Stability and Function of Extracellular Vesicles Derived from Immortalized Human Corneal Stromal Stem Cells: A Proof of Concept Study

With significant advancement and development of extracellular vesicle (EV)-based therapies, there is a growing need to understand how their storage affects their physical and functional characteristics. EVs were isolated from the conditioned medium of a corneal stromal stem cell line (imCSSC) using Total Exosome isolation kit (TEI) and ultracentrifugation (UC) combined protocol. Purified EVs were stored at 4°C, − 80°C, room temperature (RT) after lyophilization with or without trehalose for 4 weeks. EVs stored at − 80°C and RT (lyophilization with trehalose) demonstrated a comparable morphology, while the freeze-dried samples without trehalose showed aggregation and degradation under a transmission electron microscope (TEM). Lyophilized samples without trehalose demonstrated a decreased particle concentration, recovery rate and protein concentration, which was remediated by the addition of trehalose. EVs stored at − 80℃ showed no change in the protein expression of CD9, CD63, and CD81. Regardless of the storage condition, all EV samples investigated reduced inflammation, as well as inhibited expression of fibrotic markers in vitro. Lyophilization of EVs with trehalose was a feasible storage method that retained the physical property and in vitro biological activities of EVs after 4 weeks of storage, while − 80°C offered the best retention of imCSSC-derived EV physical properties. For the first time, this data demonstrated a practical and translatable method for the storage of CSSC-derived EVs for clinical use.

Small extracellular vesicles (sEVs, everything 40-150 nm in size) include a mixed population of typical endosomalderived exosomes and small microvesicles, as it is currently not possible to isolate one distinct population due to their overlap in size and the lack of sub-type specific markers (2,3). Protein, RNA (including mRNA, microRNA [miRNA], and other non-coding RNAs), DNA, and lipids are actively and selectively incorporated into EVs before they are released by the cells (1).
It is known and widely reported that mesenchymal stromal cells (MSCs) have significant benefits within tissue regeneration and wound healing (4), with potent antiinflammatory and immunosuppressive properties (5). In corneal stroma wounding animal models, EVs were shown to promote corneal repair and regeneration (reviewed in (6)) and found to be the functional unit of corneal stromal stem cells (CSSCs) after mechanical injury (7). CSSC-derived EVs suppressed the formation of corneal scars and promoted the regeneration and deposition of normal collagen fibers to close the wound (7,8). The inflammatory response was also inhibited by CSSC-derived EVs in the cornea after injury due to the decrease of early neutrophil infiltration and induced generation of anti-inflammatory M2 macrophages (7,9,10). Therefore, the potential role of CSSC-derived EVs in corneal wound healing and regeneration of corneal stroma has gained more attention within the vision science community.
EVs hold the potential to become a new and effective treatment due to their unique advantages over cell-based therapeutics including non-replicating as do viruses, low immunogenicity, and relative ease of storage and shipping (11). Most commonly, EVs are stored at − 80°C (12). Although Görgens, A. et al. have reported that PBS supplemented with human albumin and trehalose (PBS-HAT) buffer improved stability of short-term and long-term EV samples stored at − 80°C even through several freeze-thaw cycles (13), − 80°C storage normally poses challenges in storing and delivering the drug product due to the requirement of large equipment and power supply. In addition, storage at − 80°C for a long time can lead to a lack of EV stability (14,15) including morphological changes associated with activity loss, diminished biological function, and a decrease in EV contents including endogenous RNA (16,17). When stored at 4°C or 37°C for 25 days, the diameter of EVs decreased, indicating a structural change or degradation of EVs (18) and therefore alternative storage methods investigated in this manuscript could be beneficial. Lyophilization offers the potential for room temperature (RT storage and is a more desirable method to preserve EVs long-term. This process removes water from frozen samples by sublimation and desorption in a vacuum (19,20) to preserve various types of biological materials, such as protein (21), plasma (22), and EVs derived from MSCs (ready for use after reconstitution with water while maintaining their bioactivity) (14,23,24). Lyophilization can greatly increase the accessibility of EV-based therapies in developing countries (6) where medical resources and specialized equipment are scarce.
In the current study, the stability of EVs derived from an immortalized CSSC line following 4 weeks of storage at 4°C, − 80°C, − 80°C with trehalose (− 80℃ + Tre), and at RT after lyophilization (Lyo) with or without the addition of the protective agent trehalose (Lyo + Tre) was evaluated.

