Robust Differentiation of Human Pluripotent Stem Cells into Lymphatic Endothelial Cells Using Transcription Factors

The lymphatic system, a crucial component of the vascular network, plays an indispensable role in maintaining tissue fluid homeostasis, immune cell trafficking, and lipid absorption [1, 2]. Dysfunction or impairment of lymphatic vessels underlies various pathological conditions spanning neurodegenerative diseases, metabolic syndrome, cardiovascular ailments, and lymphedema [3]. Despite its critical role, therapeutic and disease modeling strategies focused on generating new lymphatic vessels remain relatively underexplored [4, 5].

One promising avenue in addressing this gap lies in harnessing human induced pluripotent stem cells (hiPSCs) for the differentiation and generation of lymphatic endothelial cells (LECs). The development of efficient protocols for generating LECs from hiPSCs is crucial for advancing our understanding of lymphatic biology and for exploring new therapeutic approaches for lymphatic-related diseases. While considerable progress has been made in directing hiPSCs toward blood endothelial cells [6‒10], the protocols to effectively differentiate hiPSCs into LECs have been somewhat elusive, hindering advancements in lymphatic-related research and therapeutic interventions. There exists a handful of differentiation protocol to generate LEC from hiPSCs mostly relying on embryonic body (EB) intermediate or the use of murine feeder layers [11‒13]. Existing protocols, however, are far from optimal. Limitations stem from the inherent complexity associated with improper cell aggregation and interaction with extraneous cell population [14]. These result in very low yield and poor isolation of final differentiated cells. Moreover, these protocols involve prolonged culture periods and lacks definite functional behavior [15].

Transcription factors, particularly ETS2 and ETV2, have emerged as pivotal regulators governing embryonic vascular development, including the specification of LECs[16]. ETV2 has shown promise in generating blood endothelial cells efficiently [17‒20]. One recent study even shows that ETV2 is required for lymphangiogenesis and directly regulates VEGFR3/FLT4 expression [21]. Using in vitro differentiated mouse embryonic stem cells, ETV2 ChIP-Seq analysis revealed specific ETV2 binding peaks present within VEGFR3 and LYVE-1 promoter/enhancer regions [22]. The VEGFR3 promoter is a likely direct target of ETV2, containing an evolutionarily conserved FOX: ETS domain that is bound by ETV2 and FOXC2 transcription factors [23]. On the other hand, the expression and colocalization of ETS2 was identified in the nuclei of LECs [24]. In addition, the work highlights the synergistic enhancement of ETS2 and PROX1 in expression of VEGFR3. Consistent with the effects on expression profile of VEGFR3, ETS2 induces LEC migration toward VEGF-C. In summary, ETS2 is reported as a pivotal pro-lymphangiogenic factor in collaboration with PROX1 during lymphangiogenesis. Though a fair number of studies exist to show that ETV2 and ETS2 are important regulatory components of LECs, the potential of these factors in the differentiation process toward LECs remains unexplored.

This study aims to address this gap by presenting a novel methodology that reliably and efficiently generates LECs from diverse hiPSC lines using transcription factors – ETS2 and ETV2. The critical temporal activation of ETV2 and ETS2 is emphasized in this protocol, as it enables the essential interaction with Prox1, a master regulator pivotal in lymphatic specification [25]. The resulting differentiated LECs express key lymphatic markers such as VEGFR3, LYVE-1, and Podoplanin, exhibiting expression levels comparable to mature LECs. Moreover, the differentiated LECs, following 3 days of culture in hydrogels, demonstrate their capacity to assemble into stable lymphatic vascular networks in vitro. Additionally, these cells exhibit the ability to secrete reelin, a crucial lymphangiocrine [26], further highlighting their functional attributes. Overall, this protocol not only offers significant implications for advancing our fundamental understanding of lymphatic biology but also holds substantial promise for various applications, including approaches in lymphatic regeneration and personalized medicine. The methodology detailed herein lays a robust foundation for further exploration and development of targeted therapies for a spectrum of lymphatic-related disorders and diseases.

