The stem cell niche finds its true north

This year, on May 22-26, the Novo Nordisk Foundation hosted its third Stem Cell Niche meeting at the beautiful Favrholm campus in Hillerød (Denmark) as part of the Copenhagen Bioscience Conference series. Once again, the organizers succeeded in creating a unique experience involving top quality science, gourmet Nordic food, beautiful surroundings and an amazing sightseeing tour of Copenhagen. All of this contributed to creating an extremely pleasant atmosphere that stimulated constant interaction, mingling and scientific networking between the speakers and meeting participants, including via social media (see Box 1). The scientific content of the meeting was organized by Anne Grapin-Botton, Joshua Brickman, Kristian Helin, Palle Serup and Elisabetta Ferretti from the Danish Stem Cell Centre at the University of Copenhagen, and the programme covered almost all aspects of stem cell biology, from pluripotency, epigenetics and single-cell transcriptome analysis to directed differentiation, in vitro organogenesis and translational research. The thread that tied all these topics together was the very high quality of the science, and in this Meeting review, we summarize some of the work presented.

The scientific part of the meeting got underway with a captivating talk by Rudolf Jaenisch (Whitehead Institute, MA, USA), who presented the work of his lab in generating naïve human pluripotent stem cells (hPSCs), with the aim of using these to generate mouse-human chimeric embryos. The group had previously identified chemically defined culture conditions for the induction and maintenance of naïve hPSCs (Theunissen et al., 2014); however, when injecting these cells into the mouse blastocyst, the efficiency of chimera formation was <0.01%, and in these embryos, the human cells accounted for <0.001% of all cells. Jaenisch further presented his group's efforts to create neural crest chimeras through in utero injection of hPSC-derived neural crest cells into the amniotic sac of E8.5 mice. Although the efficiency of chimera formation was 30-40%, the human cell contribution to neural crest lineages was <0.1%. Jaenisch therefore concluded that human-mouse chimera formation was likely to be hampered by basic biological differences between the species. Finally, Jaenisch presented an intriguing cortical organoid model revealing in vitro induction of neural convolution in PTEN mutant hiPSCs. Interestingly, neural convolution was not observed in corresponding mouse organoids, indicating that this model can be used to study anatomical species differences in brain development.

The warm pre-summer Danish weather and a beautiful setting at Favrholm campus provided a stimulating ‘niche’ for meeting participants. Similarly, the niche where stem cells reside has a profound influence over cell fate and behaviour. Studying the niche involves careful examination of multiple components, including the different types of neighbouring cells and the signals from them to the stem cells. Perhaps the best-characterized stem cell niche is that of the bone marrow. Sean Morrison (University of Texas Southwestern, USA) discussed niches of the bone marrow, and using deep imaging and 3D reconstruction showed that both dividing and non-diving haematopoietic stem cells (HSCs) reside in perivascular niches created by endothelial and perivascular stromal cells (leptin receptor positive, LEPR⁺), which were associated mainly with sinusoidal blood vessels (Acar et al., 2015). Morrison further showed that LEPR⁺ cells are the major source of osteoblasts and fat cells in adult bone marrow, as well as cartilage formed after fractures. Conditional deletion of Lepr decreased adipogenesis and increased osteogenesis in limb bones. He suggested that LEPR⁺ cells act as a diet/adiposity sensor in the bone marrow to promote the formation of fat at the expense of bone. Interestingly, Clec11a, a protein expressed by a subset of LEPR⁺ cells, was found to promote osteogenesis in vitro and in vivo, thereby constituting a promising new therapeutic target for osteoporosis treatment.

