Tolerance-induced receptor selection:

scope, sensitivity, locus specificity,

and relationship to lymphocyte-

positive selection

Djemel Aıt-Azzouzene

Patrick Skog

Marc Retter

Valerie Kouskoff

Marc Hertz

Julie Lang

Jennifer Kench

Michael Chumley

Doron Melamed

Janice Sudaria

Amanda Gavin

Annica Martensson

Laurent Verkoczy

Bao Duong

Jose Vela

David Nemazee

Authors’ addresses

Djemel Aıt-Azzouzene1, Patrick Skog1, Marc Retter1,2,Valerie, Kouskoff2, Marc Hertz2, Julie Lang2, JenniferKench2, Michael Chumley2, Doron Melamed2, JaniceSudaria1, Amanda Gavin1, Annica Martensson1, LaurentVerkoczy1, Bao Duong1,3, Jose Vela1,3, David Nemazee1,1Department of Immunology, The Scripps

Research Institute, La Jolla, CA, USA.2National Jewish Medical and Research Center,

Denver, CO, USA.3Doctoral Program in Chemical and Biological

Sciences, The Scripps Research Institute, La

Jolla, CA, USA.

Correspondence to:

David NemazeeDepartment of Immunology

The Scripps Research Institute

10550 North Torrey Pines Road

Mail drop IMM-29

La Jolla, CA 92037, USA

Tel.: þ1 858 784 9528

Fax: þ1 858 784 9554

E-mail: nemazee@scripps.edu

Acknowledgements

This research was supported by grants GM44809,

AI33608, and AI49940 from the National Institutes of

Health.

Summary: Receptor editing is a mode of immunological tolerance ofB lymphocytes that involves antigen-induced B-cell receptor signaling andconsequent secondary immunoglobulin light chain gene recombination.This ongoing rearrangement often changes B-cell specificity for antigen,rendering the cell non-autoreactive and sparing it from deletion. Wecurrently believe that tolerance-induced editing is limited to early stagesin B-cell development and that it is a major mechanism of tolerance, with alow-affinity threshold and the potential to take place in virtually everydeveloping B cell. The present review highlights the contributions fromour laboratory over several years to elucidate these features.

Clonal selection and receptor selection

The immune system uses two strikingly different strategies to

regulate its antigen receptor repertoire: clonal selection and

receptor selection. As it was first proposed by Burnet (1) [and

independently by Talmage (2)], clonal selection has been

recognized as a major, and perhaps the only, mechanism of

antigen-specific regulation of the repertoire. According to the

generally accepted model, antigen-specific tolerance and

immunity are regulated by the ability of antigen to act on

cells, through their clonally restricted antigen receptors, which

signal to stimulate cell growth and survival or apoptosis.

However, cells that express antigen receptors may also alter

their antigenic specificity by modifications of the receptor

genes and proteins themselves, a process that we refer to as

receptor selection (3). Receptor selection can take place by

ongoing V(D)J recombination, which modifies and replaces

antigen receptor genes, or by induced somatic mutation,

which introduces point mutations (or gene conversions) in

antigen receptors. In this review, we focus on studies from our

Immunological Reviews 2004Vol. 197: 219–230Printed in Denmark. All rights reserved

Copyright � Blackwell Munksgaard 2004

Immunological Reviews0105-2896

219

laboratory examining the mechanism of immune tolerance in

B cells reactive to membrane-expressed autoantigen. In these

studies, we identified V(D)J recombinase-mediated receptor

editing as a major mechanism of tolerance early in B-lymphocyte

development, whereas later at post-bone marrow stages, the

major mechanism of tolerance appeared to be clonal selection

by apoptosis or functional inactivation.

Peripheral and central B-cell tolerance

We first became convinced that editing occurs in a comparison

of peripheral and central tolerance in mice transgenic for the

immunoglobulin (Ig) M and IgD gene forms of antibody 3–83

(4). This antibody is reactive to major histocompatibility

complex (MHC) class I molecules H-2Kk, Kb, and other allo-

forms but is not reactive to H-2d molecules. In the absence of

cognate antigen, B cells in 3–83 transgenic mice retain the

transgene-encoded specificity, populating the mouse with a

virtually monoclonal B-cell population (Fig. 1, middle row). In

the experiment depicted in Fig. 1, these 3–83 Ig transgenic

mice were crossed to mice carrying liver-expressed Kb trans-

gene driven by the albumin promoter, which provides a

model of peripheral B-cell tolerance. As can be seen in the

lower panels, autoreactive B cells are eliminated to a great

extent in the lymph nodes and partially in the spleen, with

little replacement of B cells lacking the transgenic specificity

(detected with an anti-idiotypic antibody). The splenic B cells

are largely short-lived and in part rescuable by Bcl-2 over-

expression (5). Such a pattern of tolerance-induced clonal

elimination conforms to the clonal selection ideal.

In contrast, when 3–83 Ig transgenic mice are bred to H-2k

or H-2b mice, where antigen is present on the whole body, we

found a less profound B-cell loss, with considerable replace-

ment of cells carrying the transgene-encoded specificity by

long-lived B cells retaining the transgenic H-chain, but now

expressing endogenous L-chains. About half of these idiotype-

negative cells carried lL-chains (Fig. 2). A key finding was that

the appearance of these cells was clearly not the result of

outgrowth of pre-existing variants.

As shown in Fig. 3, the emerging lþ cells were specifically

found in 3–83 transgenic mice whose B cells expressed anti-

gen on the whole body, but not in mice deleting transgenic

cells in the periphery (3–83 transgenic�MT-Kb mice), which

were even more B-cell deficient. This finding suggested that

encounter of antigen at a bone marrow stage of development

promoted further L-chain rearrangements.

