PDFlib PLOP: PDF Linearization, Optimization, Protection

Page inserted by evaluation versionwww.pdflib.com – sales@pdflib.com

Production of Antigen-Specific Human Antibodies from Mice Engineered with Human Heavy and Light Chain YACs"

AYA JAKOBOVITS, LARRY L. GREEN, MARGARET C. HARDY,

DONNA M. LOUIE, MICHAEL J. MENDEZ, HAD1 ABDERRAHIM, MASATO NOGUCHI,

DOUGLAS H. SMITH, YONGJUN ZENG, NATHANIEL E. DAVID, HITOSHI SASAI, DAN GARZA,

DANIEL G. BRENNER, JOANNA F. HALES, RYAN P. McGUINNESS, DANIEL J. CAPON,

AND SUE KLAPHOLZ Cell Genesys, Inc. 322 Lakeside Drive

Foster City, California 94404

CATHERINE E. MAYNARD-CURRIE, HIROHISA TSUDA,

INTRODUCTION

One of the major obstacles in the development of therapeutic applications in humans for mouse and mouse-derivatized monoclonal antibodies is their intrinsic immunogenicity that reduces dramatically the efficacy and safety of the adminis- tered antibodies. The full therapeutic benefit of monoclonal antibodies can be properly assessed only with the availability of fully human monoclonal antibodies that could be administered repeatedly. One of the most appealing approaches towards this goal is the creation of a mouse engineered with unrearranged human immunoglobulin (Ig) chains, capable of producing a large repertoire of human antibodies, in the absence of mouse antibodies. Such a mouse could be used to generate high-affinity, and high-specificity, fully human monoclonal antibodies directed against a broad spectrum of antigens, including human proteins.

Here we describe a novel strategy to create a mouse strain in which the majority of the antibodies produced are fully human, by inactivating the mouse Ig genes and introducing into the mouse germline large fragments of the human heavy and light chain loci (FIG. 1).

Using the technology we have recently developed,' large fragments of the human heavy and kappa light chain loci cloned on yeast artificial chromosomes (YACs) have been introduced into the mouse germline via fusion of yeast sphero- plasts with mouse embryonic stem (ES) cells.' These fragments, which contain the structural genes for variable, joining, diversity, and constant domains as well

a The work was supported in part by Xenotech Inc., the joint venture between Cell Genesys Inc. and JT Immunotech.

525

526 ANNALS NEW YORK ACADEMY OF SCIENCES - Mouse ES cells with inactivated mouse Ig genes Breeding

Inactivate Mouse heavy and light chain loci by gene targeting f n

Unmodified mouse ES cells

Introduce human heavy and light chain loci contained on YACs -

Breeding

\ y H W

VH D JH CH Mouse ES cells with

YK +H-H-+ integrated human VK JK CK antibodygenes

Mouse strain incapable of producing mome antibodies

Mouse strain producing human antibodies in the absence of mouse antibodies

Mouse strain producing human antibodies

FIGURE 1. An outline of the strategy used to generate Xenomouse strain, producing fully human antibodies but not endogenous mouse antibodies.

as critical regulatory elements, proved sufficient for the production of a broad adult-like repertoire of human Ig and capable of mounting significant levels of human antibodies in mice, reflecting a utilization of a repertoire in many ways similar to adult humans.’ Moreover, the breeding of such mice with animals lacking functional mouse antibody genes3 led to a mouse strain in which fully human antibodies are produced at levels approaching those observed in normal serum.? Such mice may be exploited to develop human monoclonal antibodies for the treatment of human disease.

INTRODUCTION OF HUMAN HEAVY AND KAPPA CHAIN YACs INTO ES CELLS AND MICE

YACs containing variable and constant sequences from the human heavy and kappa chain loci (FIG. 2), identified from the Washington University human-YAC library, were characterized and determined to be in intact germline c~nfiguration.~ The human heavy chain YAC harbors a 220-kb DNA insert containing the g and 6 constant regions, all six functional joining genes (JH 1-6), the major diversity (D) cluster, the intronic enhancer, and five variable (V) genes from four human VH families: V,, , VI (2 genes: V,.:, VI.3), VIv-4 and VII.5. The 170-kb DNA insert in the human kappa YAC contains the kappa-deleting element (Kde), the intronic and 3’ enhancers, the C, constant region, all five functional joining genes (JK1-5), and the three V, variable genes in the B cluster (BI , B2, B3). Selection for the incorporation of these YACs in ES cells deficient €or hypoxanthine phosphoribosyl

