A protein chimera strategy supports production of a model "difficult-to-express" recombinant target

Abstract

Due in part to the needs of the biopharmaceutical industry, there has been an increased drive to generate high quality recombinant proteins in large amounts. However, achieving high yields can be a challenge as the novelty and increased complexity of new targets often makes them 'difficult-to-express'. This study aimed to define the molecular features that restrict the production of a model 'difficult-to-express' recombinant protein, Tissue Inhibitor Metalloproteinase-3 (TIMP-3). Building from experimental data, computational approaches were used to rationalize the redesign of this recombinant target to generate a chimera with enhanced secretion. The results highlight the importance of early identification of unfavourable sequence attributes, enabling the generation of engineered protein forms that bypass 'secretory' bottlenecks and result in efficient recombinant protein production.

Keywords: difficult-to-express; mammalian expression system; predictive computational tool; protein engineering; recombinant protein production; tissue inhibitor of metalloproteinase.

Figure 1

Alignment of TIMP‐2 and TIMP‐3 amino acid sequences. (A) Amino acid sequences for TIMP‐2 and TIMP‐3 were chemically annotated and aligned using EMBOSS Needle tools 59, 60. Amino acids are coloured according to their respective properties, Small/Hydrophobic (red), acidic (blue), basic (magenta), hydroxyl/sulfhydryl or amine (green). Conserved amino acids (I) and gaps in the alignment (−) are shown. Conserved cysteine residues for disulphide bond formation (grey boxes), the boundary between the N‐terminal and C‐terminal TIMP domains and the TIMP‐3 N‐glycosylation site are also indicated. (B) Schematic diagram summarising the construction of domain‐exchanged vectors is shown.

Figure 2

Transient expression of TIMP‐2 and TIMP‐3 domain‐exchanged sequences in CHO cell cultures. CHO‐EBNA‐GS cells were transiently transfected and sampled for secreted (culture medium) and intracellular (cell extracts) proteins on day 3, day 5 and day 6 (D3, D5 and D6) post‐transfection. Protein samples for, (A) NT2/CT3 and (B) NT3/CT2, were analysed by western blot. Nontransfected cells (NC) and intracellular TIMP‐3 (sampled day 5 post‐transfection) were loaded as controls. ERK was used as a loading control for cell extracts. The amount of TIMP‐2 and NT2/CT2 protein was assessed in crude and purified culture medium samples (sampled day 6 post‐transfection) by reducing SDS/PAGE (C). Data is representative of three biological replicates.

Figure 3

Hydrophobicity analysis of TIMP‐2 and TIMP‐3 protein structures. Structural models of TIMP‐2 and TIMP‐3 created using SWISS‐MODEL 49, 50 and processed through an algorithm developed by the Warwicker group (The University of Manchester). Predictions of the (A–B) hydrophobicity (nonpolar vs. polar regions) and (C–D) electrostatic potential (positive vs. negative charge) of protein surfaces were generated for TIMP‐2 (A,C) and TIMP‐3 (B,D). Proteins structures are represented as both tertiary structure ribbon representation (top) and surface maps (bottom) with a structural view of the front and back (180° rotation). Arrows indicate nonpolar patches unique to TIMP‐3. For electrostatic potential analysis, the predicted ranking of maximum positive electrostatic potential patch size (PosQ value) for TIMP‐2 and TIMP‐3 is detailed in Table 1.

Figure 4

Computational analysis of an engineered TIMP‐3 chimera sequence (enTIMP‐3) and characterization of secreted and intracellular protein in CHO cell cultures. (A) Sequences within the N‐terminus of TIMP‐3 (positions 26‐41) were replaced with the corresponding TIMP‐2 sequence (positions 26‐47) to generate enTIMP‐3. (B) Predicted surface hydrophobicity (top panel) and electrostatic potential (bottom panel) are shown for TIMP‐3, enTIMP‐3 and TIMP‐2. The front view is shown for each structure and arrows indicate patches of interest. enTIMP‐3‐transfected CHO‐EBNA‐GS cultures were sampled for secreted and intracellular proteins on different days (D) throughout the culture period. (C) Protein samples were analysed by western blot with secreted TIMP‐2 and intracellular TIMP‐2 and TIMP‐3 loaded as controls (sampled day 6 post‐transfection). ERK was used as a loading control for intracellular samples. Western blots were quantified using the LI‐COR imaging system and plotted for secreted enTIMP‐3, glycosylated and nonglycosylated intracellular enTIMP‐3 forms, the corresponding controls (secreted TIMP‐2, glycosylated and nonglycosylated intracellular TIMP‐3) are also shown (grey bars). The data was analysed by two way ANOVA, where a P value of < 0.05 (*), < 0.01(**) and < 0.001(***) was deemed significantly different. Error bars shown are the mean value ± SEM of three biological replicates.