Isolation and Physical Characterization of EVs from imCSSC
imCSSC at passage 15 were seeded into T875 flasks in CSSC medium supplemented with 2% human serum (HS) for 48 h by which point cells reached 70-80% confluency. Cell monolayers were washed with PBS three times removing any traces of HS and imCSSC conditioned medium collected following a 48-h culture in serum-free CSSC medium.

Lyophilization of EVs
Snap freezing of EVs was achieved using liquid nitrogen. Trehalose was dissolved in PBS to a final concentration of 50 mM and added to EV samples before snap freezing. After 30 min, samples were freeze-dried in a vacuum overnight using Freeze Zone Plus 2.5 (Labconco, Kansas City, MO, USA) and stored at RT for 4 weeks. Lyophilized EVs were rehydrated with distilled water of the original volume and vortexed immediately prior to testing.

Transmission Electron Microscopy
EVs were mounted on glow-discharged copper grids which are 200 mesh coated by formvar carbon film (Electron Microscopy Sciences, PA, Hatfield, USA) and fixed with 2% paraformaldehyde/2% glutaraldehyde for 5 min. Grids were washed three times with boiled distilled water and counterstained with 2% uranyl-oxidase solution (w/v) at pH7 for 5 min. Images were acquired using a JEM-100CX transmission electron microscopy (TEM; JEOL, Tokyo, Japan) at a voltage of 60 kV.

RAW Cell Anti-inflammatory Assay
2.5 × 10 4 RAW264.7 Cells (TIB-71; ATCC, USA) in 1.0 ml DMEM with 10% FBS were seeded into wells of a 24-well plate and cultured for 16 h. Cells were treated with RANKLE peptide (R&D Systems) and ConA (R&D Systems) at a final concentration of 50 ng/mL and 20 µg/mL with or without the addition of EVs. EVs were added at a final concentration of 1 × 10 10 particles/mL for all conditions. After 5 days, cells were harvested for RNA isolation; RT-PCR was performed for acid phosphatase 5 (Acp5), matrix metallopeptidase 9 (Mmp9), cathepsin K (Ctsk), and 18S.

RNA Extraction, cDNA Transcription, and Real-time PCR
RNA was extracted from corneal fibroblasts (CF) or RAW264.7 Cells using an RNA Mini isolation kit according to the manufacturer's instructions (Qiagen, Germantown, MD, USA). 1 μg of RNA was transcribed using Superscript™ III Reverse Transcriptase (100 U/µg of RNA, Life Technologies). Primers for real-time PCR (RT-PCR) are listed in Supplementary Table S1. RT-PCR was performed with SYBR™ Green Master Mix (Applied Biosystems™, Carlsbad, CA, USA) in an Eppendorf Mastercycler® ep realplex (v2.2, Eppendorf, Hauppauge, NY, USA). 18S as a housekeeping gene was used per sample for normalization of the data. Crossing threshold (Ct) values were directly determined. Absolute levels of transcripts were calculated via the normalization of ΔCt, noted as 2(− ΔCt).

Gene Expression of α-SMA
CF at passage 3 were seeded on 4-well chamber glass slides (Thermo Fisher Scientific) in CF media supplemented with 1% FBS for 24 h. CF were then treated in triplicate wells with control (citric acid (Sigma-Aldrich) + PBS), 2 ng/ml TGFβ1 (Peprotech, Rocky Hill, NJ, USA) + PBS, and 2 ng/ ml TGFβ1 + EVs (1 × 10 10 p/ml) in 1% FBS CF media (CF media diluted by DMEM). After 24 h, RNA samples were collected and the gene expression level of alpha-smooth muscle actin (α-SMA) and 18 s were assessed by RT-PCR.