JAKOBOVITS et al.: PRODUCTION OF AG-SPECIFIC HUMAN ABS 527

transferase (HPRT) was conferred by a human HPRT expression c a ~ s e t t e , ~ tar- geted into the right vector arm of each YAC.4

HPRT-positive (HPRT') ES cell colonies were evaluated by Southern blot analysis for the presence and integrity of the human immunoglobulin YACs (TABLE fA). Seven HPRTt ES clones obtained by fusion with yH1-containing yeast spheroplasts were analyzed for human heavy chain-specific sequences using probes spanning the 220-kb insert. Five clones analyzed were shown to contain all of the Hind111 fragments detected by probes for the Cp, C6, JH, D, and VH genes found in the parental yH1 YAC. Deletions within the D or V,, regions were detected in the other two clones (TABLE 1A). A single YAC integration was detected in all but two clones, in which multiple copies were detected (TABLE 1A). Eleven HPRT+ ES clones obtained from fusion with yK1-containing yeast spheroplasts were similarly analyzed, using probes spanning the kappa chain insert, including probes for the Kde and the C,, J,, and three V, genes (B3, B2, and B I ) (TABLE 1B). Ten HPRT' ES clones analyzed contained all of the Hind111 fragments detected by these probes in the parental yK1 YAC, and ail but four of the clones were shown to contain a single integrated YAC (TABLE IB). These results, together with pulsed-field gel analysis of these clone^,^ strongly suggested that the single YAC integrated in the yH1- and yK1-ES cell clones is structur- ally intact.

Yeast genomic sequences were detected in four yH1 and in five yK1 ES cell clones, whereas no yeast sequences were detected in the remaining clones (TABLES 1A and B).

Five yH1- and five yK1-containing ES cell clones were microinjected into C57BL/6J blastocysts, and mice containing the respective YACs in their germline were generated. Both YAC and yeast genomic DNA (when present) were trans- mitted in unaltered form.

Our results indicate that both the yH1 and yK1 YACs were integrated into ES cells at a high frequency (1 HPRT+ colony/106 fused cells), primarily as an intact, single-copy, and apparently unrearranged human insert. When present, the

Human Heavy Chain YAC yH1 ,20 kb,

T*P' VIM "w, v,, v,, V",., D J, c1 6 HPRT URA3 I I I I I~

I I - - - I w

Human Kappa Chain YAC yK1

TRP 7 B1 B2 83 J, Cr Kde Ebl HPRT URA3

I . I . I a 1. I I I

- F Bsshll Sfil

FIGURE 2. Schematic representations of the human heavy (yH1) and kappa (yK1) YACs, retrofitted with a HPRT minigene. The locations of specific Ig regions are indicated along with YAC vector elements: 4, telomere; 0 , centromere; ., EcoRI cloning site and yeast- selectable markers, TRPl and URA3 (YAC arms are not shown to scale).

TABL

E ZA.

Stru

ctur

al I

nteg

rity

of

Hum

an H

eavy

Cha

in-D

eriv

ed Y

AC

in E

S C

lone

s, G

ener

ated

from

Fus

ion

with

yH

1-C

onta

inin

g Y

east

Sph

erop

last

s"

Yea

st

YA

C

COPY

G

enom

ic

yH1

ES

C6

CY

JH

D

VV

I V

i V

lV

v I1

Cel

l C

lone

7.

8 kb

11

kb

10 k

b 9.

7, 8

.5, 7

kb

-28

kb

>12,

1.7

kb

8 kb

4.

8 kb

Se

quen

ces

Num

ber

1 2B

+

+ +

+ +

+ +

+ 2

c +

+ +

+ +

+ +

+ +

2 12

5A

+ t

+ +

+ t

+ t

+ 1 1

125E

+

+ +

+ +

+ +

+ 3A

+

+ +

+ +

+ +

t +

1 3B

+

+ +

9.7

kb

+ +

+ +

+ -8

1 5

c +

+ +

+ +

+ +

-

-

8.5

kb

-

-

a H

PRT

+-E

S cl

ones

wer

e an

alyz

ed f

or t

he p

rese

nce

of h

uman

hea

vy c

hain

-spe

cific

seq

uenc

es a

nd y

east

gen

omic

seq

uenc

es.

Sout

hern

blo

t an

alys

is, p

robe

s us

ed, a

nd d

eter

min

atio

n of

YA

C c

opy

num

ber

wer

e as

des

crib

ed i

n G

reen

et

al.'

