Antibodies generated from human immunoglobulin ... - BioMedSearch

1 downloads 0 Views 934KB Size Report
Campaign (MB) and Howard Hughes Medical Institute (MSN). ... Taylor, L.D., Carmack, C.E., Schramm, S.R., Masayekh, R., Higgins,. K.M., Kuo, C., Woodhouse, ...
1994

Oxford University Press

Nucleic Acids Research, 1994, Vol. 22, No. 8 1389-1393

Antibodies generated from human immunoglobulin miniloci in transgenic mice Simon D.Wagner, Gareth T.Williams, Tammy Larson, Michael S.Neuberger*, Daisuke Kitamura1 +, Klaus Rajewsky1, Jian Xian2 and Marianne Bruggemann2 Medical Research Council Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, UK, 'Institute for Genetics, University of Cologne, Weyertal 121, D-5000 Cologne 41, Germany and 2AFRC Babraham Institute, Babraham, Cambridge CB2 4AT, UK Received January 31, 1994; Revised and Accepted March 16, 1994

ABSTRACT One approach to the production of human monoclonal antibodies focusses on the creation of transgenic mice bearing human immunoglobulin gene miniloci. Whilst such loci undergo lymphoid-specific gene rearrangement, only a small proportion of mouse B cells express the human immunoglobulin chains; the miniloci thus contribute poorly to serum immunoglobulin. Attributing this poor performance to competition between the transgenic and endogenous immunoglobulin loci, we crossed mice bearing a human immunoglobulin heavychain (HulgH) minilocus with animals that had been rendered B cell-deficient by disruption of their endogenous heavy-chain locus. The results were dramatic: the human minilocus rescued B cell differentiation such that effectively all B cells now expressed human /t chains. The concentration of antibody in the mouse serum recognised by anti-human I increased to a concentration about one sixth that in human serum. The HulgH antibodies are heterogenous with diversity being generated by both combinatorial and junctional processes. Following antigen challenge, specific antibody is elicited but at low titre.

INTRODUCTION Since their discovery (1) in 1975, monoclonal antibodies have found a variety of uses for both the diagnosis and treatment of human disease. However, a major factor limiting administration of the existing mouse monoclonal antibodies to humans is the development of an anti-antibody response in the recipient. In an attempt to overcome this immunogenicity, chimaeric antibodies have been constructed in which either the variable (V) domains (2-4) or hypervariable regions (5,6) of rodent monoclonal antibodies replace the equivalent part of a human antibody. However, experience of using chimaeric antibodies in man (7) as well as tests in a mouse model system (8) show that such engineered antibodies still give rise to an anti-antibody response, albeit at much reduced levels. The production of whom correspondence should be addressed 'Present address: Medical Institute of Bioregulation, Kyushu University, 812 Japan

*To

completely human monoclonal antibodies might reduce this problem. Human monoclonal antibodies have been obtained through the

cloning and immortalization of human B cells (9), but this approach can only yield a very narrow range of specificities since it is not usually practicable to obtain appropriately immunised

human volunteers. Two other approaches have been considered for the production of a repertoire of human monoclonal antibodies: one exploits phage display technology (10) and the other the creation of transgenic animals carrying human immunoglobulin gene segments in germline configuration (11 - 17). For this latter approach, initial experiments were conducted using an artificially constructed plasmid (11, 16), although larger miniloci based on cosmids (12,14) and YACs (15,17) have subsequently been integrated into the mouse genome. Whilst such miniloci undergo lymphoid specific gene rearrangement, only a few percent of the B cells in the transgenic animals express human immunoglobulin chains. Consequently the miniloci contribute poorly to serum immunoglobulin, titres rarely exceeding 10 /tg/ml with values below 1 ,tg/ml being more the norm. Better expression of the transgenic loci is clearly desirable if the transgenic approach to human antibody production is to prove generally useful. We suspected that the low percentage of B cells expressing human immunoglobulin chains might be attributable to competition between the introduced minilocus and the endogenous immunoglobulin mouse locus and we show here that major improvements can indeed be achieved by crossing mice bearing a human immunoglobulin heavy chain (HuIgH) minilocus with animals (18) that have been rendered B cell deficient by a disruption of their endogenous heavy chain locus.

