key: cord-0315421-vstumsts
authors: Yu, Daniel S.; Outram, Megan A.; Crean, Emma; Smith, Ashley; Sung, Yi-Chang; Darma, Reynaldi; Sun, Xizhe; Ma, Lisong; Jones, David A.; Solomon, Peter S.; Williams, Simon J.
title: Optimised production of disulfide-bonded fungal effectors in E. coli using CyDisCo and FunCyDisCo co-expression approaches
date: 2021-08-31
journal: bioRxiv
DOI: 10.1101/2021.08.31.458447
sha: fe0db9261e293cca3593a9a93759e773452eff62
doc_id: 315421
cord_uid: vstumsts

Effectors are a key part of the arsenal of plant pathogenic fungi and promote pathogen virulence and disease. Effectors typically lack sequence similarity to proteins with known functional domains and motifs, limiting our ability to predict their functions and understand how they are recognised by plant hosts. As a result, cross-disciplinary approaches involving structural biology and protein biochemistry are often required to decipher and better characterise effector function. These approaches are reliant on high yields of relatively pure protein, which often requires protein production using a heterologous expression system. For some effectors, establishing an efficient production system can be difficult, particularly those that require multiple disulfide bonds to achieve their naturally folded structure. Here, we describe the use of a co-expression system within the heterologous host E. coli termed CyDisCo (cytoplasmic disulfide bond formation in E. coli) to produce disulfide bonded fungal effectors. We demonstrate that CyDisCo and a naturalised co-expression approach termed FunCyDisCo (Fungi-CyDisCo) can significantly improve the production yields of numerous disulfide bonded effectors from diverse fungal pathogens. The ability to produce large quantities of functional recombinant protein has facilitated functional studies and crystallisation of several of these reported fungal effectors. We suggest this approach could be useful when investigating the function and recognition of a broad range of disulfide-bond containing effectors.

We are interested in understanding the structure and function of effectors from multiple 50 plant-pathogenic fungi. Many of these are Kex2-processed pro-domain (K2PP) effectors, 51 which include cysteine-rich effectors with thiol groups of the cysteine sidechain involved in 52 disulfide bond formation (Outram et al. 2021) . To study disulfide-bond containing effectors, 53

we have sought to develop tools to enhance protein production in Escherichia coli (Zhang et 54 al. 2017; Outram et al. 2021) . To this end, we (and others), have had success using the 55 specialised strain of E. coli, SHuffle® (New England Biolabs, Ipswich, Massachusetts, United 56

States) (Maqbool et al. 2015; Zhang et al. 2017; De la Concepcion et al. 2018; Outram et al. 57 2021) . SHuffle is engineered to address unfavourable redox potential in the cytoplasm through 58 disruption of the glutaredoxin and thioredoxin pathways, and expression of a cytoplasmic 59 version of the disulfide bond isomerase protein, DsbC, which normally localises to the 60 periplasm ( Fig. 1A and B ). These manipulations have been shown to improve production of 61 correctly-folded disulfide-bonded proteins (Lobstein et al. 2012) . We have subsequently 62 utilised small solubility tags to further enhance disulfide-rich effector yields in SHuffle 63 (Outram et al. 2021) . Despite these advancements, the yields obtained for many of our effectors 64 of interest have remained low and inadequate for structural and biochemical studies. 65

To address this limitation, we have sought to further improve our production system. 66

The emergence of synthetic biology and the molecular tools that support this discipline have 67 seen an increased interest in co-expression of eukaryotic machinery and chaperones in E. coli 68 to improve recombinant protein production (Zhou et al. 2018) . This approach has also been 69 developed to enhance production of disulfide-bonded proteins. In 2014, Matos and colleagues 70 from Rhynchosporium commune (Supplementary Table S1 ). Most of these effectors could only 143 be produced in low yields from SHuffle E. coli despite the addition of fusion partners (Outram 144 et al. 2021) . SIX1, SIX4 and SnTox3 were expressed with an N-terminal 6xHisGB1 tag, but 145 GB1 was not included for SnTox1 and NIP2.1 as the tag was a similar size to the proteins of 146 interest leading to complications during downstream analysis. Proteins expressed in SHuffle 147 E. coli alone or with CyDisCo were expressed and purified (side-by-side) using the same 148 approach described for SIX6 (details in methods) and the final mono-dispersed SEC elution 149 fractions were compared (Fig. 3) . In most cases, we observed an increase in final yields 150 associated with co-expression with CyDisCo with a ~29-fold improvement for SIX1 resulting 151 The Erv1p and human PDI pair of CyDisCo was previously reported to be the most effective 162 at increasing protein yield (Gaciarz et al. 2016 ). However, this system has been used 163 predominantly to enhance the production of disulfide-rich human proteins such as antibodies, 164 human growth factor and perlecan (Matos et al. 2014; Gaciarz et al. 2016; Sohail et al. 2020) . 165

