the role of in vitro expression systems in the investigation of antibodies to dna

C L I N I C A L S C I E N C E

The Role of In Vitro Expression Systems in the Investigation of Antibodies to DNA

Anisur Rahman, David S. Latchman, and David A. Isenberg

Objectives:Antibodies to DNA are believed to be important in the development of tissue inflammation and clinical activity in systemic lupus erythematosus (SLE). Sequence analysis of monoclonal murine and human anti-DNA antibodies suggests that somatic mutations and basic residues are important features at the DNA-binding site. To test this hypothesis, it is possible to alter these residues by site-directed mutagenesis of cloned variable region cDNA. The mutagenized cDNA sequence is then expressed in the form of a protein molecule whose properties can be tested in assays of binding or pathogenicity. The purpose of this article is to provide a systematic review of the evidence derived by such methods in the study of anti-DNA antibodies. Methods: Various different expression systems are available. Experiments using bacterial and eukaryotic expression systems are considered in turn. The advantages and disadvantages of the systems are described and the results obtained are compared. Results and Conclusions: High yields of antibody fragments such as scFv and Fab can be achieved by expression in bacteria. Such studies tend to confirm that reversion of somatic mutations or removal of basic residues at the antigen binding site reduce affinity for DNA. Tests of pathogenicity can only be performed by expressing whole antibodies in eukaryotic cells. The limited data available from expression of mutagenized cDNA in such systems argue against a simple relationship between changes in DNA binding affinity and changes in pathogenic potential. Further studies are therefore required to analyze the sequence requirements for pathogenicity. Semin Arthritis Rheum 28:130-139. Copyright © 1998 by W.B. Saunders Com- pany

INDEX WORDS: Anti-DNA antibody; sequence; expression system,

I N PATIENTS WITH systemic lupus erythematosus (SLE), inflammation can occur in any of a

wide variety of tissues or organs. SLE nephritis is

From the Bloomsbury Rheumatology Unit/Division of Rheumatol- ogy, Department of Medicine, University College, London; and the Department of Molecular Pathology, University College, London.

Anisur Rahman, PhD, MRCP: Wellcome Trust Clinical Research Training Fellow; David S. Latchman, PhD, DSc: Professor of Molecular Pathology; David A. Isenberg, MD, FRCP: Arthritis Research Campaign Diamond Jubilee Profes- sor of Rheumatology.

Dr. Rahman is supported by Wellcome Trust Research Fellow- ship no. 040 366/Z/94/Z

Address reprint requests to Anisur Rahman, PhD, MRCP,, Bloomsbury Rheumatology Unit, Arthur Stanley House, 40-50 Tottenham St, London WIP 9PG, UK.

Copyright © 1998 by W.B. Saunders Company 0049-0172/98/2802-000858.00/0

particularly serious because it may lead to renal failure or to the need for treatment with potent immunosuppressive agents. Although many different autoantibodies have been reported in the serum of patients with this disease (1), levels of antibodies to double-stranded DNA (anti-dsDNA) are related most closely to fluctuation in disease activity (2,3). This is particularly true in the presence of nephritis (4). Not all anti-dsDNA antibodies, however, are equally likely to be associated with tissue inflammation and active disease. Some patients with high serum anti-dsDNA antibody levels are clinically inactive (3) whereas, in other cases, disease activity appears to be linked to the presence of a subset of anti-dsDNA antibodies defined by isotype and/or charge. Okamura et al (5) showed that, in 40 untreated Japanese patients with lupus nephritis, the degree of histological damage seen in renal

130 Seminars in Arthritis and Rheumatism, Vo128, No 2 (October), 1998: pp 130-139

IN VITRO EXPRESSION OF ANTIBODIES TO DNA 131

biopsy specimens was closely correlated with the level of immunoglobulin (Ig) G anti-dsDNA but not IgM anti-dsDNA in the patient's serum. Suzuki et al (6) found that patients with active lupus nephritis were more likely to have high titers of positively charged anti-dsDNA antibodies in their serum than those with inactive disease.

More direct evidence that anti-DNA antibodies are implicated in causing glomerular damage in SLE comes from reports that these antibodies are often found in renal biopsies from patients with lupus nephritis but not from those with other types of nephritis (7, 8). Some murine monoclonal anti- DNA antibodies have been shown to deposit in mouse (9) or rat (10) glomeruli with associated proteinuria. More recently, some human monoclonal IgG anti-dsDNA antibodies have been introduced into severe combined immunodeficiency (SCID) mice with resulting glomerular deposition and proteinuria (11, 12) suggesting that such antibodies may be capable of playing a direct pathogenic role in lupus nephritis.

Not all the monoclonal anti-DNA antibodies tested in the mouse and rat models previously mentioned could form glomerular deposits (9-12). This suggests that the ability to bind DNA is not in itself sufficient to render an antibody molecule capable of causing tissue damage in SLE. This is consistent with the finding that some anti-DNA antibodies are more likely to be associated with clinical activity than others. It is important to investigate the particular structural features that differentiate these pathogenic anti-DNA antibodies from other Ig molecules. By building up a database of sequence information from human and murine monoclonal anti-DNA antibodies, it has been possible to deduce that certain sequence features appear to occur commonly in antibodies of this specificity. These features are found particularly in the high affinity IgG antibodies, binding specifi- cally to dsDNA, which are most closely related to disease activity.