Editing in tissue culture

The ability of innocuous antigen receptors to suppress V(D)J

recombination and of autoreactive receptors to induce it was

evaluated in a number of ways. A direct ex vivo culture or

modified interleukin-7 (IL-7) culture system was used to

demonstrate that, in the absence of antigen, the 3–83 receptor

could suppress endogenous Ig-k-recombination to approxi-

mately 1–2% of cells, whereas in the presence of antigen,

40–70% underwent endogenous Ig-k recombination (6, 7)

(Figs 4–6). These changes were mirrored in the increase in

recombination-activating gene (RAG) mRNA levels in antigen-

challenged sIgþ bone marrow B cells (6, 7). Importantly, in all

of our studies, VDJ assembly at the H-chain was suppressed, as

determined by the paucity of endogenous H-chain protein

expression and also when monitored with a sensitive polymer-

ase chain reaction method (7) (Fig. 5E).

Bone marrow Spleen Lymph node

WT

3–83 ND

3–83 PD

IgM

0.4% 0.7% 0.2%

31%

0.3%15%27%

3–83

Idio

type

26% 52%

Fig. 1. Self-membrane antigen targeted to the livermediates clonal deletion. In this flow cytometryanalysis, 3–83 transgenic mice on an antigen-freebackground (ND) have abundant B cells carrying thetransgenic receptor idiotype (middle row), whereas3–83 mice expressing the cognate Kb antigen on theliver (lower panels) delete the now autoreactive Bcells in lymph node and spleen, but not in bonemarrow (56). IgM, immunoglobulin M; WT,wildtype; PD, peripherally deleting.

Aıt-Azzouzene et al �Receptor editing

220 Immunological Reviews 197/2004

It should be noted that these in vitro systems were amenable

to the analysis of non-transgenic B cells, and they allowed us

to show that editing could be readily induced in bone marrow

B cells using anti-B-cell receptor (BCR) antibodies (6–8).

Importantly, recirculating mature CD23þIgDþ B cells in

these cultures were refractory to editing, arguing that despite

sharing a similar culture environment to the CD23– cells in the

same dishes, they were editing incompetent (6). Thus, there

appears to be a ‘point of no return’ in preimmune develop-

ment, where B cells become editing incompetent, although

recent work suggests that amongst transitional B cells, editing

competence may be retained under certain conditions (9–11).

Together with considerable data from the anti-DNA trans-

genic models (12–20), in which skewed Jk usage was

observed in cells surviving tolerance, these results led to a

model suggesting that ongoing L-chain recombination could

play a major role in immune tolerance. Indeed, the structure of

the k-locus appeared to be especially suited to such a process. Asshown schematically in Fig. 7, the k-locus permits replacement

of active VJ exons with new ones by nested rearrangements.

Related work studying editing in mice with Ig-k loci tar-

geted with pre-rearranged genes found that such transgenes

suppress rearrangements in some but not all cells (21–23),

a pattern similar to that observed by Storb and colleagues

(24–26) in k-transgenic mice. These results suggested that

either feedback suppression of innocuous receptors is inef-

ficient or tolerance-induced editing is common. For example,

it was found that about one-fourth of the B cells in such

‘knockin’ mice express non-targeted L-chains, rather than the

targeted allele (22).

Editing in a knockin model

How can we explain why normal B cells rarely express two L-

chains on their surface, while functional, pre-rearranged, tar-

geted k-loci often fail to suppress endogenous recombination

Ig-Tg (H-2k)Ig-Tg (H-2d)

Dbl-Tg

Non-Tg

00

00

10

1

1FITC anti-λ

PE

ant

i-IgD

a

10

5.4% 4.8%

0.2%

10

100

100

1000

1000

Days

% la

bele

d B

220h

i B c

ells

20

20

30 40

40

60

80

A

B

H-2b Tg spleen

Fig. 2. Manifestations of receptor editing.

(A) Edited splenic B cells from H-2b 3–83transgenic mouse retain transgenic H-chainas detected with allotype-specificimmunoglobulin D (IgD) antibody, but oftenexpress endogenous l chains (4). (B) EditedB cells in Ig-transgenic H-2k mice have longlifespan as judged by BrdU uptake studiescompared to mice whose B cells encounterantigen in the periphery (Dbl transgenic) oreven non-deleting Ig-transgenic (H-2d) mice(from the doctoral thesis of J. Kench,U. Colorado Health Science Center). FITC,fluorescein isothiocyanate; PE, phycoerythrin;Tg, transgenic.

8

4

2

17

10

7

% Ig

Da /

λ+ c

ells

0H-2d

-H-2b

Whole bobyMT-Kb

LiverAntigenlocation

Fig. 3. Edited lþ B cells are induced by bone marrow antigen but notperipheral antigen. Spleen cells from 3-83 transgenic mice lackingantigen (H-2d) or bearing antigen in the liver (MT-Kb) or whole body(H-2b) were analyzed by flow cytometry for coexpression of transgenicH-chain (IgDa) and endogenous (l) L-chain. Numbers of individual micetested in each group are listed above each bar.

Aıt-Azzouzene et al �Receptor editing

Immunological Reviews 197/2004 221

(Fig. 8). In other words, how can editing be compatible with

the empiric phenomena of allelic and isotypic exclusion? One

possibility is that k-loci are monoallelically expressed (27, 28),

and another is that ongoing rearrangements often occur

because of tolerance-induced editing or because innocuous

H/L pairs often fail to mediate feedback suppression of recombin-

ation. Evidence for the latter two explanations was obtained

using the 3–83 knockin model (29, 30). In this model, hemi-

zygous versions of the targeted H and L transgenes failed to

suppress endogenous IgL rearrangements, despite their lack of

autoreactivity (31). By providing these genes in the homozygous

form, the sIg expression level was elevated to the point where

B-cell development was promoted and endogenous rearrange-

ments suppressed (29, 30). Most importantly, on an antigen-

bearing background, these same cells underwent very efficient

editing, resulting in the generation of splenic B cells in number

essentially similar to that of control mice lacking antigen (Fig. 8).

