key: cord-0323109-8dpla6jt
authors: Wu, Jun; Song, Dingyun; Li, Zhongwen; Guo, Baojie; Xiao, Yani; Liu, Wenjing; Liang, Lingmin; Feng, Chunjing; Gao, Tingting; Chen, Yanxia; Li, Ying; Wang, Zai; Wen, Jianyan; Yang, Shengnan; Liu, Peipei; Wang, Lei; Wang, Yukai; Peng, Liang; Stacey, Glyn Nigel; Hu, Zheng; Feng, Guihai; Li, Wei; Huo, Yan; Jin, Ronghua; Shyh-Chang, Ng; Zhou, Qi; Wang, Liu; Hu, Baoyang; Dai, Huaping; Hao, Jie
title: Immunity-and-Matrix-Regulatory Cells Derived from Human Embryonic Stem Cells Safely and Effectively Treat Mouse Lung Injury and Fibrosis
date: 2020-06-06
journal: bioRxiv
DOI: 10.1101/2020.04.15.042119
sha: a529826227726d194cec68b5ba55f35249edb9f2
doc_id: 323109
cord_uid: 8dpla6jt

Lung injury and fibrosis represent the most significant outcomes of severe and acute lung disorders, including COVID-19. However, there are still no effective drugs to treat lung injury and fibrosis. In this study, we report the generation of clinical-grade human embryonic stem cells (hESCs)-derived immunity- and matrix-regulatory cells (IMRCs) produced under good manufacturing practice (GMP) requirements, that can treat lung injury and fibrosis in vivo. We generate IMRCs by sequentially differentiating hESCs with serum-free reagents. IMRCs possess a unique gene expression profile distinct from umbilical cord mesenchymal stem cells (UCMSCs), such as higher levels of proliferative, immunomodulatory and anti-fibrotic genes. Moreover, intravenous delivery of IMRCs inhibits both pulmonary inflammation and fibrosis in mouse models of lung injury, and significantly improves the survival rate of the recipient mice in a dose-dependent manner, likely through paracrine regulatory mechanisms. IMRCs are superior to both primary UCMSCs and FDA-approved pirfenidone, with an excellent efficacy and safety profile in mice and monkeys. In light of public health crises involving pneumonia, acute lung injury (ALI) and acute respiratory distress syndrome (ARDS), our findings suggest that IMRCs are ready for clinical trials on lung disorders.

For osteogenic differentiation, 4.2 × 10 3 cells were seeded per well in 96-well plates. When cells 176 reached 50-70% confluency, the medium was replaced with osteogenic differentiation medium and 177 kept for 3 weeks. To assess osteogenic differentiation, immunofluorescence (Osteocalcin) and Alizarin 178

Red S (Sigma-Aldrich, A5533) staining was performed for the calcium-rich extracellular matrix. For 179 adipogenic differentiation, cells were seeded into a 24-well plate at the density of 3.7 × 10 4 cells/well, 180 and maintained in culture medium until 100% confluency. Then, cells were cultured in adipogenic 181 differentiation medium for 3 weeks. Lipid droplets of the resultant differentiated cells were detected 182 using immunofluorescence (FABP-4) and Oil red O (Sigma-Aldrich, O0625) staining. For 183 chondrogenic differentiation, 2.5 × 10 5 cells resuspended in chondrogenic differentiation medium were 184 centrifuged for 5 min at 200 g in a 15-mL conical tube (Corning, NY, USA). Then, cells were cultured 185 for 3 weeks. After 3 weeks, chondrogenic pellet was harvested and fixed in 4% paraformaldehyde 186 (Aladdin Chemical Co., Ltd., Shanghai, China; C104190). Cryosectioning was performed by OTF 187

Cryostat (Leica, Germany) and 10 µm sections were stained with immunofluorescence (Aggrecan) and 188

Alcian Blue (Sigma-Aldrich, A5268) staining. 189

IMRCs and UCMSCs were seeded into 96-well plates at a density of 8 × 10 3 cells per well. Cell 191

Counting Kit-8 (CCK8; Thermo Fisher Scientific, CK04) was used to measure the cells' growth. CCK8 192 solution was added at day 1, day 3, day 5 and day 7 by following the CCK8 kit manufacturer 's protocol. 193 After 2 h of incubation, absorbance was measured at 450 nm. 194

The IMRCs and UCMSCs were harvested at passage 5 and the cell diameters were measured with 196 Image J software (NIH) software analysis. Normal Distribution curves were fitted to the cell diameter 197 data to analyze the cell size distribution characteristics. 198

The IMRCs and UCMSCs were cryopreserved in a published clinical injection 28 Osaka, Japan; QPS-201). GAPDH was used for internal normalization. Primers for real-time PCR in 243 this study are as shown in Supplementary information, Table S2 . 244