Lentiviruses were produced as previously described [27] in HEK293T cells (ATCC) by cotransfection with three helper plasmids (pRSV-REV, pMDLg/pRRE and vesicular stomatitis virus G protein expression vector) with 12 μg of lentiviral vector DNA and 6 μg of each of the helper plasmid DNA per 75 cm² culture area) using calcium phosphate transfection. Lentiviral vector DNA plasmids used are pSIN4-EF1a-ETV2-IRES-Puro (61061, Addgene) [28], pLV-tetO-Ets2 (70272, Addgene) [29] and FUW-M2rtTA (20342, Addgene) [30]. Lentiviruses were harvested with the medium 46 h after transfection, aliquoted, and frozen at −80°C. Details of lentiviral constructs are available in Supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000539699) and online supplementary Figure 1. Only virus preparations with >90% infection efficiency as assessed by GFP expression or puromycin resistance were used for experiments (online suppl. Fig. 2).

Four different hiPSC lines derived from various tissue origins (online suppl. Table 2) and were obtained from WiCell Research Resources (WiCell, WI). The cells were maintained on growth factor reduced Matrigel (Corning) in mTeSR™ Plus medium (Stemcell Technologies). The hiPSC colonies were checked regularly and passaged when reached around 70% confluency. Accutase (Innovative Cell Technologies) was used for dislodging the cells and depending on the cell line; ROCK inhibitor (10 µm) was used on the first day. Human iPSCs were routinely examined for pluripotent markers using immunofluorescence staining and flow cytometry analysis for TRA-1-60, TRA-1-81, SSEA4, and OCT4 (online suppl. Table 3). All cell lines were routinely tested for karyotyping and mycoplasma contamination; they expressed normal karyotype and were negative for mycoplasma throughout this study.

The hiPSCs were passaged as usual and resuspended with mTeSR™ Plus (Stemcell Technologies) media and mixed with viral vector solutions. The cells were then seeded on Matrigel coated plate (20,000–30,000 cell/cm²). The next day the media was changed into STEMdiff™ APEL™2 (Stemcell Technologies) containing 4 μm CHIR99021(Stemcell Technologies). The ETS2 (co-transduced with rtTA) was activated the next day by addition of doxycycline to the cell culture media and ETV2 transduced cells were selected through puromycin treatment. This time the media was switched to APEL2 media containing 5 ng/mL VEGF-A and 10 μm SB431542 (Stemcell Technologies). After 48 h the cells were seeded on fibronectin-coated plates and cultured in MV2 media (PromoCell) containing 100 ng/mL VEGF-C. This treatment continued for 3 more days, and the cells were then used for various assays. Growth factor-based endothelial cell differentiation protocol was adopted from [31]. After EC differentiation, the cells were treated with 100 ng/mL VEGF-C similar to the treatment of transduction-based differentiation.

To visualize the lymphatic protein expression, LECs and differentiated LECs were seeded on tissue culture plastic around 60% confluency. Samples were fixed with 4% paraformaldehyde, blocked with 1% BSA, permeabilized with 0.1% Triton- X, and stained for Prox1 and ERG (online suppl. Table 3). Samples were rinsed twice in PBS and counterstained with 300 nm DAPI (Thermo Fisher). All samples were imaged using Nikon AX-R confocal at ×20 magnification.

Differentiated LECs were analyzed for lymphatic markers using flow cytometry (FACS) following standard procedures. Briefly, cells were trypsinized and centrifuged following resuspension in FACS buffer. Suspended cells (1 × 10⁶ cells) were stained with the antibodies (1 μg/mL) for 30 min at room temperature: Anti-LYVE-1 antibody (R&D systems, FAB20892A), Anti-PDPN antibody APC (Biolegend, 337004), as well as their corresponding IgG isotype controls (online suppl. Table 3). For intracellular staining, the cells were fixed and permeabilized with Foxp3/Transcription Factor Staining Buffer Set (Thermo, 00-5523-00) and then incubated with Anti-Prox1 antibody FITC (Novus Biologicals, NBP1-30045AF488) for 30 min. The cells were washed twice and resuspended in FACS buffer for analysis. Then, the cells were analyzed using flow cytometry (BD LSR FortessaX-20), and the metadata were analyzed using FlowJo.

To analyze the expression of key lymphatic genes in the differentiated LECs, quantitative reverse transcription PCR was carried out in RNA lysates prepared from cells in culture. Three biological replicates (n = 3) were collected per condition and analyzed with real-time qRT-PCR with triplicate readings as previously described [32]. RNA was reverse transcribed using a high-capacity cDNA reverse transcription kit (Thermo Fisher) according to the manufacturer’s protocol. cDNA was then used with the TaqMan Universal PCR Master Mix and Gene Expression Assays for LYVE-1, Prox1, PDPN, VEGFR3, and GAPDH (online suppl. Table 4). Each sample was prepared in triplicate, and the relative expression was normalized to GAPDH and analyzed using the ^ΔΔCt method.

