Recombinant proteins in diagnosis and therapy - Clinical Transfusion Practice

Practical Transfusion Medicine 4th Ed.

33. Recombinant proteins in diagnosis and therapy

Marion Scott

NHS Blood and Transplant, Bristol, UK

Introduction

Many potentially useful human proteins for therapeutic, diagnostic and research use are expressed in the body at very low concentrations and it is difficult, if not impossible, to isolate them by conventional biochemical methods. Other proteins, such as antibodies of a particular specificity, are difficult to purify from a complex mixture of very similar proteins. However, once the gene encoding a protein has been cloned and sequenced, it becomes possible to express the protein at high concentrations, using virally derived expression vectors that are designed to produce full-length proteins at high levels in various different in vitro culture ‘host’ cell systems. Some blood proteins, such as the coagulation factors to treat haemophilia, have been efficiently purified by fractionation of pooled human plasma, but have been shown to have the potential of transmitting diseases, such as HIV and HCV. The cloning and expression of these proteins has led to the availability of recombinant coagulation factors for the treatment of haemophilia, with reduced risk of infection. As the recombinant coagulation factors are grown in vitro, there is also the advantage of an unlimited supply of constant guaranteed product. Similar drivers have led researchers to try and develop recombinant replacements for specific immunoglobulins currently fractionated from high titre blood donations, such as anti-D. Some concern has been expressed about the safety of such recombinant products, as they may potentially contain viruses or other infectious agents arising from the host cells used to express the protein, or the culture medium components used to grow the host cells. Increasing awareness of the risks from pooled polyclonal blood components have been heightened by concerns about variant Creutzfeldt–Jakob disease (vCJD) in the UK. Are the potential risks from such biotechnology products any worse than the risks from blood components derived from pooled human plasma?

Apart from cloning and expressing such naturally occurring proteins, it is possible using recombinant DNA technology to produce modified forms of the proteins that do not occur naturally and that might have desired therapeutic effects or diagnostic advantages.

General methods for recombinant protein expression

The choice of the host cell system to use for recombinant protein expression relies on several factors. Bacterial expression systems, such as Escherichia coli, are the cheapest, simplest and most effective, but cannot be used for many types of human proteins that require eukaryotic posttranslational modifications for biological activity, e.g. glycosylation, since prokaryotes lack the enzymes that catalyse many of the posttranslational modifications found on eukaryotic proteins. Proteins produced in prokaryotes may not be folded properly and/or can be insoluble, forming inclusion bodies. Genetically modified strains of yeast that have human glycosylation pathways have been produced for the efficient production of human glycoforms of recombinant proteins. Insect cells have also been used for recombinant protein expression, using baculoviral vectors. Transgenic animals have also been produced, with targeted production of recombinant proteins in milk. A comparison of different production systems for recombinant proteins is shown in Table 33.1.

Table 33.1 Production systems for recombinant mammalian proteins.

Table033-1

For many types of human proteins, expression in a mammalian system is the best option, as this is the approach most likely to yield soluble, biologically active proteins, although it is considerably more expensive than expression in E. coli, yeast or insect cells. Cell lines commonly used are NS0 (mouse myeloma), CHO (Chinese hamster ovary) and COS-7 (African green monkey fibroblast).

A number of techniques have been developed for rapid, one-stage purification of recombinant proteins. Epitope tags are short amino acid sequences for which commercial monoclonal antibodies are available, and can be placed anywhere within the protein where it will not disrupt the protein's function. It is also common to create fusion proteins, i.e. to create a single open reading frame that encodes a well-characterized protein such as glutathione-S-transferase (GST) together with the sequence of the protein of interest. When the tag protein is produced, the protein of interest is produced as well, as one fusion protein. Fusion proteins are useful because they enable rapid purification by affinity chromatography, and the fused tag can be removed after purification using a specific protease.

Plasmids used for expression commonly contain a viral promoter sequence, an antibiotic resistance gene, a fusion tag sequence and a restriction endonuclease site for insertion of the coding sequence of interest (Figure 33.1). cDNA coding for the protein sequence of interest is normally derived by reverse transcriptase polymerase chain reaction (RT-PCR) from cells expressing the protein, using sequence specific primers to amplify the region required. This cDNA is then inserted into the expression vector and used to transfect a mammalian cell line. Growth in medium containing the antibiotic to which the vector codes resistance results in selection of transfected cells only. Production of the fusion protein can then be detected using antibodies to the fusion tag sequence, and the fusion protein purified and characterized. Some expression vectors do not insert into the host cell nuclear material and give rise to transient expression. Other vectors insert into the host cell DNA and give rise to stable expression.