As in the conventional 3–83 transgenic mice, about half of these

rescued cells expressed l light chains, but in the case of the

knockin mice, there was little overall reduction in B cells. These

studies suggest that tolerance-induced receptor editing is a

remarkably efficient process and that an insufficient level of

receptor expression can lead to alterations in B-cell specificity

that are related to a lack of positive selection. In this respect, the

B cells may to some extent behave like CD4/CD8 double-positive

thymocytes, which in the absence of positive selection, undergo

continuing T-cell receptor � recombinations [to further extend

the analogy, it also seems possible that, like B cells, autoreactive

T cells may be subject to tolerance-induced editing (32–34),

although this is a controversial topic (35)].

Editing in the normal repertoire

Although targeted Ig replacement transgenics represent a more

physiological model than conventional Ig transgenics and can

undergo nested rearrangements with the endogenous receptor

gene elements in cis, they nevertheless distort and accelerate

B-cell development, owing to their premature expression

(Fig. 9). In an attempt to assess the frequency of editing in a

non-transgenic immune system, we assessed the fate of the

Ig-k locus in B cells that expressed l-chains. Because, in Ig-lcells, the k-loci are often deleted by receptor selection (RS)

rearrangement (36–39), we focused on a subset of cells that

retained, but inactivated, the previously rearranged VJk by RS

recombination to the intronic site (Fig. 9). We found that

40–50% of the cells had inactivated in-frame, apparently

functional, rearranged k genes (40). The high frequency,

which was significantly higher than what could be achieved

by random rearrangements, suggested that tolerance often

0.25 60

50

40

30

20

10

0Treatment: – +

0.2

0.15

0.1

0.05

0

% VJκ1-bearing hybridoma in cell mix

% V

Jκ1/

actin

rel

ativ

e in

tens

ity

0

0 1 2 3 4 5 6 7 8 9 10 11

10 20 30 40 50

% B

cel

ls w

ith V

Jκ1

rear

rang

emen

t

0

3–83µ/δ 3–83µ/δ Bcl-2

0.2 2 10 20 40 Mix No tre

atm

ent

No tre

atm

ent

Anti-I

d m

Ab

Anti-I

d m

Ab

VJκ1

Actin

A

B C

% VJκ1 standard curve

Standard curve for quantitation

p = 0.0065

Fig. 4. Induction of Vk-Jk1 joining after

B-cell receptor (BCR) ligation in bone

marrow cultures. (A) Polymerase chainreaction (PCR) analysis of endogenous VJk1rearrangements (top) and control, a-actingene (bottom) after 48 h of culture. Lanes1–6 show data for control mixtures ofvarious proportions of VJk1-positive andVJk1-negative hybridoma cell lines. Lanes8 and 10 show untreated cultures for bonemarrow from a 3–83m/d mouse, and a3–83m/d/Bcl-2 mouse, respectively. Lanes 9and 11 show the anti-idiotype (Id)-treatedcultures for bone marrow from a 3–83m/dmouse and a 3–83m/d/Bcl-2 mouse,respectively. (B) Standard curve derived from(A) relating band intensity to percentage ofVk-Jk1 rearrangement. (C) Estimatedfraction of control and anti-Id-treated B cellsthat bear Vk-Jk1 rearrangement after 48 h oftreatment (6). Reproduced from: Hertz M,Namazee D. BCR ligation induces receptorediting in IgMþIgD� bone marrow cells invitro. Immunity 1997;6:429–436 (also Fig. 6reproduced from this reference).

Aıt-Azzouzene et al �Receptor editing

222 Immunological Reviews 197/2004

promoted RS rearrangement amongst normal B cells. A similar

study in the human system came up with similar results, albeit

with a somewhat lower frequency of in-frame rearrangements

(41).

These experiments were taken a step further by ‘repairing’

an inactivated k-allele that was in-frame but silenced by RS

recombination to the intronic site and expressing it alone or

together with that cell’s H-chain in transgenic mice. The cell

we studied was hybridoma 2H11, which inactivated a Vk-Jk5

gene by RS recombination to the k intron and then went on to

express a l L-chain. We found that, when expressed individu-

ally, the 2H11 mH and repaired k-L-chain transgenics sup-

pressed endogenous rearrangements relatively efficiently. The

mH-chain transgenic mouse suppressed the appearance of

endogenous IgM allotype (Fig. 10A), whereas the k transgenic

prevented (excluded) the surface expression of Ig-l when

bred to the Ig-k–/– background (Fig. 10B, center column of

panels). In contrast, in mice carrying both transgenes, lexpression was not suppressed, indicating that the specificity

of the transgenic H/L pair was stimulating editing, probably

because of autoreactivity (in this particular transgenic model,

the L-chain transgene appears to be expressed on every cell,

possibly because its ectopic chromosomal location and lack

of RS elements prevent its inactivation by editing and force

cells to express additional L-chains to dilute out its func-

tion).

Affinity threshold of editing

Taken at face value, these results suggest that receptor editing

may play a major role in antigen receptor quality control and

40

30

20

10

00 24

Rel

ativ

e R

AG

-2 e

xpre

ssio

n

Rel

ativ

e V

κ-Jκ

1 re

arra

ngem

ents

Time after IL-7 removal (h)48

40

30

20

10

00 24 48

Germline κ

Vκ

VJ23–83 Tg light chain

J1

H3H3

2.8 kb

3.2 kb

3–83 Tgκ+/–

Tg Jκ (3.2 kb)

Germline Jκ (2.8 kb)

% germline Jκ

3–83 Tg 3–83 TgNon-Tg

J2 J3

J3

J4

J4

J5

J5

Cκ

Cκ

B-cells cultureLiver DNA from

No Ab 54.1

96 31 100 217

κ+/– κ+/– κ+/–

S17 S17

1.0 7.1

S17op42

Anti-Id (54.1)

Ant

i-λ

JH1

JH2

JH1

JH2

JH1

JH2

JH2

VHJ558

VH7183

VHQ52

VHGAM3-8

Actin

A B C

D

F

ENormalspleen S17 S17 op42 S17 op42

+IL-7 No IL-73–83 Tg B-cell culture

VDJ

Fig. 5. Editing in interleukin-7 (IL-7)-cultured 3–83 transgenic (Tg)B cells. (A) Antigen-dependent recombination-activating gene (RAG)mRNA induction after IL-7 withdrawal. (B–D) k-Chain gene assemblyand expression. Semi-quantitative Southern blot in (D) using thestrategy shown estimates only a 31% retention of the germline

configuration in anti-idiotype (Id)-treated B cells. (E) Absence ofendogenous H-chain VDJ assembly in editing cells, as assessed bypolymerase chain reaction. The transgenic VH gene is a VHGAM3-8family member. (F) New sIg-l cell-surface expression is associatedwith antigen-induced editing (7).

