Structure of antibody heavy chain coding regions. The human heavy chain is encoded in 48 variable segments (V; light green),
23 diversity segments (D; purple), 6 joining segments (J; red), and 1 constant segment (C; blue).
wea25324_ch23_732-758.indd Page 742
742
12/21/10
1:53 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 23 / Transposition
This leads to extra differences in antibodies’ amino acid
sequences.
Another source of antibody diversity is somatic hypermutation, or rapid mutation in an organism’s somatic
(nonsex) cells. In this case, the mutations occur in antibody genes, probably at the time that a clone of antibodyproducing B cells proliferates to meet the challenge of an
invader.
Genetic and biochemical analysis has shown that somatic hypermutation occurs in two steps. First, a cytidine
deaminase that is induced during B cell activation deaminates cytosines to uracils during DNA replication. Next,
the uracils attract either the mismatch repair process or
uracil-N-glycosylase, which removes the uracils, leaving
abasic sites. In either case, a single-strand break occurs, and
the cell then “repairs” the break with the same auxiliary
DNA polymerases used in translesion bypass (Chapter 20):
DNA polymerases z, h, u, and possibly ι. These polymerases are error prone, so many mutations are created.
Together, imprecise joining of gene segments and somatic hypermutation magnify the number of possible antibodies tremendously. In fact, it has been estimated that the
total number of antibodies one can make in a lifetime is as
high as 100 billion. This surely seems enough to match any
attacker.
SUMMARY The immune systems of vertebrates can
produce billions of different antibodies to react with
virtually any foreign substance. These immune systems generate such enormous diversity by three basic mechanisms: (1) assembling genes for antibody
light chains and heavy chains from two or three
component parts, respectively, each part selected
from heterogeneous pools of parts; (2) joining the
gene parts by an imprecise mechanism that can delete bases or even add extra bases, thus changing the
gene; and (3) causing a high rate of somatic mutations, probably during proliferation of a clone of
immune cells, thus creating slightly different genes.
Recombination Signals
How does the recombination machinery determine where
to cut and paste to bring together the disparate parts of an
immunoglobulin gene? Susumu Tonegawa examined the
sequences of many mouse immunoglobulin genes (encoding k and l light chains, and heavy chains) and noticed a
consistent pattern (Figure 23.14a): Adjacent to each coding
region lies a conserved palindromic heptamer (7-mer), with
the consensus sequence 59-CACAGTG-39. This heptamer is
accompanied by a conserved nonamer (9-mer) whose consensus sequence is 59-ACAAAAACC-39. The heptamer and
nonamer are separated by a nonconserved spacer contain-
CACAGTG ACAAAAACC GGTTTTTGT CACTGTG
(a)
λ-chain
Vλ
7
κ-chain
Vκ
7
H-chain
VH
7
23
12
23
9
12
9
9
9
9
9
9
7
D
7
12
23
23
12
7
Jλ
7
Jκ
7
JH
9
(b)
V
D
J
C
Figure 23.14 Signals for V(D)J joining. (a) Arrangement of signals
around coding regions for immunoglobulin k and l light chain genes
and heavy chain gene. Boxes labeled “7” or “9” are conserved
heptamers or nonamers, respectively. Their consensus sequences are
given at top. The 12-mer and 23-mer spacers are also labeled. Notice
the arrangement of the 12 signals and 23 signals such that joining
one kind to the other naturally allows assembly of a complete gene.
(b) Schematic illustration of the arrangement of the 12 and 23 signals
in an immunoglobulin heavy chain gene. The yellow symbols represent
12 signals, and the orange triangles represent 23 signals. Notice again
how the 12/23 rule guarantees inclusion of one of each coding region
(V, D, and J) in the rearranged gene. (Source: (a) Adapted from Tonegawa, S.,
Somatic generation of antibody diversity. Nature 302:577, 1983.)
ing either 12 bp (a 12 signal) or 23 (61) bp (a 23 signal).
The arrangement of these recombination signal sequences
(RSSs, Figure 23.14b) is such that recombination always
joins a 12 signal to a 23 signal. This 12/23 rule stipulates
that 12 signals are never joined to each other, nor are 23 signals joined to each other, and thus ensures that one, and only
one, of each coding region is incorporated into the mature
immunoglobulin gene.
Aside from the existence of consensus RSSs, what is
the evidence for their importance? Martin Gellert and
colleagues have systematically mutated the heptamer and
nonamer by substituting bases, and the spacer regions by
adding or subtracting bases, and observed the effects of
these alterations on recombination. They measured recombination efficiency in the following way: They built
a recombinant plasmid with the construct shown in
Figure 23.15. The first element in this construct is a lac
promoter. This is followed by a 12 signal, then a prokaryotic transcription terminator, then a 23 signal, and finally a
cat reporter gene. They made mutations throughout these
RSSs, then introduced the altered plasmids into a pre-B cell
line. Finally, they purified the plasmids from the pre-B cells
and introduced them into chloramphenicol-sensitive E. coli
cells and tested them for chloramphenicol resistance. If no
recombination took place, the transcription terminator
wea25324_ch23_732-758.indd Page 743
12/21/10
1:53 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
23.3 Rearrangement of Immunoglobulin Genes
Plac
Transcription
terminator
cat
GTCGAC CTGCAG
H12
CACAGTG
S12
N12
CTACAGACTGGA ACAAAAACC
743
GGATCC CTCGGG
H23
CACAGTG
Figure 23.15 Structure of reporter construct used to measure
effects of mutations in RSSs on recombination efficiency. Gellert
and coworkers made a recombination reporter plasmid containing
a lac promoter and cat gene separated by an insert containing a
transcription terminator flanked by a 12 signal and a 23 signal.