Fig 33.1 Production of recombinant fusion proteins.

c33f001

Recombinant antibodies

Limitations of rodent monoclonal antibodies

Conventional monoclonal antibody technology uses immunization of mice or rats with antigen to yield hyperimmunized spleen cells, which are then fused with nonsecreting myeloma cell lines to yield hybridoma cell lines that can be grown in vitro to produce monoclonal antibodies [1]. Effectively the fusion process inserts the DNA from the spleen cells into the myeloma cells. Whilst many such conventional monoclonal antibodies were very successfully developed into diagnostic reagents (such as the high avidity anti-A and anti-B now used routinely worldwide for blood grouping), it was not possible to produce antibodies of certain specificities in rodents, and attempts to use rodent monoclonal antibodies in man as therapeutics rapidly ran into problems, as the recipients developed a strong human antirodent response, which rapidly cleared the antibodies from the body.

Humanizing rodent monoclonals

The early promise of monoclonal antibodies as therapeutics was not realized and many became disillusioned with the concept of the ‘magic bullet’. The success rate of rodent monoclonal antibodies that entered clinical trials was only 9% over the 20 years from 1980 to 2000. However, when the possibility of making recombinant antibodies became available in 1986, things rapidly changed. Using recombinant DNA technology, it was possible to replace the mouse constant domains of antibodies with corresponding human domains and express these chimeric recombinant immunoglobulin molecules in cell lines [2]. Further engineering work also allowed the replacement of the framework regions of the mouse variable domains with human framework regions, resulting in virtually fully humanized antibodies [3].

Human recombinant antibodies

Human circulating B cells can be selected from immune individuals and transformed into cell lines that can be grown in culture by transformation with Epstein–Barr virus (EBV). The cDNA coding for the antibodies can then be derived from these cells using RT-PCR, ligated into expression vectors and expressed in a suitable mammalian host cell line.

Alternatively phage display technology can be used (Figures 33.2 and 33.3). Bacteriophages that infect E. coli are modified such that they carry the cDNA encoding for antibody variable domains, whilst at the same time they express the antibody protein on their surface. This permits in vitro selection of antibodies of the required specificity and then expansion in E. coli. RT-PCR is used to amplify all of the heavy and light chain variable domains in a buffy coat sample. PCR is then used to assemble these randomly into VH and VL pairs, by inclusion of DNA encoding for a flexible linker chain between the heavy and light chain domains. A ‘tag’ sequence is also included to aid detection and purification. These linked heavy and light chain domains are known as single-chain Fv (scFv). The scFv constructs are then ligated into a phage display vector. The scFv domain is ligated into the vector next to regions that code for the PIII phage coat protein. The recombinant phage then expresses the scFv protein alongside their PIII coat protein at the tip of the phage. Phage libraries can be panned against antigens, and those phage selected that are displaying scFv that bind to the antigen. Selected phage are eluted from the antigen, expanded by culture in E. coli and then repanned against antigen. Selected human scFv can then removed from the phage vector and can be ligated to cloned human IgG constant domains to express full-length human recombinant antibody molecules. One large advantage of this approach is that antibodies can be derived from phage display libraries made from nonimmunized individuals and that normally restricted antibodies (e.g. anti-self) can be derived [4].

Fig 33.2 Generation of scFv phage libraries.

c33f002

Fig 33.3 scFv displayed on phage surface.

c33f003

Using these approaches, 30 recombinant antibodies are now successfully licensed for clinical use in a variety of applications with around 300 more in the pipeline, and they are now among the most commercially successful biotech drugs, with three in the top ten best selling drugs worldwide [5].

Human recombinant monoclonal anti-D

Despite the overall success in the production of rodent monoclonal antibodies to human ABO blood group antigens, no such monoclonal antibodies have been produced to the Rh antigens. Various different approaches have been developed to produce human monoclonal antibodies specific for RhD.

Early work used the immortalization of human B cells by infection with EBV. Improvements in the stability of human cell lines have been achieved by back-crossing human anti-D secreting EBV lines to a mouse–human heterohybridoma line or to a mouse myeloma line [6]. Use of these approaches has enabled the production of a large number of blood group-specific human anti-Rh monoclonal antibodies. The cDNA coding for these antibodies has then been expressed in suitable mammalian host cell lines to produce recombinant anti-D suitable for therapeutic use (i.e. free from EBV).