To determine the secretion of human MMP1, cell culture supernatants were collected after 48 hours of 246 (Sigma-Aldrich, A5228) and mouse β-actin antibody (Sigma-Aldrich, A1978). Then, the membranes 279 were washed with TBST for 3 times and incubated for 1 h with a secondary antibody: anti-mouse IgG 280 antibody (HRD) (Sigma-Aldrich, A9044) and anti-rabbit IgG antibody (HRD) (Sigma-Aldrich, A0545) 281 at room temperature. Images were obtained using the ChemiDoc XRS+ imaging system (Bio-Rad) and 282 quantified using the Quantity One software. 283

Soft agarose gels were prepared according to previously published protocols 29 . Cells were harvested 285 and diluted to a cell concentration of 2.5 × 10 4 cells/mL, before mixing with the agarose mixture in 286 6-well culture plates. The plates were incubated in a 37 °C incubator for 21 days. 287

Adult female and male cynomolgus monkeys (3-5 years old) were housed in single quarters with a Specific pathogen-free (SPF) 11-to 13-week-old C57BL/6 male mice were purchased from the Animal 295

Center of Peking University (Beijing, China). All experimental and control mice were weight-matched, 296

and their weights ranged from 25 to 30 g. All mice were housed in the animal care facility of the 297

Institute of Medical Science, China-Japan Friendship Hospital. Mice were maintained under SPF 298 conditions at room temperature (between 20-24 °C), with humidity between 35-55%, in a 12/12 h 299 light/dark cycle, with food and water ad libitum and monitored with respect to their general state, fur 300 condition, activity, and weight according to institutional guidelines. The mice were sacrificed at each 

Acute toxicity test: Low dose (2.6 × 10 6 cells/monkey), middle dose (2.6 × 10 7 cells/monkey) and high 306 dose (1 × 10 8 cells/monkey) IMRCs were infused into cynomolgus monkeys (3-5 years old) by single 307 intravenous infusion. After 6 months, their body weight, ophthalmology status, hematology, blood 308 chemistry and urine chemistry were measured. 309

Long-term toxicity test: Low dose (2.6 × 10 6 cells/monkey) and high dose (1 × 10 8 cells/monkey) 310

IMRCs, or saline, were injected into cynomolgus monkeys (3-5 years old) by intravenous infusion once 311 a week for 22 times, after which their body weight, body temperature, food intake and organ weights 312

were measured. 313

30X whole-genome sequencing data was produced by HiSeq X-Ten (Annoroad Gene Technology Co., 315

Ltd) and used to analyze single nucleotide mutation changes between IMRCs and hESCs. As 316 previously described 30 , the whole-genome sequencing data was mapped to the hg19 reference genome 317 by BWA (version 0.7.15) using the 'mem' mode with default parameters. The genome coverage was 318 calculated by bedtools 31 . The normalized coverage depth for each bin was calculated by dividing the 319 mapped reads were retained for the copy-number variation (CNV) analysis, in which chromosomal 321 sequences were placed into bins of 500 kb in length. The hg19 genome repeat regions annotated by 322

RepeatMasker (http://www.repeatmasker.org) were removed from the genomic sequence before 323 calculating the coverage. The CNV scatterplot was drawn using ggplot2. 324

Mouse lung histology was performed as follows. Briefly, the lung was dehydrated, paraffin-embedded 326 and cut into 4 µm sections. Lung sections were stained with H&E and Masson's trichrome stain for 327 assessment of pathological changes. Immunohistochemical staining was performed using antibodies 328 

Mice were lightly anesthetized and were monitored using the in vivo imaging systems at day 0, day 1, 342 day 3, day 5, day 7, day 9, day 15, day 21, day 27, day 32, day 37 and day 46 after tail intravenous 343 injection of DiR-labeled IMRCs (5 × 10 6 cells in 200 µL saline). Serial fluorescence images were also 344 obtained in major organs ex vivo. In order to reduce autofluorescence, the ideal filter conditions for 345

The flexiVentFX experimental platform was set up according to the manufacturer's instructions. At the 348 day 21, after 3% pentobarbital sodium (40 mg/kg) intraperitoneal injection of anesthesia, mice were 349 fixed supine within 5-8 minutes, and endotracheal intubation was inserted in the middle of a 350 tracheotomy. After computer processing, indicators of lung function required were obtained, including 351 the inspiratory capacity (IC), respiratory resistance (Rrs), static compliance (Crs), elastic resistance 352 (Ers), Newtonian resistance, tissue damping (G), tissue elasticity (H) and forced vital capacity (FVC) 34 . 353

The lung tissue was completely removed, weighed by electronic balance, and the lung coefficient was 355 calculated according to the formula: wet lung weight (g)/total body weight (kg). 356

The BLM mouse model was generated by intratracheal injection of 1.5 mg·kg -1 bleomycin sulphate 358 (BLM; Bioway, China; DP721) in normal saline under light anaesthesia. The IMRCs were delivered 359 intravenously at day 0 or day 1 or day 7 or day 14 after injury. Mice were euthanized with 50 mg/mL 360 sodium pentobarbital (0.6 mg/10 g weight) according to published guidelines 35 . Animals were killed at 361 day 21 after BLM injury. After perfusion with normal saline, the left lungs were used for morphometric 362 analysis while the right lungs were removed and used for other analyses. 363