Simón Méndez-Ferrer (Cambridge Stem Cell Institute, UK) added to discussions of the complexity of the bone marrow niche by presenting evidence of how autonomic sympathetic nerve fibres activate and regulate HSCs in the bone marrow to induce circadian rhythms in HSC mobilization. Mendez-Ferrer hypothesized that the sympathetic nerve terminals mediate their effect through direct activation of Nestin⁺ mesenchymal stem cells (MSCs) in the bone marrow to regulate the niche environment. This, in turn, may explain why some leukaemias cause neuropathy and apoptosis of MSCs in the bone marrow, and may constitute a mechanism by which leukaemia cells can escape the niche. Protecting the HSC niche might therefore represent a novel complementary strategy for myeloproliferative neoplasms. Also in the context of blood cancers, Emanuelle Passegué (University of California, USA) introduced the concept of ‘emergency myelopoiesis pathways’ in the bone marrow. She showed that during blood regeneration, HSCs are induced to overproduce myeloid-biased multipotent progenitors (Pietras et al., 2015). An important consequence of the activation of this myeloid regeneration axis is the formation of defined granulocyte/macrophage progenitor clusters in the bone marrow cavity, which drive the local overproduction of granulocytes. The remodelling of the multipotent progenitor compartment and the induction of clusters represent novel emergency myelopoiesis pathways that are transiently activated during regeneration and are continuously triggered in disease conditions. In leukaemia, the constant production of myeloid progenitor clusters directly contributes to myeloid disease burden and could become a key target for anti-HSC differentiation therapies.

Beyond the blood, Shosei Yoshida (National Institute for Basic Biology, Japan) introduced the intriguing concept of an open stem cell niche in the mammalian testis. Taking advantage of intravital live-imaging, Yoshida showed how spermatogenic stem cells in the seminiferous tubules are not confined to a certain anatomical space. Rather, stem cells are instead highly motile between differentiating progenies and somatic supporting cells, and scatter along the tubules at a constant density. Yoshida presented convincing evidence that the mechanism behind this density-dependent open niche lies in competition for a finite supply of mitogens between the spermatogenic stem cells. Moving to the lung, Joo-Hyeon Lee (Cambridge Stem Cell Institute, UK) showed that stromal cells are required for the formation of lung organoids that are derived from adult distal epithelial stem or progenitor cells in vitro. Clonal three-dimensional (3D) co-culture of lung epithelial stem cells with endothelial cells revealed heterotypic organoids that retain multi-lineage differentiated cells. Of note, branching organoids are closely associated with endothelial cells that are actively engaged in vascular tube formation. In addition, Lee also showed that LGR6-expressing peri-airway mesenchymal cells serve as a ‘niche’ for the club cells, which have previously been suggested to have the properties of lung stem cells. Isolated club cells and LGR6⁺ cells were sufficient to form lung organoids in culture, but the precise role of the LGR6⁺ cell type within the lung stem cell niche has not yet been defined. Extending the theme of 3D tissue reconstruction in vitro, Agnete Kirkeby (Lund University, Sweden) presented an alternative approach to the organoid models, namely a microfluidic gradient system for producing controlled rostro-caudal patterning of neural cells derived from hPSCs. In this system, hPSCs were differentiated to produce a single piece of tissue comprising forebrain, midbrain and hindbrain cells located in a spatially ordered manner, thereby mimicking the rostro-caudal patterning of the early neural tube. Kirkeby further showed how the system could be used to assess region-specific responses of neural cells to genes and growth factors.

The various signals that derive from the niche may be interpreted differently by different types of stem cells. This was the take-home message of the work presented by Michael Elowitz (Caltech, CA, USA), who discussed the challenge of synthetic circuit design in multicellular organisms using signalling integration in the bone morphogenetic protein (BMP) pathway as a model. The BMP pathway has 20 different ligands and numerous receptors that can interact in a ‘many-to-many’ fashion. Using a BMP reporter cell line and testing combinations of ligands, Elowitz showed that the effect of each ligand generally depends on the level of other ligands and that receptor expression can ‘reprogram’ ligand interaction modes. Elowitz presented a mathematical model to enable the analysis of such promiscuous ligand-receptor interactions, which may open up more sophisticated strategies to understand and re-engineer complex ligand-signalling modes in the niche. Continuing the theme of BMP signalling, a short talk by Alison McGarvey from Alexander Medvinsky's group (MRC Centre for Regenerative Medicine, UK) outlined a transcriptomic approach to elucidate niche regulators of HSC emergence from the embryonic aorta-gonad-mesonephros region. By coupling gene expression signatures with an ex-vivo re-aggregate culture followed by transplantation, McGarvey identified a novel candidate that increases the efficiency of HSC maturation from their direct precursors. She proposed that through its action as a modulator of BMP signalling, it supports the maturation of HSCs by fine-tuning the level of BMP signalling in the niche.