Candidate monoclonal anti-Ds for immunoprophylaxis are selected, first, on their ability to bind to the RhD antigen via the Fv part of the molecule and, second, on their ability to interact with Fc receptors via the Fc part of the molecule to bring about immunomodulation. The exact mechanism of immunosuppression by anti-D is not known, but it is clear that it involves interaction of anti-D with Fc receptors. To be effective, prophylactic antibody must be capable of not only binding to the RhD antigen on the red cells via its Fv regions but also interacting with the effector cells of the immune system via its Fc region. Selection of recombinant monoclonal anti-D for therapeutic use therefore depends on not only the antigen specificity and avidity of the monoclonal antibody but also its functional activity in interacting with effector cells. To suppress immunization, IgG-coated RBC need to be rapidly cleared from the maternal circulation and localized in the spleen. It has been suggested that D antigen-specific B cells in the spleen are then deactivated by the simultaneous binding of the Fc region of the anti-D to FcγRIIb together with binding of the B-cell receptor to the D antigen. Interactions of anti-D with FcγRI, FcγRIIb and FcγRIIIa may thus all be required for effective immuno-suppression.

IgG monoclonal anti-D antibodies have been evaluated in various in vitro systems to test how effective the antibodies are at interacting with immune system effector cells. Each assay tests efficacy at binding to different Fc receptors. Rosette formation of sensitized cells with monocytes and phagocytes, adherence of sensitized cells to monocyte monolayers and chemiluminescent measurements of the oxidative burst caused when monocytes react with sensitized red cell are all in vitro measures of interaction with FcγRI. Antibody-dependent cellular cytotoxicity measurements by radiolabelled chromium release from natural killer cells measures interaction with FcγRIIIa. It is not clear at present how well performance in these various in vitro assays will predict in vivo efficacy.

Two monoclonal anti-D antibodies, BRAD-3 and BRAD-5, were selected for clinical study because of their high activity in in vitro functional assays, high avidity and specificity for the immunodominant epitope region of the RhD antigen and initial studies in D-negative male volunteers showed expected half-lives and pharmacokinetics after injection. Further studies on the antibodies administered with ⁵¹Cr-labelled D-positive red cells demonstrated accelerated red cell clearance in all subjects and provided preliminary evidence for protection from immunization [7]. One determinant of affinity for FcγRIII is the glycosylation pattern of the immunoglobulin heavy chains, which may be determined by the nature of the host cell in which the antibody is expressed. A glycosylation pattern of low fucose content has been shown to confer enhanced ADCC activity. A human recombinant monoclonal anti-D selected to have such high affinity FcγRIII binding activity has been shown to promote a similar clearance of antibody-coated RhD RBCs in healthy volunteers to polyclonal anti-RhD immunoglobulin [8]. Roledumab, a commercial product based on this antibody is currently in phase 2 clinical trials.

It is clear from the clinical trials to date that recombinant anti-D has the potential to replace polyclonal prophylactic anti-D. There is a case for universal antenatal prophylaxis if sufficient supplies of anti-D are available. How quickly recombinant anti-D becomes available will largely be determined by commercial investment and regulatory procedures.

Antiprion recombinant antibodies

A range of monoclonal antibodies to human recombinant prion proteins has been produced by immunizing prion knockout mice. Selected antibodies have been developed into a diagnostic test for bovine spongiform encephalopathy using homogenized bovine brain post mortem. Work is currently underway to try and increase the sensitivity of the assay to make it suitable for screening human blood for vCJD. It has been shown that these mouse monoclonal antibodies can prevent the spread of vCJD prion disease in a mouse model [9]. Currently efforts aim to engineer human chimeric and humanized versions of these mouse antibodies to progress this work into clinical trials for the potential treatment of vCJD.

Anti-HPA-1a recombinant antibodies

scFv specific for the human platelet antigen HPA-1a have been derived from a phage display library and ligated to scFv specific for the RhD antigen on red blood cells, and the novel bispecific recombinant antibody can be used in a mixed passive haemagglutination test for the HPA-1a antigen on platelets. The scFv has also been expressed as a full-length human IgG antibody by ligation to the constant domains of human IgG1, and this antibody had been used either fluorescently labelled or enzyme labelled in other diagnostic tests for the HPA-1a antigen on platelets.

