key: cord-1009626-ft7juqgo
authors: Kiefer, Anna; Niemeyer, Justus; Probst, Anna; Erkel, Gerhard; Schroda, Michael
title: Production and secretion of functional full-length SARS-CoV-2 spike protein in Chlamydomonas reinhardtii
date: 2021-12-14
journal: bioRxiv
DOI: 10.1101/2021.12.13.472433
sha: 6251fdfc15e110636dcab4a652f1b5f8bfbd0861
doc_id: 1009626
cord_uid: ft7juqgo

The spike protein is the major protein on the surface of coronaviruses. It is therefore the prominent target of neutralizing antibodies and consequently the antigen of all currently admitted vaccines against SARS-CoV-2. Since it is a 1273-amino acids glycoprotein with 22 N-linked glycans, the production of functional, full-length spike protein was limited to mammalian and insect cells, requiring complex culture media. Here we report the production of full-length SARS-CoV-2 spike protein – lacking the C-terminal membrane anchor – as a secreted protein in the prefusion-stabilized conformation in the unicellular green alga Chlamydomonas reinhardtii. We show that the spike protein is efficiently cleaved at the furin cleavage site during synthesis in the alga and that cleavage is abolished upon mutation of the multi-basic cleavage site. We could enrich the spike protein from culture medium by ammonium sulfate precipitation and demonstrate its functionality based on its interaction with recombinant ACE2 and ACE2 expressed on human 293T cells. Chlamydomonas reinhardtii is a GRAS organism that can be cultivated at low cost in simple media at a large scale, making it an attractive production platform for recombinant spike protein and other biopharmaceuticals in low-income countries.

In late 2019, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was 32 identified as the causative agent for the coronavirus disease 19 (COVID-19) 1 . Since 33 then, the virus is spreading throughout the world, causing not only humanitarian but 34 also severe economic crisis. One of the most important proteins for the pathogenicity 35 of the SARS-CoV-2 virus is the trimeric spike protein on its surface 2-4 . This 1273-amino 36 acids protein contains 22 N-linked glycans and only traces of O-linked glycans 5 . Each 37 spike protein monomer is built of two main subunits S1 and S2, with S1 mediating 38 receptor binding and S2 membrane anchoring [6] [7] [8] . Cleavage between the S1 and S2 39 subunits occurs during spike protein synthesis in the Golgi apparatus by a furin-like 40 protease at a multi-basic cleavage site, leaving the S1 and S2 subunits linked by non-41 covalent interactions 8, 9 . The S1 subunit contains the receptor binding domain (RBD), 42 which binds to the cellular surface receptor angiotensin-converting enzyme 2 (ACE2) 43 10-12 . Binding to ACE2 requires at least one RBD in the spike protein trimer to be in the 44 "up" conformation. Accordingly, roughly half of the ~25 spike protein trimers present 45 expose the S2' sites located immediately upstream of the fusion peptide, leading to 48 cleavage by the transmembrane protease serine 2 (TMPRSS2) 13, 14 . S2' sites can also 49 be cleaved by cathepsin L in the endosome after endocytosis of virions 15 . Cleavage 50 at S2' leads to the shedding of S1 subunits and triggers a cascade of folding events 51 resembling a "jackknife mechanism", during which the fusion peptides are inserted into 52 the target cell membrane. Further folding bends the viral and host membranes toward 53 each other, leading to membrane fusion and viral entry 8, 16 . 54

As the major protein on the virion surface, the spike protein is the prominent target of 55 neutralizing antibodies and, therefore, all currently admitted vaccines use the spike 56 protein as antigen [16] [17] [18] [19] . While mRNA-and vector-based vaccines have proven efficient 57 and fast to establish, they do have side effects including heart inflammation and blood 58 clots. Moreover, they are difficult to produce and to handle in lower-income countries 59 17 . In contrast, recombinant protein based vaccines have less side effects and are 60 easier to produce once their production platform is established 20 . While potent 61 neutralizing antibodies bind to the RBD, the RBD lacks other neutralizing epitopes 62 present on the full-length spike protein [16] [17] [18] . Hence, it appears advisable to produce 63 the prefusion-stabilized, full-length spike protein for protein-based vaccines and for 64 serological tests. Accordingly, a phase III clinical trial for the Novavax NVX-CoV2373 65 vaccine based on recombinant prefusion-stabilized full-length spike protein trimers has 66 been completed recently with the vaccine showing 90% protection against COVID-19 67 21 . Because of its size and its 22 N-linked glycans, the production of full-length spike 68 protein has only been achieved in mammalian and insect cells 5-7, 10, 21-23 , while yeast, 69 plant, and microalgal expression hosts have been limited to the production of spike 70 protein fragments, mostly the RBD 24-32 . Moreover, in the plant and microalgal 71 expression hosts, the RBD was expressed as a cytosolic or ER-resident protein, 72 requiring purification from whole cells [24] [25] [26] [27] . The development of simple production 73 platforms for the secretion of complex therapeutic glycoproteins such as the full-length 74 spike protein is desirable to facilitate their production in low-income countries. 75