Copenhagen is a city that maintains its historic charm whilst at the same time undergoing rapid development. An evening boat tour provided the opportunity for participants to witness this first hand, cruising the ancient canals whilst surveying the exciting progress being made throughout the city. Rapid progress was also evident from talks that utilized research tools for single-cell analysis. Methods for genome-wide RNA sequencing of single cells have been well established for several years, but a remaining challenge has been to retrieve spatial information in single-cell RNA-seq data sets. With this in mind, Alexander van Oudenaarden (Hubrecht Institute, The Netherlands) described a strategy whereby bone marrow cells were incompletely dissociated, allowing cell duplets to be separately isolated. Single cells were then manually dissociated and sequenced, allowing information on cell-cell interactions to be traced in the data. This method revealed close interactions between neutrophil and megakaryocyte lineages, and this partnership is believed to influence cell maturation. Future robust strategies for high-throughput isolation of cell duplets followed by single-cell RNA sequencing may therefore serve to retrieve important information of key spatial interactions between cells within a given niche. Staying in the bone marrow, Franziska Paul (Weizmann Institute, Israel) discussed recent progress in using single-cell RNA-seq to profile myeloid progenitor cells from the murine bone marrow (Paul et al., 2015). Surprisingly, Paul and colleagues found that there does not seem to be a common myeloid progenitor population in the bone marrow. Instead, the progenitor population was highly heterogeneous and could be grouped into 19 progenitor subgroups that showed unexpected transcriptional priming toward seven different cell lineages: erythrocytes, megakaryocytes, monocytes, dendritic cells, neutrophils, basophils and eosinophils.

Continuing the theme of single-cell analyses, Thomas Perlmann (Karolinska Institutet, Sweden) showed how single-cell RNA-seq can be used to reconstruct lineage development in the central nervous system at the genome-wide level. The focus was on developing dopamine neurons, which degenerate in Parkinson's disease and are therefore of major interest for potential stem cell-based therapies. Perlmann's results uncover an unexpectedly close lineage relationship between developing dopamine neurons and more rostral non-dopaminergic neurons. The talk also illustrated the importance of combining developmental biology and stem cell biology for clinical translation.

Kat Hadjantonakis (Sloan Kettering Institute, NY, USA) presented her group's efforts to use quantitative single-cell-resolution imaging of the developing mouse blastocyst in order to pinpoint the very first molecular changes that take place in the inner cell mass (ICM), which ultimately lead to cell fate segregation into either primitive endoderm or the pluripotent epiblast. Combined with single-cell transcriptomics, Hadjantonakis showed how lineage segregation is incremental and demonstrated that FGF4 is the only factor showing differential expression in the ICM cells prior to lineage segregation. She also showed that the combinatorial activity of FGFR1 and FGFR2 was essential for these early embryo fate decisions. Still in the embryo, Angela Stathopoulos (Caltech, CA, USA) used sophisticated tools to follow the dynamic transcription factor behaviour that governs the patterning of the rapidly dividing Drosophila embryo. By following gene expression of dorsal, a transcription factor expressed in a dorsal-ventral nuclear gradient, Stathopoulos described how oscillations drive the step-wise progression of early development. Interestingly, she showed how the dorso-ventral boundaries of the embryo shift up and down as development progressed. A particularly crucial moment in the development of the embryo was found to occur at mitotic cycle no. 13, where expression of key signalling pathway components is initiated and required to support proper gene regulatory network progression.

