Building Human Renal Tracts

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


Introduction
Worldwide around 2.8 million individuals receive renal replacement therapy (RRT) for end-stage kidney failure (ESKF) [1]. It has been estimated that at least two million further individuals need transplantation or long-term dialysis to survive but are unable to access these therapies [1].
Moreover, even in high-income developed countries, transplants are in short supply, and life expectancy on long-term dialysis is severely curtailed compared with life expectancy in the average aged-matched population [2]. Based on surveys, there are around 8,000 children in Europe and 9,000 children in the US treated with RRT, with an estimated worldwide prevalence of 150,000 [3,4]. With regard to young children with ESKF, around half were born with abnormal renal tracts i.e. with kidney malformations, often accompanied urinary tract (ureter, bladder and urethra) anomalies [3]. Moreover, it has been estimated that up to one fifth of young adults with ESKF were born with renal tract malformations [5].
Given the above information, there are pressing needs to discover new therapies with which to either treat ESKF or to slow the progressive loss of kidney excretory function that generally precedes ESKF. It is in these contexts that human pluripotent stem cell (PSC) technology offers promise. Here, we review progress towards making functional kidney tissues from human PSCs. The substantial challenges to achieving this end will be discussed. We also indicate how human PSCs are being used to create 'diseases in a dish' to understand the pathobiology underlying congenital kidney disease.

Introduction to PSCs and kidney development.
By definition, PSCs have the ability to replicate themselves in perpetuity. In addition, they have the potential to differentiate into the three germ layers (mesoderm, endoderm and ectoderm) of the early embryo and thus form all organs. As recently reviewed [6,7] and depicted in Figure 1, there are two sources of PSCs for research studies. The first is the inner cell mass of very early human embryos that have been generated, but eventually not placed in the uterus, during in vitro fertilization therapy. This type of PSC is called an embryonic SC (ESC). More recently, it has become possible to make 'induced' PSC (iPSCs) by taking a sample of differentiated cells (e.g. white blood cells, skin fibroblasts or even renal cells shed into the urine) and driving these into PSCs by viral vector-mediated transient expression of pluripotency molecules.
Given the ability to source human PSCs, the question is how to drive these to become kidney tissues in the laboratory. Here, stem cell researchers have been much informed by classical studies of the anatomy and developmental biology of kidney development. Over half a century ago, Edith Potter made detailed morphological and histological descriptions of normal human embryonic and fetal kidneys, from their inception at five weeks gestation to the end of their development in the late third trimester [8,9]. She also documented the perturbed anatomy of human dysplastic kidneys, either arising from primary disturbances in the ureteric bud (UB) and metanephric mesenchyme (MM) compartments, or arising from the physical disruption of impairment of fetal urinary flow caused by urinary tract obstruction [8,9].
A second line of studies were initiated at a similar time by Clifford Grobstein [10] and then continued by Lauri Saxen and other pioneers [11]. These researchers showed that embryonic mouse kidney rudiments could be explanted into organ culture and maintained for around a week during which time the UB elongated into a ureteric stalk, the top of which branched recurrently to form collecting ducts, and some MM cells underwent a mesenchymal to epithelial transition to form nephrons (glomerular podocytes and Bowman capsule, proximal tubules and distal tubules), as depicted in Figure 2. Other MM cells differentiate into interstitial cells located between the tubules and at least a subset of kidney capillaries [12]. Conversely, if the UB or MM was cultured in isolation, each component failed to differentiate and instead died by apoptosis.
Subsequent experiments using organ culture or wild-type murine tissues and also mutant mice [13] showed that the MM and UB nurtured each other, a process called mutual induction ( Figure   2), by secreting a variety of paracrine growth factors. Moreover, other types of molecule, especially transcription factors and extracellular matrix proteins, were found to be expressed by the developing kidney and were essential for its normal development [13]. Of note, the genes encoding several of these molecules have been found to be mutated in individuals with congenital kidney disease [7]. The UB and MM themselves each originate from separate compartments within the intermediate mesoderm [14], that itself has formed during gastulation, the name of the event when the three germ layers form early on in embryogenesis. This gives insight into growth factors and other molecules that, during normal development, generate the building blocks of the kidney rudiment [14].

