Recapitulate Human Cardio-pulmonary Co-development Using Simultaneous Multilineage Differentiation of Pluripotent Stem Cells

The extensive crosstalk between the developing heart and lung is pivotal for their proper morphogenesis and maturation. However, there remains a lack of model systems for investigating the critical cardio-pulmonary mutual interaction during human embryogenesis. Here, we reported a novel stepwise strategy for directing simultaneous induction of both mesoderm-derived cardiac and endoderm-derived lung epithelial lineages within a single differentiation of human pluripotent stem cells (hPSCs) via temporal specific tuning of WNT and TGF-β signaling in the absence of exogenous growth factors. Using 3D suspension culture, we established concentric cardio-pulmonary micro-Tissues (μTs), and observed expedited alveolar maturation in the presence of cardiac accompany. Upon withdrawal of WNT agonist, the cardiac and pulmonary components within each dual-lineage μT effectively segregated from each other with concurrent initiation of cardiac contraction. We expect our multilineage differentiation model to offer an experimentally tractable system for investigating human cardio-pulmonary interplay and tissue boundary formation during embryogenesis.


179
The action of CHIR treatment on hPSC differentiation depends on not only its dosage but also the length of 180 exposure time. 44, 45 We evaluated the efficiency of cardio-pulmonary induction following exposure to CHIR (7 181 µM) for different time spans (24, 48 and 72 hrs), and found that extended CHIR exposure for 48 or 72 hrs was 182 required to induce robust cardio-pulmonary programs (Fig. 1f). Specifically, CHIR favored cardiac specification 183 in a time-dependent manner and plateaued at 48 hrs of treatment (Fig. 1h); while the induction of pulmonary 184 program peaked at 48 hrs of CHIR treatment (Fig. 1g) and declined with further extension of the treatment.

185
Based on these observations, for all subsequent experiments, we used 48-hr treatment of CHIR (7 µM) during 186 Stage-1 of the co-differentiation program. Furthermore, we showed that maintaining hPSCs in mTESR1 Plus 187 during the initial CHIR treatment appeared to be critical for enabling effective cardio-pulmonary differentiation 188 ( Supplementary Fig. 3), as compared to using RPMI1640 supplemented with B-27 minus insulin as the basal 189 medium during CHIR treatment.

191
Exogenous activation of TGF-b and BMP signaling during the very initial steps of hPSC specification has been 192 widely utilized for cardiac 9,13,41 and pulmonary 15,19,23,42,46 specification from hPSCs. Here, we investigated how 193 exogenous and endogenous TGF-b and BMP signaling regulates cardio-pulmonary induction during germ 194 layer induction (Stage-1). TGF-b inhibition (using A8301, Day-2 to Day-4) immediately following CHIR 195 treatment abolished both cardiac and pulmonary induction; while TGF-b activation through Activin A 196 supplementation (Day-2 to Day-4) led to pulmonary-only differentiation outcome ( Fig. 2a,b,c). This suggests 197 the requirement of endogenous TGF-b for cardio-pulmonary induction and that high-level TGF-b activation 198 favors pulmonary instead of cardiac induction. In parallel, BMP inhibition (using DMH-1) during the same time 199 period compromised cardiac induction and mildly reduced pulmonary specification; while exogenous BMP4 200 supplementation enhanced cardiac induction but inhibited pulmonary specification (Fig. 2d,e,f). This indicates 201 that endogenous BMP signaling is primarily required for cardiac induction and that exogenous augmentation 202 of BMP signaling further favors the cardiac lineage at the expense of the pulmonary lineage.
insert for air-liquid interface (ALI) culture or on 2D plastic surface for regular submerged culture, and failed to 272 detect obvious AT2 induction by Day 18 (Fig. 3c Supplementary Fig. 6). Consistent with the observations using 273 fluorescence reporters, NKX2.1 and SFTPC gene expression was significantly upregulated in 3D suspension 274 culture on Day-18 compared to the starting Day-15 cells or cells following ALI maturation (Fig. 3d,e). Our results 275 demonstrated 3D suspension culture as a robust platform to expedite alveolar maturation.