The

pre

senc

e of

the

exp

ecte

d fr

agm

ent s

izes

fo

r the

spe

cific

pro

be (

as s

how

n) a

nd o

f va

ried

leve

ls o

f ye

ast

geno

mic

seq

uenc

es a

re in

dica

ted

as (+

). A

ltere

d-si

zed

frag

men

ts a

re in

dica

ted.

In

the

case

of

yH1-

ES c

lone

3B

, the

7-k

b do

uble

t was

del

eted

. ND

, not

det

erm

ined

.

JAKOBOVITS el al.: PRODUCTION OF Ac-SPECIFIC HUMAN ABS 529

TABLE IB. Structural Integrity of Human Kappa Light Chain Derived YAC in ES Clones, Generated from Fusion with yK1-Containing Yeast Spheroplastsa

Yeast YAC yK1 ES Kde C, J, B3 B2 B1 Genomic Copy

Cell Clone 2.5 kb 5.8 kb 5 kb 4.8 kb 5.1 kb 2.8 kb Sequences Number

4.4 5.2 8.2A 9.2E 9.2c

10.2B 9.2H 9.2F

10.2c 10.2A 10.2CA

+ + + + + + + + + + +

t + + + + + + + + + +

5.5 kb + + + + + + + + + +

4.7 kb + + + + + + + + + +

4.9 kb + + + + + +

ND ND + +

+ + + + + + + + + + +

+ - - + + + + + + +

-

1 1 I 2 1 1

-3 1 1 2 2

HPRTt-ES clones were analyzed for the presence of human kappa chain-specific se- quences and yeast genomic sequences. Southern blot analysis, probes used, and determina- tion of YAC copy number were as described in Green et al.' The presence of the expected fragment sizes for the specific probe (as shown) and of varied levels of yeast genomic sequences are indicated as (+ ). Altered-sized fragments are indicated. ND, not determined.

cointegrated yeast genomic DNA apparently did not interfere with proper mouse development. These results demonstrate the utility and reproducibility of ES cell-yeast spheroplast fusion technology for the functional introduction of large and complex loci into the mouse germline.

PRODUCTION OF ANTIGEN-SPECIFIC HUMAN ANTIBODIES BY yHl-, gK1-BEARING MICE

Expression of the inherited human heavy and kappa chain YACs by mouse B cells was demonstrated by flow cytometry and ELISA. In all yHl-containing mouse strains analyzed, 1.5-2.5% of the B220+ cells expressed human p chain (hp). Approximately half of this population had no detectable surface mouse p, indicating that the production of human heavy chains can at least partially exclude the expression of the mouse p chains. The majority of the B220+, hp+ population also expressed surface human 6 chain (ha). Similarly, in all yKl-bearing mouse strains analyzed, 5 to 9% of the B220+ splenocytes expressed surface human K chain (hK). The majority (80%) of the hK+ population did not express mouse kappa (mK) or lambda (mh), indicating that the human kappa chain was able to exclude expression of the mouse light chain genes.

Human heavy and kappa chain proteins were detected by ELISA in sera derived from the respective yH1- and yK1-containing mice. Human p protein was detected in all yH1-bearing mice at concentrations of 0.2-3.8 pglml. Similarly, human K protein, at levels of 8 to 30 pg/ml, was detected in all mice containing intact and unrearranged yK1. Thus, significant immunoglobulin chain expression was observed in mice derived from independent YAC-containing ES cell clones, which are likely to represent different YAC integration sites.

530 ANNALS NEW YORK ACADEMY OF SCIENCES

Heavy and kappa chain-expressing mouse strains were bred to generate human antibody-producing (HuAb) mice containing one copy each of yH1 and yK1. Flow cytometry analysis of B splenocytes, derived from five different HuAb strains, revealed B220' populations containing 0.8 to 3.9% hp' or 1.4 to 4.8% hK+ cells, as well as a population of B220' cells (0.06-0.27%) that simultaneously expressed both human p and K chains. The presence of fuliy human antibodies (hp/hK), at levels of 0.3 to 0.7 mg/ml, in sera from HuAb mice was demonstrated by ELISA, based on the capture of the h p chain and the detection of the hK chain.