THE IMPORTANCE OF SOMATIC MUTATIONS

AND BASIC RESIDUES IN HIGH AFFINITY

ANTI-dsDNA ANTIBODIES

The basic unit of an Ig molecule consists of four peptide chains that are linked by disulphide bonds. There are two identical light chains and two identical heavy chains (Fig t). Each chain is divided into variable (V) and constant (C) domains.

VL

VH

CHI

CL ~I

CH2

CH3

VH

CHI

CL

CH2

i cH3

VL

Papain

I Fc

4-

/ / NN

2X Fab Fig 1. Basic structure of an antibody molecule. VL, variable domain of light chain; VH, variabme domain of heavy chain; CL, constant domain of light chain; CH, constant domain of heavy chain.

132 RAHMAN, LATCHMAN, AND ISENBERG

The V regions at the amino terminal end show marked variation in sequence between one antibody and another and are responsible for forming the antigen binding site. The C regions have practically the same sequence in all antibodies of a given isotype and mediate the effector functions of the molecule, such as complement fixation or attachment to Fc receptors. Digestion with papain separates the molecule into two Fab fragments, capable of binding antigen, and one Fc fragment, which cannot bind antigen since it does not contain the V regions (Fig 1).

Within the V regions, variability of sequence is not uniform. Three areas of very high variability, the complementarity-determining regions (CDRs), are separated by areas of more conserved sequence known as framework regions (FRs). Crystallo- graphic and computer modelling studies of the three-dimensional structure of antibodies show consistently that the CDRs form the majority of the antigen binding site in most Igs (13).

Because Ig binding properties are dependent only on the V regions, the ability of the immune system to produce antibodies binding to many different antigens arises from its capacity to produce a vast repertoire of different V-region sequences. The diversity of sequence in the CDRs is particularly great. The sources of this diversity have been reviewed extensively elsewhere (14, 15). One of the most important of these is the presence of a hypermutation mechanism in B cells (16). This mechanism leads to a high frequency of somatic mutations in the V region, but not the C region sequences of Ig molecules. Somatic mutations are those that are not inherited through the germline, but develop only within a particular expanded clone of cells. In both mice and humans, IgG molecules tend to carry more somatic mutations than IgM (17, 18). This is often associated with increased affinity for antigen.

It is believed that there is a direct relationship between the accumulation of somatic mutations and the development of high antigen binding affinity. This theory arose from a consideration of the pattern of mutations in monoclonal antibodies derived from immunized (17) or autoimmune (19) mice. In high affinity IgG antibodies there is a concentration of replacement mutations (which alter amino acid sequence) in the CDRs but not the FRs. Silent mutations, which do not alter amino acid sequence, are not concentrated in this way.

This pattern arises because, in the presence of the cognate antigen, antibody secreting cells are stimulated to divide. As the clone of cells expands, somatic mutations occur in some of the daughter cells. Cells in which the mutations act to increase affinity for antigen will be stimulated more strongly by surface antibody-antigen interactions and will divide faster. After several generations, the population of cells derived from the original clone will therefore be dominated by the descendants of cells in which such preferred mutations occurred. These mutations must be replacements (because they alter phenotypic properties) and are likely to be in the CDRs (because they affect the antigen binding site). Thus, the process of antigen-driven clonal expansion leads to the production of high affinity antibodies with accumulated replacement mutations in CDRs but not FRs.

Consideration of the sequences of murine and human monoclonal anti-DNA antibodies reveals that those of high affinity (especially IgG anti- dsDNA antibodies) do indeed show evidence of somatic mutations concentrated in the CDRs (20, 21). In many cases, these mutations lead to an increase in the number of basic residues such as arginine (R), asparagine (N), and lysine (K) in the CDRs. Thus it has been postulated that DNA (or a complex of DNA and protein) drives clonal expansion of antibody-producing cells in such a way as to select those which have many basic residues in the CDRs. This idea is supported by the finding that basic residues in VHCDR3, produced by processes other than mutation (eg, junctional diversity and addition of nontemplated nucleotides) are also found commonly in anti-DNA antibodies. Indeed, R, N, and K residues are found so commonly in VHCDR3 that this region is felt to be of particular importance in conferring the ability to bind DNA (20, 21). Basic residues can enhance binding to the negatively charged DNA molecule by charge interactions, formation of hydrogen bonds, or both (22).

Given that both somatically mutated and basic amino acids appear to be important at the DNA- binding site, a number of groups have sought to identify the positions of these amino acids in three-dimensional images of complexes between monoclonal anti-DNA antibodies and either ssDNA or dsDNA. Using computer modelling programs, Eilat et al (23) and Barry et al (24) showed that murine monoclonal anti-ssDNA antibodies bound ssDNA in shallow clefts and were able to identify


particular basic residues in the walls of these clefts. Kalsi et al (25) used similar methods to show that the human monoclonal antibody (mAb) B3 bound dsDNA in a cleft and that the interaction was stabilized by contacts between the double helix and three R residues from VLCDR1, VLCDR2, and VHCDR2. Two of these arginines arose by somatic mutation. Crystallographic studies of a complex between a trinucleotide and a murine monoclonal anti-ssDNA antibody BV-0401 showed few contacts with basic residues but important stacking interactions between the planar rings of tyrosine and tryptophan residues and those of thymine molecules in the trinucleotide (26).