Microalgae appear to be a good choice here, as they can be grown at large scale in 76 very simple media and do not produce toxic compounds 33 . In general, glycosylation is 77 essential for protein folding, stability, and functionality and is the most eminent post-78 translational modification in biopharmaceuticals 34 . 79 is 137 kDa. However, no signal was obtained for proteins from the culture medium. We 113 therefore conclude that Chlamydomonas can produce the full-length recombinant 114 spike protein but the endogenous secretion peptide cannot target the protein to the 115 secretory pathway in Chlamydomonas. 116

To enable secretion of the spike protein, the MoClo level 1 module was reconstructed 117 by adding the coding regions for three different N-terminal secretion signals of native 118 extracellular Chlamydomonas proteins, namely carbonic anhydrase (cCA) 41 , 119 arylsulfatase 2 (ARS2) 42 and gamete lytic enzyme (GLE) 43 . In the following, the 120 constructs are named by the secretion signal (cCA-S, ARS2-S, and GLE-S). 121

Additionally, the coding region for a C-terminal synthetic glycomodule of 20 122 serine/proline repeats was added, which was reported to enhance secretion yields by 123 up to 12-fold in Chlamydomonas 43 . To increase the detectability of the spike protein, 124 the glycomodule was equipped with a triple HA motif (SP20 3xHA). As before, the 125 resulting level 1 modules were assembled into level 2 devices with the aadA cassette. 126

Proteins in the culture medium of various spectinomycin resistant transformants were 127 precipitated and analyzed by immunoblotting. We noticed that the chemiluminescence 128 signal of secreted spike protein was much stronger in transformants generated with 129 the cCA-S and ARS2-S constructs when compared with transformants generated with 130 GLE-S ( Figure 2C ), pointing to a more efficient secretion via the cCA and ARS 131 secretion peptides. Moreover, the HA antibody specifically detected three protein 132 bands in the culture medium of cCA-S and ARS2-S transformants with apparent 133 molecular masses of ~240, ~120 and ~90 kDa. The calculated molecular mass of the 134 mature spike protein lacking the secretion signal is 142 kDa, which does not fit to any 135 of the detected bands. It has been shown that the synthetic SP20 glycomodule is 136 efficiently glycosylated during its passage through the secretory pathway in 137 Chlamydomonas 43 . Moreover, the spike protein contains 22 N-linked glycans 5 . 138

Hence, it appears likely that glycosylation of the SP20 module and the spike protein 139 itself takes place and that the protein migrating at ~240 kDa represents the 140 glycosylated full-length protein. The latter appears to be efficiently processed, giving 141 rise to two C-terminal fragments, accounting for the weak signal for the full-length ~240 142 kDa protein and the strong signals at ~120 and ~90 kDa. In the weakly expressing 143 GLE-S transformants, the full-length protein appears to be below the detection limit 144 and only the two C-terminal cleavage products are detected ( Figure 2C ). 145 During its synthesis, the SARS-CoV-2 spike protein is cleaved into two functional 146 subunits S1 and S2 by a furin-like protease in the Golgi apparatus 8, 9 ( Figure 1A) . 147

Cleavage at the S1/S2 boundary is strongly enhanced in the SARS-CoV-2 spike 148 protein by the presence of a "RRAR" furin cleavage site that is missing in the SARS-149

CoV spike protein 6, 7, 9, 10, 44 . We wondered, whether processing of the SARS-CoV-2 150 spike protein produced in Chlamydomonas was also mediated by a furin-like protease. 151

