Membrane-based inverse-transition purification facilitates a rapid isolation of various spider-silk elastin-like polypeptide fusion proteins from extracts of transgenic tobacco

Abstract

Plants can produce complex pharmaceutical and technical proteins. Spider silk proteins are one example of the latter and can be used, for example, as compounds for high-performance textiles or wound dressings. If genetically fused to elastin-like polypeptides (ELPs), the silk proteins can be reversibly precipitated from clarified plant extracts at moderate temperatures of ~ 30 °C together with salt concentrations > 1.5 M, which simplifies purification and thus reduces costs. However, the technologies developed around this mechanism rely on a repeated cycling between soluble and aggregated state to remove plant host cell impurities, which increase process time and buffer consumption. Additionally, ELPs are difficult to detect using conventional staining methods, which hinders the analysis of unit operation performance and process development. Here, we have first developed a surface plasmon resonance (SPR) spectroscopy-based assay to quantity ELP fusion proteins. Then we tested different filters to prepare clarified plant extract with > 50% recovery of spider silk ELP fusion proteins. Finally, we established a membrane-based purification method that does not require cycling between soluble and aggregated ELP state but operates similar to an ultrafiltration/diafiltration device. Using a data-driven design of experiments (DoE) approach to characterize the system of reversible ELP precipitation we found that membranes with pore sizes up to 1.2 µm and concentrations of 2-3 M sodium chloride facilitate step a recovery close to 100% and purities of > 90%. The system can thus be useful for the purification of ELP-tagged proteins produced in plants and other hosts.

Keywords: Downstream processing; Plant molecular farming; Process optimization; Spider silk proteins; Surface plasmon resonance spectroscopy; Ultrafiltration/diafiltration.

Fig. 1

Schematic representation of the membrane-based inverse transition purification (mITP) process. Starting with a clarified plant homogenate (A), the temperature is increased in the presence of salt to trigger the precipitation of ELP-fusion proteins (B, here: fused to spider-silk proteins). The suspension is then applied to a membrane of suitable pore size (e.g. 0.2–2.0 µm), for example in an ultrafiltration/diafiltration device, so that the precipitate is retained whereas the bulk homogenate passes into the flowthrough (C). Next, the membrane is flushed with a hot, salt-rich buffer to remove residual impurities whereas the ELP-fusion proteins remain in a precipitated state (D). Lastly, a cold buffer (without salt) or plain water is used to re-dissolve the ELP precipitate and to elute the product from the membrane (E)

Fig. 2

Spider silk elastin-like polypeptides (ELP) fusion proteins and their quantification of with a surface plasmon resonance (SPR) spectroscopy competitive binding assay. A Schematic representation of the five fusion proteins used in this study. The c-myc part of the fusion protein is shown in blue, whereas the spider-silk domain is colored in green and the ELP part is orange. B Competition assay principle. Anti-c-myc antibody (red) pre-incubated with c-myc-tagged ELP fusion protein (domain color code as in A) containing sample or standard is brought in contact with a surface decorated with peptides containing a c-myc epitope or variant thereof. Only antibodies with at least one unoccupied valency can bind to the surface resulting in a response signal. C Response resulting from antibody (green—9E10, n = 1; orange—A00704, n = 3) binding to a surface decorated with peptide 3 in dependence of the concentration of ELP standard the antibody was pre-incubated with. Data were fitted to a site competition model (Eq. 1) to derive inflection points

Fig. 3

Screening of depth filters and ELP fusion protein recovery during clarification. A Western blot of process samples using depth filter P1 and anti-c-myc for detection of VSO1ELP. Elution fractions originated from 1&3—0.2 µm membrane pore size, 2.0 M sodium chloride during aggregation, 30 °C during wash; 2&4—1.2 µm membrane pore size, 2.4 M sodium chloride during aggregation, 45 °C during wash; primary elution was carried out using 15 mM sodium phosphate buffer pH 7.5 whereas de-ionized water (indicated by “w”) was used for a second elution step. B Western blot of process samples using centrifugation and anti-c-myc for detection of MaSp1ELP. Elution fraction conditions as in A. C ELP fusion protein recovery achieved with different depth filters (Table S1). D Filter capacity in dependence of filter layer combinations and ELP fusion protein. E Turbidity observed after clarifying ELP fusion protein containing extract with different depth filters. Error bars in C–E indicate the standard deviation (n ≥ 3). F Western blot of process samples using depth filter PDR1 and anti-c-myc for detection of ELP fusion proteins. ELP elastin-like polypeptide, ITP inverse transition purification. Error bars indicate the standard deviation of replicate runs with n ≥ 2

Fig. 4

Screening for mITP conditions ensuring a high recovery and purity of ELP fusion proteins using VSO1ELP as a model protein. A Response surface model for VSO1ELP recovery in dependence of sodium chloride concentration during aggregation and wash using a 0.2-µm membrane for aggregate retention. B Same model as in A but using a 1.2-µm membrane. C Response surface model for VSO1ELP purity increase as a multiple of the starting purity in dependence of sodium chloride concentration during aggregation and wash using a 0.2-µm membrane for aggregate retention. D Same model as in C but using a 1.2-µm membrane. The aggregation temperature did not have a significant effect in the 30–45 °C range and was set to 37.5 °C in panels A–D. Dots indicate actual measurements. E Overlay of Coomassie-stained LDS-PAA gel and corresponding western blot using anti-c-myc antibody for VSO1ELP detection in process samples for the verification of the process optimum using a 0.2-µm membrane. F Same setup as in E but with samples from runs confirming the optimum for a 1.2-µm membrane