Aıt-Azzouzene et al �Receptor editing

Immunological Reviews 197/2004 223

immune tolerance. Editing can occur in most, if not all, B cells,

and it appears to occur frequently in a normal immune system,

often as a response to autoreactivity. Various theoretical and

experimental estimates of the precursor frequency of autoreact-

ive B cells in the generating repertoire range from 50 to 90%

(22, 40, 42–45). This frequency raises the question of the

affinity threshold of editing. We have addressed this point

using the MHC alloantigen system and 3–83 transgenic mice,

taking advantage of the fact that the 3–83 antibody recognizes

different H-2K alloforms with a wide range of affinities. Using

readouts for editing, including bone marrow RAG expression

and emergence of Ig-lþ/3–83Hþ cells (Fig. 11), we found that

cell-bound MHC molecules with affinities of <5� 104M�1

for the 3–83 receptor were able to stimulate very efficient

editing (46).

These findings suggest that massive receptor editing stimu-

lated by negative selection in the bone marrow could make a

major impact on the immune repertoire. Because of the likely

massive scope of negative selection along with kinetic delays

and cell losses associated with editing, an initially small non-

autoreactive subset should increase in frequency as develop-

ment progresses. Comparisons between the pre-selected and

post-selected repertoires could therefore easily lead to the

erroneous conclusion that the cells not subject to negative

selection are ‘positively selected’.

Limits to editing and the sparing of a B-1 receptor

Although the low-affinity threshold for editing in our

transgenic system suggests that most autoreactive B cells

should be efficiently eliminated, functional autoreactive

B cells are demonstrable in normal mice, especially those

cells reactive to intracellular antigens (47–49). Some auto-

reactive B cells appear to be actively positively selected, at

least into the B-1 and marginal zone compartments (16,

50–53). Many B-1 cells appear to be polyreactive, and

these cells possibly represent a unique cell lineage or

set of antigenic specificities (54). In addition, B-1 cells

arise predominantly from fetal precursors. We wondered

whether adult bone marrow does not normally generate B-

1 type cells, because bone marrow precursors with B-1

specificity are subject to editing. The work of Michael

Chumley and colleagues (55) addressed this problem in a

novel way, by comparing B cells of transgenic mice carry-

ing the 3–83 receptor with those carrying a VH11/Vk9

B-1-derived receptor, which can bind to liposomes made

2.5

2

1.5

1

0.5

0

0

5

10

* *

*15

% λ

+ B

cel

ls%

λ+ B

cel

ls

20

No

trea

tmen

t

Con

trol

mA

b

Ant

i-Id

mA

b

Ant

i-κ m

Ab

Ant

i-IgD

mA

b

No

trea

tmen

t

Con

trol

mA

b

Ant

i-κ (

1 µg

/ml)

Ant

i-κ (

10 µ

g/m

l)

Ant

i-κ (

100

µg/m

l)

Ant

i-IgD

(10

0 µg

/ml)

A

B

P < 0.0001

P < 0.0001

Fig. 6. Anti-B-cell receptor (BCR)-induced editing in ex vivo bone

marrow cultures as measured by the appearance of lþ cells. Otherexperiments verified that the increases seen were not the result ofselective death of kþ cells or proliferation of lþ cells (6). (A) 3–83transgenic total bone marrow cells were challenged in vitro with 10 mg/mlof the indicated antibodies and the appearance of lþ cells measured. (B)Non-transgenic bone marrow cells were challenged with antibodies asshown and the appearance of lþ cells enumerated.

Aıt-Azzouzene et al �Receptor editing

224 Immunological Reviews 197/2004

up of phosphatidylcholine, a constituent of the plasma

membrane of all mouse cells. This study had four interest-

ing findings, illustrated in Fig. 12. First, bone marrow from

the particular VH11/Vk9 line used could give rise to B-1 cells

when transferred to irradiated mice, as judged by cell-surface

markers and cell-survival characteristics, whereas mice reconsti-

tuted with 3–83 cells had a B-2 phenotype. This finding sug-

gested that cells constrained by transgenesis to express a B-1

receptor can generate B-1 cells frommouse bone marrow in vivo,

upon reconstitution of lethally irradiated mice. Second, there

was no evidence that VH11/Vk9-expressing cells were

negatively selected. Indeed, the VH11/Vk9 transgenic bone

marrow cells appeared to suppress RAG expression, an indica-

tion that editing was not occurring, and the cells actually

had improved in vivo and in vitro survival compared to 3–83

cells, i.e. B-2 type cells. Third, IL-7-expanded bone marrow B

cells carrying the B-1 receptor rapidly acquired B-1 character-

istics upon in vitro differentiation. Finally, we found that the

VH11/Vk9 transgenic bone marrow B cells were competent

to edit, if challenged with anti-BCR antibody in vitro. These

studies suggest that cells with certain autoreactive specificities

are not negatively selected by editing, possibly because their

affinities are too low or because these specificities are associated

with unique coreceptor signals that drive B-1 type differentia-

tion.

Why should the receptor-editing mechanism exist?