α-SMA Immunofluorescence Staining and Quantitative Analysis
CF at passage 3 were treated in triplicate wells with control (citric acid + PBS), 2 ng/ml TGFβ1 + PBS or 2 ng/ml TGFβ1 + EVs (1 × 10 10 p/ml) in 1% CF media for 72 h. The cells were then washed with PBS and fixed with ice-cold 50% acetone/50% methanol for 20 min at − 20℃. Monolayers were washed with PBS and blocked with 1% BSA for 30 min. Cells were then probed with α-SMA antibody (Abcam, Cambridge, MA, USA) diluted 1:200 overnight at 4℃. Following three PBS washes, the cells were incubated with an anti-mouse secondary antibody conjugated to Alexa Fluor 488 (Invitrogen) diluted 1:4000 for 1 h at RT. The slides were then labeled with 4 μg/ml Hoescht 33,342 (Invitrogen) nuclear stain for 15 min at RT and mounted with Fluoromount mounting medium (Sigma-Aldrich). Images were taken using a Keyence BZ-X710 fluorescence microscope (Keyence, Osaka, Japan). α-SMA-positive cells were quantified using custom Python3 code.

Statistics
All experiments were independently repeated at least three times. All data was shown as mean ± SD. The statistical analysis was performed by using GraphPad Prism software (version 8.0, GraphPad Software, San Diego, CA, USA). For a two-pair comparison of the means, a one-way ANOVA test followed by 2-sided t-test or one-way ANOVA with multiple comparisons was performed. A p value of < 0.05 was considered statistically significant.

Physical Properties of EVs After Storage
PDI was used to characterize the size distribution of particles in EV solutions. A PDI value below 0.05 indicates a single-distributed standard solution with a high degree of mono-dispersion (highly uniform particle size) (28). On the contrary, a PDI value greater than 0.7 indicates a wide range of particle size distribution in the sample (28). PDI of fresh EV samples from imCSSC was on average 0.28 ± 0.01 (Fig. 1), indicating heterogeneity in the particle size but as to be expected with EV samples (28). After 4 weeks of storage at 4℃ (0.26 ± 0.01) or at − 80℃ (0.277 ± 0.01), there was no significant change in the PDI (p > 0.05) when compared to the fresh EV samples (Fig. 1). However, the addition of trehalose did cause a significant increase in PDI in all instances. The highest PDI was observed in EVs stored at RT after lyophilization (0.43 ± 0.02). This was reduced with the addition of trehalose significantly (0.38 ± 0.03; p < 0.001) (Fig. 1). Although the PDI values changed following different storage conditions, all of them were lower than 0.7.

Preservation of EV Structure
Freshly isolated EVs demonstrated heterogeneous size particles with a defined double lipid membrane ( Fig. 2A), which was largely preserved after storage under all conditions. There was an increase in particle aggregation in Lyo EVs ( Fig. 2A), which was reduced with the addition of trehalose.
The particle concentration of fresh EVs was similar after storage at 4℃, − 80℃ with/without trehalose (Fig. 2B). A significant reduction of particle concentration was found in the Lyo EV group (p < 0.001, Fig. 2B) compared to fresh samples. After the addition of trehalose during the lyophilization process, particle concentration was effectively improved (p < 0.05, Fig. 2B). The particle recovery rate (percentage particles recovered from the initial fresh EV isolation) was similar regardless of storage conditions at 4℃, − 80℃ with/without trehalose (all p > 0.05, Fig. 2C). EVs undergoing lyophilization with or without trehalose went through a significant reduction in recovery rate as compared to freshly isolated EVs (p < 0.0001, p < 0.05, Fig. 2 Physical characteristics of EVs. A Morphology of fresh EVs and stored EVs for 4 weeks under the different conditions captured by TEM: intact EVs (black arrows) and aggregated EVs (white arrows). Scale bar = 200 nm. Effect of storage for 4 weeks on particle concentration (B), particle recovery rate (C), and modal particle size (D). (E) Representative particle distribution plot. Data were expressed as mean ± SD (n = 3). Tre, trehalose and Lyo, lyophilization. Oneway ANOVA with multiple comparisons, *p < 0.05, ***p < 0.001, ****p < 0.0001 compared to fresh EVs; # < 0.05, Lyo EVs vs. Lyo + Tre EVs respectively, Fig. 2C). However, the addition of trehalose could also rescue the decreasing tendency (p < 0.05, Fig. 2C).
The modal particle diameter was assessed using nanoparticle tracking analysis (NTA) and demonstrated freshly isolated EVs had a modal size of 98.6 nm (± 4.2 nm; Fig. 2D and E). Storage at 4℃, − 80℃ with/without trehalose and lyophilization with trehalose demonstrated no significant differences in particle size when compared to fresh EVs (p > 0.05). Lyophilization without trehalose did however increase the modal particle size to 108.6 nm (± 6.7 nm, Fig. 2D) and an increase in the number of particles > 200 nm (Fig. 2E). Similarly, the addition of trehalose before lyophilization process could rescue the increased particle size (p < 0.05, Fig. 2D). Particle size distribution remained similar among freshly isolated EVs, EVs stored at 4℃, − 80℃ with/without trehalose, and lyophilization with trehalose.