Recombination between the two RSSs either inverts or deletes the
terminator, allowing expression of cat. Transformation of bacterial cells
prevented cat expression, and therefore chloramphenicol
resistance was almost nonexistent. On the other hand, if
recombination between the 12 signal and the 23 signal occurred, the terminator was either inverted or deleted, and
therefore inactivated. In that case, cat expression occurred
under control of the lac promoter, and many chloramphenicol-resistant colonies formed. This experiment showed
that many alterations in bases in the heptamer or nonamer reduced recombination efficiency to background level.
The same was true of insertions and deletions of bases in
the spacer regions. Thus, all these elements of the RSSs are
important in V(D)J recombination.
SUMMARY The recombination signal sequences
(RSSs) in V(D)J recombination consist of a heptamer and a nonamer separated by either 12-bp or
23-bp spacers. Recombination occurs only between
a 12 signal and a 23 signal, which guarantees that
only one of each coding region is incorporated into
the rearranged gene.
S23
N23
GTAGTACTCCACTGTCTGGCTGT ACAAAAACC
with the rearranged plasmid yields many CAT-producing colonies
that are chloramphenicol-resistant. On the other hand, transformation
of bacteria with the unrearranged plasmid yields almost no
chloramphenicol-resistant colonies. (Source: Adapted from Hesse, J., M. R.
Lieber, K. Mizuuchi, and M. Gellert, V(D)J recombination: a functional definition of
the joining signals. Genes and Development 3:1053–61, 1989.)
better, so something seemed to be missing. Baltimore’s
group sequenced the whole genomic fragment containing
most of RAG-1 and found another whole gene tightly
linked to it. They wondered whether this other gene might
also have something to do with V(D)J joining, so they
tested this genomic fragment plus a RAG-1 cDNA in the
same transfection experiment. When they introduced the
two DNAs together into the same cell, they found many
more drug-resistant cells. In this way, they discovered that
two genes are responsible for V(D)J recombination, and
they named the second RAG-2.
RAG-1 and RAG-2 are expressed only in pre-B and
pre-T cells, where V(D)J joining of immunoglobulin and
T-cell receptor gene segments, respectively, are occurring.
The T-cell receptors are membrane-bound antigen-binding
proteins with an architecture similar to that of the immunoglobulins. The genes encoding the T-cell receptors rearrange according to the same rules that apply to the
immunoglobulin genes, complete with RSSs containing
12 signals and 23 signals. Thus, RAG-1 and RAG-2 are
apparently involved in both immunoglobulin and T-cell
receptor V(D)J joining.
The Recombinase
Mechanism of V(D)J Recombination
David Baltimore and his colleagues searched for the gene(s)
encoding the V(D)J recombinase using a recombination reporter plasmid similar to the one we just discussed, but
designed to operate in eukaryotic cells by conferring resistance to the drug mycophenolic acid. They introduced this
plasmid, along with fragments of mouse genomic DNA,
into NIH 3T3 cells, which lack V(D)J recombination activity, and tested for recombination by assaying for drugresistant 3T3 cells. This led to the identification of a
recombination-activating gene (RAG-1) that stimulated
V(D)J joining activity in vivo.
However, the degree of stimulation by a genomic clone
containing most of RAG-1 was modest—no more than
that obtained with whole genomic DNA. Furthermore,
cDNA clones containing the whole RAG-1 sequence did no
V(D)J joining is imprecise, which contributes to the diversity of products from the process. Both loss of bases and
addition of extra bases at the joints are frequently observed.
This is good for immunoglobulin and T-cell receptor production, because it adds to the variety of proteins that can
be made from a limited repertoire of gene segments.
How do we explain this imprecision? Figure 23.16 illustrates the mechanism of cleavage at the RSSs that flank
an intervening segment between two coding segments. We
see that the products of the RAG-1 and RAG-2 genes,
Rag-1 and Rag-2, respectively, first nick the DNAs at the
joints. Then the new 39-hydroxyl groups attack phosphodiester bonds on the complementary strands, liberating the
intervening segment and forming hairpins at the ends of
the coding segments. These hairpins are the key to the
wea25324_ch23_732-758.indd Page 744
744
12/21/10
1:53 PM user-f469
/Volume/204/MHDQ268/wea25324_disk1of1/0073525324/wea25324_pagefile
Chapter 23 / Transposition
Coding
region
1
Coding
region
2
Hairpins
1
3-OH—
Intervening
region
2
—3-OH
Lost from
cell
New joint
coding
region
Figure 23.16 Mechanism of cleavage at RSSs. Nicking of opposite strands (vertical arrows) occurs at RSSs at the junctions between coding
regions (red) and the intervening region (yellow). The new 39-hydroxyl groups (blue) attack and break the opposite strands, forming hairpins and
releasing the intervening segment, which is lost. Finally, the hairpins open, and the two coding regions are joined by an imprecise mechanism.