The evidence from sequence and structural analy- ses previously outlined enables one to predict which residues in the sequence of a particular monoclonal antibody are likely to be essential in confelTing the ability to bind DNA with high affinity. To test such predictions it is then necessary to produce altered forms of the antibody in which only those supposedly crucial amino acids have been changed. These altered forms would be ex- pected to have markedly altered affinity for DNA and this might also affect their pathogenicity in biological assays. It is important to undertake such experiments not only to test the validity of the hypotheses of DNA-antibody interaction previously described, but to produce a clearer picture of the DNA-binding site itself. Such a picture might lead to the development of drugs capable of blocking this interaction in high affinity pathogenic antibodies and thereby ameliorating the disease process.

PRINCIPLES OF ANTIBODY EXPRESSION SYSTEMS

Antibody expression systems convert antibody cDNA, cloned into suitable expression vectors, into protein molecules that can bind antigen. In comparison to methods such as hybridoma production or Epstein Barr virus (EBV) transformation of B cells, the in vitro expression of cloned antibody cDNA is an inefficient way of producing Ig molecules. The great advantage of expression systems, however, is that the sequence of the DNA being expressed can be controlled. This is precisely the requirement for the production of variants of monoclonal anti-DNA antibodies into which small alterations of sequence have been introduced by site-directed mutagenesis.

The major difficulty associated with expression

of antibody molecules from cloned cDNA is that both the heavy and light chain variable regions must be expressed in the same cell. This may entail cloning both VH and VL into the same expression vector or cotransfecting cells with two expression vectors, one carrying VH and the other VL. In either case, expression of a functional molecule requires appropriate folding of both chains so that they can associate with each other to form an antigen binding site.

Expression vectors can be transfected into cells of various types. These include bacterial cells (27-31), yeasts (32), insect cells (33), and mammalian cells (34-39). Of these, only bacterial and mammalian cells have thus far been used to express antibodies with DNA-binding specificity.

The major drawbacks of bacterial expression of eukaryotic proteins are that post-translational modi- fications such as glycosylation are absent, and that the molecules may not be folded properly'. This applies particularly to large molecules like Ig heavy chains, and it has not been possibIe to express a stable, functional whole antibody molecule in bacteria (33). Because binding properties of an anti~ body depend only on the variable regions, however, bacterial expression can be used to produce smaller fragments of antibodies that can still bind antigen. These fragments are Fab (see Fig 1) or single chain fused variable regions (scFv). The advantages of bacterial systems are that it is possible to produce large amounts of antibody fragments by bulk culture of the transfected bacteria and that, by using libraries of expression vectors, large numbers of combinations of Va and VL sequences can be rapidly screened for the ability to bind a particular antigen.

BACTERIAL EXPRESSION OF ANTI-DNA ANTIBODY cDNA AS scFv

ScEv consist of the VH and VL domains joined by a small, flexible, linker segment that is typically designed to be rich in glycine and serine residues. This flexibility allows the VH and VL domains to adopt a range of different conformations relative to each other and thus to create an antigen binding site. ScFv are produced by cloning the VH and VL cDNA sequences into a vector in which they are separated by an oligonucleotide encoding the linker. No C region cDNA is included in the vector so the scFv (containing only variable regions) does not resemble any molecule found in nature. Neverthe-


less, expressed scFv can be purified from bacterial supernatants in large quantities and have been shown, in some cases, to have binding properties similar to those of the monoclonal antibodies from which they were derived (27).

Brigido et al (27) used expression of scFv to study the properties of the murine monoclonal anti-Z-DNA antibody Z-22. Z-DNA is an alternative conformational form of dsDNA in which the double helix is left handed, whereas the more common B-DNA is a right-handed helix. ScFv containing both Vn and Vc of Z-22 bound Z-DNA but not B-DNA or ssDNA. These are the same binding properties shown by whole Z-22 molecules. By cloning a mixture of total mouse VK cDNA into an scFv expression vector containing Z-22 VH, it was possible to produce a plasmid library in which this VH was paired with a wide variety of different VL. Plasmids from the library were used to transform bacteria that were then screened for production of anti-Z-DNA scFv. Of several hundred colonies screened, only one produced any Z-DNA binding activity. This contained a VL region very similar to that of Z-22 itself. Conversely, only 2 of 11 VH sequences tested could combine with Z-22 VL to give anti-Z-DNA. These experiments suggested that the sequences of both Vn and V L were important contributors to the binding properties of this antibody.