If tolerance can occur by deletion in the periphery, why is

editing necessary? The answer may lie in the high frequency of

autoreactivity in the pre-selected compartment and the con-

sequences that deletion would have on cell loss. Editing may

be a thrifty approach to tolerance. A second reason may be that

peripheral B cells must be ready to respond to antigen much

more quickly than immature B cells, and tolerance in these

cells may need to be correspondingly slower and contingent

on T-cell signals. Therefore, peripheral tolerance must be slow

and perhaps can never be complete.

The macroself antigen approach

A major future direction of research will be into the molecular

mechanisms of editing and B-cell positive selection. Studies of

signaling protein gene knockouts have begun to suggest

important players in B-cell receptor signaling. However, muta-

tions that block the development may not always be easily

interpretable, because, as we have seen, at the immature B-cell

stage, excessive BCR signaling may arrest development and

promote ongoing rearrangements similarly to subnormal

BCR signaling. As a method of screening signaling mutants

quickly for effects on B-cell tolerance, we have recently

V1

V1

V1

V1

V1

V2

V2

V2

V2

V2

V3

V3

V3

V3

V4

V4

Incompatible signal sequences

Compatible signal sequences facilitate receptor editing

V4

V4

D

D

D J

J

J

J

J

J

J

J

J C

C

C

C

C1°

1°

2°

A

B

‘The first chain’ IgH, TCR-γ, TCR-β?

‘The editing locus’ IgL (κ + λ), TCR-α, TCR-γ?

Fig. 7. Role of gene organization in facilitating or inhibiting receptor

editing. V-, D-, and J-coding elements are flanked by differentrecombination signal sequence that constrain the range of possiblerearrangements in cis. (A) In the initially rearranging locus, such asimmunoglobulin H (IgH), the presence of D elements along with Vgenes in the same transcriptional orientation as the J/C cluster forcesdeletional rearrangements. Primary VDJ assembly cannot be replaced by

recombination using conventional signal sequences. (B) In contrast, inloci without D elements, such as the Ig-k locus, sequentialrearrangements are often possible. In this example, a primary (1�) V4-to-J join is replaced by a subsequent secondary (2�) rearrangement betweenV2 and the downstream J. Such secondary rearrangement permits thereplacement of potentially functional V4-to-J joins, i.e. receptor editing.TCR, T-cell receptor.

Aıt-Azzouzene et al �Receptor editing

Immunological Reviews 197/2004 225

developed a macroself antigen approach. As shown in Fig. 13, a

macroself antigen is a synthetic superantigen based on single-

chain antibodies. For the study of central tolerance, we express

an anti-k specificity as a ubiquitous membrane protein. Such

mice lack kþ cells in the peripheral lymphoid organs and

have an increase in the frequency and absolute numbers of

lþ cells. Such macroself antigen-expressing mice are interest-

ing adoptive hosts for bone marrow transfers from mutant

strains of mice, potentially allowing a rapid assessment of

mutations that affect editing and other types of negative selec-

tion.

Conclusion

From this survey of work carried out over a number of

years, the picture emerges that tolerance-induced receptor

editing is a major player in the somatic learning process,

which is a unique feature of the immune system. Editing is

a process that occurs often with great efficiency and min-

imal cell loss. In addition, the affinity threshold of editing

is such that it may eliminate many potentially dangerous

specificities. Editing is also a pathway available in the case

of receptors that fail positive selection, adding to its poten-

tial value as a quality control and repair mechanism and

raising questions about the signals from the antigen recep-

tor that are involved in this process. We still have much to

learn about the parameters of these signals, how they are

regulated, and what might go wrong in disease. For example,

we do not really know what phenotype to predict in a

mutant individual who is unable to carry out editing.

It may well be that such defects contribute to certain auto-

immune conditions. It remains possible that B cells are

heterogeneous with respect to their antigen receptor signal-

ing thresholds for editing, which might underlie the func-

tional and phenotypic heterogeneity of B-cell subsets such as

marginal zone, follicular, and B-1 populations.

An important insight arising from these studies is into the

interplay between antigen receptor gene structure and editing.

Relative to the heavy-chain locus, the k-locus structure pro-

vides only a minor contribution to diversity, but these same

104

103

13.5%

42.1%44.2%

40.9%

102

101

104

103

102

101

0

Donors

Idio

type

0

IgM

Recipients

H-2d H-2b

3–83KI→

3–83KIH→

Fig. 8. Both underexpressed and harmfulB-cell receptors are subject to editing. Datashown are from a bone marrow chimeraexperiment in which donor bone marrowfrom 3-83 knockin heterozygous (3–83KI)or homozygous (3–83KIH) mice developedin the presence (H-2b) or absence (H-2d) of3–83 ligand. 3–83KI B cells only poorlyretained the transgenic specificity on aninnocuous, H-2d background (top left),whereas 3–83KIH B cells express the 3–83receptor well on the H-2d background (lowerleft) and efficiently edit these receptors in thepresence of H-2Kb autoantigen (lower right).Note that tolerance alters receptor specificity,eliminating the autoreactive idiotype, butdoes not alter B-cell numbers in the spleen(29). IgM, immunoglobulin M.

V’s VJ J’s IRS1 iE Cκ RS

RS probeIVS probe

3′E

AVRS VJ-intron-RS

A CB BPCR primers PCR primers

A

B C

Fig. 9. How receptor selection (RS) rearrangements to the Jk-Ckintron inactivate the k-locus and retain VJ joins. A rearranged,potentially functional k-locus (A) can be silenced by two types of RSrecombination (B, C). Type (C) retains the prior Vk-Jk join. (C) RSrecombination eliminates the known enhancers that are critical forefficient rearrangement and expression, thus freezing the locus to furtherV-J recombination. These silenced VJ joins are often in-frame andapparently functional (40). PCR, polymerase chain reaction.