EV Markers are Preserved Following Storage
Protein concentration and expression of EV-associated markers were examined to evaluate the effects of storage on EV protein integrity. Storage in any of the conditions had no significant effect on the protein concentration (Fig. 3A, all p > 0.05).
Ponceau S staining of SDS-PAGE-separated EV protein extracts showed that the protein molecular weight distribution of samples in each group was similar (Fig. 3B). Expression levels of the EV-associated surface antigens including CD63, CD9, CD81, and the endosomal protein ALIX were examined by western blot. After storage, the expression levels of CD63, CD9, and CD81 were reduced in the EV-Lyo samples compared to the fresh EVs and EVs stored in all other conditions (Fig. 3C). Consistent protein loading demonstrated in Fig. 3B suggested this is a reduction in the EV

Fig. 3 Protein contents of EVs.
A Protein concentration of fresh and stored EVs. Data expressed as mean±SD (n = 3). Oneway 425 ANOVA followed by 2-sided t-test, ns = no significance. B Ponceau S staining of protein extract of EVs samples. C Protein expression 426 of EV-specific markers using Western blot: CD63, CD9, CD81 and ALIX specific proteins and not due to differences in sample loading. The value of total fluorescence for each of the groups was quantified in Fig. 4. EVs stored at 4℃, − 80℃ with/ without trehalose and RT with Lyo + Tre showed no obvious differences (all p > 0.05, Fig. 4A-D). However, a total decreasing tendency of EV protein markers (CD9, CD63, CD81, and ALIX) in the Lyo EV group was found compared to fresh samples and the expressions of CD9 and CD63 were significantly reduced in the Lyo EV group (p < 0.01, Fig. 4A and B).
Flow cytometry was used to investigate expression levels of EV-specific surface antigens after storage using CD63 + captured EVs. CD63 + captured EVs were then labeled with fluorescent antibodies specific for CD9, and CD81. In addition, CD73 was used as a control for microvesicle contamination in the samples (29)(30)(31), as it is expressed by the imCSSC but not in their sEVs (7). CD73 mean fluorescence intensity (MFI) was low and similar in all the EV groups, demonstrating minimal contamination from cell membrane fragments or microvesicles (Fig. 5A, p > 0.05). CD9 and CD81 MFI were not significantly affected when EVs were stored at 4°C, − 80°C with/without trehalose or at RT following lyophilization with trehalose ( Fig. 4B and C; p > 0.05). Lyophilization without trehalose did result in a . One-way ANOVA followed by 2-sided t-test, *p < 0.05, **p < 0.01, compared to fresh EVs; ##p < 0.01, Lyo EVs vs. Lyo + Tre EVs significant reduction in MFI for both CD9 (Fig. 5B; p < 0.05) and CD81 (Fig. 5C; p < 0.01). A significant increase in MFI was observed between lyophilized samples with the addition of trehalose (p < 0.01 for both CD9 and CD81). Representative images of isotype control specifications, gating strategies, and FCM controls in flow cytometry experiments have been shown in Figure S1, and our data showed that all the PBS populations were negative.