‘Null’ recombinant antibodies

Using site directed mutagenesis, the Fc domains of human IgG antibodies have been mutated to have as little biological function as possible. Recombinant anti-D and anti-HPA-1a antibodies have been produced with this ‘null’ Fc region. In vitro studies have shown that these ‘null’ antibodies can effectively compete with clinically significant antibodies and prevent them, causing immune destruction of red cells and platelets, respectively. A clinical trial in male volunteers showed that the ‘null’ anti-D protected D-positive red cells from clearance by anti-D in vivo [10]. The aim is to see if the ‘null’ anti-HPA-1a antibody can be administered to HPA-1a-negative pregnant women who are carrying HPA-1a-positive fetuses and prevent neonatal alloimmune thrombocytopaenia by crossing the placenta and competing with maternal anti-HPA-1a that can cause destruction of fetal platelets (see also Chapter 5).

Recombinant phenotyping reagents

Many monoclonal human IgG antibodies have been produced that react with blood group antigens, but these must be used in enzyme or antiglobulin techniques that are not suited to hig throughput, automated blood grouping machines. By cloning the variable regions of these antibodies, it is possible to ligate them to the constant domains of human IgM antibodies and express hybrid recombinant molecules in myeloma cells. These antibodies are highly potent because they combine the high affinity of the affinity-matured IgG antibodies with the polymeric structure of IgM [11]. They are very potent direct agglutinins that can readily be used in automated blood grouping machines.

Recombinant antigens

Blood group antigens

Detection and identification of clinically significant blood group, platelet and granulocyte antibodies currently rely on the availability of high quality antibody screening and identification cells that cover all clinically significant antigens and carry them in combinations such that the specificity of antibodies can be deduced. The quality of these panels of cells is critical and their variability has been shown in UK National External Quality Assessment Scheme exercises to be the main cause of error in the detection and identification of antibodies. Quantitation of antibodies during pregnancy is carried out using titration or autoanalyser technology, both of which show high levels of variation, such that it is difficult to set levels at which clinical action is required.

Most of the relevant antigens have been sequenced and cloned. Some have been inserted in expression vectors and expressed in the membranes of in vitro cultured cells, e.g. expression of the Rh protein in K562 human erythroleukaemia cells. For some antigens it is possible to amplify just the extracellular domain of the protein that carries the antigen and express a soluble recombinant protein that carries antigenic activity. This has been demonstrated for the Kell, Lutheran, Duffy, MNSs and Cartwright red cell antigens, HPA-1a and HPA-1b platelet antigens and HNA-1 and HNA-2 granulocyte antigens [12].

The target of this work is to be able to produce microarrays of recombinant antigens that could then be used for high throughput antibody screening, identification, subclass determination and quantitation of antibodies in transfusion recipients and pregnant women.

Microbial antigens

In a similar way, recombinant microbial antigens have been produced to test blood donor plasma for the presence of antimicrobial antibodies, and thus exposure to transfusion transmitted diseases. Viral antigens and tests have been produced in this way, and more recently tests for parasites, such as those causing malaria (Plasmodium falciparum and P. vivax) and Chagas (Trypanosoma cruzi), have been produced and shown to have good specificity and sensitivity for screening blood donors [12].

Recombinant enzymes

It has long been known that it is possible to treat group A, B or AB red cells with glycosidase enzymes and convert them to group O, such that they would theoretically be a safe blood transfusion product for any ABO blood group recipient. However, the naturally occurring enzymes have poor kinetic properties and difficult-to-achieve pH optima, such that the process was not economically viable or able to pass rigorous quality assurance requirements for clinical use. New recombinant enzymes have now been produced from bacterial glycosidases with remarkably improved kinetic properties, such that the enzymes reproducibly cleave the A and B antigens with low enzyme protein consumption, short incubation times and at neutral pH [13]. Clinical trials evaluating the safety and efficacy of such recombinant enzyme-treated red cells are looking promising.

Recombinant coagulation factors

Recombinant coagulation factors have been successfully used for the treatment of haemophilia for several years. Recombinant protein technology has virtually eliminated transmissible disease risk from these products, such that the recombinant products are the products of choice for haemophiliacs. In the UK, most patients with severe haemophilia now receive recombinant factor VIII and factor IX. Recombinant factor VIIa was originally developed for the treatment of haemophilia patients who had developed inhibitory alloantibodies to factors VIII and IX, and is licensed for this application (see Chapter 37).