277
To elucidate how the co-induced cardiac lineage modulates the alveolar maturation process, we introduced 278 activin A (20 ng/mL) during germ-layer specification (Fig. 3f), which effectively inhibited mesoderm specification 279 and led to pulmonary-only differentiation outcome on Day-15 (Fig. 2a,b). In the absence of accompanying cardiac 280 cells, although NKX2.1 GFP+ lung progenitors can be robustly induced and maintained, their alveolar maturation 281 (as indicated by SFTPC tdTomato reporter) following 3 days of maturation in 3D suspension culture was dramatically 282 diminished compared to the cardio-pulmonary group (Fig. 3g). Whole mount imaging of µTs on Day 18 showed 283  NKX2.5 + cells (Fig. 3h). This was 289 further supported by gene 290 expression analysis of NKX2.1 291 (Fig. 3i) and SFTPC (Fig. 3j).

292
Further extension of CKDCI 293 maturation period for 2 weeks up 294 to Day-29 in the pulmonary-only 295 group failed to produce AT2 296 induction to a level comparable to 297 the cardio-pulmonary group 298 ( Supplementary Fig. 8 following AT2 establishment on 322 Day-18 in 3D suspension culture 323 (Fig. 4g), we transitioned the 324 maturation medium from CKDCI 325 to KDCI without CHIR (Fig. 4a). 16 326 To our surprise, upon CHIR 327 removal, the cardiac and 328 pulmonary components within 329 each dual-lineage µT, which 330 initially arranged in the 331 pulmonary-centered, concentric 332 manner (Fig. 4b), effectively 333 reorganized over time and 334 eventually segregated from each 335 other (Fig. 4a,b).

337
To quantitatively assess this 338 segregation process, we 339  (e) Histological analysis of cTnT expression on the segregated cardiac and pulmonary µTs, with co-staining of NKX2.5 and NKX2.1. (f) Calcium influx measured using calcium indicator, Xrhod-AM. Scale bar = 125 µm for 20X images. All data are mean ± SD. *p < 0.05; **p < 0.01; ***p < 0.001. performed time-lapse single-µT tracking and calculated the percentage of overlapping between the cardiac and 340 pulmonary tissues by measuring the length of overlapping border between the GFP + and non-GFP components 341 and normalizing it by the total perimeter of the GFP + pulmonary component (Fig. 4c). We compared the 342 segregation process in the presence (CKDCI) and absence (KDCI) of CHIR, and found that although cardio-343 pulmonary segregation took place in both medium conditions, it was significantly expedited by the withdrawal of 344 CHIR (Fig. 4d). To investigate the requirement of endogenous WNT signaling for this segregation process, we 345 introduced inhibitors of canonical (IWP4) and non-canonical (NSC668036, a Dishevelled inhibitor) WNT signaling, 346 54 and did not detect obvious difference in the segregation process as compared to the control KDCI condition 347 (Fig. 4d). In parallel with the cardio-pulmonary segregation, cardiac contraction was observed 7 days following 348 CHIR withdrawal (Supplementary Video 1). Immunohistochemical analysis demonstrated specific co-expression 349 of NKX2.5 and cardiac troponin T (cTnT) in the segregated cardiac µT (Fig. 4e). The contractile function of the 350 segregated cardiac µT was further confirmed via the detection of calcium influx (Fig. 4f, Supplementary Video  351 2). 352 353 354 Discussion 355 356 Here, we described a novel strategy to model human cardio-pulmonary co-development using multi-lineage 357 hPSC differentiation. We demonstrated that upon co-induction of mesoderm and endoderm, a series of shared 358 signaling events were capable of driving simultaneous cardiac and pulmonary specification from their 359 respective germ-layer progenitors. Transitioning the co-induced cardiac and pulmonary progenitors to 3D 360 suspension culture, we observed expedited alveolar maturation within 3 days, which was supported by the 361 accompanying cardiac lineage. In 3D suspension culture, each cardio-pulmonary µT effectively segregates 362 into separate cardiac and pulmonary µTs, which was partially inhibited by WNT activation. This study therefore 363 delivers an effective in vitro model for studying the mechanistic interplay between developing heart and lung 364 during human embryogenesis.

