Molecular Determinants of Vascular Heterogeneity for Engineering Tissue-Specific Endothelium for Organ Regeneration Sina Y. Rabbany, Michael Ginsberg, Dan Nolan, and Shahin Rafii - Weill Cornell Medicine/Hofstra Univerity
Abstract Organ regeneration promises unlimited access to replacement tissues. Cultured stem cells can repopulate damaged organs but the regeneration is short-lived and is usually accompanied by fibrosis. To circumvent these hurdles, we have found that tissue-specific adult endothelial cells (ECs) are unique instructive vascular niche cells that produce paracrine “angiocrine factors” to directly induce organ regeneration (Figure 1). This notion has also revealed the remarkable heterogeneity of the adult vasculature that is underscored by production of tissue-specific angiocrine factors necessary for orchestrating organ regeneration. We show here that tissue-specificity of the ECs are determined by hierarchical expression of transcription factors (TFs) and organ-specific microenvironmental cues. We have generated generic ECs by reprogramming of amniotic cells into vascular ECs (rAC-VECs). These generic rAC-VECs have allowed us to identify TFs and cues that generate organ specific ECs. We expect that transplantation of tissuespecific ECs will establish a novel approach to promote organ regeneration without provoking fibrosis.
Figure 1: Cross-talk between tissue-specific capillaries and repopulating stem and progenitor cells support organregeneration. Tissue-specific repopulating stem cells produce angiogenic factors and cues to regulate the activation state of ECs. Primed tissue-specific ECs respond by deploying angiocrine factors to promote organ regeneration. Transplantation of ECs engraft into injured tissues and support organ regeneration.
Models for deciphering the transcriptional hierarchy that regulates tissue-specificity of ECs: Pluripotent embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSC) can be coaxed to differentiate into generic ECs. These ECs have a signature similar to mature ECs, allowing identification of TFs that regulate EC specification. By comparing the molecular profile of mature ECs to their mesodermal precursors, we identified three key E-twenty six (ETS) TFs, ER71, Fli1 and Erg1 that specify ESCs into ECs (Figure 4). However, only a limited number of stable ECs can be derived using this approach, restricting their utility for high throughput analyses. Furthermore, direct reprogramming of adult fibroblasts into ECs is inefficient.
Figure 4. Pluripotent stem cells and adult fibroblasts fail to reprogram into functional endothelial cells. Transduction of ESCs, iPSC or adult somatic cells with TFs, ETV2, FLI1 and ERG1 are ineffective in reprogramming these cells into functional endothelial cells.
Alternative approach: Transcriptional reprogramming of mature amniotic cells into vascular ECs (rAC-VECs): Amniotic cells obtained from the fluid of mid-gestation fetuses contain lineage-committed epithelial/mesenchymal cells that are amenable to transcriptional reprogramming into rAC-VECs (Reference 1) (Figure 5) . We show that transient expression of ETV2 for two weeks and TGFb inhibition for 3 weeks along with constitutive co-expression of FLI1 and ERG1 is sufficient to reprogram amniotic cells into rAC-VEC.
to TF binding sites, we identified groups of TFs that might serve as tissue-specifying EC TFs. We then assessed the expression of these TFs in the various tissue-type specified ECs, to identify those that are differentially expressed and whose expression could explain organspecific programming of the ECs. We have begun to validate these data using ChIP analyses to show organ-specific TF occupancy of these motifs in tissue-type specified ECs.
Induction of TFs by Microenvironmental Cues Extravascular cues confer ECs their tissue-specific attributes by switching on organotypic TFs. To study the influence of biomechanical cues on EC behavior, we use a parallel-plate flow chamber that phenocopies the in vivo microenvironment experienced by ECs. This approach subjects rAC-VECs to laminar or pulsatile/oscillatory shear stress, recreating the dynamic environment of blood vessels in vitro. In addition, we have demonstrated that subjecting cells to a more compliant substrate, rather than the traditional cell culture practice of seeding cells onto rigid polystyrene dishes, yields more robust, physiologically relevant results. In this way, microenvironmental cues—via induction of tissue-specific TFs—can “educate” transplanted, naïve ECs by specifying their organotypic function. Our novel approach to induce TFs that trigger programming of rAC-VECs into organ-specific endothelium opens the door to therapeutic approaches for organ regeneration. Shear stress: Substrate elasticity: Matrix composition: Oxygen tension:
Steady state Laminar flow Rigid Collagen/fibronectin Normoxia
Organ regeneration Oscillatory Soft Laminins Hypoxia
Approach Blood vessels in adult organs are not just passive conduits for delivering oxygen and nutrients, but active participants in organ function. Tissuespecific angiocrine factors are produced at a steady state to sustain organ homeostasis. Organ injury stimulates ECs to alter their angiocrine repertoire and elaborate tissue-specific factors that drive the regeneration of repopulating cells, including liver hepatocyte, lung epithelial cells, spermatogonial stem cells and hematopoietic stem and progenitor cells. These data collectively point to the functional and angiogrine heterogeneity of ECs in adult organs. Notably, ECs alien to a specific organ will fail to induce regeneration in a different organ because they do not possess the proper homing zip codes, functional and angiocrine repertoire. Indeed, we found that lung ECs, and not liver ECs, support lung regeneration. These data illustrate the remarkable heterogeneity of the vasculature.
Figure 8: Induction of transcriptional factors by microenvironmental cues including shear stress, substrate elasticity, ECM composition, and oxygen tension.