Mouse CT scans were performed according to the manufacturer's instructions. 365

We conducted a pilot study using GMP-grade IMRCs, under an expanded access program for 367 compassionate use in COVID-19 patients. The pilot study was approved by the Ethics Committee of 

In this study, hESC-derived IMRCs were generated by passaging cells that are migrating from human 383 embryoid bodies (hEBs), using serum-free reagents (Fig. 1a, b) . The clinical hESC line 384 (Q-CTS-hESC-2) was prepared as described previously 20 . Clinical hESCs were maintained in Essential 385 8 TM basal medium (E8) on vitronectin-coated plates, before dissociation into small clumps to form 386 hEBs for 5 days. Subsequently, hEBs were transferred onto vitronectin-coated plates and cultured for 387 14 additional days. The hEBs outgrowth cells were dissociated and passaged continuously in IMRCs 388

Medium. After 5 passages, IMRCs were harvested for characterization. IMRCs possessed 389 fibroblast-like morphology (Fig. 1b) and maintained diploid karyotypes at passage 5 (Fig. 1c) . 390

Moreover, copy-number variation analysis by whole genome sequencing data also indicated that no 391 chromosomal aneuploidies, large deletions nor duplication fragments were detected (Fig. 1d) . Next, we 392 analyzed the expression profile of MSC-specific genes (Fig. 1e) . IMRCs showed a pattern that greatly 393 differed from hESCs, and closely resembled primary umbilical cord-derived MSCs (UCMSCs). All 394 pluripotency genes (POU5F1, SOX2, NANOG, ZFP42, SALL4), mesendoderm genes (MIXL1), and 395 ectoderm genes (GAD1) were extinguished in both IMRCs and UCMSCs (Fig. 1e) . IMRCs expressed 396 (CD73), ENG (CD90), THY1 (CD105) and ITGB1 (CD29). Flow cytometry analysis further confirmed 398 this surface marker profile ( Fig. 1f; Supplementary information, Fig. S1a, b) . By contrast, IMRCs were 399 negative for the hematopoietic surface markers PTPRC (CD45) and CD34. IMRCs displayed the 400 ability to undergo tri-lineage differentiation into mesenchymal tissues, such as adipocytes, 401 chondroblasts and osteoblasts ( Fig. 1g; Supplementary information, Fig. S1c ). The proliferation rate of 402

IMRCs was higher than that of UCMSCs at passage 15, suggesting that IMRCs have a stronger 403 capacity for long-term self-renewal than primary MSCs (Fig. 1h) . Interestingly, IMRCs were generally 404 smaller than UCMSCs (Fig. 1i) , suggesting that IMRCs can pass through small blood vessels and 405 capillaries more easily, and are thus perhaps less likely to cause pulmonary embolism. To evaluate the 406 clinical potential of the IMRCs, we measured the viability of IMRCs suspended in a published clinical 407 injection buffer at 4 °C. We found that the viability of IMRCs remained higher (93%) than UCMSCs 408 (73%) after 48h (Fig. 1j) . 409

To clarify the degree of similarity between hESCs, IMRCs and primary UCMSCs at the whole 411 transcriptome level, we performed genome-wide profiling of IMRCs and UCMSCs and compared their 412 gene expression with hESCs 23 . Whole-transcriptome analysis confirmed that IMRCs clustered together 413

with UCMSCs in an unsupervised hierarchical clustering (Fig. 2a) . The global differentially expressed 414 gene analysis showed that highly expressed genes in IMRCs and UCMSCs compared with hESCs, 415

were enriched with angiogenesis, inflammatory response and extracellular matrix disassembly related 416 processes (Supplementary information, Fig. S2 ). Accordingly, MSC-specific genes such as NT5E, ENG, 417 PDGFRA, SPARC and ITGB1 were upregulated, whereas pluripotency genes such as POU5F1, SOX2, 418 SALL4 and ZFP42 were extinguished in IMRCs relative to hESCs, and the overall correlation with 419 hESCs was weak (R 2 = 0.66; Fig. 2b ). Next, we analyzed the expression of genes specific to IMRCs, 420 compared to UCMSCs (Fig. 2c) . While the overall correlation with UCMSCs was stronger (R 2 = 0.87), 421 we also found that many genes were differentially expressed in IMRCs compared to primary UCMSCs. 422

The upregulated genes promote immunomodulation (LIF), tissue repair (VEGFA, GREM1), cell 423 division (CDC20) and anti-fibrosis (MMP1). By contrast, the downregulated genes predominantly 424 promote inflammation (IL-1B, CXCL8, CCL2 and CXCL1; Fig. 2c ). Gene set enrichment analysis 425 (GSEA) of the differentially expressed genes confirmed that IMRCs manifest reduced inflammation 426 and stronger proliferative capacity as their top gene signatures, compared to primary UCMSCs (Fig. 2d , 427 e; Supplementary information, Fig. S3) . 428