366
The extensive cardio-pulmonary mutual interaction during organogenesis has been well documented in the 367 mouse model, 1,2,4 however, the translatability of these findings to human embryogenesis remains elusive due 368 to the lack of proper model systems. Human PSC differentiation has offered an effective means for 369 recapitulating and investigating human organogenesis, and tremendous progresses have been made towards 370 directed cardiac or pulmonary specification. 7-23,39 However, almost all existing models have been focusing on 371 one parenchymal lineage at a time, and therefore lack the ability to support investigation on inter-organ 372 crosstalk. Here, building on the established understanding of signaling events necessary for cardiac and 373 pulmonary induction, 7-23,39 we have developed a robust protocol for simultaneous cardio-pulmonary co-374 differentiation from hPSCs. Within our co-differentiation system, unrestricted interaction between cells of both 375 lineages is enabled even before their lineage commitment.

377
Most current attempts on pulmonary induction from hPSCs relies on initial TGF-b activation using growth factor 378 (Activin A) supplementation, which is critical for definitive endoderm specification. Here, we showed that by 379 fine tuning of WNT signaling using a small-molecule inhibitor of GSK-3b (CHIR), robust induction of endoderm 380 and subsequently lung progenitors can be achieved without any exogenous growth factor. This is consistent 381 with the observation that CHIR was capable of inducing cardiac differentiation in replacement of combined 382 effect of exogenous Activin A and BMP4. 7 Nonetheless, Nodal and BMP signaling remains crucial in mesoderm 383 and endoderm specification, as inhibition of these signaling abolished effective cardio-pulmonary co-induction.

385
Our study demonstrated the requirement of endogenous TGF-b signaling for effective cardio-pulmonary 386 induction, as well as the critical role of endogenous BMP signaling for cardiogenesis. Furthermore, we found 387 that temporal-specific action of the same set of small molecules regulating TGF-b and WNT signaling was 388 capable of driving mesoderm-to-cardiac and endoderm-to-pulmonary specification in a concurrent manner. 389 Moreover, BMP4 has been shown to improve NKX2.1 + lung progenitor specification in both mouse and human 390 iPSCs. 15,16,47 In our system, endogenous instead of exogenous BMP signaling was required during a 391 developmental stage corresponding to foregut ventralization for effective co-emergence of cardiac and 392 pulmonary progenitors. This is in line with the close spatial positioning of developing heart and lung primordia 393 within embryonic body patterning, which implies their exposure to a similar paracrine microenvironment. 4,55 394 395 To achieve alveolarization, NKX2.1 + lung progenitors are usually embedded in extracellular matrices, such as 396 Matrigel and collagen. 42,46,48 Here, we established an effective approach that enabled AT2 cell maturation 397 within 3 days in suspension culture of 3D cell aggregates spontaneously formed from Day-15 cardiac and 398 pulmonary progenitors. We further demonstrated that the presence of accompanying cardiac lineage is critical 399 for robust alveolar induction. This observation is consistent with the recently reported inter-dependence 400 between cardiac and pulmonary lineages during embryogenesis. 4 In addition, the presence of mesoderm-401 derived stromal cells has been shown to be essential for effective alveolarization in vivo 55-57 and in vitro. 19,43 402 Furthermore, cells of the mesodermal lineage are known to be robust producers of extracellular matrix, which 403 may also contribute to the effective alveolar maturation in the absence of external extracellular matrix support. 404 The ability to enable effective alveolar induction from hiPSC-derived lung progenitors in a convenient 405 suspension culture also opens the door to large scale production of alveolar cells, a critical enabling step for 406 regenerative medicine applications. 407 408 Using dual-lineage cardio-pulmonary µTs formed from the co-induced progenitors, we observed a novel 409 process of cardio-pulmonary tissue segregation. The human body cavities are highly crowded spaces, filled 410 with different tissues and organs that are in close contact with each other. It remains enigmatic how inter-411 organ boundaries are maintained to prevent undesired cell migration or tissue merging. Our cardio-pulmonary 412 tissue segregation model suggests an intrinsic mechanism that effectively establishes a boundary between 413 two distinct parenchymal lineages even when they are initially mingled together. Although no model of 414 collective migration has been described in the context of cardio-pulmonary development, studies in other 415 model systems suggest cell-cell communication and paracrine signaling (e.g. WNT) to be crucial for directed 416 cell migration during development. 58-60 Here we found that exogenous WNT activation via GSK-3b inhibition 417 effectively slowed down the cardio-pulmonary segregation, while inhibition of endogenous WNT (canonical 418 and non-canonical) did not obviously affect the process. In consistence with our observation, it has been shown 419 that inhibition of non-canonical WNT signaling does not stop collective cell migration but distorting migration 420 direction. 54 421 422 In conclusion, our work offers a novel model for investigating the molecular and cellular mechanisms underlying 423 human cardio-pulmonary co-development and tissue boundary formation. We also expect this work to be of 424 potential use for studying congenital diseases affecting both cardiovascular and pulmonary systems, such as Co-maturation of cardio-pulmonary progenitors in air-liquid interface (ALI) culture 475 On Day-15 of cardio-pulmonary co-differentiation, cells were dissociated into single cells using TrypLE Express, 476 and re-plated at 500,000 cells/cm 2 onto the apical side of each 24-well Transwell insert (pore size of 0.4 µm, pre-477 coated with 1% growth factor-reduced Matrigel) in 100 µL maturation medium. Basolateral side of the transwell 478 insert was filled with 500 µL of maturation medium. The maturation medium was basal medium supplemented 479 with 3 µM CHIR99021, 10 ng/mL Keratinocyte growth factor (KGF), 50 nM Dexamethasone, 0.1 mM 8-480 bromoadenosine 3', 5'-cyclic monophosphate (cAMP, AMP-activated protein kinase activator) and 0.1 mM 3-481 isobutyl-1-methylxanthine (IBMX, PKA activator), which was referred to as CKDCI medium. 10 µM Y-27632 was 482 added during the initial 24 hrs following re-plating. The next day, all medium on the apical side was removed. 483 200 µL of fresh CKDCI medium without Y-27632 was added to the basolateral side to establish ALI culture, and 484 was replaced daily. Red fluorescence from the SFTPC TdTomato reporter was examined daily using EVOS Imaging 485 System to monitor the emergence of alveolar type 2 (AT2) cells. On Day-3 of ALI maturation, Transwell 486 membrane were excised from the insert, and analyzed by qPCR (NKX2.1, SFTPC). 487 488 Co-maturation of cardio-pulmonary µTs in 3D suspension culture 489 On Day-15 of cardio-pulmonary co-differentiation, cells were dissociated into single cells using TrypLE Express.