To test this, we reconstructed the level 0 parts of the coding sequence. At first, we 152 removed sequences coding for the N-terminal endogenous signal peptide, since it was 153 unable to mediate secretion in Chlamydomonas ( Figure 1C ). Next, we exchanged the 154 sequences encoding the furin cleavage site (S682RRAR↓S685) via site directed 155 mutagenesis by nucleotides coding for S682GSAS S685, which is supposed to 156 completely inactivate the furin cleavage site 44 . Furthermore, we substituted codons for 157 K986V987 situated at the beginning of the central helix by two codons for proline, which 158 was reported to increase expression yields and to stabilize the prefusion conformation 159 of coronavirus spike proteins resulting in higher immunogenicity 7, 45, 46 . Finally, we 160 equipped this new variant of the CoV-2 spike protein with a SP20 glycomodule carrying 161 a single HA tag and an octa-histidine tag. We employed only the cCA secretion signal, 162 as it proved to be efficient (construct S-GSAS/PP in Figure 1A and B). Immunoblot 163 analyses of proteins precipitated from the culture medium of a S-GSAS/PP 164 transformant with the HA antibody revealed a prominent protein band at ~240 kDa and 165 very minor ones at ~120 kDa and ~90 kDa ( Figure 1C ). Hence, eliminating the furin 166 cleavage site appears to almost completely abolish proteolytic cleavage of the spike 167 protein during its passage through the secretory pathway, as has been observed in 168 human cells 6, 10, 44 . 169

To estimate the amounts of full-length spike protein secreted into the culture medium 170 by the S-GSAS/PP transformant, we performed quantitative immunoblotting. To this 171 end, proteins in the medium of S-GSAS/PP cultures grown for five days after 172 inoculation were precipitated and separated on an SDS-gel next to known amounts of 173 recombinant mCherry carrying a C-terminal HA tag ( Figure 1D ). Immunodetection with 174 the HA antibody and quantification of the signals revealed an average concentration of 175 full-length spike protein of 11.2 ± 1.8 µg/L (n = 3, ± SD). 176

After the successful production of full-length SARS-CoV-2 spike protein in 179 Chlamydomonas, we aimed to purify the secreted protein from culture medium of a S-180 GSAS/PP transformant via immobilized metal affinity chromatography, taking 181

advantage of the C-terminal octa-histidine tag. However, the spike protein failed to bind 182 to the nickel resin and largely remained in the flow through ( Figure 2A ). The same 183 result was obtained when the medium was supplied with 2% Triton or 250 mM NaCl, 184 with longer incubation times, or when we used cobalt beads instead of nickel beads. 185

An affinity purification with immobilized HA antibodies also failed (not shown). To enrich 186 the protein by other means, we concentrated proteins in the culture medium with a 187 rotary evaporator, followed by two runs through centrifugal filters. Although an 188 enrichment of the spike protein was achieved, about half of it was found in aggregates 189 ( Figure 2B ). As an alternative, we precipitated proteins in the culture medium with 190 ammonium sulfate and could recover virtually all of the spike protein, with very little 191 found in aggregates ( Figure 2B ). Hence, ammonium sulfate precipitation appears most 192 suitable for enrichment. 193

Next, we wanted to test the functionality of the spike protein produced in 195

Chlamydomonas. The ability of the protein to bind to the human ACE2 receptor was 196 described previously 10-12 and binding to ACE2 is an accepted assay to verify the 197 functionality of recombinant spike protein 29 . We incubated proteins enriched by 198 ammonium sulfate precipitation from a S-GSAS/PP culture with or without recombinant 199 hACE2 containing a C-terminal deca-histidine tag. Nickel resin was added and bound 200 proteins eluted. As shown in Figure 3A , full-length spike protein co-eluted with hACE2. 201

No spike protein was eluted if hACE2 was absent. We noticed that less spike protein 202 was recovered with hACE2 than supplied in the input, indicating that only part of the 203 supplied spike protein bound to the hACE2 receptor. Moreover, unbound spike protein 204 in the supernatant was almost completely converted to the ~90-kDa breakdown 205 product, presumably during the 30-min incubation at 37°C. As a control, we produced 206 the SARS-CoV-2 RBD fused to sfGFP in transiently transfected human 293T cells. 207

Secreted proteins were concentrated with centrifugal filters and incubated with 208 recombinant hACE2. As expected, we observed the RBD-sfGFP fusion protein to co-209 elute with hACE2 ( Figure 3B ). Hence, the in vitro pull-down assay works equally well 210 with human-derived RBD and Chlamydomonas-derived full-length spike protein. 211