Supernatant samples were collected from cell culture and standardized using Coomassie Protein Assay Kit (ThermoFisher Scientific). Then standardized protocol for ELISA kits (ab100664 – VEGFC Human ELISA Kit, ab284620 – Human Reelin SimpleStep ELISA^® Kit) were followed with the samples diluted to correct values. Briefly, standard solutions of target protein were prepared in a two-fold dilution series in assay diluent (online suppl. Fig. 3). Supernatant samples and standards were added to respective wells in triplicate and incubated. Then subsequently detection antibodies, enzyme conjugate, and substrate solution were added and finally the absorbance was measured at 495 nm using a microplate reader.

Tube formation assay was performed as previously described [33, 34]. Briefly, human differentiated LECs were stained CellTracker™ Green (ThermoFisher) plated at a density of 100,000 cells/cm² in one well of a µ-Slide 15 Well 3D plate (ibidi) on top of solidified Matrigel (10 μL) with differentiation media [34, 35]. After 6 h, cells were imaged using a fluorescence microscope. Numbers of branches were counted by AutoTube script in MATLAB [36].

Unless otherwise stated, data were expressed as means ± SD of the mean. Statistical analyses were performed with GraphPad Prism 10.1.2(324). For each condition, at least three independent experiments were performed with three biological replicates. Statistical comparisons were made using Student’s t test for paired data, analysis of variance (ANOVA) for multiple comparisons, and with Tukey post hoc analysis for parametric data. Specifically, the Student’s t test was used to analyze differences between protein expression and gene expression among different transduced groups. Significance levels were set at the following: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

We developed a two-dimensional, feeder-free, and chemically defined protocol that relies on a timely transition of hiPSCs through three distinct stages (Fig. 1). First is the conversion of hiPSCs into intermediate mesodermal progenitor cells (MPCs), which is mediated by the activation of Wnt and Nodal signaling pathways using the glycogen synthase kinase 3 inhibitor CHIR99021. Second, we converted the mesodermal progenitor cells into iECs (transduced endothelial cells). This is done by the transduction and activation of exogenous ETV2/ETS2. Third, we treat the cells with vascular endothelial growth factor C (VEGF-C) and transforming growth factor-β (TGF-β) inhibitor. Our stepwise protocol rapidly and uniformly converted hiPSCs into iLECs (transduced LECs). We achieved maximum 88.5% efficiency (CD144⁺/Podoplanin⁺) in driving the cells into LEC lineage. In contrast when endogenous ETV2 was activated via VEGF signaling, the differentiation efficiency was much lower (less than 30%, data not shown). Conversion efficiency of iECs was dependent on amount of ETV2/ETS2 utilized and thus affecting the EC-LEC conversion as well. We tested 3 different conditions to see how the differentiation efficiency was impacted by ETV2, ETS2 or combination of both. Transduction using ETV2 produced 69.5% Prox1⁺ cells (Fig. 2a), while transduction using ETS2 produced 95.0% Prox1⁺ cells (Fig. 2b). Transduction using both ETV2 and ETS2 produced 81.0% Prox1⁺ cells (Fig. 2c). Collectively, we found that ETS2 transduced group had the most efficiency followed by the combination of ETS2 and ETV2 group, and lastly ETV2 alone resulted in the least number of differentiated cells.

Next, we further examined the lymphatic endothelial differentiation of the hiPSCs with FACS analysis for key lymphatic markers. Upon differentiation, the resulting iLECs showcased a robust and mature profile of multiple transmembrane key lymphatic markers, including Podoplanin, LYVE-1 and VEGFR3. Importantly, the levels of these markers in the differentiated LECs closely paralleled those observed in mature LECs, indicating a high fidelity of phenotype and maturity in the generated cell population (Fig. 2d–f). Interestingly, we did see differences in the degree of Podoplanin and LYVE-1 expression in the three transduced groups. From the FACS data, it is clearly observed that ETV2 transduction produced 37.93% Podoplanin⁺ cells (Fig. 2d) whereas ETS2 transduction produced 95.65% Podoplanin⁺ cells (Fig. 2e). The combination of ETV2 and ETS2 group produced 48.23% Podoplanin⁺ cells (Fig. 2f). In terms of LYVE-1 expression, we see the opposite trend. Combination of ETV2 and ETS2 produced 23.3% LYVE-1⁺ cells, ETV2 produced 22.63% LYVE-1⁺ cells, and ETS2 produced 9.0% LYVE-1⁺ cells (Fig. 2d–f).

