Engineering mesenchymal stem cells with neutralizing and anti- inammation dual-capability against SARS-CoV-2 infection

The emergence of the novel human coronavirus SARS-CoV-2 has caused a worldwide epidemic of coronavirus disease 2019 (COVID-19), which markedly affected the global health and economy. Both uncontrolled viral replication and proinammatory cytokine storm can cause severe tissue damage in COVID-19 patients. SARS-CoV-2 utilizes angiotensin-converting enzyme 2 (ACE2) as its entry receptor. In this study, we generated ACE2 extracellular domain-Fc and scFv-IL6R-Fc fusion protein to differentially neutralize virus and temper cytokine storm. The hACE21-740-Fc fusion protein showed potent inhibitory effect on pseudotyped SARS-CoV-2 entry and good safety prole in mice. The scFv-IL6R-Fc showed strong blockade effect on IL-6 signal activation. In addition, we established a mesenchymal stem cells (MSCs)-based hACE21-740-Fc and scFv-IL6R-Fc delivery strategy, which provided a potential rapid option for urgent clinic therapeutic need of COVID-19 patients. Pharmacokinetics were determined in C57BL/6 mice after a single injection of 100 μg of ACE21–740-hFc. Serum concentrations at indicated time points were determined by ELISA with immobilized RBD-mFc. C57BL/6 mice received a single injection of 100 μg ACE21–740-hFc. Seven days later, the hematology, body weight, and H&E staining of indicated organ were analyzed. H&E staining was performed by Servicebio (Wuhan, China).

Phylogenetic analysis revealed that it is approximately 80% similar to SARS-CoV in the genome sequence. Speci cally, these two viruses exhibit ~74% amino acid sequence similarity in the receptor binding domain (RBD) of spike gene, suggesting a possible common host receptor. Indeed, angiotensin converting enzyme 2 (ACE2) has been recognized as the critical entry receptor of both SARS-CoV-2 and SARS-CoV 7,8. Compared to the spike of SARS-CoV, the higher binding a nity of SARS-CoV-2 spike to hACE2 may contribute to the higher transmission rate of SARS-CoV-29,10, implying the importance of ACE2-dependent viral entry into cells. ACE2 is primarily expressed in lung alveolar epithelial type II cells, which are crucial for the gas exchange function of lung11,12. Severe lung injure in some COVID-19 patients may stem from dysfunction of these cells13 ACE2 is also expressed in multiple other organs, such as heart, liver, intestine, testis, kidney and blood vessels14-18, which may give rise to the multiple organ dysfunction syndrome (MODS) induced by SARS-CoV-2 infection6. Apart from being the entry receptor of SARS-CoV, ACE2 was shown to facilitate tissue repair after lung injury19. Most importantly, soluble recombinant ACE2 protein has exhibited therapeutic potential for the SARS-CoV20.
About 5% of patients infected with SARS-CoV-2 have developed a sever lung injury symptom, which is accompanied with overwhelmed immune activation. Some anti-virus and in ammation blocking drugs have been tested in clinic for treating COVID-1921. However, these single targeting drugs showed minimal clinical effect. Drugs targeting both virus itself and tissue damage are urgently needed for treating severe COVID-19.
The multi-potent differential potential, the tissue repair ability and the anti-in ammatory property of MSCs potentiate their clinical development for the treatment of autoimmune disease22,23. Besides, MSCs exert the paracrine functions through secreting soluble factors and releasing extracellular vesicles, and possess the innate ability to tra c to the in ammation sites, which made them a promising drug delivery vector in preclinical models24. Here we represent a MSC-based cell therapy to simultaneously block SARS-CoV-2 entry and inhibit virus induced cytokine storm. This dual targeting strategy provided a rapid clinic-ready, multiple layers of resolution to COVID-19 therapy.