To complement the in vitro binding assay, we performed a second functionality assay 212 based on the binding of spike protein to hACE2 on human cells. For this, we 213 concentrated secreted proteins from culture medium of a S-GSAS/PP transformant by 214 ammonium sulfate precipitation. Concentrated proteins were then applied to 293T cells 215 constitutively overexpressing hACE2 and hTMPRSS2 (293T+AT) and as a control to 216 regular 293T cells. Importantly, we did not observe any cytotoxic or other negative 217 effects of this treatment on the cells by microscopy. After washing, cells were lysed 218 and proteins in the lysates analysed by SDS-PAGE and immunoblotting. As shown in 219 Figure Here we report on the construction of Chlamydomonas strains that secrete functional 230 full-length SARS-CoV-2 spike protein into the medium. To our knowledge, this has only 231 been achieved with mammalian and insect cell-based expression systems. Recent 232 reports on the expression of SARS-CoV2 spike protein in Chlamydomonas were 233 limited to intracellularly expressed RBD 24, 26 . The advantage of Chlamydomonas is 234 that it is a GRAS organism that can be cultivated at low cost in simple media at a large 235 scale, making Chlamydomonas an attractive production platform for human 236 therapeutical proteins in low-income countries 33, 47 . 237

During our endeavor to produce the spike protein in Chlamydomonas, we have learned 238 that its native signal peptide does not guide the protein to the secretory pathway in 239

Chlamydomonas. For this, a signal peptide of a native Chlamydomonas protein was 240 required, with those from cCA and ARS2 being more efficient than that of GLE ( Figure  241 1C). We observed that during its passage through the secretory pathway the spike 242 protein is processed at the multi-basic furin cleavage site, just like in mammalian cells reported for other human therapeutic proteins, but far below those reported for the 296 simpler reporter proteins. A potential factor limiting yield might be the efficiency of N-297 glycosylation and associated protein folding, which must be a challenge for the spike 298 protein given its large size and 22 N-linked glycans 5 . Accordingly, the ER chaperone 299

BiP was found as a major contaminant of spike protein preparations 8 . Therefore, 300 Cocktail, Roche). Cell debris was removed by centrifugation (8,000 g, 10 min, 4 °C). 499

For immunoblot analysis, the lysates were separated on an 8% SDS-polyacrylamide 500 gel and transferred to a nitrocellulose membrane. The membranes were blocked and 501 incubated with specific antibodies: HA-tagged spike protein was detected with anti-HA 502 antibody (H9658, Sigma-Aldrich, 1:10,000), hACE2 was detected with anti-ACE2 503 antibody (sc-390851, Santa Cruz Biotechnology, 1:2,000), RBD was detected using 504 anti-RBD antibody (MAB105401, R&D Systems, 1:2,000). m-IgGκ BP-HRP was used 505 as secondary antibody for detection (sc-516102, Santa Cruz Biotechnology, 1:10,000). 506

We thank Benedikt Venn for helpful comments to the manuscript. 508 

Sec S1/S2 S2' S1 subunit S2 subunit 500 bp 

A novel coronavirus from patients with pneumonia in China

Molecular architecture of the SARS-CoV-2 virus

Structures and distributions of SARS-CoV-2 spike proteins on intact 581 virions

In situ structural analysis of SARS-CoV-2 spike reveals 583 flexibility mediated by three hinges

Site-specific 585 glycan analysis of the SARS-CoV-2 spike

Structure, function, and antigenicity of the SARS-CoV-2 spike 587 glycoprotein

Cryo-EM structure of the 2019-nCoV spike in the prefusion 589 conformation

Distinct conformational states of SARS-CoV-2 spike protein

A multibasic cleavage site in 593 the spike protein of SARS-CoV-2 is essential for infection of human lung cells

SARS-CoV-2 cell entry depends on ACE2 and TMPRSS2 596 and is blocked by a clinically proven protease inhibitor

Structural basis of receptor recognition by SARS-CoV-2

Functional assessment of cell entry and 601 receptor usage for SARS-CoV-2 and other lineage B betacoronaviruses

TMPRSS2 and furin are both essential for proteolytic activation 604 of SARS-CoV-2 in human airway cells

Efficient activation of the severe acute respiratory 606 syndrome coronavirus spike protein by the transmembrane protease 607 TMPRSS2

Inhibitors of cathepsin L prevent severe acute respiratory 609 syndrome coronavirus entry

Mechanisms of SARS-CoV-2 612 entry into cells

A comprehensive review of the global efforts on COVID-19 vaccine 614 development

SARS-CoV-2 vaccines in development

SARS-CoV-2 vaccines based on the spike 617 glycoprotein and implications of new viral variants

How protein-based COVID vaccines could change the pandemic

Efficacy and safety of NVX-CoV2373 in adults in the United 622 States and Mexico. medRxiv

A recombinant baculovirus-expressed S glycoprotein vaccine 624 elicits high titers of SARS-associated coronavirus (SARS-CoV) neutralizing 625 antibodies in mice

Structural analysis of full-length SARS-CoV-2 spike protein 627 from an advanced vaccine candidate