Recombinant haemoglobin is considered in Chapter 36 and recombinant erythropoietin in Chapters 28 and 34.

Conclusions

Recombinant protein technology has rapidly advanced over the last 25 years and we are now starting to see the routine use of recombinant proteins in transfusion medicine. Recombinant proteins will probably totally replace coagulation factors and specific immunoglobulins that are currently produced from fractionated pooled plasma. However, it is unlikely that recombinant products will replace intravenous immunoglobulin or albumin. Intravenous immunoglobulin works because of its broad specificity – it would be very difficult/impossible to mimic this successfully with a recombinant product. Albumin could be produced as a recombinant protein, but this is unlikely to be economically viable, compared to the ease of production from plasma. Only evidence of disease transmission by plasma-derived albumin could drive the production of recombinant albumin.

Further specific recombinant immunoglobulins are being produced that are not currently available as blood components – anti-HCV and anti-vCJD – and the efficacy of these needs to be investigated in clinical trials. Blood group antigens are now available as recombinant molecules, such that there may no longer be a need to use red cells, platelets and granulocytes for antibody screening, identification and quantitation.

Key points

1. Proteins expressed at low levels naturally can be cloned and expressed as recombinant proteins at high levels.

2. The sequence of recombinant proteins can be altered to give properties not found in naturally occurring proteins.

3. Human recombinant antibodies can be produced from nonimmune donors.

4. Murine monoclonal antibodies can be humanized for clinical use.

5. Recombinant antibodies with inactive Fc regions can be produced as blocking antibodies.

6. Recombinant antigens can be used for screening, identification and quantification of clinically significant antibodies.

7. Recombinant coagulation factors have largely replaced those derived from pooled plasma.

References

1. Kohler G & Milstein C. Continuous cultures of fused cells secreting antibody of predefined specificity. Nature 1975; 256: 495–497.

2. Morrison SL. Chimeric human antibody molecules: mouse antigen binding domains with human constant region domains. Proc Natl Acad Sci USA 1984; 81: 6851–6855.

3. Jones PT. Replacing the complementarity determining regions in a human antibody with those from a mouse. Nature 1986; 321: 522–525.

4. Marks JD, Hoogenboom HR, Bonnert TP et al. By-passing immunisation. Human antibodies from V-gene libraries displayed on phage. J Molec Biol 1991; 222: 581–592.

5. Reichert JM. Antibody-based therapeutics to watch in 2011. mAbs 2011; 3: 76–99.

6. Thompson KM, Melamed MD, Eagle K et al. Production of human monoclonal IgG and IgM antibodies with anti-D Rhesus specificity using heterohybridomas. Immunology 1986; 58: 157–160.

7. Kumpel BM, Goodrick MJ, Pamphilon DH et al. Human RhD monoclonal antibodies (BRAD-3 and BRAD-5) cause accelerated clearance of RhD red blood cells and suppression of RhD immunization in RhD– volunteers. Blood 1995; 86: 1701–1709.

8. Beliard R, Waegemans T, Notelet D et al. A human anti-D monoclonal antibody selected for enhanced FcgammaRIII engagement clears RhD+ autologous red cells in human volunteers as efficiently as polyclonal anti-D antibodies. Br J Haematol 2008; 141: 109–119.

9. White AR, Enever P, Tayebi M et al. Monoclonal antibodies inhibit prion replication and delay the development of prion disease. Nature 2003; 422: 80–83.

10. Armour KL, Parry-Jones DR, Beharry N et al. Intravascular survival of red cells coated with a mutated human anti-D antibody engineered to lack destructive activity. Blood 2006; 107: 2619–2626.

11. Gilmour JEM, Pittman S, Nesbitt R & Scott ML. Effect of the presence or absence of J chain on expression of recombinant anti-Kell immunoglobulin M. Transfus Med 2008; 18: 167–174.

12. Ridgwell K, Dixey J & Scott ML. Production of soluble recombinant proteins with Kell, Duffy and Lutheran blood group activity, and their use in screening human sera for Kell, Duffy and Lutheran antibodies. Transfus Med 2007; 5: 384–394.

13. Olsson ML & Clausen H. Modifying the red cell surface: towards an ABO-universal blood supply. Br J Haematol 2008; 140: 3–12.