ACE2-Fc bound to SARS-CoV-2 spike with high a nity
It is reported that SARS-CoV-2 enters into permissive cells through ACE27,25 Competitive soluble extracellular domains of ACE2 thus might be potential viral entry blocker. To test whether ACE2 extracellular domain could block SARS-CoV-2 entry, two recombinant ACE2-Fc fusion proteins were generated, consisting of the amino acid 1-615 or 1-740 of ACE2 fused with human IgG1 Fc portion respectively. We then characterized their binding pro le with SARS-CoV-2 spike protein. By the similar strategy, another two recombinant SARS-CoV-2 spike-Fc fusion proteins were generated as well, containing amino acid 1-685(S1) or amino acid 319-541(RBD) fused to mouse IgG2a Fc portion ( Figure   1A). Through an ELISA binding assay, both ACE21-615-hFc and ACE21-740-hFc bound to S1-mFc and RBD-mFc with similar a nity ( Figure 1C&D). These data suggested that S1 and RBD region are essential for the binding of ACE2 and SARS-CoV-2 spike. To test the binding ability of ACE2-hFc fusion protein to SARS-CoV-2 spike in a more physiological condition, we established the 293-spike cell line, which stably expressed SARS-CoV-2 spike protein. By ow cytometry analysis, both ACE21-615-hFc and ACE21-740-hFc bound to 293-spike with similar a nity ( Figure 1E). These data collectively suggested that the extracellular domain of ACE2-Fc fusion protein can bind to SARS-CoV-2 spike protein with high a nity and could be used as a potential entry blocker.
ACE2-Fc blocked the entry of pseudotyped SARS-CoV-2 virus into ACE2+ cells Due to the potential risk of severe COVID-19 induced by SARS-CoV-2 infection, researches on SARS-CoV-2 must be conducted in laboratories with a biosafety level (BSL) of 3 or higher, which becomes a hurdle to SARS-CoV-2 research. To overcome this limitation, we have established a pseudotyped SARS-CoV-2 virus system, which replaced the VSV-G with SARS-CoV-2 spike to produce a SARS-CoV-2 pseudotyped virus utilizing the well-de ned 3rd generation of lentivirus system (Figure 2A). We rst checked the infection ability of SARS-CoV-2 pseudotyped virus in both ACE2+ and ACE2-293 cells. SARS-CoV-2 pseudotyped virus can only infect ACE2+ 293 cells, rather than ACE-293 cells, suggesting a SARS-CoV-2 spike-mediated, ACE2-dependent entry process ( Figure 2B). By using this pseudotyped virus system, we next tested the blocking e cacy of ACE21-615-Fc and ACE21-740-Fc. Indeed, ACE21-740-Fc showed a strong blockade effect by inhibiting the entry of SARS-CoV2 pseudotyped virus into 293-ACE2 cell with an IC50 of ~ 0.68 nM ( Figure 2C). To our surprise, the ACE21-615-Fc showed very low blocking effect on the entry of SARS-CoV-2 pseudotyped virus despite its similar binding pro le to S1-mFc, RBD-mFc and 293-spike cells ( Figure 2D & Figure 1C-E). To further test the speci city of ACE21-740-hFc-mediated blockade of pseudotyped SARS-CoV-2 virus, we investigated the blocking effect of ACE21-740-hFc on the entry of VSV pseudotyped virus. Although ACE21-740-Fc inhibited ~80% infection of SARS-CoV-2 pseudotyped virus, it had no effect on the infection of VSV pseudotyped virus ( Figure 2E). To test whether ACE21-740-Fc -mediated blockade could be applicable to other types of cell, we performed similar assay on lung epithelial A549-ACE2 cell line. Consistent with the data from 293- Safety pro le of ACE21-740-hFc in vivo ACE2 is a carboxypeptidase that potently degrades angiotensin II to angiotensin 1-728. ACE2 antagonizes the activation of the classical RAS system and protects against organ damage, hypertension, diabetes, and cardiovascular disease29. However, whether exogenous exposure of ACE21-740-hFc could affect the normal physiology status is not clear. Firstly, we tested the in vivo half-life of hACE21-740-hFc in mice. The half-life of intact hACE21-740-hFc in vivo is about 8.22 hours ( Figure 4A).
Subsequently, different tissues were collected to investigate the potential effect of ACE21-740-hFc infusion on tissue damage. There were no difference in hematological feature, body weight change, ALT and immune cell in ltration in the organs of heart, liver, spleen, lung, and kidney ( Figure 4B-E). Although the enzyme activity of hACE21-740-hFc functions well in mice30, two potential factors may affect the evaluation of side effects, namely, potential immune response against human protein hACE2 in mice and different interaction network of human ACE2 compared with mouse ACE2. To avoid these limitations, we performed similar experiments by using mACE21-740-hFc and did not observe adverse effects in seven days after treatment ( Figure 4B-E). Collectively, ACE21-740-hFc was well tolerated in mice without obvious side effect.