MATERIALS AND METHODS Mice The construction of the HuIgH minilocus and the production of MT transgenic mice has previously been described (11). The liMT mice harbour a disruption of one of the mouse Itm exons and have been described elsewhere (18). Mice bearing the HuIgH

1390 Nucleic Acids Research, 1994, Vol. 22, No. 8 locus (Figure la) derived from founder F17 were crossed with I MT animals. The HulgH LmT heterozygotes were crossed with each other to obtain an F2 generation containing individuals with the genotype 'HuIgH AMIT' DNA analysis Tail DNA was digested with BamHlI and EcoRI, Southern blotted and hybridised with a mouse ,um probe (18). This probe detects both the HuIgH minilocus and the endogenous wild type and 1tMT mouse alleles.

PCR and sequencing Rearranged VH26 genes were cloned into M 13 following amplification from genomic DNA by PCR using an oligonucleotide (12) priming in the leader-intron and another (HJ36; ref. 10) which primes back on human JHS but with preference for JH3 and JH6.

FACS analysis Splenic B cells were obtained by Ficoll density gradient centrifugation from 8 week old animals and were stained with biotinylated rabbit anti-human ,t (Jackson) followed by FITCstreptavidin or with FITC-conjugated goat anti-mouse it (Southern Biotech) and analysed by flow cytometry on a FACScan instrument (Becton Dickinson, Mountain View, CA). Spleen cells were also double stained (Figure 2) with antihuman IgM together with a phycoerythrin-conjugated antiCD45R antibody (RA3-6B2 [ref. 19]; Immunoselect). ELISA Sera from 7-9 week old mice were titred by enzyme linked immunosorbent assay (ELISA) on microtitre plates coated with rabbit anti-human IgM (Jackson) and developed with biotinylated rabbit anti-human IgM (Jackson) and horseradish peroxidase. The assay was calibrated by using pooled human sera (protein calibrant SPS-01) obtained from PRU Procurement, Department of Immunology, Sheffield.

Immunisations Pairs of 8 week old animals were immunised intra peritoneally with either alum-precipitated 2-phenyloxazolone coupled to chicken serum albumin (phO x CSA) (100 ytg) together with 109 heat-killed Bordetella pertussis or with recombinant hepatitis B

core antigen (20 ,ug) in complete Freund's adjuvant. The titres of human i-bearing antigen-specific antibodies in dilutions of prebleed and day 21 immune sera were monitored by ELISA on antigen-coated plates and developed with biotinylated rabbit anti-human IgM antiserum and horseradish peroxidase.

RESULTS

Crossing HulgH mice into a 1tI background Disruption of the mouse endogenous heavy chain has been achieved by targeting the CIA membrane exon (18). This disruption is called the JtMT mnutation. Homozygous AMT animals cannot make the membrane form of C,u which is essential for B cell development and thus lack B cells and circulating antibody ( 1 8). The H animals were crossed ciossed with /`MTT mice and the Hu~~H ~ iMT HuIgH ~- animals in the F generation crossed with each other. Analysis of tail DNA samples from the F2 generation by Southern blotting (Figure 2) allowed animals of different genotypes to be distinguished. Densitometry and breeding analysis reveals that the right hand of the two HulgH AtMT mice in Figure 2 is homozygous at the HuIgH locus.

Expression of HulgH antibody in serum HuIgH nMmice showed a very large increase in the titre of serum antibody expressed by the minilocus over that seen in either HuIgH +tT animals or HuIgH + animals (Table 1). Thus, whilst animals heterozygous for the ,uMT disruption showed only

about a fivefold increase in HuIgH antibodies over that seen in the wild type background, increases of over 100 fold were seen on breeding into the homozygous ,uMT background.