Aıt-Azzouzene et al �Receptor editing

226 Immunological Reviews 197/2004

A

B

104

104

103

103

102

102

101

100

101100 100100

104

104

103

103

102

102

101

100 100 100

100

100100

100100

100

101100 100 100

100 100 100

100 100 100

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101

Spleen

Lymphnode

Bonemarrow

λ

κ

H-only L-only

IgMb

H + L B220+ gatedTransgene

51% 0.5%

1%55%

14%

92%

11% 84%

79%

16%5%

90%

51% 35%

61%

2% 14%

67%

14%

IgMa

Non-transgenic H-chain transgenicFig. 10. Recapitulation of L-chainrearrangements in transgenic mice

carrying antibody H-chain andreconstructed k-chain from recombining

sequence (RS)-edited cell 2H11. (A)H-chain transgene of 2H11 excludesendogenous H-chain expression. Thetransgene fragment was engineered to havethe 2H11 VDJ and flanking genomicsequences upstream of a genomic Cm cassettederived from BALB/c (‘a’ allotype). Thegenetic background of the mice was B10.D2(‘b’ allotype). (B) Allelic exclusion in 2H11L-chain-only transgenic but not in HþL-transgenic mice (arrow) suggests aspecificity-dependent induction of receptorediting. All transgenic mice were crossed tomice lacking endogenous k loci; therefore allk protein is transgenic and all l protein isendogenously encoded. For clarity, cellsanalyzed were gated on B220þ populations.In both 2H11 H-chain only and HþLtransgenics, endogenous H-chain expressionwas suppressed (Fig. 3).

8

6

4

2

0

% Ig

Da

λ+ c

ells

d

H-2 of 3–83 transgenic mice

17

2

107

45

2

5

7 13

k b bm3 bm6 bm8 bm11 Dκ MT-Kb

Non-Tg

10

12

Fig. 11. Low-affinity ligands that mediate efficient central deletion

induce receptor editing as measured by appearance of 3–83-idiotype-negative B cells that coexpress transgenic H-chain (IgDa)and endogenous L-chain (l) (46).

Aıt-Azzouzene et al �Receptor editing

Immunological Reviews 197/2004 227

features, such as the lack of D elements and the propensity for

locus-inactivating rearrangements, significantly facilitate edit-

ing. Our studies point to the RS element in the k-locus as a

potentially important player in allowing editing, while also

facilitating the maintenance of B-cell allelic and isotypic

exclusion.

104

104

103

103

102

102

10

101

1

104

104

103

103

102

102

10

101

1

104103102101

104103102101

104103102101

104

104

103

103

102

102

10

101

1

VH11/Vκ9 BMchimera

Mixed BMchimera

3–83µδ BMchimera

CD43 CD5

CD23

B10

.D2

BM

3–83

md

BM

(H

-2d )

Thy

mus

Thy

mus

IIA1.

6

IIA1.

6

CD19

Rag-2

CD19

Rag-2

No

trea

tmen

t

Gate A

Gate B

anti-

κ

anti-

Ars

No

trea

tmen

t

anti-

κ

anti-

Ars

No

trea

tmen

t

anti-

κ

anti-

Ars

VH11

/Vκ9

BM

3–83

µδ B

M (

H-2

k )

B10.D2BM

3–83µδBM

VH11/Vκ9

3–83

µδ

VH11/Vκ9 BM

A B

DC

96%

96%

Fig. 12. B-1 receptor with a self-reactive affinity sufficient todrive bone marrow (BM) B-1 development fails to induce

receptor editing, yet permits editing to strong B-cell receptor(BCR) stimulus. (A, B) Analysis of splenic B cells in mixed(VH11/Vk9 transgenicþ 3–83 md transgenic)(!)B10.D2 radiationBM chimeras demonstrates the ability of the B-1- and B-2-derivedreceptors to drive cell autonomous development of B-1 and B-2 celltypes, respectively. Mixed chimeras received an equal dose of bothdonor BM types. (A) Transgene-encoded BCR expression wasanalyzed with idiotypic antibodies in spleen cells of the followingchimeras: VH11/Vk9!)B10.D2 (left panel), 3–83md!) B10.D2(center), and (VH11/Vk9þ 3–83 md)!)B10.D2 (lower left).The binding of the anti-idiotypes defined two analysis gates, a and bused in panel B. (B) Splenic B cells of the mixed BM chimera

[gated as indicated in (A)] were further analyzed for the expressionof CD43, CD23, and CD5. (C, D) Reverse-transcriptase-polymerasechain reaction analysis of RAG-2 expression in BM of transgenic mice.(C) Lack of spontaneous RAG-2 expression in the BM of VH11-Vk9mouse. Results using BM from 3-83 md transgenic mice on theantigen-bearing (H-2k) or antigen-free (H-2d) backgrounds indicatethe effect of autoantigen on RAG expression. Other controls includethymus and the RAG-2-non-expressing cell line, IIA1.6.(D) VH11-Vk9 B cells express RAG-2 in vitro following receptorcrosslinking. Whole BM isolated from 3-83 md, VH11-Vk9, andnon-transgenic B10.D2 mice were cultured with anti-k monoclonalantibody (10 mg/ml) or media supplemented with a non-specificmonoclonal antibody (10 mg/ml) for 24 h and analyzed for RAG-2mRNA expression (55).

Aıt-Azzouzene et al �Receptor editing

228 Immunological Reviews 197/2004

104

104

103

6%

25%

26%

47%

>0.2%

>0.2%

103

102

102

101

100

100

100

100

100

100

101

104

104

103

103

102

102

101

101

104

104

103

103

102

102

101

101104103102101

0

20

40

60

80

100

0

20

40

60

Cou

nts

Cou

nts

80

100

104103102101100 100

Rat-IgG1PE Rat-IgG1PEAnti-rat-IgG Fc

Anti-κ Fv

Rat γ1-CH1,2, hinge

H-2Kb Tm/cyto

Igκ macroself Tgline 11

Ig-κ macroself Tgline 11

Transgene surface expression

Igκ macroself Tgline 2

Igκ macroself Tgline 2

Wildtype

A C

B

B220

Tg

Tg

WT

Ig-κ

Fig. 13. Anti-k macroself antigen-transgenic (Tg) mice. (A) Apredesigned ligand that is expressed on cell surfaces and reacts withall immunoglobulin (Ig)-kþ cells. Transgenic mice expressing sucha molecule ubiquitously were generated. (B) Expression pattern of

transgene-encoded protein on blood leukocytes. (C) Elimination of Ig-kcells in the blood of the transgenic mice. A large population of Ig-l cellsis present (not shown), which accounts for the high frequency of B220þ

cells. WT, wildtype.