Engineered MSCs as a drug development strategy for urgent need of COVID-19 treatment It is time consuming for protein-based therapeutics to enter the clinic due to good manufacturing practices (GMP) production, and quality control, which cannot meet the urgent need for treating SARS-CoV-2 induced COVID-19. To this end, other drug delivery systems should be considered to save the time on development. Mesenchymal stem cells have been widely used for treating autoimmune disease and as the vector for protein drug delivery, which could be a good option. We then generated the MSC cell line stably expressing ACE21-740-hFc (MSC-ACE21-740-hFc) by lentiviral infection. As expected, this cell line secreted ACE21-740-hFc into supernatant persistently ( Figure 5A). The concentration of ACE21-740-hFc in supernatant is about 190ng/ml after a 4-day cell culture. To recognize whether ACE21-740-hFc in MSC-derived supernatant is bio-reactive, we rst measured its binding ability to spike positive 293spike cells. Actually, the MSC-secreted ACE21-740-hFc showed similar binding strength to the puri ed ACE21-740-hFc ( Figure 5B). Since it has the equivalent binding ability, we next sought to identify whether ACE21-740-hFc in MSC-derived supernatant could block pseudotyped virus entry. Indeed, the MSCderived supernatants showed signi cant blockade effect on the entry of pseudotyped SARS-CoV-2 virus into 293-ACE2 cells ( Figure 5C). To test the possibility of using MSC-ACE2-1-740-hFc as a live drug in vivo, we then injected these engineered MSCs into mice. Interestingly, the expression of ACE2-Fc in serum lasted for more than 14 days after one single infusion Figure 5D . Most importantly, when we checked its bio-distribution, the MSC preferentially located in lung at 24 hours after injection Figure 5E), suggesting a potential usage for treating SARS-CoV-2-induced pneumonia. Therefore, MSC-based ACE21-740-Fc delivery strategy provided a time-saving option for treating clinical SARS-CoV-2-induced disease.
MSC expressing scFv-IL6R-Fc as a potential strategy to inhibit SARS-CoV-2-induced immune damage Proin ammatory cytokine storm can cause tissue damage in multiple disease models. Blocking of proin ammatory cytokine such as IL-6 has shown therapeutic effect in autoimmune disease and CAR-T cell-induced cytokine storm31,32. For example, Tocilizumab has been used for treating COVID-19 and improved patient outcome33,34. Considering that MSC is capable of producing proteins rapidly and preferentially migrating to in ammatory tissues, we aim to generate scFv-IL6R-Fc expressing MSC as a treatment option for SARS-CoV-2-induced cytokine storm. To accelerate this process, we rst generated scFv-IL6R-Fc protein. We puri ed it and compared its activity with Tocilizumab. The scFv-IL6R-Fc showed similar blockade effect on IL-6-induced TF-1 cell proliferation ( Figure 6A). We then established the MSC cell line stably expressing scFv-IL-6R-Fc. After one-month culture, this cell line was still capable of secreting scFv-IL6R-Fc. The concentration of scFv-IL6R-Fc in supernatant was about 300ng/ml after a 4day cell culture ( Figure 6B). To test the persistence of MSC-derived scFv-IL6R-Fc in vivo, we then injected these engineered scFv-IL6R-Fc expressing cell into mice. Interestingly, the expression of ACE2-Fc in serum lasted for more than 14 days after one single infusion Figure 6C . Therefore, the scFv-IL-6R-Fcexpressing MSCs can be potentially used for blocking SARS-CoV-2-induced cytokine storm, which can be combined with viral neutralizing MSC-ACE21-740-Fc to provide multiple layers of protection against SARS-CoV-2 infection.