Expression of HuIgH antibody

on the B cell surface We surmised that this increase in expression of serum HuIgH antibody could be ascribed to the fact that the differentiation of B cells requires the expression of membrane immunoglobulin (18). Thus, in a tMT (but not a tlT) background, nearly all mature B cells would need to express human It chains on their surface. This is borne out by the pattern of staining of spleen cells with an anti-human IgM antiserum (Figure 3). Human ,. chains were expressed on the surface in over 80% of those HuIgH AMT spleen cells strongly positive for the B cell marker CD45R(B220) (Figure 4). The level of surface IgM expression

HulgH VH26 VH186 D J1-6 HuMo trUi A

II

I -L,.E

1 kb LJI

CHl En_ a

HulgHcos TVH VH6 III

MULI LI-,

-- .

10 kb

D4 D23/7 D1 VH26 D21j10 D3 D219 III

I

D052 J1-6 CpMp

II I I1I

Figure 1. Structure of the HuIgH and HuIgHCO)S loci. Transgenic animals carr' ing 3 -5 copies of the HulgH locus as well as anliiials carring o-integrated copies of the cosmids constituting the HuIgHCos locus have been described previously (1.2). Hunman sequences are depictecl by filled liins, mouse by open lines and s'ntbetic D elements are cross-hatched. The IgH intron enhancers are indicated by E.

Nucleic Acids Research, 1994, Vol. 22, No. 8 1391 appears similar to that obtained with cells in human peripheral blood (not shown).

Expression of mouse immunoglobulin The ,tMT disruption prevents expression of the membrane isoform of mouse It. Thus, as anticipated, B cells from the HuIgH MT mice show negligible fluorescence when incubated with WITC-conjugated anti-mouse ,t antiserum (Figure 3) and similarly decreased staining with anti-mouse 6 antiserum (not shown). The /tMT mouse IgH locus is, however, still capable of undergoing productive V-D-J joining (18). Such rearrangements would not normally lead to production of secretory mouse IgM in the serum of M-Tanimals since, in the absence of the membrane IgM isoform, B cell maturation does not occur. However, one would expect that the rescue of mouse B cell development caused by provision of the HuIgH locus in HuIgH I MTmice should allow the development of some B cells that not only express HuIgH immunoglobulin on their surface but which also happen

_

~

~

....~

to harbour a productive V - D - J integration on one one of the ,tMT alleles thereby leading to co-expression of the secretory (but not the membrane) isoform of mouse IgM. This is indeed the case; the titre of mouse IgM in the sera of HuIgH IMT animals whilst much less than that in wild type animals (averaging around 20 jxg/ml rather than 700 /Ag/ml) is higher than that in uncrossed zMT homozygotes (less than 0.1 ,ug/ml). Nevertheless, as expected, the secretory mouse u chains in HuIgH "M animals are not found on the B cell surface (Figure 3).

Diversity of V gene segments To assess the degree of structural diversity which could be achieved with the minilocus (11), we cloned rearranged VH26 genes from the spleen of HuIgH L- mice. Thirteen out of eighteen of these rearrangements were non-productive in that they yielded frameshifts or stop codons. This can be understood in terms of the fact that the HuIgH minilocus is present in 3-5 copies; a B cell would only need (or, indeed, possibly wish) to express one functional HuIgH rearrangement. Analysis of the potentially functional rearrangements (Figure 5) reveals that diversity is achieved by the use of multiple JH segments as well as at the V - D - J border and confirms the expression of heavy chains composed of solely human V, D and J segments. Recognisable D sequences are underlined (human DQ52 in clone 17, DSP2 and DFL16 in clones 2 and 18, Figure 5). The non-functional rearrangements used JH2 (three times), JH3 (twice); JH4 (five times); JHS (once); JH6 (twice) with (six

...... ~~~~~~~~~

HuigH [MT

HulgH +

WT 30

~~~~~~~~~~~~~Anti-

%_

n E

Mouse IgM

j

a)

7 T7~ ~~~ ~~~~~ ~Anti.1 Human IgM

C)30 -7

zL Figure 2. Southern blot analyses of HuIgH x,uMT crosses. Bands derived from the HulgH locus are indicated by '0'; from the mouse /iMT allele by 'it' and from wild type mouse it allele by '+'. The deduced genotype is indicated above each lane (H=HuIgH; jt=jtMT).

Table

I.

Endogenous mouse IgH locus

-/t

102

104

Fluorescence Intensity

104

HuIgM serum titre (Ag/ml)