To characterize the identity of the cells derived from hiPSCs we investigated whether the differentiated cells expressed lymphatic markers at the protein level with proper localization. For endothelial cells, ERG (ETS-related gene) is expressed in the nuclei of endothelial cells [37], whereas in LECs, PROX-1, a transcription factor, is expressed in the nuclei [25, 38, 39]. As expected, immunostaining results showed that ERG and PROX-1 both were exclusively localized in the nuclei of the differentiated cells confirming correct localization of the key EC and LEC markers in the differentiated cells (Fig. 3).

Apart from protein level expression, we also looked at the gene expression of the common LEC markers, such as LYVE-1, PDPN, Prox1, and VEGFR3 (Fig. 4). Compared to growth factor derived LECs, ETV2 transduced LECs express higher level of lymphatic markers LYVE-1 (1.92 ± 0.78-fold), PDPN (2.84 ± 0.44), Prox1 (2.06 ± 0.64-fold), and VEGFR3 (2.20 ± 0.65-fold), ETV2 and ETS2 transduced LECs express higher level of lymphatic markers LYVE-1 (2.14 ± 0.60-fold), PDPN (7.75 ± 0.56-fold), Prox1 (3.28 ± 0.54-fold), and VEGFR3 (2.57 ± 0.28-fold), and ETS2 transduced LECs express the highest level of lymphatic markers LYVE-1 (3.03 ± 0.78-fold), PDPN (9.12 ± 0.61-fold), Prox1 (4 ± 0.64-fold), and VEGFR3 (3.90 ± 0.37-fold). Consistent with the differentiation efficiency data, ETS2 transduced group express the highest lymphatic markers followed by the combination of ETS2 and ETV2 group, and lastly the ETV2 group express relatively higher lymphatic markers compared to the growth factor derived LECs.

To validate the functionality of the LECs differentiated from hiPSCs under our culture conditions, we performed a series of in vitro studies [40]. First, we tested in vitro activities of iLECs using a tube formation assay (Figure 5a–c, online suppl. Fig. 4). Cells positive for LYVE-1 and Podoplanin were isolated by MACS from differentiating cell population at day 14, labeled with a fluorescent dye, CellTracker™ Green and subjected to tube formation assay. After 12 h of culture, all hiPSC-derived LECs readily formed tube-like structures (Fig. 5a–c). Although all the cells showed some degree of tube formation capability, ETS2 group formed 33.072 ± 1.268 mm tube length and 109 ± 12.73 branches, followed by ETS2+ETV2 group that formed 28.271 ± 2.381 mm tube length and 106.5 ± 13.44 branches, and ETV2 group that formed 24.005 ± 0.235 mm tube length and 66.5 ± 7.78 branches (Fig. 5d, e). A number of branches are not significantly different among the different conditions, but tube length of ETS2 group was significantly greater than ETV2 group.

Further, we looked at the secretory properties of our differentiated cells. LECs are known for secreting reelin, one of the major lymphangiocrine factors [26, 41, 42]. In our study, we found that our differentiated LECs secrete reelin comparable to primary LECs in the ranges between 10 and 20 ng/mL (Fig. 5f). We also investigated the level of VEGF-C secreted by iLECs. We confirmed that all the iLECs and primary LECs secreted a very negligible amount of VEGF-C, in the range of pg/mL (Fig. 5g). Overall, the hiPSC-derived LECs showed phenotypic characteristics of mature LECs and faithfully recapitulated functional lymphatic behavior.

Promoting the development of new lymphatic vessels has been postulated as an innovative therapeutic strategy for various disease phenotypes [5]. Yet, LECs are difficult to isolate, and they can lose their lymphatic phenotypes during in vitro culture [32]. Therefore, generating LECs from hiPSCs represents an appealing strategy not only for lymphatic regeneration, but also for modeling human diseases in vitro. Thus far, methods to generate LECs from hiPSCs have mainly relied on either EB intermediate or the use of murine feeder layers (i.e., OP9) [43, 44]. While generation of EB recapitulate early embryonic development, they require sorting of differentiated cells and therefore produce lower overall yield [15, 43]. On the other hand, co-culture with murine feeder layers introduces xenogeneic components to the final product [4, 5].