Discussion
As the most crucial global health crisis of the century, the ongoing COVID-19 pandemic has created a much greater social and economic impact than either of its predecessors, SARS outbreak and MERS epidemic. To date, no clinically approved antiviral drugs or vaccines have been proven effective against COVID-19. Although some drugs showed in vitro anti-SARS-CoV activity, the clinical e cacy of these drugs are dismal. A safe and effective treatment for COVID-19 is urgently needed. SARS-CoV-2 enters into the permissive cells mainly through ACE27. Both neutralizing antibodies against SARS-CoV-2 spike and soluble ACE2 receptors are potential drug candidates with the ability to block viral entry 13,35.The good pro le of stability, speci city and manufacture process has made neutralizing antibody a promising candidate for drug development. Theoretically, it only affects virus or virus-infected cells and thus will have limited side effect in vivo. However, with the high mutation rate, virus may escape neutralizing antibody via single mutations. It is supported by the observation that neutralizing antibodies against SARS-CoV have limited blocking effect on SARS-CoV-2, although they are highly similar in the structure of RBD domain. However, there is no such trouble in the soluble ACE2-based neutralization strategy, since mutant virus escaping soluble ACE2 will also lose the binding ability to ACE2 positive cells. There is a limitation for both neutralizing antibody and ACE2-Fc fusion protein: antibody dependent enhancement effect to virus infection. Further experiments need to be performed to clarify these concerns. Due to biosafety issue, our experiment was performed with pseudotyped virus. Although the entry step of pseudotyped virus is similar to live virus, further investigation with live virus and animal model will provide more information on the e cacy of ACE2-Fc based therapy.
ACE2 is a carboxypeptidase that potently degrades angiotensin II to angiotensin 1-728. It plays a central role in the homeostatic control of cardiorenal actions. ACE2 is a membrane protein, its distribution is determined by the expressing cells. While soluble ACE2 will deliver to all possible tissue through system circulation. This raised the concern that whether soluble ACE2 has potential side effect to the cardiorenal system. Our preliminary mouse data showed that there is no obvious side effect being induced after ACE2-Fc treatment. Further investigation on the long-term side effect of ACE2 needs to be conducted on animal models before clinical study. Despite this limitation, there is one potential advantage of soluble ACE2 as a promising therapy. ACE2 was reported to play protective role in lung injury19. SARS-CoV-2 infection induces internalization of ACE2 receptor, which may reduce the ACE2 enzymatic activity and the protective effect on lung injury. Supplement of soluble ACE2 may provide additional tissue repair effect beyond viral neutralization.
To meet the urgent need for treating SARS-CoV-2 induced COVID-19, we have developed a MSC-based cell therapy strategy. MSC has been widely used for treating autoimmune diseases for more than twenty years in clinic36. There are some ongoing clinical trials using MSC to treat COVID-19 due to its innate anti-in ammation and tissue repair ability. Now we armed it with virus neutralization and anti-cytokine storm capability, which will signi cantly boost its therapeutic effect. MSCs preferentially locate in in ammation tissue, providing a targeted delivery of ACE-2 fusion proteins to SARS-CoV-2 infected tissue. This will signi cantly enhance the drug concentration in targeted tissue and lower the potential side effect. However, there is a potential disadvantage for MSC-based therapy that it is challenging to regulate the dosage of soluble ACE2 secreted from cells. Further engineering of MSC with on-off switchable promoter or suicide gene will provide possible solution to overcome these hurdles.
Collectively, our data indicated that ACE2-Fc has the potential to be a therapeutic drug for treating COVID-19 with a relatively good safety pro le. MSC-based viral blockade and in ammation control (Shanghai, China). Plasmids encoding SARS-CoV-2 spike and ACE2 were obtained from MolecularCloud (Nanjing, China) and sub-cloned to pCDH-EF (System Biosciences) lentiviral vector plasmid with puromycin resistance marker. To establish SARS-CoV-2 spike or ACE2 expressing cell lines, Lenti-X 293, A549 or Raji were infected with SARS-CoV-2 spike or ACE2-expressing lentivirus. After selection with puromycin, pooled resistant cells were identi ed by ow cytometry analysis. Cell culture medium were supplemented with 10% heat-inactivated fetal bovine serum, 2 mmol/L L-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin. Lenti-X 293, A549 and their derivatives were cultured with complete DMEM medium. Raji and its derivatives were cultured with complete RPMI medium. Human AB serum were purchased from Gemini Bio-products (West Sacramento, CA).