In a subsequent experiment, Z-22 was compared with the murine mAb Z3-3, which shares the same light chain sequence but does not bind Z-DNA (28). When VH CDR3 and VH FR 4 of Z3-3 were replaced with those of Z-22, the resulting scFv could bind Z-DNA, confirming that VHCDR3 is a major contributor to this affinity in Z-22. Z-22 scFv molecules in which particular residues in CDR3 of either VL or Vn had been altered by mutation showed reduced affinity for Z-DNA. These residues included an N in Vn CDR3 which may be involved in hydrogen bond formation.

BACTERIAL EXPRESSION OF ANTI-DNA ANTIBODY cDNA AS Fab

An alternative to the use of plasmid vectors for bacterial expression is to clone VH and VL into a phagemid. Phagemids are plasmid-based vectors that can replicate as ssDNA and be packaged into bacteriophage particles. The advantage of this is that the recombinant antibody fragment can be produced as a fusion protein, connected covalently

to a surface protein of the phage particle. This is known as phage display. Phage display vectors can contain VH and VL linked in the form of an scFv. Alternatively, they can contain Vu and VL in separate cassettes followed by a Cnl sequence (ie, cDNA encoding the first heavy chain constant domain) and a CL sequence, respectively. Two different foreign peptides will be produced by the host phage in this case. One will be the whole light chain of the antibody whereas the other will comprise half of its heavy chain. One of these foreign peptides (usually the light chain) is connected to the phage surface protein. The other will be secreted into the bacterial periplasm and will interact with the surface bound component to form Fab.

The bacterial-growth medium containing the free-phage particles can be exposed to a surface coated with the antigen of interest; phage particles that bind to this surface can be selected for a further round of transfection and selection. By repeating this process several times it is possible to obtain a population of phage particles highly enriched for the presence of Fab, which binds this antigen with high affinity. The phage display system can therefore be used to screen a large number of VH/VL combinations to discover which combinations con- fer high affinity binding. This is known as repertoire cloning (30).

Barbas et al (29) used this approach to clone unselected human V~ and Vc genes in a phage display library from which antibodies were selected for binding to human placental DNA. Six high affinity anti-dsDNA Fab were isolated from V genes derived from lymphocyte IgG cDNA of a patient with SLE, and one from bone marrow IgG cDNA of a healthy donor. Some of the antibodies derived from the patient with SLE showed evidence of hypermutation concentrated in the CDRs, however, this was not associated with increased affinity for DNA. Indeed, the highest affinity antibodies had 98% to 99% homology to germline VH genes. There was, however, accumulation of basic residues in Vu CDR3 of these high affinity antibodies.

The main difficulty in interpreting these results is that there is no evidence to show that the high affinity VH/VL combinations isolated from the phage display library bear any relevance to antibodies produced in vivo. Because the cloned V genes were derived from cDNA of the donors, it is clear


that they must be expressed at the time of donation. However, this does not prove that the particular V~{ and VL genes paired in tile high-affinity phages were coexpressed in any B cell of the donor. This is in contrast to mAb produced by hybridomas or EBV-transformed B cells. Because these are derived ultimately from single lymphocytes, it fop lows that any VH/VL combination seen in such a mAb must occur in at least one cell in the donor.

One way to show that the combinations identi- fied by repertoire cloning have some clinical relevance is to identify idiotypes they possess. Idiotypes are structural features within the variable regions of antibodies that can be recognized by specific antisera. Roben et al (30) showed that a pt~age bound anti-dsDNA Fab molecule (desig- nated AD4-37) produced from lymphocyte IgG cDNA of a patient with SLY carried particular idiotypes on VH and VL. They then showed that the blood of the patient concerned contained anti-DNA antibodies that carried both these idiotypes (B6 and MIIa) simultaneously. This suggested the presence of a clone of B cells secreting antibodies in which both chains resembled those of the phage-displayed Fab. However, AD4-37 was polyreactive and contained very few mutations in either VH or VL. It may therefore not be representative of pathogeneti- caIly important anti-dsDNA antibodies found in SLY. A second high affinity anti-dsDNA antibody (AD4-18), derived from the same patient, was much more specific for DNA and contained many somatic mutations in both VH and VL. However, no idiotype studies were performed with this Fab. Both AD4-37 and AD4-18 contained several basic residues in VHCDR3.

Mockridge et al (31) used phage expression of Fab in a different way. Rather than cloning a whole repertoire, this group studied a single well- characterized human anti-DNA mAb, D5 (40, 41). The VH region of D5 is encoded by the germline gene V4_34. All human monoclonal cold agglutinins (which bind the Ii antigen on the surface of erythrocytes) are also encoded by this gene. How- ever, sequence analysis shows that V4_34-encoded anti-DNA antibodies (including D5) have concen- trations of basic residues in VHCDR3 whereas cold agglutinins do not (41). Mockridge et al (31) showed that D5 V~{ and VL could be expressed as phage-bound Fab, which would bind both ssDNA and dsDNA. When the D5 heavy chain was replaced by that of a cold agglutinin that had no

mutations in VH and no basic residues in CDR3, DNA-binding activity was lost completely. How- ever, if the VH CDR3 sequence of this cold agglutinin was then replaced by that of D5, DNA binding activity was restored. This showed that somatic mutations in the FRs, CDR1, and CDR2 of D5 VH were not essential for binding to DNA, but that the basic residues in V~ CDR3 were very important. Replacement of D5 V~ by the correspond- ing unmutated germline gene reduced affinity for DNA. Thus mutations in the light chain must also play a role in this antibody.