(Source: Adapted from Craig, N.L., V(D)J recombination and transposition: closer than expected. Science 271:1512, 1996.)
imprecision of joining; they can open up on either side of
the apex of the hairpin, and bases can then be added or
subtracted to make the DNA ends blunt for joining. The
Rag-1 and Rag-2 proteins hold both hairpins together in a
complex so they can join covalently with each other.
How do we know hairpins form? They were first found
in vivo, but in very low concentration. Gellert and his colleagues later developed an in vitro system in which they
could be readily observed. Figure 23.17a illustrates one of
the labeled substrates these workers used. It was a 50-mer
labeled at one 59-end with 32P. It contained a 12 signal, represented by a yellow symbol, flanked by a 16-bp segment on
the left; the right-hand end of the fragment was, therefore, a
34-bp segment, which included the 12 signal. A similar substrate contained the same flanking segments, but had a 23
signal instead of a 12 signal. Thus, it was 61 bp long.
Gellert and colleagues incubated these substrates with
RAG1 and RAG2, the human homologs of mouse Rag1
and Rag2, respectively, then electrophoresed the products
under nondenaturing conditions to see if any DNA cleavages had occurred (Figure 23.17b). They found a 16-mer,
demonstrating that a double-stranded cleavage had occurred. However, nondenaturing gel electrophoresis could
not distinguish between a true double-stranded 16-mer
and a 16-mer with a hairpin end, so these workers subjected
the same products to denaturing polyacrylamide gel electrophoresis in the presence of urea and at an elevated
temperature (Figure 23.17c). Under these conditions, a
double-stranded 16-mer would give rise to two singlestranded 16-mers. On the other hand, a 16-mer with a
hairpin at the end would give rise to a single-stranded 32-mer.
This is what Gellert and coworkers observed whenever
the DNA contained either a 12 signal or a 23 signal and
both RAG1 and RAG2 proteins were present. A DNA with
no 12 or 23 signal gave no product, hairpin or otherwise,
and reactions lacking either RAG1 or RAG2 protein gave
no product (Figure 23.17d). Thus, RAG1 and RAG2
recognize both the 12 signal and the 23 signal and cleave
the DNA adjacent to the signal, forming a hairpin at the
end of the coding segment.
Moreover, the 16-mer product from the nondenaturing
gel yielded only hairpin product on the denaturing gel,
demonstrating that no simple double-stranded 16-mer
formed. But labeled DNA migrating with the substrate in
the nondenaturing gel yielded a small amount of 16-mer in
the denaturing gel. This cannot have come from a doublestranded break, or it would not have remained with the
substrate in the nondenaturing gel. Thus, it must have come
from a nick in the labeled strand. The 16-mer created by
the nick would have remained base-paired to its partner
during nondenaturing electrophoresis, but would have migrated independently as a 16-mer during denaturing electrophoresis. Thus, single-stranded nicking is apparently
also part of the action of RAG1 and RAG2 proteins.
To investigate further the relationship between nicking
and hairpin formation, Gellert and colleagues ran a timecourse study in which they incubated the substrate for increasing lengths of time with RAG1 and RAG2 proteins
and then subjected the products to denaturing gel electrophoresis. They found that the nicked species appeared first,
followed by the hairpin species. This suggested that the
nicked species is a precursor of the hairpin species. To test
this hypothesis, they created nicked intermediates and incubated them with RAG1 and RAG2. Sure enough, the
RAG1 and RAG2 converted the nicked DNAs to hairpins.
Subsequent work by Gellert’s group has shown the sequence of events seems to be: RAG1 and RAG2 nick one
DNA strand adjacent to a 12 signal or a 23 signal; then the
newly formed hydroxyl group attacks the other strand in a
transesterification reaction, forming the hairpin, as was
illustrated in Figure 23.16.
What enzyme opens up the hairpins created by RAG1
and RAG2? Michael Lieber and colleagues demonstrated
in 2002 that an enzyme called Artemis carries out this
function. On its own, Artemis has exonuclease activity.
However, in conjuction with DNA-PK cs, Artemis gains
endonuclease activity that can cleave hairpins. You may
recongnize DNA-PK cs from our discussion in Chapter 20 of
nonhomologous DNA end-joining (NHEJ) for repair
of double-strand DNA breaks. In fact, joining of the
opened hairpins resembles NHEJ and relies on the NHEJ
machinery.
Artemis is also required to cleave the hairpins created
during the rearrangement of T cell receptor genes, which