To determine whether HuAb mice can mount a specific human antibody re- sponse, mice were immunized with tetanus toxin C fragment (tet C) or with a human protein, IgE antibody. Upon immunization, tet C-specific or IgE-specific hp and hK antibodies were readily detected in the serum of immunized mice. The human origin of these antigen-specific antibodies was confirmed by an ELISA in which each of the antigens was used to capture an anti-hp or anti-hK employed to detect the bound species (FIG. 3). Furthermore, upon fusion of splenocytes derived from these immunized mice with P3X63-Ag8.653 myeloma cells, mouse hybridomas producing tet C or IgE-specific fully human antibodies were gener- ated.' These results indicate that HuAb mice, although containing only a small fraction of B cells coexpressing hp and hK, can mount an antigen-specific response leading to the generation of fully human monoclonal antibodies.

HUMAN IG YACs DIRECT THE PRODUCTION OF A DIVERSE,

SEQUENCES IN MICE ADULT-LIKE REPERTOIRE OF HUMAN ANTIBODY

The diversity of the human antibody repertoire, directed by the human heavy and light chain YACs, was determined by analysis of immunoglobulin transcripts from mouse spleen. Colony hybridization analysis using V and J region-specific probes (TABLES 2A and B) and sequencing of human heavy and kappa chain cDNA clones revealed broad usage of V , D, and J segments contained in the YACS.~ All JH and J, segments were represented, with a frequency similar to that detected in adult human B In 18 heavy chain cDNAs, 10 different D segments were represented (NI , 21/9, XP4, M I , XPI, K1, N4, LR2, A4, Al).'Three of the four V, families contained in yHI (V,,, V,,, V,) were represented. All three of the V, variable segments (Bl , B2, and B3) were represented, with B3 the most fre- quently used. Frequent N segment addition (83%) was observed in the heavy chain cDNAs, with a majority having N segments at both the 5' and 3' D segment junctions. The average length of N regions was 6.1 nucleotides. The majority of the in-frame heavy chain clones contained a complementary determining region (CDR3) sequence of 10 or more amino acids, with some as long as 18 residues (average 12). CDR3 sequences, for the in-frame K transcripts were 9 to 10 amino acids in length.

The pattern of V, , D, and JH use demonstrated is reminiscent of adult human B cells6.' in several respects. First, there is an absence of position-biased usage of V and D genes, in particular, the Vv, and DQ52 segments, which are preferentially employed during human fetal d e v e l ~ p m e n t . ~ , ~ In addition, the frequency of usage of specific JH genes, the extent and length of N segments, and the length of the heavy and kappa CDR3 region, all appear to recapitulate the diversity seen in adult humans. These results suggest that the human immunoglobulin sequences contained on these YACs include all the elements necessary to produce a normal adult human heavy chain repertoire.

JAKOBOVITS et al.: PRODUCTION OF AG-SPECIFIC HUMAN ABS 531

u

d

532 ANNALS NEW YORK ACADEMY OF SCIENCES

TABLE 2A. Repertoire Analysis of Human Heavy Chain Transcripts Expressed in Transgenic Mice"

VH VIV Vl VVI Total Human PBL'

J l 0 0 1 0 1 ( 1%) 1% J? 0 0 1 0 1 ( 1%) 0% 53 0 5 7 3 15 (18%) 9.1% 5 4 0 8 16 21 45 (54%) 52.5% J 5 0 0 3 1 4 ( 5%) 15.2% J, 0 3 7 7 17 (21%) 22.2%

(0%) (19%) (42%) (39%) (100%) 100% Total 0 16 35 32 83

~~ ~~

Human p-specific mRNAs were amplified by PCR, cloned and analyzed by hybridization to VH- and JH- region specific probes, respectively, as described in Green et a/.* Results of colony hybridizations showing VH and JH (as compared to that detected in human peripheral B cells described in Yamada et gene utilization. Results are given only for colonies hybridizing to a V,, a J H , and a Cp probe. A similar J , V utilization pattern was obtained by nucleotide sequences of unique human heavy chain cDNA clones.2

Yamada er

THE HUMAN IC YACs RESTORE B-CELL DEVELOPMENT AND ANTIBODY PRODUCTION IN IG-INACTIVATED MICE

The results described above demonstrate that the YAC-encoded human immu- noglobulin genes are productively rearranged and expressed in the presence of functional mouse antibody genes. Nevertheless, the preferential expression of the mouse Ig genes indicated the need for their inactivation to increase the production of fully human antibodies in mice.