EXPRESSION OF ANTI-DNA ANTIBODY cDNA AS WHOLE ANTIBODIES

IN MAMMALIAN CELLS

Expression of antibody cDNA as whole antibodies in eukaryotic cells is essential if the effects of site-directed mutations on pathogenicity are to be investigated. Fab and scFv fragments, lacking the full complement of constant region domains, are biologically nonfunctional and cannot be used in such assays.

Cells from lower eukaryotes, such as yeasts or insects, have been used in expression of antibodies, and high yields can be achieved by these methods (32, 33). However, the glycosylation of antibody molecules by these cells is not as complex as in mammalian cells, and this may alter the functional capabilities of the antibodies produced (32). Where it is necessary to study the effects of changes in sequence on the activity of antibodies in biological assays (such as deposition in mouse or rat kidney) expression of whole, fully glycosylated antibodies in mammalian cells remains the ideal system. However, the overall yield of antibody is lower than in bacteria because it is difficult to grow mammalian cells in such large quantities.

Two types of expression of antibodies in mammalian cells have been described. Transient expression implies that the foreign genes inserted on the plasmid vector are active for a short period after which no more antibody is produced. Only small amounts of Ig are produced in this way, but selection of cells for drug resistance markers and long-term maintenance of cell lines are not required. Stable expression relies on the fact that a small minority of cells will continue to express genes on the plasmid for many generations, usually because of integration of the vector DNA into the host-cell chromosome. This minority popu][ation


can be selected and maintained by using markers such as drug resistance that are present in the plasmid but not in the host-cell chromosomes. Only cells that maintain the plasmid will survive in the presence of the appropriate drug. Stable expression allows the production of large quantities of antibody, but can be difficult to achieve.

Most of this work has been performed with murine antibodies and has been designed to investigate the importance of R and other positively charged residues, particularly in V~ CDR3. A number of groups have used heavy chain loss variants. These are hybridoma cells that have lost the ability to secrete heavy chain. They can be prepared by growing hybridoma cells in culture and selecting for inability to secrete antibody which binds rabbit anti-mouse heavy chain reagents (34, 37). The variants selected secrete only light chains. Stable expression of whole IgG can be achieved by transfecting these cells with vectors containing VH and Cn sequences under the control of an appropriate promoter.

Radic et al (34) used this method to study the Vu sequence of the murine anti-DNA mAb 3H9. Initially, they transfected 3H9VH into a number of different heavy chain loss variants and showed that it could produce anti-ssDNA antibodies in combination with a wide range of Vc sequences (34). This suggested that the heavy chain of 3H9 was the major contributor to DNA-binding specificity. 3H9VH contains three somatic mutations in CDR2. When the transfected 3H9VH sequence was modi- fied by mutagenesis, it was found that reversion of one of these somatic mutations, leading to loss of an R at position 53, dramatically reduced affinity for dsDNA (35). Reversion of the other mutations in CDR2 (not involving basic residues) had little or no effect on affinity for dsDNA. 3H9VH also contains a single R in CDR3 (arising from addition of nontemplated nucleotides at the junction between VH and DH genes). When this R was changed to glycine by mutagenesis, ability to bind DNA was lost completely. The crucial role of R residues was underlined by experiments in which extra Rs were introduced into the transfected 3H9VH sequence at sites where they were known to occur in CDRs of other murine anti-DNA mAb. In most cases, the addition of a single extra R increased the affinity for both ssDNA and dsDNA by between 5- and 10-fold. Addition of two Rs could increase affinity by up to 70-fold, but this did not occur in all cases.

It is important to stress, however, that there was no simple relationship between absolute number of CDR Rs and DNA-binding affinity. Rs in some positions may interact with other residues in such a way as to reduce this affinity. For example, the introduction of an R at position 64 of 3H9VH eliminated binding to DNA, probably because of the formation of a salt bridge with the neighbouring aspartic acid residue at position 65.

In a similar study, Katz et al (36) studied the VH sequence of the murine anti-dsDNA antibody R4A. This antibody is encoded by the unmutated murine germline gene VH11. A number of mutant forms of R4AVH were produced in all of which a number of charged residues were altered. These residues were in CDR2, FR3, or CDR3. The germline or mutated sequences were transfected into heavy chain loss variants secreting the light chain of R4A. It was found that DNA-binding affinity could be altered by a single amino acid change, even if this was in FR3. Although loss of an R in CDR3 was found to reduce affinity for DNA, there were other sites at which a loss of positive charge enhanced binding or a loss of negative charge reduced binding. In fact, the mutant with highest dsDNA binding affinity was identical to R4A in the CDRs but had lost two R residues from FR3. These results show that charge interactions alone cannot explain antibody- DNA binding.