Anti-inflammatory Function of EVs After Storage
CSSC with high regenerative potential can effectively inhibit the differentiation of RAW 264.7 cells toward osteoclasts (32) and the function of EV mirrors that of the cells in which they are derived (7). Thus, the RAW cell differentiation test (RAW cell anti-inflammatory assay) was performed to identify any functional differences in EVs following storage. The addition of EVs inhibited osteoclast differentiation, with no dramatic change in morphology as compared to untreated RAW cells (Fig. 6A). Results demonstrated the gene expression of osteoclast markers was significantly up-regulated (from 1.4 ± 1.0 to 9.2 ± 1.4 for Acp-5, from 1.1 ± 0.5 to 41.0 ± 22.6 for Ctsk, from 1.0 ± 0.2 to 9.8 ± 3.6, Mmp9, p < 0.01 for all transcripts, Fig. 6B-D) following the addition of RANKL and ConA. Fresh EVs or stored EVs treatment decreased the expression of osteoclast marker genes (Acp5, Ctsk, and Mmp9) compared to the vehicle group ( Fig. 6B-D). The expression level of lyophilized EVs without trehalose was 1.3 ± 0.1 for Acp5, 4.8 ± 1.4 for Ctsk, and 1.6 ± 0.3 for Mmp9. Notably, lyophilized EVs with trehalose showed that the lowest expression level of Acp5 and Ctsk (0.7 ± 0.9 for Acp5, 3.3 ± 4.5 for Ctsk, p < 0.01 and p < 0.05, respectively, Fig. 6B and D) and a significant reduction in Mmp9 expression (1.8 ± 0.5, p < 0.01, Fig. 6C).

In Vitro Anti-fibrotic Property of EVs
In the presence of TGFβ1, corneal fibroblasts differentiate toward a myofibroblast phenotype demonstrating a significantly increased expression level of α-SMA, a myofibroblast marker (33,34). The addition of EVs reduced α-SMA expression in TGFβ1-treated corneal fibroblasts at both mRNA and protein levels ( Fig. 7A-C). Notably, the highest antifibrotic effect was observed in EVs stored at RT following lyophilization with trehalose based upon the lowest number of α-SMA positive myofibroblasts compared to TGFβ1-treated corneal fibroblasts (Fig. 7B, p < 0.05). At the mRNA level α-SMA was most strongly inhibited with EVs stored at − 80℃ (Fig. 7C, p < 0.01). EVs stored at 4℃, − 80℃ and lyophilization without trehalose were also capable of reducing α-SMA gene expression ( Fig. 7C; p < 0.05) but at a lesser degree. There is a trend of a lower mRNA expression of α-SMA when treated with EVs stored at − 80℃ with trehalose and following lyophilization with trehalose at RT, but did not reach significance (Fig. 7C, p > 0.05).

Discussion
Therapeutic development and research involving EVs is gaining increasing attention. However, the optimization and effect of storage on EVs is often overlooked. In this study, we isolated EVs from imCSSC and assessed their characteristics and in vitro function following 4 weeks of storage under different conditions. For the first time, comprehensive characterization and functional testing of EVs stored using "typical" storage conditions commonly used in many laboratories (4℃ and − 80℃) have been compared to freshly isolated EVs, as well as EVs stored in a way that would allow for 4 weeks storage at RT. NTA analysis, western blot, and flow cytometry demonstrated all EVs, other than Lyo-EVs, retained their physical characteristics and integrity as well as their protein components. Functional testing demonstrated EVs stored at RT following lyophilization with trehalose retained their anti-inflammatory and anti-fibrotic properties at comparable levels to those of freshly purified EVs. It is interesting to note that other published studies have demonstrated a significant alteration in EV integrity following storage at − 80℃ (35). The experiments conducted here did only assess the effect of 4 weeks of storage which could be responsible for some of the differences. Additionally, the current study shows that while there was some loss of EV integrity and function following storage at − 80℃, this storage condition was still better than storage at 4℃. This finding is consistent with those reported by others.
Lyophilization has great potential to become an ideal solution for the storage of EVs allowing for storage at RT. The CSSC-conditioned medium has previously demonstrated to retain its anti-inflammatory function following lyophilization (36). However, the effect of lyophilization on CSSC EVs has not previously been assessed. The effect of lyophilization has been investigated on EVs derived from other cell types (37,38). Akers et al. demonstrated that freeze-drying without a protective agent caused a decrease in the particle number of EVs derived from cerebrospinal fluid resulting from aggregation, which is consistent with our findings (37). Trehalose is a disaccharide known to preserve proteins during lyophilization (39,40). Stark and colleagues' work reported that protein concentration and EV surface markers could be sustained with the addition of trehalose during lyophilization (38,41). The main protective effect of trehalose on EVs is during the lyophilization process or freeze-thaw cycling as explained above. EVs directly stored at RT have been shown to have lower DNA concentration, miRNA level, and CD63 expression after short-term storage (42). We therefore did not include EV stored at RT with trehalose. Nevertheless, adding a control group in which trehalose is added at RT would have provided a confirmatory result. In future long-term storage studies, we will include this as a control. In terms of the bioactivity, lyophilization of EVs with trehalose could retain their biological function in vitro (14,23,41,43). More interestingly, Neupane, Yub Raj et al. reported that lyophilization with lyoprotection agents like sucrose and trehalose could preserve the intrinsic cardioprotective activity of bioinspired cell-derived nanovesicles in a mouse model of myocardial ischemia-reperfusion injury (41).
The lipid bilayer of EVs can be damaged by ice crystals generated during freezing and results in vesicle fusion or phase change during the lyophilization process (19). Since miRNAs are the functional unit of CSSC corneal repair and regeneration in animal models (7), one can argue that wellpreserved EVs preserve their function by avoiding degradation of their cargo during storage. Interestingly, Aker et., al (37) reported that total RNA and representative miRNA levels of EVs from cerebrospinal fluid were stable after lyophilization, even without protective agents like trehalose for 7 days. Future experiments assessing the relative expression, degradations, and abundancies of the miRNAs within EVs after storage would be necessary to confirm the integrity Fig. 6 Anti-inflammation effect of EVs in RAW cell anti-inflammatory assay. A Representative images of RAW cell anti-inflammatory assay in different groups. After the treatment of RANKL and ConA, RAW cells differentiated into osteoclasts (black arrows) related to inflammation. Scale bar = 100 μm. Relative gene expression levels of Acp5 (B), Mmp9 (C), and Ctsk (D) in RANKL-and ConA-treated RAW cells with or without EVs. EV concentration applied was 10 10 particles/ml. Acp5, acid phosphatase 5; Mmp9, matrix metallopeptidase 9; Ctsk, cathepsin K. Data were expressed as mean ± SD (n = 3). One-way ANOVA with multiple comparisons, *p < 0.05, **p < 0.01, compared to control; #p < 0.05, ##p < 0.01, ###p < 0.001, compared to PBS group of EVs and their miRNAs after the EVs following storage. However, for the purpose of this manuscript, maintaining the function of the EVs indicates the miRNA within the EVs are still stable and capable of a functional response. Though short-term storage did not significantly alter EV physical integrity and function in vitro, the effects of long-term storage (> 1 year) as well as their in vivo capabilities to repair and regenerate corneal scarring need to be assessed.