Recently, a modified CyDisCo system was utilised to produce disulfide-rich conotoxins from 166 cone snails, whereby an additional conotoxin-specific PDI from Conus geographus was 167 We have shown that CyDisCo benefits the production of numerous disulfide-rich 169 fungal effectors. Despite this advance, the yield for some effectors, such as FovSIX6, remained 7 low (0.2 mg/L) and we wanted to investigate whether the CyDisCo system could be modified 171 and improved to benefit the production of recalcitrant disulfide-rich fungal effectors. 172

We substituted the human PDI with a PDI from Fol, as the amino acid sequence of 173 human PDI is substantially divergent from fungal PDIs ( Supplementary Fig. S3A ). We also 174 selected a sulfhydryl oxidase (Erv2) that localises in the endoplasmic reticulum (ER) of fungi 175 to co-express with PDI in place of ERV1p. Erv2 is a fungal-specific membrane-bound 176 sulfhydryl oxidase that catalyses disulfide bonds de novo within the ER ( indicated that four putative PDI proteins and two Erv2-like proteins were present in Fol 185 Fig. S3A and B). To select the most appropriate proteins for co-expression 186 studies in E. coli, we made use of RNAseq data from Fol infections of tomato (Fig. 4A ). This 187 demonstrated that FOXG_00140 (FolPDI) and FOXG_09255 (FolErv2) were upregulated 188 during infection, and these proteins were subsequently selected for expression trials (Fig. 4B) . 189

To assess whether we could improve the CyDisCo system for production of disulfide- and FolPDI lacking the signal peptide (Fig. 4B) , and purified them (side-by-side) using the 195 same approaches detailed above. To confirm CyDisCo/FunCyDisCo components were 196 expressed in a soluble form, total and clarified lysates were analysed by SDS-PAGE (Fig. 4C) . 197

For FolSIX6, the co-expression of CyDisCo or FunCyDisCo were equally effective, each 198 resulting in a yield of ~2 mg per litre of culture, a 5-fold increase compared to SHuffle alone 199 (Fig. 4D ). FovSIX6 co-expressed with FunCyDisCo resulted in a yield of ~0.6 mg per litre of 200 culture, a 3-fold improvement in yield compared to co-expression with CyDisCo and 15-fold 201 improvement compared to SHuffle alone (Fig. 4E ). For SIX1, co-expression with FunCyDisCo 202 resulted in a 13-fold improvement in yield compared to SHuffle alone and a 2-fold decrease in 203 protein yield when compared to CyDisCo (Fig. 4F ). NIP2.1 co-expressed with FunCyDisCo 204 resulted in a yield of ~0.06 mg/L, which was similar to the yields obtained from SHuffle alone, 205 but a 2.5-fold decrease compared to CyDisCo (Fig. 4G) . Collectively, these results suggest the 206 use of CyDisCo or FunCyDisCo co-expression systems can both improve yields of disulfide-207 rich effectors compared to SHuffle E. coli alone, however the choice of which system works 208 best is protein specific. 209 210

We have shown the CyDisCo and FunCyDisCo co-expression systems are effective at 212

improving yields for numerous disulfide-rich fungal effectors produced in SHuffle E. coli 213 compared to SHuffle alone. In a previous report, the CyDisCo system could be used to produce We therefore investigated if improvements to the yield of disulfide-rich fungal effectors 218

can be made using the CyDisCo and FunCyDisCo systems expressed in non-redox mutant 219 strains such as BL21(DE3). FolSIX6 lacking the signal peptide with an N-terminal 6xHisGB1 220 tag was expressed in BL21(DE3) by itself, or co-expressed with CyDisCo or FunCyDisCo 221 systems, and was purified simultaneously from both using nickel affinity chromatography. 222

However, we were unable to produce high quantities of FolSIX6 with either co-expression 223 system in BL21(DE3). We were also unable to confirm the soluble production of 224 CyDisCo/FunCyDisCo components ( Supplementary Fig. S4 ). Collectively, in our hands, the 225 CyDisCo and modified FunCyDisCo systems were not transferable into BL21(DE3) E. coli. 226 227

The adoption of the CyDisCo/FunCyDisCo system has facilitated the structural elucidation of 229 numerous fungal effectors in our lab (structures to be presented elsewhere). Here, however, we 230 present data to show that these high-quality/purity proteins have applications outside of 231 structural studies. Previously, we used a protein-mediated phenotyping approach to study the 232 necrotrophic effector SnTox1 and SnTox3 in wheat (Zhang et al. 2017; Outram et al. 2021; 233 Sung et al. 2021 ). Here, we were interested in determining whether the effectors produced 234 using our enhanced production system could be used to study effector recognition. We 235 demonstrated that purified SIX4 (Avr1) protein infiltrated into cotyledons caused cell death in 236 a tomato cultivar that contained the I-resistance gene (M82). Importantly, cell death was not 237 observed when the same protein was infiltrated into a tomato cultivar lacking I (Moneymaker) 238 (Fig. 5) . This demonstrates the capacity for E. coli-produced SIX4 (Avr1) to be recognised by 239 the I resistance protein in the native tomato system. 240 241