Architecture and design play an important role in shaping our world: not just in the biological sense but culturally too. A memorable visit to the Copenhagen Opera House, designed by the renowned architect Henning Larsen, solidified this point. Meeting participants listened with fascination as staff described how both the site of the opera house and the building itself were meticulously designed to maximize the geographic and artistic impact of this masterpiece of Danish design, superbly located just opposite Amalienborg castle. The architecture and control of stem cell chromatin was also a fascinating subject discussed by several speakers. Rick Young (Whitehead Institute, MA, USA) introduced the concept of ‘insulated neighbourhoods’: loop structures in the DNA formed by nested protein complexes of the CCCTC-binding factor CTCF, which forms boundaries around important enhancer-promoter regions that are necessary for gene control. He showed how CRISPR/Cas9 deletion of CTCF-binding sites altered the expression of local genes and lead to new topological interactions. Moreover, Young showed mapping data of insulated neighborhoods in T-cell acute lymphoblastic leukaemia and described how tumour cell genomes contain recurrent microdeletions that eliminate the CTCF boundary sites (Hnisz et al., 2016). Directed perturbation of such boundaries near the TAL1 and LMO2 genes in non-malignant cells was sufficient to activate these proto-oncogenes. The identification of these genomic anchors may be important for the interpretation of genome-wide association studies in cancer.

Ken Zaret (University of Pennsylvania School of Medicine, USA) stressed the importance of understanding how transcription factors access and bind to chromatin. So-called pioneer factors can bind to unprogrammed chromatin – chromatin that is unmarked for activation or repression – yet there remain heterochromatic regions not accessible to pioneer factors. Zaret showed that these regions, which are marked by H3K9me3, represent an impediment to transcription factor-based reprogramming. Mapping of H3K9me3 domains throughout development showed unexpectedly high levels in germ layer cells and a marked diminution during terminal differentiation. Zaret proposed that the high levels of heterochromatin in early endoderm and mesoderm germ layers create less chromatin that is available to scan, which would allow relatively fast cell programming in the embryo. Using sucrose gradients coupled with quantitative proteomics, heterochromatin proteins that may impede gene activation were identified. Knocking down these proteins may represent a way to safely break down heterochromatin to increase reprogramming efficiency.

Achieving reprogramming through the use of small molecular drugs rather than by conventional reprogramming factors is a major focus of Hongkui Deng (Peking University, China), and he presented an overview of their work to generate iPSCs (Zhao et al., 2015) and induced neurons via a purely chemical approach. Deng showed how the compound I-BET151, which is used in chemical reprogramming to induce efficient neuronal transdifferentiation, works by repressing the core fibroblast transcription programs, potentially through blocking of super-enhancer-driven transcription. Only a few hours of incubation with I-BET151 is required to significantly downregulate the fibroblast core transcriptional network, which may prove to be a useful strategy to enhance all lineage conversions from a fibroblast starting population. In a different direct reprogramming approach, Kiran Batta from Georges Lacaud's group (CRUK Manchester Institute, UK) reported the identification of five transcription factors required for reprogramming of mouse and human dermal fibroblasts to haematopoietic progenitors. Batta showed that co-culturing the reprogrammed progenitor cells with endothelial cells led to long-term multilineage potential in vivo, suggesting that the endothelial niche cells protected the progenitors against cell death and differentiation induced by oxidative stress. Turning the system upside down, Carlos-Filipe Pereira (University of Coimbra, Portugal) showed how information acquired during haematopoietic reprogramming informed the complex process of HSC specification during development. The analysis of induced haemogenic cells from fibroblasts revealed a phenotype that could be used to isolate precursors of HSCs in vivo from mid-gestation mouse placentas. These early precursor cells of haematopoiesis were characterized by single-cell RNA-seq and matured in vitro into HSCs.

