Three unrelated protease inhibitors enhance accumulation of pharmaceutical recombinant proteins in Nicotiana benthamiana

Abstract

Agroinfiltrated Nicotiana benthamiana is a flexible and scalable platform for recombinant protein (RP) production, but its great potential is hampered by plant proteases that degrade RPs. Here, we tested 29 candidate protease inhibitors (PIs) in agroinfiltrated N. benthamiana leaves for enhancing accumulation of three unrelated RPs: glycoenzyme α-Galactosidase; glycohormone erythropoietin (EPO); and IgG antibody VRC01. Of the previously described PIs enhancing RP accumulation, we found only cystatin SlCYS8 to be effective. We identified three additional new, unrelated PIs that enhance RP accumulation: N. benthamiana NbPR4, NbPot1 and human HsTIMP, which have been reported to inhibit cysteine, serine and metalloproteases, respectively. Remarkably, accumulation of all three RPs is enhanced by each PI similarly, suggesting that the mechanism of degradation of unrelated RPs follows a common pathway. Inhibitory functions HsTIMP and SlCYS8 are required to enhance RP accumulation, suggesting that their target proteases may degrade RPs. Different PIs additively enhance RP accumulation, but the effect of each PI is dose-dependent. Activity-based protein profiling (ABPP) revealed that the activities of papain-like Cys proteases (PLCPs), Ser hydrolases (SHs) or vacuolar processing enzymes (VPEs) in leaves are unaffected upon expression of the new PIs, whereas SlCYS8 expression specifically suppresses PLCP activity only. Quantitative proteomics indicates that the three new PIs affect agroinfiltrated tissues similarly and that they all increase immune responses. NbPR4, NbPot1 and HsTIMP can be used to study plant proteases and improve RP accumulation in molecular farming.

Keywords: Agrobacterium tumefaciens; Nicotiana benthamiana; activity-based protein profiling; agroinfiltration; protease inhibitor; transient expression.

Figure 1

Four of 29 tested candidate protease inhibitors (PIs) enhance recombinant protein (RP) accumulation. (a) Four candidate PI selection strategies. (b) Details of the selected candidate PIs. *, presumed class of target proteases. References: [1] Luckett et al., ; [2] Arkadash et al., ; [3] Fluhr et al., ; [4] Kim et al., ; [5] Martinez et al., ; [6] Goulet et al., ; [7] Shindo et al., ; [8] Tian et al., ; [9] Tian et al., ; [10] Lozano‐Torres et al., ; [11] Kim et al., . (c) Expression cassette for PIs containing a 35S promoter and terminator, NtPR1 signal peptide (SP) and PIV2 intron, flanked by T‐DNA borders. (d) PIs were transiently co‐expressed in N. benthamiana with three RPs [α‐Galactosidase (αGal), erythropoietin (EPO), and antibody VRC01] by agroinfiltration. (e) Screening results. Four PIs enhanced RP accumulation (black boxes) and 25 PIs had minor or no effect on RP accumulation (grey boxes). Effects of six PI‐RP combinations were not determined (ND).

Figure 2

Individual co‐expression with NbPR4, NbPot1 or HsTIMP enhances accumulation of αGal (a), EPO (b) and VRC01 (c). Leaves were infiltrated with 1/1 (v/v) mixes of A. tumefaciens strains carrying plasmids for expression of αGal (a) or EPO (b) and PI or 1/1/1 (v/v) mixes of A. tumefaciens strains carrying plasmids for expression of VRC01 heavy chain, VRC01 light chain and PI (c). Full leaf extracts were harvested at 3 dpi. Proteins were subjected to reducing (a–b) or nonreducing (c) SDS‐PAGE and transferred onto PVDF membranes. RP accumulation was visualized using the indicated antibodies. Closed and open triangles in (c) indicate the full‐length VRC01 and putative degradation products, respectively (V_L /C_L , variable/constant domain of the light chain, V_H /C_H , variable/constant domain of the heavy chain). The blots are representative of at least five biological replicates. (d) The top band in VRC01 blots was quantified using ImageJ and normalized to the SlCYS8‐Q47P control (± SD, n = 5, ANOVA P < 0.0001 with post hoc Tukey test, P < 0.05). The 55 kDa Rubisco protein stained by Coomassie is shown as a loading control.