References

1. Burnet FM. The clonal selection theory of

acquired immunity. Cambridge: The

University Press, 1959.

2. Talmage DW. Immunological specificity:

unique combinations of selected natural

globulins provide an alternative to the classical

concept. Science 1959;129:1643–1648.

3. Nemazee D. Receptor selection in B and T

lymphocytes. Annu Rev Immunol

2000;18:19–51.

4. Tiegs SL, Russell DM, Nemazee D. Receptor

editing in self-reactive bone marrow B cells.

J Exp Med 1993;177:1009–1020.

5. Lang J, et al. Enforced Bcl-2 expression

inhibits antigen-mediated clonal elimination

of peripheral B cells in an antigen dose-

dependent manner and promotes receptor

editing in autoreactive, immature B cells. J Exp

Med 1997;186:1513–1522.

6. Hertz M, Nemazee D. BCR ligation induces

receptor editing in IgMþIgD– bone marrow B

cells in vitro. Immunity 1997;6:429–436.

7. Melamed D, Nemazee D. Self-antigen does not

accelerate immature B cell apoptosis, but

stimulates receptor editing as a consequence

of developmental arrest. Proc Natl Acad Sci

USA 1997;94: 9267–9272.

8. Melamed D, Benschop RJ, Cambier JC,

Nemazee D. Developmental regulation of B

lymphocyte immune tolerance

compartmentalizes clonal selection from

receptor selection. Cell 1998;92:173–182.

9. Sandel PC, Monroe JG. Negative selection of

immature B cells by receptor editing or

deletion is determined by site of antigen

encounter. Immunity 1999;10:289–299.

10. Sandel PC, Gendelman M, Kelsoe G, Monroe

JG. Definition of a novel cellular constituent

of the bone marrow that regulates the

response of immature B cells to B cell antigen

receptor engagement. J Immunol

2001;166:5935–5944.

11. Tze LE, Hippen KL, Behrens TW. Late

immature B cells (IgM (high) IgD (neg))

undergo a light chain receptor editing

response to soluble self-antigen. J Immunol

2003;171:678–682.

12. Chen C, Prak EL, Weigert M. Editing disease-

associated autoantibodies. Immunity

1997;6:97–105.

13. Erikson J, Radic MZ, Camper SA, Hardy RR,

Carmack C, Weigert M. Expression of anti-

DNA immunoglobulin transgenes in non-

autoimmunemice. Nature 1991;349: 331–334.

14. Gay D, Saunders T, Camper S, Weigert M.

Receptor editing: an approach by autoreactive

B cells to escape tolerance. J Exp Med

1993;177:999–1008.

15. Li H, Jiang Y, Prak EL, Radic M, Weigert M.

Editors and editing of anti-DNA receptors.

Immunity 2001;15:947–957.

16. Li Y, Li H, Weigert M. Autoreactive B Cells in

the marginal zone that express dual receptors.

J Exp Med 2002;195:181–188.

Aıt-Azzouzene et al �Receptor editing

Immunological Reviews 197/2004 229

17. Prak EL, Trounstine M, Huszar D, Weigert M.

Light chain editing in kappa-deficient

animals: a potential mechanism of B cell

tolerance. J Exp Med 1994;180:1805–1815.

18. Radic MZ, Erikson J, Litwin S, Weigert M. B

lymphocytes may escape tolerance by revising

their antigen receptors. J Exp Med 1993;177:

1165–1173.

19. Pewzner-Jung Y, Friedmann D, Sonoda E,

Jung S, Rajewsky K, Eilat D. B cell deletion,

anergy, and receptor editing in ‘knock in’

mice targeted with a germline-encoded or

somatically mutated anti-DNA heavy chain. J

Immunol 1998;161:4634–4645.

20. Yachimovich N, Mostoslavsky G, Yarkoni Y,

Verbovetski I, Eilat D. The efficiency of B cell

receptor (BCR) editing is dependent on BCR

light chain rearrangement status. Eur J

Immunol 2002;32:1164–1174.

21. Luning Prak E, Weigert M. Light chain

replacement: a new model for antibody gene

rearrangement. J Exp Med 1995;182:

541–548.

22. Casellas R, et al. Contribution of receptor

editing to the antibody repertoire. Science

2001;291:1541–1544.

23. Pelanda R, Schaal S, Torres RM, Rajewsky K. A

prematurely expressed Ig (kappa) transgene,

but not V (kappa) J (kappa) gene segment

targeted into the Ig (kappa) locus, can rescue

B cell development in lambda5-deficient

mice. Immunity 1996;5:229–239.

24. Gollahon KA, Hagman J, Brinster RL, Storb U.

Ig lambda-producing B cells do not show

feedback inhibition of gene rearrangement. J

Immunol 1988;141:2771–2780.

25. Ritchie KA, Brinster RL, Storb U. Allelic

exclusion and control of endogenous

immunoglobulin gene rearrangement in

kappa transgenic mice. Nature

1984;312:517–520.

26. Storb U. Transgenic mice with

immunoglobulin genes. Annu Rev Immunol

1987;5:151–174.

27. Goldmit M, Schlissel M, Cedar H, Bergman Y.

Differential accessibility at the kappa chain

locus plays a role in allelic exclusion. EMBO J

2002; 21:5255–5261.