To elucidate the heterogeneity of gene expression amongst IMRCs, single cell RNA sequencing 429 (scRNAseq) was performed for both IMRCs and primary UCMSCs. A total of 16,000 single-cell 430 transcriptomes were obtained from two samples. Approximately ~ 100,000 reads were obtained per cell, 431

which generated a median of 31,000 unique molecular identifiers per cell, ~ 4,800 expressed genes per 432 cell and more than 22,000 total genes detected in the population. Our scRNAseq data also showed that 433

IMRCs were relatively homogenous in expression for the MSC-specific markers PDGFRA, SPARC, 434 NT5E, ITGB1 and ENG ( Fig. 2f; Supplementary information, Fig. S4 ). IMRCs were also relatively 435 homogenous in their suppression of pluripotency and non-mesenchymal markers, relative to hESCs 436 and UCMSCs, suggesting they are likely to be similar in biological activity to primary MSCs after 437 transplantation ( Fig. 2g; Supplementary information, Fig. S4 ). These results provided insight into the 438 clinical applicability of IMRCs, especially with regards to their complete loss of pluripotency and their 439 gain in hyper-immunomodulatory potential. 440

To test the immunomodulatory capacity of IMRCs, we exposed them to the pro-inflammatory cytokine 442 IFN-γ. We found that after stimulation with IFN-γ, both primary UCMSCs and IMRCs displayed 443 similar characteristic morphological changes (Fig. 3a) . IFN-γ stimulation also potently upregulated 444 IDO1 (indoleamine 2, 3-dioxygenase 1) expression in both primary UCMSCs and IMRCs, but much 445 more significantly so in IMRCs (Fig. 3b) . This is important because IDO1 has been shown to mediate 446 immunosuppression in T cell-inflamed microenvironments through its catabolism of tryptophan and 447 thus suppression of the tryptophan-kynurenine-aryl hydrocarbon receptor (Trp-Kyn-AhR) pathway in T 448 cells. To further characterize IMRCs at a molecular level, we conducted genome-wide RNA profiling 449 to compare IMRCs with UCMSCs, after IFN-γ stimulation (Fig. 3c) . Hierarchical clustering revealed 450 that IMRCs clustered separately from UCMSCs, although they appeared more similar after IFN-γ 451 stimulation (Fig. 3d) . To analyze IMRCs after IFN-γ stimulation in greater detail, we separated their 452 analysis (Fig. 3e) . The global differential expressed gene analysis showed that highly expressed genes 454

in IMRCs and UCMSCs after IFN-γ stimulation, compared with IMRCs and UCMSCs before IFN-γ 455 stimulation, were enriched with immune response and interferon-gamma-mediated signaling pathway 456 (Supplementary information, Fig. S5 ). Accordingly, we found that some pro-inflammatory genes 457

showed lower expression while some pro-regenerative genes showed higher expression levels in 458

IMRCs, compared to primary UCMSCs (Fig. 3f) . 459

To confirm these findings, we performed a focused ELISA analysis of 48 biologically relevant 460 chemokines and cytokines in the secretomes of both stimulated IMRCs and UCMSCs. We found that at 461 least nine pro-inflammatory cytokines were lower in IMRCs than UCMSCs, including interleukin-6 462 (IL-6), IL-1β, tumor necrosis factor alpha (TNF-a), GRO-a, IFN-a2, IL-1a, IL-3, and IL-8 ( Fig. 3g ; 463

Supplementary information, Fig. S6a ). Furthermore, while the anti-inflammatory IL-4 was lower, the 464 immunomodulatory cytokines IL-1RA, LIF, and RANTES were higher in IMRCs than UCMSCs (Fig.  465   3h) . Amongst the pro-regenerative cytokines, we found that GM-CSF was similar while VEGFA, MIG 466 and SDF-1α were higher in IMRCs than UCMSCs (Fig. 3i) . To determine the immunomodulatory 467

properties of IMRCs, we examined the effects of IMRCs on PBMCs' proliferation and found that 468

IMRCs significantly inhibited PHA-stimulated PBMCs' proliferation when cocultured together at a 469 ratio of 1:1 (Supplementary information, Fig. S6b ). These results indicate that IMRCs manifest much 470 stronger immunomodulatory and pro-regenerative functions than UCMSCs. 471