Conclusion
While storage at − 80℃ offered the best preservation of imCSSC-derived EV physical properties, lyophilization supplemented with trehalose was efficient in preserving imCSSC-derived EVs' physical quality and function in vitro. Lyophilization with trehalose offers a feasible cost-effective approach to store and transport EVs for clinical application.

Acknowledgements
The UCLA Integrated Molecular Technologies Core is supported by CURE/P30 DK041301.  Anti-fibrotic effect of EVs. A Representative images of α-SMA staining of primary corneal fibroblasts after treatment with TGFβ1 in different groups. EV concentration used was 10 10 particles/ml. Scale bar = 100 μm. B percentage of cells staining positive for α-SMA in TGFβ1-treated corneal fibroblasts with or without EVs. Relative α-SMA (C) gene expression levels in TGFβ1-treated corneal fibroblasts with or without fresh EVs or stored EVs. EV concentration was 10 10 particles/ml. Data were expressed as mean ± SD (n = 3). One-way ANOVA followed by 2-sided t-test, *p < 0.05, **p < 0.01, compared to fresh EVs; #p < 0.05, ##p < 0.01, compared to PBS group work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: N.L., R.K., S.X.D.; All authors have read and agreed to the published version of the manuscript.

Author Contribution
Funding This work is supported in part by an unrestricted grant from Research to Prevent Blindness to the Department of Ophthalmology at the University of California, Los Angeles. S.X.D received grant support from the National Eye Institute (R01 EY021797 and R01 EY028557) and from the California Institute for Regenerative Medicine (TR2-01768, CLIN1-08686, CLIN2-11650). N.L. was supported by the program of the China Scholarships Council (NO. 201906100167).

Declarations
Ethics Approval and Consent to Participate Experimentation on human tissue adhered to the tenets of the Declaration of Helsinki. The experimental protocol was evaluated and exempted by the University of California, Los Angeles Institutional Review Boards (IRB#12-000363).
Conflict of Interest S. X. D. is a consultant for Dompe US, Novartis, and Claris Biotherapeutics, Inc.; Other authors: none.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.