490
A total of 250,000 cells in 500 µL CKDCI maturation medium was transferred into each well of 24-well ultra-low 491 adherence plate and cultured with agitation at 125 rpm to form cardio-pulmonary µTs. 10 µM Y-27632 was added 492 during the initial 24 hrs following re-plating. Following 3 days of culture in CKDCI medium, CHIR99021 was 493 removed and µT culture was continued in KDCI medium for an additional 7 days. At desired time points of 3D 494 suspension maturation, µTs were analyzed by histology (NKX2.1, NKX2.5, cTnT) and qPCR analysis (NKX2.1, 495 SFTPC).

497
Single µT time-lapse imaging and analysis 498 To investigate the segregation of cardio-pulmonary µTs into their respective cardiac and lung µTs, following 3 499 days suspension culture in CKDCI medium in 24-well ultra-low adherence plate, single µT was transferred into 500 each well in 96-well ultra-low adherence plate and cultured for an additional 7 days. The following medium 501 recipes were examined for cardio-pulmonary segregation: KDCI medium, KDCI medium supplemented with 3 502 µM CHIR99021, KDCI medium with 5 µM IWP4, and KDCI medium with 50 µM NSC668036. Time-lapse imaging 503 was performed on Day-18, Day-22 and Day-25 following µT transfer to monitor the segregation process. The 504 pulmonary compartment within each cardio-pulmonary µT was tracked based on the NKX2.1 GFP reporter. To 505 quantify the segregation between the two compartments within each µT. Image J was used to measure the 506 overlapping perimeter between GFP + (pulmonary) and non-GFP (cardiac) compartments, which was then 507 normalized to total perimeter of GFP + compartments and expressed as the percentage of overlapping. 508 509 510 511 512 qPCR analysis 513 Total RNA was extracted using TRIzol, processed by chloroform extraction, precipitated using 1 volume of 514 absolute isopropanol with 50 µg/mL of RNase-free glycoblue as carrier, washed with 75% ethanol, air-dried, 515 solubilized in RNase-free water and quantified using NanoDrop 2000 spectrophotometer. cDNA was synthesized 516 via reverse transcription of 1 µg total RNA with random hexamers and the High-Capacity cDNA Reverse 517 Transcription kit according to manufacturer's instruction. Real-time qPCR analysis was performed on CFX96 518 Touch Real-Time PCR Detection System using TaqMan probes. Each reaction mixture was prepared by 519 combining 1 µL of probe, 10 µL of TaqMan Master Mix, 1 µL of cDNA (equivalent to 50 ng), and the final volume 520 was brought up to 20 µL. The final Ct value was normalized to housekeeping gene (b-actin), using comparative 521 Ct method. Unless otherwise specified, baseline, defined as fold change =1, was set as undifferentiated hiPSCs, 522 or if undetected, a cycle number of 40 was assigned to allow fold change calculations. 42 List of TaqMan probes 523 was summarized in Supplementary Table 3.