Our scRNAseq results also indicated that more than 99% of IMRCs expressed MMP1 compared with 473 primary UCMSCs (< 1%) ( Fig. 4a; Supplementary information, Fig. S4) . Similarly, gene expression 474 analysis by RT-qPCR showed that IMRCs express much higher levels of MMP1 than hESCs, UCMSCs 475 or human foreskin fibroblasts (HFF; Fig. 4b ). MMP1 could be detected as a highly secreted protein in 476 the conditioned media of IMRCs by ELISA (Fig. 4c ). In addition, the secreted MMP1 was an 477 enzymatically active and correctly processed isoform (Fig. 4d, e) . To investigate the role of the IMRCs 478 on fibrosis, the human epithelial cell line A549 was used to establish an in vitro fibrosis model. The 479 morphology of A549 cells became myofibroblast-like after 48 h treatment with 10 ng/mL TGF-β1 (Fig.  480 4f). Consistent with the morphological changes, the expression of myofibroblast markers such as 481 ACTA2, Collagen I, Collagen II, Fibronectin and TGFB1 increased, while the expression of the 482 epithelial identity marker CDH1 was extinguished, suggesting that the A549 cells had undergone 483 TGF-β1-induced myofibroblast transdifferentiation (Fig. 4g) . Immunofluorescence and western blot 484 analyses confirmed that TGF-β1 induced the upregulation of collagen I and the downregulation of 485 E-cadherin (Fig. 4h, i) . These results indicated that the A549 alveolar epithelial cells could simulate 486 lung fibrosis in vitro. Thereafter, A549 cells were cultured with or without the conditioned media of 487

IMRCs, along with TGF-β1 treatment. Our results showed that the IMRC conditioned media 488 significantly ameliorated the induction of Collagen I, ACTA2 and TGFB1 expression ( Fig. 4j;  489 Supplementary information, Fig. S7a ). Immunofluorescence and western blot analyses further 490 confirmed that IMRC conditioned media could extinguish Collagen I protein expression and inhibit 491 α-SMA protein expression during the A549-to-myofibroblast transdifferentiation (Fig. 4k-m;  492 Supplementary information, Fig. S7b, c) . These results suggested that IMRCs might be able to inhibit 493 lung fibrosis. 494

Tumorigenicity has long been an obstacle to the clinical application of cells derived from hESCs, due 496 to the contamination of residual hESCs that could form teratomas. Our single cell RNA sequencing 497 showed that no residual OCT4 + /SOX2 + /NANOG + /TERT + /DPPA5 + hESCs remained amongst the 498 IMRCs, and that all pluripotency gene expression had been extinguished (Fig. 5a) . To ensure their 499 short-term and long-term safety, a series of biosafety-related experiments were performed according to Table S1 ). These biosafety-related experiments included testing for bacteria, fungi, mycoplasma, virus 504 (by in vivo and in vitro methods), pluripotent cell residuals, tumorigenicity and biopreparation safety 505 (endotoxin and bovine serum albumin residuals). According to these tests, the safety of the IMRCs has 506 been verified as required by the National Institutes for Food and Drug Control (NIFDC) in China. 507

To track the biodistribution and long-term engraftment of IMRCs in vivo, whole-animal imaging of 508 after injection with DiR-labeled IMRCs (Supplementary information, Fig. S8a ). Our data showed that 510 DiR fluorescence intensity declined by half on the 5th day, and dropped steadily over at least 15 days, 511 disappearing around day 46 (Fig. 5b, c; Supplementary information, Fig. S8c ). No DiR fluorescence 512 signals were observed in the control mouse. These results indicate that IMRCs never show long-term 513 engraftment in vivo. However, IMRCs were predominantly distributed to the lung, with small amounts 514 in the liver and spleen (Supplementary information, Fig. S8b ). To assess the recruitment of IMRCs to 515 the lung in greater detail, GFP-labeled IMRCs were used to track the distribution of IMRCs. A number 516 of GFP-labeled IMRCs were observed in the interstitial spaces in the lung, but not in the lung 517 capillaries (Fig. 5d ). Immunofluorescence analysis of the expression of the alveolar type II epithelial 518 cell marker SPC and the endothelial cell marker CD31 showed that neither marker was co-expressed 519 with GFP-labeled IMRCs in the lungs of mice by day 21 (Fig. 5d) . These data suggested that IMRCs 520 are unlikely to engraft nor transdifferentiate into endothelial or epithelial cells, after homing to the 521 interstitium of lung tissues in vivo. 522

To give some indication of the tumorigenic potential of IMRCs in vitro, a soft agar assay was 523 performed (Fig. 5e) . As a positive control, the PANC-1 pancreas tumor cell line showed a colony 524 formation rate of about 30%. A small number of clones were also formed by hESCs, and the colony 525 formation rate was about 0.5%. However, no colonies were formed by IMRCs. These results indicated 526 that IMRCs had no tumorigenic potential as determined by this assay. Somatic mutation analysis by 527 whole genome sequencing also showed zero mutations in all coding and noncoding exon regions of the 528

IMRCs genome (Fig. 5f ). Tumor formation assays also confirmed that IMRCs could not form any 529 tumors in immunodeficient mice after injection in vivo (Supplementary information, Table S1 ). 530

We also transfused different doses of IMRCs into cynomolgus monkeys (Macaca fascicularis), to 531 evaluate the short-and long-term toxicity of IMRCs in primates. After 6 months, acute toxicity data 532 showed that all blood biochemistry and urinalysis markers remained in the normal range (Fig. 5g, h;  533 Supplementary information, Fig. S9a) , indicating normal liver, kidney, heart, muscular, pancreatic and 534 overall metabolic functions. Moreover, no abnormalities were observed in the long-term toxicity tests 535 either, suggesting minimal xenogeneic rejection responses (Supplementary information, Fig. S9b ; n = 536 18). These results indicated that IMRCs have a robust safety profile in vitro and in vivo, and could 537 potentially provide therapeutic treatments with good safety levels for clinical potential. 538