Maintaining pluripotency for the correct period of time in vivo is essential for embryonic development, and significant progress has been made towards understanding how this is regulated at the molecular and biophysical level. Sally Lowell (University of Edinburgh, UK) presented convincing evidence that ID1 makes Nanog^low cells resistant to differentiation and thereby functions as a safeguard for pluripotency during the transition from naïve to primed states. Furthermore, Lowell emphasized that tissue organisation is not just a passive consequence of differentiation but that it actively feeds back into the decision-making process. By using a novel 3D-imaging and data interrogation tool, Lowell showed how her group is able to track the orientation of cell divisions in the early embryo, and how they use this to study the interplay between tissue organization and differentiation competence. Another safeguard for maintaining pluripotency was presented by Pablo Navarro (Institute Pasteur, France), who discussed the ‘bookmarking’ properties of the transcription factor ESRRB, which is an important regulator of pluripotency and reprogramming. Navarro showed that ESRRB binds and coats chromosomes during mitosis – both in PSCs and in the embryo – and that this ESRRB coating might be necessary for the maintenance of epigenetic marks within the daughter cells. These data demonstrated the importance of transcription factor bookmarking to help the transcriptional machinery to perpetuate stem cell fate throughout cell division.

One of the main goals of the stem cell field is to improve health outcomes in ageing and disease. Various approaches have been taken to achieve this, ranging from the use of stem cell-based models of disease to the use of stem cells for cell-based therapies. But in order to accurately model adult diseases or to provide adult cell types, it is absolutely crucial to develop highly specific differentiation protocols, not just for one cell type, but potentially for the whole organ. To this end, Gordon Keller (McEwen Centre for Regenerative Medicine, Canada) presented an impressive tour-de-force in controlling accurate differentiation of hPSCs into populations of all the various subtypes of cells in the heart. He showed that conventional protocols give rise to mainly ventricular cardiomyocytes intermingled with a smaller population of pacemaker cells. Addition of retinoic acid shifted cell fate towards an atrial phenotype. In contrast, activation of the WNT pathway in combination with the BMP pathway was able to shift myocardial cells towards epicardium, while the production of the endocardium was still being optimized. Keller and his group were able to purify and transplant human hPSC-derived pacemaker cells into the rat heart, and upon subsequently lowering the rat heartbeat, the ectopically placed human pacemaker cells were able to take control of the heartbeat, thereby demonstrating full in vivo functionality.

Another way to generate mature cell types for translational research is via direct cell fate reprogramming, also known as lineage reprogramming. To this end, Oliver Brüstle (University of Bonn, Germany) reported on his group's latest developments using this method. They have been able to generate distinct neural stem cell populations with defined neuronal and glial differentiation propensities, and have used this to model late-onset neurodegenerative disease. Combined with in vitro calibration of proteostatic pathways, this approach enables highly standardized phenotypic assays, for example, the assessment of pathological protein aggregation in polyglutamine diseases. Brüstle also demonstrated how trans-synaptic rabies virus-based tracing combined with light sheet microscopy could enable in vivo connectome studies of grafted hPSC-derived neurons in a whole-brain scenario.

PSC-based approaches are now integral to our understanding of disease and represent a theoretically infinite source of cells with which to treat it. But not all disease scenarios require PSC-based therapies. Adult tissue-residing stem cells – in the skin and eye, for example, and in various internal organs such as the intestine – are also critical for homeostasis and for understanding disease. Paloma Ordóñez from Joerg Huelsken's group (École Polytechnique Féderale de Lausanne, Switzerland) identified HOXA5 as an important repressor of intestinal stem cell fate in vivo. HOXA5 is downregulated by the WNT pathway to maintain stemness and becomes active only when cells leave the intestinal crypt and undergo differentiation (Ordóñez-Morán et al., 2015). In support of this, she showed how HOXA5 expression reduces intestinal LGR5⁺ stem cells and that intestinal organoids lose the potential to be passaged. In colon cancer, HOXA5 is downregulated and its re-expression induces the loss of the cancer cell phenotype, preventing tumour progression and metastasis. Ordóñez also showed how retinoids could induce HOXA5 upregulation and block tumour growth, opening the way for possible retinoid therapy for colon cancer.