525
Immunofluorescence staining on 2D cell samples 526 Cells were fixed with ice-cold methanol, air-dried, rehydrated with phosphate-buffered saline (PBS), 527 permeabilized with 1% (v/v) Triton X-100, blocked in 1% (w/v) bovine serum albumin in PBS (blocking buffer), 528 incubated with primary antibodies diluted in blocking buffer at 4 o C overnight, and incubated with corresponding 529 fluorescence-conjugated secondary antibodies in blocking buffer at room temperature (RT) for 45 min. Nuclear 530 counterstain was performed using Hoechst-33342 (1:500) in PBS. Fluorescence images were acquired using 531 EVOS Imaging System. All antibodies used and their respective dilution were summarized in Supplementary 532 Table 4. 533 534 Histology 535 The µTs were fixed with 4% paraformaldehyde, embedded in HistoGel and then in paraffin. Tissue processing 536 and paraffin embedding was performed in Research Histology Lab of Pitt Biospecimen Core at the University of 537 Pittsburgh Medical Center (UPMC) Shadyside Hospital. Paraffin blocks were sectioned at 5 µm thickness, 538 transferred onto glass slides, rehydrated by sequential incubation in Histoclear, 100% ethanol, 95% ethanol and 539 distilled water. To unmask antigen, slides were treated with Antigen Unmasking Solution at 95 o C for 20 min and 540 cooled down to RT. Immunofluorescence staining was performed as described above for 2D cell samples. After 541 the final wash, slides were mounted with DAPI Fluoromount-G, and imaged using EVOS Imaging System. All 542 antibodies used and their respective dilution were summarized in Supplementary Table 4. 543 544 Contraction and calcium signal 545 To assess contraction of cardiac µT, segregated cardiac µT was stained with 5 µM of Cal-520 AM (AAT Bioquest, 546 21130), a calcium indicator dye. Calcium imaging (500 frames per second) was performed using a Prime 95B 547 Scientific CMOS camera (Photometrics) mounted on an epifluorescent stereomicroscope (Nikon SMZ1000) with 548 a GFP filter and an X-cite Lamp (Excelitas). 549 550 TEM 551 Cardio-pulmonary µTs were fixed in 2.5% glutaraldehyde in 0.1 M PBS (pH7.4) for at least 1 hr. After 3 washes 552 in 0.1 M PBS for 10 min each, the µTs were post fixed in 1% Osmium tetroxide containing 1% potassium 553 ferricyanide at 4 o C for 1 hr, followed by 3 washes in 0.1 M PBS for 10 min each. µTs were dehydrated in graded 554 series of ethanol starting from 30%, 50%, 70%, 90% and finally 100% of ethanol for 10 min each. µTs were 555 further dehydrated epon for 1 hr at RT. This step was repeated for another three times prior to embedding in 556 pure epon at 37 o C for 24 hrs. Finally, the µTs were cured for 48 hrs at 60 o C. The presence of lamellar body in 557 cardio-pulmonary µTs were identified using JEM 1400 Flash TEM. 558 559 Statistics 560