To evaluate the therapeutic effects of IMRCs on lung injury and fibrosis, IMRCs were administered 540 intravenously into a bleomycin-induced model of lung injury (Fig. 6a) . IMRCs ameliorated the total 541 body weight reduction in mice exposed to bleomycin (BLM)-induced lung injury, in a dose-dependent 542 manner (Fig. 6b) . Kaplan-Meier survival curves indicated that IMRC treatment prolonged the overall 543 survival rates (BLM group 12.5% vs. 1 × 10 6 IMRCs 25%, 3 × 10 6 IMRCs 50%, 5 × 10 6 IMRCs 62.5% 544 (P < 0.01)) and the median survival time (BLM group 11.5 d vs. 1 × 10 6 IMRCs 15.5 d, 3 × 10 6 IMRCs 545 18.0 d, 5 × 10 6 IMRCs 21.0 d) in mice exposed to bleomycin-induced lung injury. There was a 546 statistically significant difference between the BLM group and the 5 × 10 6 IMRCs group, but not the 547 other groups (Fig. 6c) . Histological staining of the lung at day 21 after bleomycin injection 548 (inflammation phase) showed diffuse pneumonic lesions with loss of the normal alveolar architecture, 549 septal thickening, enlarged alveoli, and increased infiltration of inflammatory cells in the interstitial 550 and peribronchiolar areas, in the BLM lung compared with normal lung. IMRC treatment reduced 551 alveolar thickening in the lung in a dose-dependent manner (Fig. 6d-f) . Moreover, IMRC treatment 552 reduced the number of macrophages in the lung (Fig. 6f) . These results indicated that IMRC treatment 553 can reduce inflammation in the lung after acute injury. Moreover, IMRC treatment also improved the 554

Ashcroft score for pulmonary fibrosis in a dose-dependent manner (Fig. 6g) . In particularly, IMRC 555 treatment decreased collagen deposition in the BLM lung in a dose-dependent manner (Supplementary 556 information, Fig. S10b, f) . The expression levels of COL I, FN and α-SMA were also significant lower 557 after IMRC treatment (Supplementary information, Fig. S10 ). ELISA showed that IMRC treatment 558 reduced both TNF-α and TGF-β1 levels in the BLM lung in a dose-dependent manner (Fig. 6h, i) . 559

These results suggested that IMRCs can significantly reduce inflammation and fibrosis after lung injury, 560 in a dose-dependent manner. 561

Pirfenidone (PFD) is an FDA-approved drug for the treatment of idiopathic pulmonary fibrosis. It 563 works by reducing lung fibrosis through downregulating the production of growth factors and 564 procollagens I and II. To evaluate the efficacy of IMRC treatment, we compared it side-by-side with 565

UCMSCs and PFD treatments of BLM-injured mice (Fig. 6j) . Our results showed that IMRC treatment 566 led to an improved amelioration of the weight reduction after BLM-induced lung injury, compared to 567 UCMSCs or PFD treatments (Fig. 6k) . Kaplan-Meier survival curves indicated that the IMRC 568 treatments also resulted in the highest overall survival rates compared to UCMSCs or PFD treatments 569 (BLM group 39.02%, vs. IMRCs 70% (P < 0.01), UCMSCs 52.63%, PFD 65% (P < 0.05); Fig. 6l ). 570

Moreover, IMRC treatment showed the best improvement in lung morphology (Supplementary  571   information, Fig. S11a ) and the best reduction in BLM-induced edema, as measured by the lung 572 coefficient (lung wet weight/total body weight), compared to UCMSCs or PFD treatments 573 (Supplementary information, Fig. S11b) . 574

Functionally, mice treated with IMRCs showed the best improvement in indices for lung capacity, 575

including pressure-volume (PV), inspiratory capacity (IC), static compliance (Crs) and forced vital 576 capacity (FVC) (Supplementary information, Fig. S12a-d) . IMRC treatment also showed the best 577 reduction in functional indices for lung fibrosis, including respiratory resistance (Rrs), elastic resistance 578 (Ers), tissue damping (G) and tissue elasticity (H) were observed (Supplementary information, Fig.  579 S12e-h). Computed tomography (CT) scans also showed that IMRC treatment led to the best 580 improvement in lung morphology, compared to UCMSCs or PFD treatments (Supplementary 581 information, Fig. S12i, j) . In fact, measurements of the lung volume indicated that mice treated with 582

IMRCs had the best improvement in lung volume compared to UCMSCs or PFD treatments. These 583 results indicated that the IMRC therapy is superior to primary UCMSCs and PDF in treating lung 584 injury and fibrosis. 585