Production of ACE2 and spike fusion protein
For hACE21-740-hFc hACE21-615-hFc, mACE21-740-hFc, S1-mFc and RBD-mFc fusion protein, DNA sequences encoding the indicated protein (supplemental gure 1) were cloned into pCDH-EF vector (System Biosciences, Mountain View, CA). The plasmids containing indicated fusion protein were transfected into Lenti-X 293 cells and supernatants were collected and puri ed by Diamond Protein A column according to the manufacturer's protocol (Bestchrom, Shanghai, China).
ELISA and Flow cytometry analysis of hACE21-740-hFc and hACE21-615-hFc ELISA plates (Jet Bio l, Guangzhou, China) were coated with 2μg/ml of RBD-mFc at 4 °C overnight. Plates were washed three times with PBS c ontaining 0.05% Tween-20 and blocked with 2% fetal bovine serum in PBS at room temperature for 1 hour. Indicated diluted hACE21-740-hFc and hACE21-615-hFc containing samples were added and plates were incubated for 1 hour at room temperature. Plates were washed three times and incubated with alkaline phosphatase (AP)-conjugated goat anti-human Fc secondary antibody (Jackson Immuno. Research) diluted 1:2000 in blocking buffer for 1 hour at room temperature. AP activity was measured at 405 nanometer using p-Nitrophenyl Phosphate (Guangzhou Howei Pharmaceutical Technology Co. Ltd Guangzhou, China) by an ELISA plate reader (Molecular Devices).Half-maximum effective concentration (EC50) binding values were calculated by GraphPad Prism (version 8).

Lentivirus production
Lentivirus was produced by transient transfection of Lenti-X 293 with a four plasmid system. Supernatants containing lentivirus particles were collected 48 and 72 h post-transfection and used for establishing stable cell lines.

In vitro ADCC and CDC assay
Raji and Raji-spike cells were labeled with 5 or 50 μM of CFSE, respectively according to manufacturer's protocol (MedChemExpress, Shanghai, China). Approximately 2x104 CFSE labeled Raji and Raji-spike cells were co-cultured with 4x105 primary NK cells or supplemented with human AB serum in the presence of various concentration of hACE21-740-hFc. Forty-eight hours later, cell were analyzed by ow cytometry. The ratio of CFSEhigh/CFSElow was used to indicate the relative killing to Raji-spike than Raji.
In vivo characterization of hACE21-740-hFc Pharmacokinetics were determined in C57BL/6 mice after a single injection of 100 μg of ACE21-740-hFc. Serum concentrations at indicated time points were determined by ELISA with immobilized RBD-mFc. C57BL/6 mice received a single injection of 100 μg ACE21-740-hFc. Seven days later, the hematology, body weight, and H&E staining of indicated organ were analyzed. H&E staining was performed by Servicebio (Wuhan, China).
Bio-distribution of MSC-ACE2-Fc NSG mice were injected with 5-7.5x105 MSC-ACE2-Fc intravenously through retro orbital sinus. Twentyfour hours later, heart, liver, spleen, lung and kidney organs were collected. Genomic DNA was isolated from these organs using an E. Z. N. A.® Tissue DNA Kit I (Omega Bio-tek). The relative distribution of human MSC was quanti ed by real-time PCR with primers (5'-aagtcataagtcggttgaggggagat-3' and 5'ccagattctgcagttttagcatctgtgt-3') against human IL2RG gene, which was de cient in NSG mice.
TF-1 proliferation assay TF-1 cells were plated in at bottomed 96-well plates at 10,000 cells/well in the presence of 1, 10 or 50ng/ml of IL-6 (Sino Biological, Beijing, China). Then 0.1, 1 or 10 μg/ml of scFv-IL6R-Fc or Tocilizumab (Roche) were added to the cell culture. Forty eight hours later, Cell proliferation was measured by CCK-8 assay (Dojindo Molecular Technologies, Shanghai, China) according to manufacturer's protocol.

Statistical analysis
Data are expressed as means ± standard deviation (SD) or standard error of the mean (SEM). The data were compared using two-tailed unpaired Student's t-test compared with two groups. Statistically signi cant differences p<0.05, p<0.01, and p<0.001 are noted with *, **, and *** respectively.