28. Mostoslavsky R, et al. Kappa chain monoallelic

demethylation and the establishment of allelic

exclusion. Genes Dev 1998;12:1801–1811.

29. Kouskoff V, Lacaud G, Pape K, Retter M,

Nemazee D. B cell receptor expression level

determines the fate of developing B

lymphocytes: receptor editing versus

selection. Proc Natl Acad Sci USA 2000;97:

7435–7439.

30. Braun U, Rajewsky K, Pelanda R. Different

sensitivity to receptor editing of B cells from

mice hemizygous or homozygous for targeted

Ig transgenes. Proc Natl Acad Sci USA

2000;97:7429–7434.

31. Pelanda R, Schwers S, Sonoda E, Torres RM,

Nemazee D, Rajewsky K. Receptor editing in a

transgenic mouse model: site, efficiency, and

role in B cell tolerance and antibody

diversification. Immunity 1997;7:765–775.

32. McMahan CJ, Fink PJ. RAG reexpression and

DNA recombination at T cell receptor loci in

peripheral CD4þ T cells. Immunity

1998;9:637–647.

33. McGargill MA, Derbinski JM, Hogquist KA.

Receptor editing in developing T cells. Nat

Immunol 2000;1:336–341.

34. Wang F, Huang CY, Kanagawa O. Rapid

deletion of rearranged T cell antigen receptor

(TCR) Valpha-Jalpha segment by secondary

rearrangement in the thymus: role of

continuous rearrangement of TCR alpha chain

gene and positive selection in the T cell

repertoire formation. Proc Natl Acad Sci USA

1998;95:11834–11839.

35. Buch T, Rieux-Laucat F, Forster I, Rajewsky K.

Failure of HY-specific thymocytes to escape

negative selection by receptor editing.

Immunity 2002;16:707–718.

36. Durdik J, Moore MW, Selsing E. Novel kappa

light-chain gene rearrangements in mouse

lambda light chain-producing B lymphocytes.

Nature 1984;307:749–752.

37. Moore MW, Durdik J, Persiani DM, Selsing E.

Deletions of kappa chain constant region

genes in mouse lambda chain-producing B

cells involve intrachromosomal DNA

recombinations similar to V-J joining. Proc

Natl Acad Sci USA 1985;82:6211–6215.

38. Siminovitch KA, Moore MW, Durdik J, Selsing

E. The human kappa deleting element and the

mouse recombining segment share DNA

sequence homology. Nucleic Acids Res

1987;15:2699–2705.

39. Hieter PA, Korsmeyer SJ, Waldmann TA,

Leder P. Human immunoglobulin kappa

light-chain genes are deleted or rearranged in

lambda-producing B cells. Nature

1981;290:368–372.

40. Retter MW, Nemazee D. Receptor editing

occurs frequently during normal B cell

development. J Exp Med 1998;188:

1231–1238.

41. Brauninger A, Goossens T, Rajewsky K,

Kuppers R. Regulation of immunoglobulin

light chain gene rearrangements during early

B cell development in the human. Eur J

Immunol 2001;31:3631–3637.

42. Nemazee D. Antigen receptor ‘capacity’ and

the sensitivity of self-tolerance. Immunol

Today 1996;17:25–29.

43. Nemazee D. Theoretical limits to massive

receptor editing in immature B cells. Curr Top

Microbiol Immunol 1998;229:163–171.

44. Wardemann H, Yurasov S, Schaefer A, Young

JW, Meffre E, Nussenzweig MC. Predominant

autoantibody production by early human B

cell precursors. Science 2003;301:

1374–1377.

45. Louzoun Y, Luning PE, Friedman T, Litwin S,

Weigert M. Comment on Langman and Cohn.

Semin Immunol 2002;14:231–232.

46. Lang J, Jackson M, Teyton L, Brunmark A,

Kane K, Nemazee D. B cells are exquisitely

sensitive to central tolerance and receptor

editing induced by ultralow affinity,

membrane-bound antigen. J Exp Med

1996;184:1685–1697.

47. Dighiero G, et al. Murine hybridomas

secreting natural monoclonal antibodies

reacting with self antigens. J Immunol

1983;131:2267–2272.

48. Rolink AG, Radaszkiewicz T, Melchers F. The

autoantigen-binding B cell repertoires of

normal and of chronically graft-versus-host-

diseased mice. J Exp Med 1987;165:

1675–1687.

49. McHeyzer-Williams MG, Nossal GJ. Clonal

analysis of autoantibody-producing cell

precursors in the preimmune B cell repertoire.

J Immunol 1988;141:4118–4123.

50. Hayakawa K, et al. Positive selection of natural

autoreactive B cells. Science 1999;285:

113–116.

51. Chen X, Kearney JF. Generation and function

of natural self-reactive B lymphocytes. Semin

Immunol 1996;8:19–27.

52. Chen X, Martin F, Forbush KA, Perlmutter

RM, Kearney JF. Evidence for selection of a

population of multi-reactive B cells into the

splenic marginal zone. Int Immunol

1997;9:27–41.

53. Gu H, Tarlinton D, Muller W, Rajewsky K,

Forster I. Most peripheral B cells in mice are

ligand selected. J Exp Med 1991;173:

1357–1371.

54. Hardy RR. Development of VH11þ B cells: a

model for selection of B cells producing

natural autoantibodies. Curr Dir Autoimmun

2003;6:196–211.

55. Chumley MJ, Dal Porto JM, Kawaguchi S,

Cambier JC, Nemazee D, Hardy RR. A VH11V

kappa 9 B cell antigen receptor drives

generation of CD5þ B cells both in vivo and in

vitro. J Immunol 2000;164:4586–4593.

56. Kouskoff V, Lacaud G, Nemazee D. T cell-

independent rescue of B lymphocytes from

peripheral immune tolerance. Science

2000;287:2501–2503.

Aıt-Azzouzene et al �Receptor editing

230 Immunological Reviews 197/2004