In this report we have exploited promising, alternative expression systems to produce and test ScFv800E6 as a candidate molecule for applications in oncology. ScFv800E6 is the first ScFv to ErbB-2 produced in plants. Along with its 5 tagged variants, it is also one of the few ScFvs to be developed for expression in second-generation, high-yield, cell-free transcription-translation systems capable of establishing disulfide link formation.
We have shown that the binding of ScFv800E6 obtained from different platforms is antigen-specific, saturable, titratable, and can be competed by the parental antibody, i.e. it recapitulates the canonical features and the fine specificity of a natural ligand-target interaction. Radiobinding studies and flow cytometry data were consistent with a remarkably robust, stable, versatile, and modular ScFv800E6 'backbone' that tolerates extensive modifications at both the N- and C-termini, although the position of the tag is crucial for its availability upon incubation with secondary reagents. The apparently low staining efficiency of ScFv800E6 was largely due to the use of conventional secondary anti-Ig reagents and not to a low binding affinity, since ScFv binding, in flow cytometry, was at least as high as that of the monovalent Fab' in spite of the use of secondary reagent that preferentially bound to the latter. Accordingly, equilibrium binding studies revealed a binding affinity slightly higher than that of the Fab' with no major drop as compared to that of the parental, bivalent antibody. This is remarkable, since bivalent binding is known to greatly stabilize antigen-antibody complexes . These results suggest that the antigen binding site of the recombinant ScFv has undergone no major derangements as compared to that of the natural antibody, whereas enzymatic fragmentation may moderately hamper the performance of the Fab'. Thus, expression of recombinant ScFv800E6 bypasses a potential obstacle that would preclude size reduction of the parental 800E6 antibody.
ScFv800E6 can be produced in all the expression platforms at concentrations (see below) sufficient, or greater than required, for all the major indirect trace binding assays, and all the ScFv variants perform satisfactorily with no need to modify or adapt commercially available immunodiagnostic reagents and kits (e.g. in immunohistochemistry). ScFvs can be tagged for detection by an extremely sensitive secondary reagent, such as Strep-Tactin, that outperforms even sensitive streptavidin-based detection systems and largely compensates for monovalent binding. ScFvs can be radiolabeled to high specific activity for in vivo radioimaging by a standard Chloramine T iodination, with no need for special procedures or dedicated protocols. In summary, ScFv800E6 variants are all ready for application in oncology. In this respect, two issues are of particular interest: yield and folding.
We observed that the yield of ScFv800E6 from stable transgenic plants did not exceed the microgram per ml range, i.e. it was low as compared to other ScFvs produced in tobacco plants [reviewed in ]. Strikingly, an improvement of three orders of magnitude was obtained by recovering the ScFv800E6 from leaves exhibiting systemic symptoms in transiently modified plants, indicating that the features of ScFv800E6 are not intrinsically incompatible with its efficient expression in plants. Because preliminary data indicate that transgene silencing may affect ScFv expression in stable transgenic plants, we are currently improving ScFv yield by taking advantage of plant expression systems that alleviate this problem and in addition dispatch antibody fragments to specific plant compartments such as roots and seeds.
As an alternative to plant expression, we produced ScF800E6 in a cell-free transcription-translation system that, as described in methods, differs under several respects from other cell-free systems previously used to express disulfide-bonded ScFvs [6, 8]. ScFv800E6 was obtained with a yield of 200 μg/ml, approximately 20 times higher than that of a different ScFv produced in a conventional format . A systematic comparison involving transcription-translation of many different ScFvs in the available formats is mandatory to determine whether optimal conditions must be worked out individually for every construct or, alternatively, similar protocols can be used for different ScFvs.
Cell-free expression of ScFvs is also relevant in the context of the 'ribosome display' approach. This approach takes advantage of transcription-translation for the phenotypic selection of ScFvs with increased affinity upon their immobilization on polysomes [27, 28]. In principle, it should be possible to incorporate the present semi-continuous format in current ribosome display protocols, offering a high-yield alternative to experiments aimed at ScFv improvement.
A surprising finding becomes apparent when the results in the various expression platforms are compared. ScFv800E6 was functional when expressed in reducing cytosolic environments (bacteria and plants), but not in transcription-translation systems unable to establish disulfide linkage (Figs. 3 and 4). At least two interpretations may be proposed: (a) some or all of the ScFv molecules synthesized in vivo are functional because they somehow manage to get disulfide-linked in the cytosol or other cellular locations; (b) disulfide linkages are essential in vitro but not in vivo, resulting in two different (but similarly active) ScFv folds. These issues are of bio-technological relevance in view of large-scale production. However, to be addressed they require structural studies on large amounts of purified ScFvs from different sources. Whatever the exact folding mechanism, and the role of disulfide linkage, ScFv800E6 is stable and active in different expression platforms. This property is highly desirable and unusual among previously described recombinant antibody fragments . This results in unique versatility and flexibility in the choice of expression platforms. Therefore, ScFv800E6 appears to be an ideal candidate for a three-step development of bio-technological processes leading to progressive improvement of the reagent. As outlined in this report, reagents of this kind may be pre-screened for activity in different hosts, modified and tested on a small-medium scale in a convenient platform (e.g. cell-free systems), and then moved (if required) to another one (e.g. plants) for mass production, and the optimization loop may be repeated as many times as needed.
Plant and cell-free expression systems are largely complementary. The advantages of mass production of recombinant antibody fragments in plants have been long known . Now, with the production of functional ScFvs in vitro, transcription-translation systems may also become extremely attractive. In this respect, it may be noted that although the available cell-free systems are designed to produce amounts of recombinant proteins in the milligram range, yield and final ScFv concentrations in our experimental system exceed those contained in hybridoma supernatants (Fig. 3). Moreover, scale-up is not a concern, since there is no theoretical upper limit to the reaction volume. This is in contrast to mammalian cell bioreactors that require sophisticated equipments in order to maintain adequate gas permeation and nutrient diffusion in the liquid phase, and at the same time prevent cell shearing.