When R4A hybridoma cells or cells expressing the mutant R4AVH were injected intraperitoneally into SCID mice, the mutations were found to alter pathogenicity as well as bind dsDNA. Injection of R4A cells led to proteinuria and glomerular deposition of antibody. A point mutation in CDR3 that abolished dsDNA binding also abolished renal sequestration and ability to induce proteinuria. Surprisingly, the highest affinity mutant previously described showed more deposition in the tubules than the glomeruli, although proteinuria still re- sulted. Thus mutations that enhance binding to dsDNA do not necessarily increase the potential for glomerular deposition and damage.

The murine anti-DNA antibody D42 and the antiphosphorylcholine antibody 6G6 are also derived from germline gene VH11. The two antibodies differ at two sites in VH CDR1 and VnCDR2 caused by replacement mutations in D42 (42). D42VH has three Rs in CDR3 whereas 6G6VH has none. Pewzner-Jung et al (37) transfected D42Vn, 6G6VH, and sequences intermediate between them into


heavy chain loss variants. They found that only transfection into loss variants secreting the D42 light chain produced anti-DNA antibodies and that this was only possible if the transfected DNA contained D42V~ CDR3. The presence of the two replacement mutations in CDR1 and CDR2 was not essential, but increased affinity for DNA. It is interesting to note that these findings, suggesting tha~ both heavy and light chains are important but that VH CDR3 is dominant, are similar to those of Mockridge et al (31) using human antibodies in a completely different expression system.

Heavy chain loss variants have not been used to express whole human anti-DNA mAb. It has been very difficult to produce and maintain human hybridoma cells, particularly those secreting IgG (40). Because loss variants form a very small proportion of the total hybridoma clone, it may prove even more difficult to produce such lines for expression studies. In addition, this method is not well suited to studying the effects of changes in the light chain. Some information can be deduced by transfecting the same heavy chain into variants secreting different light chains (34, 37), but this does not allow for the design of point mutations in the light chain itself. Studies of the light chain may be more important in human anti-DNA mAb that (unlike murine ones) show no strong preference for use of particular VH genes and in which somatic mutations in VL may- play a major role (43-45).

An alternative to the use of heavy chain loss variants is to transfect both heavy and light chain vectors into a cell type that produces no Ig. Zack et al (38) achieved transient expression of whole routine IgG anti-DNA antibodies in COS cells by ligating the entire V-C coding regions of the heavy and light chains of the mAb 3El0 into a mammalian expression vector. The yield of IgG was low (30 to 50 ng/mL) but the antibody produced bound ssDNA and dsDNA. Mutations in all CDRs of VH were found to alter affinity for DNA. In particular, loss of an R in CDR3 reduced affinity whereas gain of an N in CDR1 increased it.

Recently, we have produced whole human IgG molecules by transient expression of cDNA from two human monoclonal anti-DNA antibodies, B3 and WRI176, in COS cells (Rahman et al, submit- ted). The yield of antibody was low (5 to 50 ng/mL) but it was possible to show binding to DNA in a direct ELISA. In this expression system, the Vn and VL cDNA sequences are cloned into separate

plasmid vectors, which contain CH and C~ cDNA sequences, respectively. This makes it easy to express the same heavy chain in combination with different light chains (or vice-versa) in different experiments. It was found that DNA-binding antibodies could only be produced when the heavy chain of either B3 or WRI176 was coexpressed with the light chain of the same antibody. The heterologous combinations B3VH/WRI176VL and WRI176VH/B3VL produced similar quantities of whole IgG, but this IgG did not bind DNA. These results suggested that the sequences of both VH and VL were important in creating the DNA-binding sites of these antibodies, which was consistent with computer models of their structures (25).

The main criticism of these transient expression systems is their inability to produce sufficient whole antibody for use in biological assays. Stable expression in cell types other than heavy chain loss variants, however, entails selecting those rare trans- fectants that have stably incorporated both the heavy chain and light chain expression plasmids into their genomic DNA. Irigoyen et al (39) have shown that this is possible by simultaneous transfection of human VL and mouse Vn expression vectors into a nonproductive mouse myeloma line to give stable expression of chimeric antibodies can'ying the human lambda chain idiotype K12. The purpose of this experiment was to investigate the effect of various mutations in Vx on expression of the idiotype. By achieving stable expression, this group was able to achieve yields of antibody of up to 200 ng/mL and to show that point mutations could alter the binding of anti-idiotype.

CONCLUSION

After use of sequence and structural data to construct hypotheses concerning the nature of amino acids important in antibody-DNA interactions, it is now necessary to test these hypotheses. The expression of cloned mutagenized autoantibody cDNA sequences provides a powerful method of doing this. Bacterial expression of scFv and Fab has shown the importance of both VH and VL in contributing amino acids to the DNA-binding sites of several mAb.