The ability of yH1 to restore proper B-cell development and antibody produc- tion was examined in mice in which both heavy chain alleles were functionally inactivated by gene-targeted JH region (AJH/AJH).3 In AJH/AJH mice, bearing one copy of yH1 ( yHI;AJH/AJH), reconstitution of mature B cells was detected in

TABLE ZB. Repertoire Analysis of Human Kappa Chain Transcripts Expressed in Transgenic Mice"

BI B2 B3 Total

J, 4 13 I08 125 (47%) Jz 10 5 53 68 (26%)

3 4 1 9 25 35 (13%) J 3 0 2 29 31 (12%)

JS 0 0 4 4 ( 1%)

Total 15 (6%) 29 (11%) 219 (53%) 263 (100%)

Human H-specific mRNAs were amplified by PCR, cloned. and analyzed by hybridization to V,- and J,- region specific probes, respectively, as described in Green et a/.? Results of colony hybridizations showing V, and J, gene utilization. Results are given only for colonies hybridizing to V, and J, probes. A similar J, V utilization pattern was obtained by nucleotide sequences of unique human kappa chain cDNA clones.?

JAKOBOVITS ef af.: PRODUCTION OF AG-SPECIFIC HUMAN ABS 533

bone marrow, spleen, and blood at levels corresponding to 58%, 55%, and 30% of the B220t population in wild-type mice, respectively. Virtually all mature B cells detected expressed h p on their surface, with no detectable mouse p, with the majority of the B220+/hpf cells also expressing ha, indicating proper expres- sion and regulation of the human constant region in these mice.

The orderly differentiation of bone marrow-derived B cells in yH1 ;AJH/AJH mice was evaluated by three-color flow cytometry using antibodies to the cell surface markers CD43, HSA, and BP-1, which define early B-cell subpopulatioqs.'O Such analysis revealed that proper B-cell development was substantially restored by yH1. The maturation of pro-B cell to pre-B cell, however, a point at which V-DJ rearrangement is initiated, was found to be less efficient than in normal mice.* These findings suggested that lower frequency of rearrangement, which may result from the limited number of V genes in yH1, can lead to the detected incomplete B-cell development.

ELISA analysis of serum from yH1 ;AJ,/AJH mice demonstrated the production of h p antibodies at a concentration of 350 pg/ml, a level 510-fold greater than that produced by yH1-transgenic mice, and only several-fold below normal human IgM serum levels.

These results demonstrate that the human heavy chain YAC can reconstitute normal B-cell development in AJH-inactivated mice and induce the proper expres- sion and assembly of human h p chains with mouse light chains to form func- tional antibodies.

We further examined the ability of the heavy chain YAC, in conjunction with the human kappa chain YAC, to restore normal B-cell development and antibody production in the absence of functional mouse heavy and kappa light antibody genes. HuAb strains were bred with mice that are homozygous for inactivated heavy (AJHIAJH) and kappa (AC,/AC,, derived by deletion of the mouse CK region) chain alleles, in which the expression of mouse heavy and kappa chain expression is completely blocked. The generated mouse strain, Xenomouse, contained the human heavy and kappa YACs and lacked functional mouse heavy and kappa chains. Analysis of peripheral blood lymphocytes and splenocytes from Xeno- mouse indicated the presence of mature B220t cells, 45% of which are coexpressing exclusively h p and hK chains, and 55% expressing the hp and mh chains. No coexpression of mh and hK was detected, indicating that each of the light chains can completely exclude the expression of the other. The majority (75%) of the antibodies detected in Xenomouse serum were fully human antibodies (350 pg/ml) with a minority of h p h h antibodies. The level of B cells expressing only fully human antibodies, and the level of serum hp/hK antibodies in Xenomouse are several hundred-fold higher than those detected in HuAb strains, confirming that inactivation of the mouse Ig genes greater increased the level of fully human antibodies and suggesting Xenomouse as a valuable source for the production of human monoclonal antibodies.

SUMMARY

Our paper describes the introduction of large fragments of both the human heavy and light chain Ig genes into the mouse germline to create a mouse strain capable of producing a broad repertoire of antigen-specific, fully human antibodies.

The human immunoglobulin gene sequences were functional in the context of the mouse machinery for antibody recombination and expression, either in the

534 ANNALS NEW YORK ACADEMY OF SCIENCES

presence or absence of functional endogenous genes. This was demonstrated by their ability to undergo diverse rearrangement, to be expressed at significant levels, and to exclude expression of mouse immunoglobulins irrespective of their copy number or site of integration. The decrease in susceptibility to influence by adjacent genomic sequences may reflect the greater size, variable gene content, or structural integrity of the human Ig YACs and/or the presence of unidentified but important regulatory elements needed for optimal expression of the human immunoglobulin genes and their correct regulation.