Alternatively, direct differentiation can be achieved using lentiviral transduction of ETV2 and ETS2, which are known to be important for blood endothelial cells [45]. In this study, we compared the differentiation efficiency, lymphatic markers, and functionality of differentiated LECs using transduction of ETV2, ETS2, and combination of ETV2 and ETS2. We discovered that transduction with ETS2 produced the highest efficiency of Prox1⁺ cells, which also express high lymphatic markers, such as LYVE-1 and Podoplanin. These results are consistent with previous studies that show ETS2 interaction with Prox1, the master regulator of lymphatic genes [24, 46]. Transduction with ETV2 produced lower efficiency of Prox1⁺, which express lower lymphatic markers. Interestingly, the combination of ETS2 and ETV2 produced lower efficiency of Prox1⁺ cells, which may suggest the opposite effect of ETS2 and ETV2.

LECs are also known to express different lymphatic markers depending on their locations [47]. Lymphatic capillaries highly express LYVE-1, a receptor for hyaluronic acid, which is important for leukocytes trafficking [48]. Lymphatic collecting vessels express Podoplanin, which can bind to platelet receptor CLEC-2 and is important for blood and lymphatic separation [35]. ETS2 transduced cells highly express Podoplanin, which reflects LECs that reside at the lymphatic collecting vessels [49]. On the other hand, ETV2 transduced cells highly express LYVE-1, which corresponds to LECs that reside at the lymphatic capillaries. Depending on the final applications, different transcription factors may be more suitable to generate different LECs with varying degree of Podoplanin and LYVE-1 expression.

Regardless of the different lentivirus vectors used, all the transduced cells express key lymphatic markers comparable to primary LECs and are able to form lymphatic networks in vitro. The transduced cells also express reelin in the ranges between 10 and 20 ng/mL, which are comparable to primary LECs. Reelin is a key lymphangiocrine, which has been attributed to cardiac regeneration following heart attack [26]. We also confirmed that the transduced LECs secrete low level of VEGF-C, comparable to primary LECs. While VEGF-C is a key soluble factor to promote lymphangiogenesis [50, 51], it is well known that VEGF-C is not secreted by LECs, rather by macrophages and fibroblasts [51]. All these results suggest that the differentiated LECs exhibit the characteristics and functionalities of LECs.

Overall, the current study demonstrated an efficient method of deriving LECs, which may be useful for modeling human diseases in vitro and basic understanding of human development. For instance, differentiated LECs can be cultured in synthetic hydrogels and microfluidic devices to investigate the molecular mechanism underlying lymphatic disorders in patients with down syndrome and lymphatic malformations [4, 52‒55]. Future studies can also generate differentiated LECs that can secrete VEGF-C to provide pseudo-autocrine signaling and further enhance lymphatic regeneration in patients with lymphedema [34, 51]. It is worth noting that while direct differentiation using lentiviral vectors may introduce lentiviral components, future studies can take advantage of modulating ETV2 and ETS2 expression with modified mRNA [20]. Furthermore, promising results from the current work warrants future studies to investigate a stepwise and well-defined method to differentiate LECs from hiPSCs, useful for broad applications in basic study of lymphatic biology, as well as toward various approaches in lymphatic regeneration and personalized medicine [5].

This study protocol was reviewed and approved by the University of Notre Dame Institutional Biosafety Committee, Approval No. 21-05-6605. Written informed consent was obtained from donors to participate in the study to generate hiPSC, which are available through WiCell Research Institute.

The authors have no conflicts to disclose.

We acknowledge support from the University of Notre Dame through “Advancing Our Vision” Initiative in Stem Cell Research, Harper Cancer Research Institute – American Cancer Society Institutional Research Grant (IRG-17-182-04), American Heart Association through Career Development Award (19-CDA-34630012 to D.H.-P.), National Science Foundation (2047903 to D.H.-P.) and from National Institutes of Health (1R35-GM-143055-01 to D.H.-P.).

S.S., C.P., and D.H.-P. conceived the ideas, designed the experiments, interpreted the data, and wrote the manuscript. S.S., F.G., and J.K. conducted the experiments and analyzed data. All authors have approved of the manuscript.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.