The important roles played by somatically mutated and/or basic residues at the DNA-binding site are demonstrable by consideration of the properties of expressed whole antibodies or fragments in which these residues have been altered by muta-


tion. In most (but not all) cases, such alterations result in reduced affinity for DNA. The limited data on the behavior of these altered antibodies in assays of pathogenicity, however, they do not suggest that there is a simple relationship between affinity for DNA and ability to damage the kidney. Other properties, such as the ability to cross-react with protein constituents of the glomerular cell mem- branes, may also be important. Future experiments are likely to focus on the effects of sequence changes on pathogenicity, and for this purpose large amounts of whole antibodies expressed by

stably transfected mammalian cells will be necessary. Bacterial expression systems will still play a major role, particularly in the production of large quantities of mutant scFv or Fab for use in crystallographic or nuclear magnetic resonance studies designed to investigate effects on structure rather than directly on function.

ACKNOWLEDGMENT

Dr. Rahman is supported by Wellcome Trust Research Fellowship no. 040 366/Z/94/Z.

REFERENCES

1. Rahman MAA, Isenberg DA. Autoantibodies in systemic lupus erythematosus. Curt Opin Rheumatol 1994;6:468-73.

2. Koffler D, Carr R, Agnello V, Thoburn R, Kunkel HG. Antibodies to polynucleotides in human sera: Antigenic specificity and relation to disease. J Exp Med 1971;134:294-312.

3. Swaak AJG, Aarden LA, Statius van Eps LW, Feltkamp TEW. Anti-dsDNA and complement profiles as prognostic guides in systemic lupus erythematosus. Arthritis Rheum 1979; 22:226-35.

4. Ter Borg EJ, Horst G, Hummel EJ, Limburg PL, Kallen- berg CGM. Measurement of increases in anti-double-stranded DNA antibody levels as a predictor of disease exacerbation in systemic lupus erythematosus. Arthritis Rheum 1990;33:634- 43.

5. Okamura M, Kanayama Y, Amastu K, et al. Significance of enzyme linked immunosorbent assay (ELISA) for antibodies to double stranded and single stranded DNA in patients with lupus nephritis: Correlation with severity of renal histology. Ann Rheum Dis 1993;52:14-20.

6. Suzuki N, Harada T, Mizushima T, Sakane T. Possible pathogenic role of cationic anti-DNA antoantibodies in the development of nephritis in patients with systemic lupus erythematosus. J Immunol 1993;151:1128-34.

7. Koffler D, Schur PH, Kunkel HG. Immunological studies concerning the nephritis of systemic lupus erythematosus. J Exp Meal 1967;126:607-24.

8. Winfield JB, Faiferman I, Koffler D. Avidity of anti-DNA antibodies in serum and IgG glomerular eluates from patients with systemic lupus erythematosus. Association of high avidity anti-native DNA antibody with glomerulonephritis. J Clin Invest 1977;59:90-6.

9. Madaio ME Carlson J, Cataldo J, Ucci A, Migliorini R Pankewycz O. Murine monoclonal anti-DNA antibodies bind directly to glomerular antigens and form immune deposits. J Immnnol 1987;138:2883-9.

10. Raz E, Brezis M, Rosenmann E, Eilat D. Anti-DNA antibodies bind directly to renal antigens and induce kidney dysfunction in the isolated perfused rat kidney. J Immunol 1989;142:3076-82.

11. Ehrenstein MR, Katz DR, Griffiths MH, et al. Human IgG anti-DNA antibodies deposit in kidneys and induce protein- oria in SCID mice. Kidney Int 1995;48:705-11.

12. Ravirajan CT, Rahman MAA, Papadaki L, et al. Genetic,

structural and functional properties of a human IgG anti-DNA autoantibody. Eur J Immunol 1998;28:339-50.

13. Amzel LM, Poljak RJ, Saul F, Varga JM, Richards FF. The three-dimensional structure of a combining region-ligand complex of immunoglobulin New at 3.4A resolution. Proc Natl Acad Sci U S A 1974;71:1427-30.

14. Tonegawa S. Somatic generation of antibody diversity. Nature 1983;302:575-81.

15. Alt FW, Blackwell TK, Depinho RA, Reith MG, Yanco- poulus GD. Regulation of genome rearrangement events during lymphocyte differentiation. Immunol Rev 1986;89:5-30.

16. Neuberger MS, Milstein C. Somatic hypermutation. Curt Opin Immunol 1995;7:248-54.

17. Berek C, Mitstein C. The dynamic nature of the antibody repertoire. Immunol Rev 1988; 105: 5 -26.

18. Huang C, Stollar BD. A majority of human IgH chain cDNA of normal human adult blood lymphocytes resembles cDNA for fetal Ig and natural autoantibodies. J Immunol 1993;151:5290-300.

19. Shlomchik MJ, Marshak-Rothstein A, Wolfowicz CB, Rothstein TL, Weigert MG. The role of clonal selection and somatic mutation in autoimmunity. Nature 1987;328:805-11.

20. Radic MZ, Weigert M. Genetic and structural evidence for antigen selection of anti-DNA antibodies. Annu Rev Immu- nol 1994;12:487-520.

21. Isenberg DA, Ravirajan CT, Rahman A, Kalsi JK. The role of antibodies to DNA in systemic lupus erythematosns. Lupus 1997;6:290-304.

22. Seeman NC, Rosenberg JM, Rich A. Sequence-speclfic recognition of double helical nucleic acids by proteins. Proc Natl Acad Sci U S A 1976;73:804-8.