Recombinant production of a functional SARS-CoV-2 spike 629 receptor binding domain in the green algae Chlamydomonas reinhardtii

SARS-CoV-2 antigens expressed in plants detect antibody 632 responses in COVID-19 patients

634 Efficient transient expression of recombinant proteins using DNA viral vectors 635 in freshwater microalgal species

Rapid production of SARS-CoV-2 receptor binding domain 637 (RBD) and spike specific monoclonal antibody CR3022 in Nicotiana 638 benthamiana

Yeast-expressed SARS-CoV recombinant receptor-binding 640 domain (RBD219-N1) formulated with aluminum hydroxide induces protective 641 immunity and reduces immune enhancement

Expression of SARS-coronavirus spike glycoprotein in Pichia 643 pastoris

Engineered SARS-CoV-2 receptor binding domain improves 645 manufacturability in yeast and immunogenicity in mice

Structural and functional comparison of SARS-CoV-2-648 spike receptor binding domain produced in Pichia pastoris and mammalian 649 cells

Yeast-produced RBD-based recombinant protein vaccines elicit 651 broadly neutralizing antibodies and durable protective immunity against SARS-652

CoV-2 infection

Current advances in the algae-made 655 biopharmaceuticals field

The sweet tooth of 657 biopharmaceuticals: importance of recombinant protein glycosylation analysis

Intron-containing algal 660 transgenes mediate efficient recombinant gene expression in the green 661 microalga Chlamydomonas reinhardtii

Good News for Nuclear Transgene Expression in 664 Chlamydomonas

Birth of a photosynthetic chassis: a MoClo toolkit enabling 666

Synthetic Biology in the microalga Chlamydomonas reinhardtii

A modular 669 cloning system for standardized assembly of multigene constructs

An epigenetic gene silencing pathway selectively acting on 672 transgenic DNA in the green alga Chlamydomonas

Generation of Chlamydomonas strains that 675 efficiently express nuclear transgenes

Efficient recombinant 677 protein production and secretion from nuclear transgenes in Chlamydomonas 678 reinhardtii

Strategies to 680 facilitate transgene expression in Chlamydomonas reinhardtii

High-yield secretion of 683 recombinant proteins from the microalga Chlamydomonas reinhardtii

Furin cleavage of SARS-CoV-2 Spike promotes but is not 686 essential for infection and cell-cell fusion

Immunogenicity and structures of a rationally designed 688 prefusion MERS-CoV spike antigen

Stabilized coronavirus spikes are resistant to 691 conformational changes induced by receptor recognition or proteolysis

Green biologics: The algal chloroplast as 694 a platform for making biopharmaceuticals

N-and O-glycosylation of the SARS-696

CoV-2 spike protein

The viral spike protein is not involved in the polarized 698 sorting of coronaviruses in epithelial cells

Towards autotrophic tissue engineering: Photosynthetic 700 gene therapy for regeneration

Engineered fusion proteins for efficient protein secretion and 702 purification of a human growth factor from the green microalga Chlamydomonas 703 reinhardtii

O-glycosylation pathways in the microalgae polyphyletic group. 706 Frontiers in plant science

Absolute 708 quantification of major photosynthetic protein complexes in Chlamydomonas 709 reinhardtii using quantification concatamers (QconCATs). Frontiers in plant 710 science

Targeted expression of nuclear 712 transgenes in Chlamydomonas reinhardtii with a versatile, modular vector 713 toolkit

Investigating the dynamics of recombinant protein 715 secretion from a microalgal host

The bZIP1 transcription factor regulates lipid remodeling and 717 contributes to ER stress management in Chlamydomonas reinhardtii

Protein folding and conformational stress in microbial cells 720 producing recombinant proteins: a host comparative overview

High cell 723 density cultivation enables efficient and sustainable recombinant polyamine 724 production in the microalga Chlamydomonas reinhardtii

A revised mineral nutrient supplement increases biomass and 727 growth rate in Chlamydomonas reinhardtii

Real-729 time monitoring of subcellular H2O2 distribution in Chlamydomonas reinhardtii

Engineering human ACE2 to optimize binding to the spike 732 protein of SARS coronavirus 2

Structural basis for VIPP1 oligomerization and maintenance 734 of thylakoid membrane integrity

A multicolored set of in vivo organelle 736 markers for co-localization studies in Arabidopsis and other plants. The Plant 737

Determination of accurate 739 extinction coefficients and simultaneous-equations for assaying chlorophyll-a 740 and chlorophyll-b extracted with 4 different solvents -verification of the 741 concentration of chlorophyll standards by atomic-absorption spectroscopy