Figure 3

PIs accumulate in N. benthamiana leaves upon transient expression. Inhibitor‐derived peptides detected by MS in extracts obtained at 4 dpi from agroinfiltrated leaves co‐expressing PIs with P19. Peptides (blue) are mapped to the inhibitor sequences (black) which carry the NtPR1 signal peptide (SP, grey). For HsTIMP and SlCYS8, mutations are indicated in red.

Figure 4

Combinations of NbPR4, HsTIMP and SlCYS8 enhance RP accumulation. Leaves were infiltrated with 1/1 (v/v) mixes of A. tumefaciens strains carrying plasmids for expression of αGal (a) or EPO (b) and PI or 1/1/1 (v/v) mixes of A. tumefaciens strains carrying plasmids for expression of VRC01 heavy chain, VRC01 light chain and PI (c). The PI part of the mixture contained three volumes of A. tumefaciens strains for expression of the indicated PIs, with one part mutant PI (Ala‐HsTIMP) used to replace each missing PI. Full leaf extracts were harvested at 3 dpi. Proteins were subjected to reducing (a–b) or nonreducing (c) SDS‐PAGE and transferred onto PVDF membranes. RP accumulation was visualized using the indicated antibodies. The blots are representative of six biological replicates each. (d) The top band in all VRC01 blots was quantified using ImageJ and normalized to the control (± SD, n = 6, ANOVA P < 0.0001 with post hoc Tukey test, P < 0.05). The 55 kDa Rubisco protein stained by Coomassie is shown as a loading control.

Figure 5

Activity‐based protein profiling of cellular proteases indicates that NbPR4, NbPot1 and HsTIMP differ from SlCYS8 in their mode of action. Activity profiles of papain‐like Cys proteases (PLCPs, (a), Ser hydrolases (SHs, (b) and vacuolar processing enzymes (VPEs, (c) Leaves were infiltrated with A. tumefaciens harbouring the indicated PI expression plasmid, mixed 1/1 (v/v) with A. tumefaciens harbouring the P19 expression plasmid. Leaf extracts (pH 5) were obtained at 4 dpi, adjusted to the same protein concentration, and 48 μL of each sample was pre‐incubated with or without 0.2 mm of inhibitor (E‐64, DFP or YVAD) for 30 min and then incubated with or without the indicated probe for 4 h (MV201, JOPD1) or 1 h (FP) at room temperature. Labelled proteins were visualized by in‐gel fluorescence scanning. The 55 kDa Rubisco protein stained by Coomassie is shown as a loading control.

Figure 6

NbPR4, NbPot1 and HsTIMP similarly affect leaf proteomes. Label‐free quantitative MS was performed on leaf extracts (4 dpi) from leaves overexpressing the PIs in the presence of P19. (a) Complete linkage clustering was performed using Euclidean distances between samples, based on all endogenous differential proteins. Endogenous differential proteins are all proteins that differ significantly (t‐test, P < 0.05) and >2‐fold in abundance between PI‐overexpressing and P19‐expressing control leaves, but not P19 and PIs. (b) Differentials associated with PI overexpression are endogenous proteins that differ significantly (t‐test, P < 0.05) and >2‐fold when comparing NbPR4 and NbPot1 expressing leaves to the P19 control and HsTIMP and SlCYS8 expressing leaves to the respective mutant PI control. Proteins were grouped according to whether they increased for at least one PI and never decreased (group I), decreased for at least one PI and never increased (group II), increased for some, but decreased for other PIs (group III) or never changed significantly and more than twofold (group IV). PFAM domains named with groups I to III are enriched (geometric test, Benjamini‐Hochberg adjusted P < 0.05) among the proteins in the respective group, representative PFAMs are shown.

Figure 7

Models for RP degradation mechanisms.