Encouraged by these results, and in response to the urgency of the COVID-19 crisis in China, we 586 pilot-tested IMRC transfusion as compassionate treatments in two severely ill COVID-19 patients who 587

were diagnosed with ALI, as part of an expanded access program. After patient consent, IMRCs were 588 administered intravenously in both severely ill patients. Within 14 days after IMRC transfusion, both 589 the severely ill COVID-19 patients showed significant recovery from pneumonia, tested negative for 590 SARS-CoV-2, and were recommended for discharge. Cytokine analysis of both patients' plasma 591

showed very high levels of inflammatory cytokines initially in days 1-2, as expected of the 592 ALI-induced cytokine release syndrome that was manifesting at the point of treatment (Fig. 6m) . 593

However, by days 4-8, many pro-inflammatory cytokines were suppressed after IMRC infusion, 594

In this study, IMRCs were derived from self-renewing hESC cultures with serum-free reagents. Our 599 results showed that IMRCs, while similar to primary UCMSCs, were superior in their long-term 600 proliferative capacity, hyper-immunomodulatory and anti-fibrotic functions. In addition, the cell 601 diameters of IMRCs were generally smaller than UCMSCs, suggesting that they pose lower risks for 602 Products" and the IMRCs were verified as suitable for use in human therapy. Our data showed that 614 there was no teratoma formation observed after IMRC injection into the testes of NOD-SCID mice. In 615 addition, single cell RNA sequencing data showed no residual hESCs were detected in IMRCs and soft 616 agar assays also showed that no colonies were formed by IMRCs. In addition, the somatic mutation 617 analysis by whole genome sequencing also showed no mutations in proto-oncogenes and tumor 618 suppressor genes, nor any coding and non-coding exon regions of IMRCs. These results indicated that 619

IMRCs have no demonstrable potential to form tumors. Moreover, IMRCs could not undergo 620 long-term engraftment and disappeared around day 46. IMRCs did not transdifferentiate into 621 endothelial cells nor alveolar epithelial cells in the lung. To evaluate the risks of clinical application, 622

short-and long-term toxicity tests were performed in cynomolgus monkeys. After 6 months, both acute 623 toxicity tests and long-term toxicity test data showed that no abnormalities were observed. These 624 results proved that IMRCs could have an excellent safety profile and clinical potential. Moreover, the 625 above data suggested that the therapeutic efficacy of IMRCs may not be due to long-term engraftment 626 and direct repair of tissue, but due to their hyper-immunomodulatory and anti-fibrotic functions at 627 inflammatory sites of tissue interstitium. stimulation with the pro-inflammatory cytokine IFN-γ, even more than primary MSCs. Genome-wide 636 RNA sequencing revealed that IMRCs and UCMSCs had different expression profiles after exposure to 637 IFN-γ. Analysis of the secretomes of IMRCs and UCMSCs confirmed at least 9 pro-inflammatory 638 cytokines that were lower, and several anti-inflammatory cytokines and pro-regenerative factors that 639 were higher in IMRCs than UCMSCs. All these results suggested that IMRCs possess a 640 hyper-immunomodulatory capacity compared to primary UCMSCs, after stimulation with IFN-γ. Thus, 641 these cells can be administered to immunocompetent animals and human patients without the need for 642 further immunosuppression. 643

Moreover, higher levels of MMP1 were secreted by IMRCs than primary UCMSCs. MMP1 plays a 644 very important role in the process of fibrosis, by degrading existing collagens and promoting the early 645 stages of tissue remodeling that are critical for the progression of fibrogenesis 39,40 . In general, an 646 imbalance between MMPs and TIMPs is the direct cause of fibrosis and tissue scarring. It has been 647 reported that transplantation of human MMP1-overexpressing bone marrow-derived mesenchymal 648 stem cells can attenuate CCL4-induced liver fibrosis in rats 41 . Therefore, it is conceivable that IMRCs 649 that highly express MMP1 may also reverse lung fibrosis. Accordingly, our data showed that the 650 secretomes from IMRCs could reduce collagen I levels during fibrogenesis induced by TGF-β1. for the hESC-derived IMRCs. These efforts provide considerable grounds for hope that this artificial 679 cell-type could provide significant clinical benefit in the treatment of inflammatory and fibrotic disease 680 conditions. 681

inflammation and fibrosis after acute lung injury in vivo. IMRCs significantly improved the survival 684 rate in a dose-dependent manner. Furthermore, the mechanism for amelioration of pulmonary injury 685 may be mediated by the IMRCs' paracrine action rather than their potential to differentiate and replace 686 the damaged alveolar epithelial cells. The IMRCs' functional inhibition of TGF-β1-induced fibrosis 687 could also be part of its mechanism. IMRCs were superior to UCMSCs and pirfenidone in therapeutic 688 efficacy against lung injury and fibrosis. Furthermore, IMRCs showed an excellent safety profile in 689 both mice and monkeys. In light of recent public health crises involving pneumonia, respiratory failure, 690 ALI and ARDS, our pre-clinical results indicate that IMRCs are ready to be carefully considered for 691 human trials in the treatment of lung injury and fibrosis. 692 