Here, we demonstrate that the CyDisCo co-expression strategy has the capacity to significantly 243 increase the yield of functional disulfide-rich effectors when produced in SHuffle E. coli. Of 244 the eight effectors we trialled, all could be expressed and purified using CyDisCo and seven 245 displaying improved yields and purity compared to SHuffle alone. Our tailored FunCyDisCo 246 outperformed SHuffle alone for the three F. oxysporum effectors studied, but showed effector-247 specific differences compared to CyDisCo. 248

The basis of the CyDisCo co-expression approach is to mimic (albeit loosely) 249 eukaryotic secretory pathways within a prokaryotic host. PDI and sulfhydryl oxidases proteins 250 are known to function together to assist protein folding through disulfide-bond formation and 251 correct pairing of disulfide bonds (Sevier 2010). There is some evidence that these proteins are 252 also important in pathogenic fungi. For example, PDI1 from Ustilago maydis is crucial for the 253 correct folding of a pool of secreted disulfide-rich proteins important for virulence (Marin-254

FunCyDisCo co-expression, using Fol PDI and Erv2 isoforms identified in RNAseq data from 256

Fol-infected tomato. Our data for FunCyDisCo showed mixed success compared to CyDisCo. 257

One potential reason for this variation is isoform specificity. In Saccharomyces cerevisiae, 258 there are more than five PDI-like proteins and two sulfhydryl oxidases localised to the 259 endoplasmic reticulum, each preferentially aiding the disulfide-bond formation of different 260 proteins (Frand and Kaiser 1999; Norgaard et al. 2001; Sevier and Kaiser 2006) . In Fol, four 261 PDI and two Erv2-like homologs can be identified. While RNAseq data from host infection 262 were used to guide our selection, it is plausible other homologs or combinations could result in 263 better effector yields. Table S2 ). All genes were cloned into the modified, Golden Gate-compatible, pOPIN 300 expression vector (Bentham et al. 2021) . The final expression constructs contained either a N-301 terminal 6xHis-tag or 6xHis-GB1-tag followed by a 3C protease recognition site. The Golden 302

Gate digestion/ligation reactions and cycling were carried out as described by Iverson et al. 303 (2016) . 304 DNA sequences that encode the yeast Erv1p and human PDI (CyDisCo) and Fol Erv2 305

Fol PDI (FunCyDisCo) were codon optimised using the tool provided by IDT, and synthesised 11 by Twist Bioscience (San Francisco, California, USA) (Supplementary Table S2 ). The Yeast 307

Erv1p and Human PDI pair, and Fol Erv2 and Fol PDI pair were cloned into a modified Golden 308 

Black arrows indicate 598 overexpression of PDI or sulfhydryl oxidase components

SDS-PAGE analysis of the Fusarium oxysporum effectors (D) FolSIX6, (E) FovSIX6, (F)

Red arrows indicate the protein of interest 602 peak on the size-exclusion chromatogram and band on SDS-PAGE gel

*FOXG_09255 was incorrectly annotated. The FOXG_09255 sequence has been corrected 604 based on RNASeq data and can be found in Supplementary Table S2

Escherichia coli-produced SIX4 (Avr1) causes cell death when infiltrated into 607 tomato cultivars containing the resistance gene I

μg/mL) and a buffer control were syringe-infiltrated into 10-day old tomato cotyledons from 609

cultivars M82 (containing I) and Moneymaker (lacking I). Cotyledons were harvested and 610 imaged 4 days post infiltration

Tomato seeds were sown in seed raising mix and grown in a controlled environment chamber 375 with a 16-h day/8-h night cycle at 22°C. Purified SIX4 (Avr1) protein was diluted in water to 376 0.1 mg/mL. Syringe infiltrations of the cotyledons of 10-d old tomato seedlings were conducted 377 with 100 µl of protein or buffer (10 mM HEPES pH 8, 150 mM NaCl diluted 1/100). 378Cotyledons were harvested and imaged at 4 days post-infiltration (dpi). bioRxiv:2021 bioRxiv: .2008 bioRxiv: .2010 Blondeau, K., Blaise, F., Graille, M., Kale, S.D., Linglin, J., Ollivier, B., Labarde, A., Lazar, 400 N., Daverdin, G., Balesdent, M.H., Choi, D.H., Tyler, B.M., Rouxel, T., van Tilbeurgh, 401 H., and Fudal, I. 2015 the cytoplasm is altered to be more oxidising. 1: The cytoplasm of SHuffle has a lower GSH: 537 GSSG ratio due to the Gor knockout weakening the reduction pathway. 2: TrxB knockout 538 prevents the reduction of Trx1, which is usually maintained to reset the redox state of DsbD. 539In conjunction with a weaker reduction pathway, the higher proportion of oxidised Trx1 540 strengthens the oxidation pathway and newly translated proteins can be oxidised in the 541 cytoplasm. 3: Newly oxidised proteins in the cytoplasm may be incorrectly disulfide-bonded.