23. Eilat D, Webster DM, Rees AR. V region sequences of anti-DNA and anti-RNA autoantibodies from NZB/NZW FL mice. J Immunol 1988; 141:1745-53.

24. Barry MM, Mol CP, Anderson WF, Lee JS. Sequencing and modelling of anti-DNA immunoglobulin Fv domains: Comparison with crystal structures. J Biol Chem 1994;269:3623- 32.

25. Kalsi JK, Martin ACR, Hirabayashi Y, et al. Functional and modeling studies of the binding of human monoclonal anti-DNA antibodies to DNA. Mol Immunol 1996;33:471-83.

26. Herron JN, He XM, Ballard DW, et al. An autoantibody to single-stranded DNA: Comparison of the three dimensional


structures of the unliganded Fab and a deoxynucleotide-Fab complex. Proteins 1991; 11:159-75.

27. Brigido MM, Polymenis M, Stollar BD. Role of mouse VHI 0 and VL gene segments in the specific binding of antibody to Z-DNA, analyzed with recombinant single chain Fv molecules. J Immunol 1993; 150:469-79.

28. Polymenis M, Stotlar BD. Critical binding site amino acids of anti-Z DNA single chain Fv molecules: Role of heavy and light chain CDR3 and relationship to autoantibody activity. J Immunol 1994;152:5318-29.

29. Barbas SM, Ditzel HJ, Salonen EM, Yang W-P, Silver- man GJ, Burton DR. Human autoantibody recognition of DNA. Proc Natl Acad Sci U S A 1995;92:2529-33.

30. Roben R Barbas SM, Sandoval L, et al. Repertoire cloning of lupus anti-DNA autoantibodies. J Clin Invest 1996;98: 2827-37.

31. Mockridge CI, Chapman CJ, Spellerberg MB, Iseuberg DA, Stevenson FK. Use of phage surface expression to analyze regions of a human V4-34 (VH4-21)-encoded IgG autoantibody required for recognition of DNA: No involvement of the 9G4 idiotope. J Immunol 1996;157:2449-54.

32. Horwitz AH, Chang CR Better M, Hillstrom KE, Robin- son RR. Secretion of functional antibody and Fab fragment from yeast cells. Proc Nati Acad Sci U S A 1988;88:8678-82.

33. Hasemann CA, Capra JD. High level production of a functional immunoglobulin heterodimer in a baculovirus expression system. Proc Natl Acad Sci U S A 1990;87:3942-6.

34. Radic MZ, Mascelli MA, Erikson J, Shan H, Weigert M. Ig H and L chain contributions to autoimmune specificities. J Immunol 1991;146:176-82.

35. Radic MZ, Mackle J, Erikson J, Mol C, Anderson WF, Weigert M. Residues that mediate DNA binding of autoimmune antibodies. J Immunol 1993; 150:4966-77.

36. Katz JB, Limpanasithikul W, Diamond B. Mutational analysis of an autoantibody: Differential binding and pathogenicity. J Exp Med 1994; 180:925-32.

37. Pewzner-Jung Y, Simon T, Eilat D. Structural elements controlling anti-DNA antibody affinity and their relationship to anti-phosphorylcholine activity. J Immunol 1996;156:3065-73.

38. Zack DJ, Yamamoto K, Wong AL, Stempnak M, French C, Weisbart RH. DNA mimics a self-protein that may be a target for some anti-DNA antibodies in systemic lupus erythematosus. J Immunol 1995;154:1987-94.

39. Irigoyen M, Kowal C, Young ACM, Sacchettini JC, Diamond B. Molecular mapping of the 8.12 SLE associated idiotype specificity at the single amino acid level. Mol Immunol 1996;33:1255-65.

40. Ehrenstein M, Longhurst C, Isenberg DA. Production and analysis of IgG monoclonal anti-DNA antibodies from systemic lupus erythematosus (SLE) patients. Clin Exp Immu- nol 1993;92:39-45.

41. Stevenson FK, Longhurst C, Chapman CJ, et al. Utiliza- tion of the VH4-21 gene segment by anti-DNA antibodies from patients With SLE. J Autoimmun 1993;6:809-26.

42. Eilat D, Hochberg M, Tron F, Jacob L, Bach J-E The V• gene sequences of anti-DNA antibodies in two different strains of lupus-prone mice are highly related. Eur J Immunol 1989;19: 1241-6.

43. Winkler TH, FeN H, Kalden JR. Analysis of immunoglobulin variable region genes from human IgG anti-DNA hybridomas. Eur J Immunol 1992;22:1719-28.

44. Manheimer-Lory A, Katz JB, Pillinger M, Ghossein C, Smith A, Diamond B. Molecular characteristics of antibodies bearing an anti-DNA associated idiotype. J Exp Med lC~91 ;174: 1639-52.

45. Ehrenstein MR, Longhurst CM, Latchman DS, Isenberg DA. Serological and genetic characterization of a human monoclonal immnnoglobulin G anti-DNA idiotype. J Clin Invest t 994;93:1787-97.

the role of in vitro expression systems in the investigation of antibodies to dna

Documents