Cell-based therapy in lung regenerative medicine. Regenerative medicine 712 research

Acute Respiratory Distress Syndrome: Advances in 714 Diagnosis and Treatment

Incidence and outcomes of acute lung injury

Epidemiology, Patterns of Care, and Mortality for Patients With Acute 718

Respiratory Distress Syndrome in Intensive Care Units in 50 Countries

Tumor necrosis factor-alpha and angiostatin are mediators of endothelial 721 cytotoxicity in bronchoalveolar lavages of patients with acute respiratory distress 722 syndrome

An official ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fibrosis: 724 evidence-based guidelines for diagnosis and management

The acute respiratory distress syndrome: fibrosis in 727 the fast lane

Transforming growth factor-beta: a mediator of 729 cell regulation in acute respiratory distress syndrome

A phase 3 trial of pirfenidone in patients with idiopathic pulmonary fibrosis

Efficacy and safety of nintedanib in advanced 734 idiopathic pulmonary fibrosis

Research progress of mesenchymal stem cells in the treatment of

Cell Therapy in Idiopathic Pulmonary Fibrosis( †)

Interleukin 1 receptor antagonist mediates the antiinflammatory and 740 antifibrotic effect of mesenchymal stem cells during lung injury

Mesenchymal stem cells: mechanisms of potential 744 therapeutic benefit in ARDS and sepsis

Mesenchymal stem (stromal) cells for treatment of ARDS: a phase 1 clinical trial

Embryonic stem cell lines derived from human blastocysts

Derivation of multipotent mesenchymal 750 precursors from human embryonic stem cells

A mesoderm-derived precursor for mesenchymal stem and 755 endothelial cells

Accreditation of Biosafe Clinical-Grade Human Embryonic Stem Cells According 757 to Chinese Regulations

Multilineage potential of adult human mesenchymal stem cells

Mesenchymal stem cell perspective: cell biology to clinical progress. 761 NPJ Regenerative medicine

Mapping human pluripotent stem cell differentiation pathways using high 763 throughput single-cell RNA-sequencing

HISAT: a fast spliced aligner with low memory 765 requirements

Transcript assembly and quantification by RNA-Seq reveals unannotated 767 transcripts and isoform switching during cell differentiation

Systematic and integrative analysis of large 770 gene lists using DAVID bioinformatics resources

Gene set enrichment analysis: a knowledge-based approach for 772 interpreting genome-wide expression profiles

Efficacy and safety of ex vivo cultured adult human mesenchymal stem 775 cells (Prochymal) in pediatric patients with severe refractory acute graft-versus-host 776 disease in a compassionate use study. Biology of blood and marrow transplantation : 777 journal of the American Society for Blood and Marrow Transplantation

Utilization of the

Soft Agar Colony Formation Assay to Identify Inhibitors of Tumorigenicity in Breast 781

SIRT6 deficiency results in developmental retardation in cynomolgus 783 monkeys

BEDTools: a flexible suite of utilities for comparing genomic 785 features

Fluorescence molecular tomography of DiR-labeled mesenchymal 787 stem cell implants for osteochondral defect repair in rabbit knees

Tumor tropism of intravenously injected human-induced pluripotent stem 790 cell-derived neural stem cells and their gene therapy application in a metastatic breast 791 cancer model

Nonlinearity of respiratory 793 mechanics during bronchoconstriction in mice with airway inflammation

euthanasia on metabolomics of mammalian tissues: studies in a C57BL/6J mouse model. 797 PloS one

Derivation of clinically compliant MSCs from CD105+, CD24-differentiated 799 human ESCs

Novel embryoid body-based method to derive mesenchymal stem cells from 801 human embryonic stem cells

How mesenchymal stem cells interact with tissue immune responses. Trends in 803 immunology

Differential expression of matrix-metalloproteinase-1 and -2 genes in 805 normal and fibrotic human liver. The American journal of pathology

Expression of tissue 808 inhibitor of metalloproteinases 1 and 2 is increased in fibrotic human liver

Transplantation of human matrix metalloproteinase-1 gene-modified bone 811 marrow-derived mesenchymal stem cell attenuates CCL4-induced liver fibrosis in rats. 812 International journal of molecular medicine

Idiopathic pulmonary fibrosis: pathogenesis and management. Respiratory 814 research

Adult stem cells for acute lung injury: 816 remaining questions and concerns

Route of Delivery Modulates 818 the Efficacy of Mesenchymal Stem Cell Therapy for Myocardial Infarction: A

Cell therapy trials for lung diseases: progress and cautions

Gene set enrichment analysis (GSEA) of the top up-regulated gene signature in IMRCs, compared with 856 primary UCMSCs. e GSEA of the top down-regulated gene signature in IMRCs

Heatmaps of specific gene expression amongst single IMRCs groups. g Quantification of 858 non-mesenchymal marker gene expression amongst single IMRCs, UCMSCs and hESCs, as measured 859 by single cell RNAseq