In a different adult stem cell-based system, this time in the eye, Graziella Pellegrini (University of Modena, Italy) presented her decade-long efforts to produce a treatment for patients with impaired vision caused by damage to the cornea. Transplantation of autologous, cultured limbal stem cells has been shown to cause significant and long-term clinical benefits on vision in transplanted patients (Rama et al., 2010). Now, Pellegrini has embarked on a mission to treat not only superficial corneal damage, but also deeper lesions of the eye. Pellegrini showed their progress in designing physiologically relevant extracellular matrix scaffolds of orthogonally located layers of biopolymers. Although they have now succeeded in producing scaffolds with optimized rigidity and stiffness, some work still remains to render the scaffolds ideal for cell adhesion and differentiation. Finally, in the skin, Valentina Greco's talk (Yale University, CT, USA) was a beautiful example of how powerful live-imaging techniques can expose well-kept stem cell secrets. In her lab, two-photon microscopy is used in live imaging of the mouse hair follicle. By combining imaging with transgenic reporters and functional genetics, she showed how the skin stem cell niche has the potential to eliminate mutant cells with defects that otherwise would have tumorigenic potential. Understanding the mechanistic basis of this method of disease prevention may bring insight into possible targets for skin cancer therapies.

Generally speaking, mammalian species do not compare well with non-mammalian models in terms of their capacity to regenerate and much can be learned from the latter. Brigitte Galliot (University of Geneva, Switzerland) introduced the model organism Hydra, which can be used to study regeneration and ageing. It is impressive that the freshwater Hydra vulgaris polyps can regenerate any missing body part – even from cellular aggregates. Using transcriptomic analysis, she showed that epithelial cells adapt to the loss of interstitial stem cells and neurogenesis by upregulating a large subset of neurogenic genes (Wenger et al., 2016). In contrast, epithelial cells in the cold-sensitive strain of Hydra oligactis lack this adaptive plasticity, as a result of deficient autophagy, thereby providing a model where epithelial autophagy appears to support stem cell self-renewal as an anti-ageing mechanism. In a different model organism, this time the fly, Allan Spradling (Carnegie Institution, MD) described an unexpected, alternative source of cell replacement by wound-induced polyploidization in Drosophila. Following wounding, the cells that surround the injury site do not divide, but instead become polyploid, that is, they increase their chromosomal number. Interestingly, Hippo signalling controls polyploidization and inversely JNK signalling through AP-1 limits it. Spradling showed how the induction of polyploidy occurs also during wounding of the mouse corneal endothelium and suggested that the large size and aneuploidy tolerance of polyploid cells represents an evolutionarily conserved mechanism and may provide advantages over diploid cells to maintain homeostasis.

Just like the orchestra at the Copenhagen Opera House, the multiple levels of regulation and components of the stem cell niche is an ensemble that needs to be regulated in order to balance stem cell self-renewal and differentiation. It was clear from the 12 sessions that unlocking the fundamental mechanisms and therapeutic potential of stem cells and their particular niches will continue to be a major focus of stem cell biology. Efforts to deconstruct and reconstruct the niche with emerging technologies are providing new information to help understand niche composition and function, and further highlight potential pathways for therapeutic manipulation (Fig. 1).

In addition to the inspiring talks, the poster sessions were also remarkable. Of the 134 posters, the two poster prize winners were Birgit Ritschka (Center for Genomic Regulation, Spain) with her poster entitled ‘Senescence is a regenerative process that instructs stem cell function’ and Jakub Sedzinski (University of Texas, USA) with his poster entitled ‘How to fit in? Emergence of an apical epithelial surface in vivo’. After four days of lively discussion, the meeting theme was a well-placed reminder that it was indeed time to return to our own familiar laboratory ‘niche’, hopefully to generate new and interesting data in preparation for the next Stem Cell Niche meeting in two years’ time.

We thank the organizers for a stimulating scientific programme, the speakers for sharing unpublished work, and the Novo Nordisk Foundation staff for their excellent services. We apologize to the speakers whose work we were unable to discuss due to space restrictions. We thank Sally Lowell and Guillaume Blin (University of Edinburgh) for kindly sharing images and Fábio Rosa (University of Coimbra) for assistance with figure preparation.