Future Directions Figure 5. Efficient direct reprogramming of amniotic cells into functional generic ECs. Lentiviral transduction of mature monlayers of mid-gestation human amniotic cells with ETV2, ERG1 and FLI1 in the presence of neutralizing monoclonal antibodies to TGFb 1 to 3 generates clonal rAC-VECs that express VE-cadherin, VEGFR2 and CD31. Note, junctional localization of the VE-cadherin (Green Fluorescense).
Clonally derived rAC-VECs display a transcriptome profile similar to generic ECs: Transcriptome sequencing (RNA-seq) was performed on rACVECs derived from ACs. These genome-wide analyses demonstrated of rAC-VECs were then compared to the transcriptomes of HUVECs, LSECs, and non-EC types, including CD34+ hematopoietic cells (‘CD34+’), bone marrow stromal cells (‘BMS’), and naïve control ACs (‘Amni ctrl’) (Figure 6). A significant number of vascular genes were upregulated in non-clonal (‘rACVEC’) and clonal (‘rAC-VEC clone-3’ and ‘rAC-VEC clone-4’) derived rACVECs compared to naïve ACs. Genome-wide analyses demonstrate that enforced expression of ETV2, FLI1 and ERG1 with TGFβ inhibition reprogram ACs into mature rAC-VECs (while silencing non-vascular genes) that resemble authentic mature ECs in global gene expression.
Figure 2: Tissue-specific ECs promote organ regeneration by producing defined angiocrine factors. Each tissue-specific ECs produce defined angiocrine factors. Only transplantation of lung ECs, but not liver ECs restores regeneration of the epithelial cells in mice that are resistant to organ regeneration.
Intrinsic and microenvironmental determinants of vascular heterogeneity: Tissue-specific combinatorial activities of transcription factors (TFs) and extra-vascular cues play key roles in dictating vascular and angiocrine heterogeneity (Figure 3). Tissue-specific extra-vascular cues may specify ECs. For example, we have shown that hematopoietic cells determine the morphology and phenotypic attributes of marrow ECs via production of angiopoietins, thrombospondins, and the protease MMP9. Biomechanical forces and oxygen tension alter EC phenotypes, but the roles of these processes in EC specification are unclear.
Identification of TFs that may drive tissue-type EC specification: To identify candidate TFs that might drive EC organ specificity, we mined the transcriptional repertoire of ECs isolated from multiple organs. A de novo DNA motif discovery analysis of the promoters (1kb from the initiation codon) of all differentially transcribed genes revealed groups of known and unknown DNA motifs that were over-represented in tissue-genes of various organs (Figure 7). By comparing high scoring DNA motifs in each cell type
Results To exploit the potential of ECs for therapeutic organ regeneration, we need to determine how ECs attain and perform tissue-specific supportive functions. Indeed, the TFs and microenvironmental cues that establish adult vascular and angiocrine heterogeneity have eluded scientists for decades. Prior work has been handicapped by a lack of technologies and genetic models to purify and culture abundant tissuespecific ECs for molecular profiling and by absence of models to assess the instructive function of ECs during organ regeneration.
Figure 9: Transcriptional programming of naive ECs into tissue-specific ECs: Human and mouse ESC-derived and amniotic cells-derived ECs will be transduced with combinat-ions of TFs to confer tissuespecificity.
Conclusions
Figure 6. Genome wide molecular profiling of rAC-VECs.
Figure 3: Determination of tissue-specific vascular heterogeneity by TFs and organ-specific external microenvironmental cues.
In vivo tissue-specific vascular education: Labeled rAC-VECs will be injected into mice who have undergone hepatectomy, pneumonectomy or chemotherapy. After ~one month -labeled ECs will be repurified from various organs for molecular (RNA-Seq and proteomic) and functional analyses. We will evaluate the extent to which the “educated” ECs have acquired molecular phenotypes of native tissue-specified ECs using standard bio-informatic/statistical measures. We will focus on the expression of putative tissuespecifying TFs in the “educated” ECs and determine to what extent microenvironmental education confers tissue-specific function. For instance, naïve ECs that have homed to liver are expected to express liver-type EC transcripts and to rescue liver regeneration in Id1deficient mice. This approach should permit iterative discovery of tissue specifying TFs by validating the putative TFs identified in Figure 9 and uncovering new factors that correlate with tissue specificity.
• Each organ is vascularized by specialized capillary ECs producing defined angiocrine growth factors. • The expression of tissue- and vascular- specific TFs and external cues establishes the unique structural, phenotypic and functional heterogeneity of the endothelial cells. • Generic rAC-VECs and iVECs provide for generic programmable ECs to identify TFs and cues dictating vascular-type tissue specificity. • Transplantation of engineered tissue-specific ECs stimulates organ regeneration and repair (Figure 10).
Purify organ specific endothelial cells (8 organs)
Molecular profiling RNA-Seq De novo motif discovery Figure 7. Identification of tissue-specific TFs via de novo motif discovery.
Figure 10. Tissue-specific EC transplantation and engraftment model: Mice with EC-specific regenerative defects will be stressed to undergo repair. The capacity of tissue-specific ECs to restore organ regeneration will be examined via studying the following stages of engraftment: Stage 1: Sustain vascular identity during propagation. Stage 2: Home and engraft into injured organ. Stage 3: Deploy the proper stoichiometry of stimulatory and anti-fibrotic angiocrine factors to initiate organ repair.
Ginsberg et al., Efficient direct reprogramming of mature amniotic cells into ECs by ETS factors and TGFβ suppression. Cell 151:559-75, 2012.