Our results show that mouse B cells coexpressing human heavy and kappa chains, upon immunization, can produce antigen-specific, fully human antibodies. Furthermore, the human heavy and kappa chain YACs induced differentiation and maturation of the growth-arrested B-cell lineage in mice with inactivated endogenous Ig genes, leading to the production of a diverse repertoire of fully human antibodies at levels approaching those in normal serum. These results suggest the potential value of these mice as a source of fully human antibodies for human therapy. Furthermore, it is expected that such mice would lack immuno- logical tolerance to and thus readily yield antibodies to human proteins, which may constitute an important class of targets for monoclonal antibody therapy.

Our findings suggest that the introduction of even larger portions of the human heavy and light chain loci, which should be achievable with the ES cell-yeast spheroplast fusion technology described, will result in strains of mice ultimately capable of recapitulating the full antibody repertoire characteristic of the human humoral response to infection and immunization. The present and future mouse strains may prove to be valuable tools for studying the molecular mechanisms and regulatory sequences influencing the programmed assembly and expression of human antibodies in the normal immune response, as well as the abnormal response characteristic of autoimmune disease and other disorders.

The strategy we have described for the introduction of large segments of the human genome into mice in conjunction with the inactivation of the corresponding mouse loci may also have broad applicability to the investigation of other complex or uncharacterized loci.

ACKNOWLEDGMENTS

We thank Eva Ramirez and Veronica Lubeck for figure preparations.

REFERENCES

1. JAKOBOVITS, A., A. L. MOORE, L. L. GREEN, G. J . VERGARA, C. E. MAYNARD- CURRIE, H. A. AUSTIN & S. KLAPHOLZ. 1993. Nature 362: 255-258.

2. GREEN, L. L., M. H. HARDY, C. E. MAYNARD-CURRIE, H. TSUDA, D. M. LOUIE, M. J. MENDEZ, H. ABDERRAHIM, M. NOGUCHI, D. H. SMITH, Y. ZENG, N. E. DAVID, H. SASAI. D. GARZA, D. G. BRENNER, J . F. HALES, R. P. MCGUINNESS, D. J. CAPON, S. KLAPHOLZ & A. JAKOBOVITS. 1994. Nature Genetics 17: 13-21,

3. JAKOBOVITS, A., G. J. VERGARA, J . L. KENNEDY, J. F. HALES, R. P. MCGUINNESS. D. E. CASENTINI-BOROCZ, D. G. BRENNER & G. R. OTTEN. 1993. Proc. Natl. Acad. Sci. USA 90: 2551-2555.

4. MENDEZ, M. J., H. ABDERRAHIM, M. NOGUCHI, N. E. DAVID, M. H. HARDY, L. L. GREEN, H. TSUDA, S. YOAST, C. E. MAYNARD-CURRIE, D. GARZA, R. GEMMIL, A. JAKOBOVITS & S. KLAPHOLZ. 1995. Genomics 26 294-307.

JAKOBOVITS et al.: PRODUCTION OF AG-SPECIFIC HUMAN ABS 535

5 . REID, L., R. G . GREGG, 0. SMITHIES & B. KOLLER. 1990. Proc. Natl. Acad. Sci.

6. YAMADA, M., R. WASSERMAN, B. A. REICHARD, S . SHANE, A. J. CATON & G. ROVERA.

7. MARKS, J. D., M. TRISTEM, A. KARPAS & G . WINTER. 1991. Eur. J . Immunol. 21:

8. SCHROEDER,H. W.,JR. & J . Y . WANG. 1990. Proc.Nat1. Acad. Sci. USA87:6146-6150. 9 . CUISINIER, A.-M., V. GUIGUO, L. BOUBLI, M. FOUGEREAU & C. TONNELLE. 1989.

Scand. J . Immunol. 30: 493-497. 10. HARDY, R. R. , C. E. CONDIE, S . A. SHINTON, J. D. KEMP & K . HAYAKAWA. 1991.

Resolution and characterization of pro-B and pre-pro-B cell stages in normal bone marrow. J. Exp. Med. 173: 1213-1225.

USA 87: 4299-4303.

1991. J . Exp. Med. 173: 395-407.

985-991.