Making kidney-like tissues from PSCs
In the last decade, as reviewed [6,7], several laboratories around the world have published protocols in which human PSC can be driven by sequential addition of specific growth factors and other chemicals to form kidney-like tissues in 2D and in 3D, or organoid, culture ( Figure 1). During these protocols, PSCs are differentiated to mesodermal precursors of either MM or UB cells, and then these precursors are given further stimulation to become nephrons [15,16] or collecting ducts [14,17]. In general, the protocols generate just one of these two kidney lineages, as demonstrated by single cell RNA sequencing to interrogate the molecular phenotypes of individual cells within organoids [15]. There does, however, appear to be some plasticity in these systems and, for example, nephron distal tubule precursors can be reprogrammed in culture to transdifferentate into collecting ducts [16].
We have experience with the nephron culture protocol and, like other investigators, have demonstrated that the resulting organoids contain glomeruli, proximal tubules and distal tubules [18 19].
A limitation, however, is that the resulting glomeruli contain only epithelial cells (podocytes and the Bowman capsule) and lack capillary loops and mesangial cells in their tufts [18,19]. This is despite the fact that capillaries are detected between tubules in the organoid.
Clearly, avascular glomeruli have no chance of filtering blood to make urine and, moreover, the organoids anyhow lack a blood supply. We had previously documented that, after native embryonic mouse kidney rudiments were transplanted into the nephrogenic cortex of newborn mice, the transplant differentiated into glomeruli containing capillary loops [12].
Furthermore, after fluorescently labelled low molecular dextran was intravenously injected into the host mouse, we detected fluorescent signal in proximal tubules that had formed in the transplant [20]. This was consistent with the occurrence of ultrafiltration by glomeruli followed by reuptake by tubules.
With these results in mind, we reasoned that implanting human PSC-derived kidney precursor cells subcutaneously into immunocompromised mice would lead to the creation of more structurally mature and functional nephrons. As depicted in Figure 3, 12 weeks after implantation, we were able to visualise living implants, the PSC having been molecularly-labelled with a luciferase expressing transgene. On histology, the implanted cells had formed structures up to around 1 cm in length and they contained glomeruli and tubules. The former contained capillary loops and mesangial-like cells. Not all parts of the implant were so mature and these other areas contained thin tubules surrounded by stromal cells; moreover, some implants contained islands of cartilage-like cells, 'off-target' 8 mesodermal derivatives (Figure 3). It was also demonstrated that, after injection of fluorescently-labelled dextran into the host, a subset of tubules in the implant contained florescent granules, suggestive of endocytosed ultrafiltrate [18]. Other investigators were able to show that, after implantation of human PSC-derived organoids under the kidney capsule of host mice, the arising glomeruli also contain capillary loops [21].

Remaining limitations to creating functional renal tracts from PSCs
Although the above advances in organoid culture and after implantation are encouraging, it is worth reflecting on the remaining challenges before clinically useful functional intact renal tracts might be generated from human PSCs. As reviewed [6], the first problem is one of scale: an organoid might contain up to around 100 glomeruli but a healthy human kidney contains over three magnitudes more. Second, although organoid implants are perfused by host blood, there are only small calibre connecting vessels. Presumably, given that glomerular filtration is positively correlated with blood flow and hydrostatic pressure, without a more robust arterial supply implant glomeruli would never generate much urine.
With this in mind, we undertook experiments in which human PSC-derived kidney precursor cells were paced inside perforated plastic chambers which we implanted into the thighs of immunocompromised mice [22]. In some implants we surgically fashioned an 'arteriovenous flow through' in which the host's femoral artery and vein were mobilised and threaded through the chamber. The rationale here was that these might enhance the vascularity and growth of the implanted cells, as had been described for other, non-renal, cell implants [23]. By comparing histology of kidney precursor implants with or without flowthroughs, it was demonstrated that the former enhanced the number of small vessels around the human implants at three weeks after surgery. While kidney-like tissues were seen in all implants (Figure 4), quantification did not reveal a significant difference between the two conditions. Moreover, at 12 weeks after surgery, we documented an overabundance of scar-like interstitial tissues of human origin in implants with an arteriovenous flow through [22], suggesting that pathological differentiation might follow a period of increased vascularity.
A third challenge is that human PSC-derived kidney nephron generated in culture or after implantation lack a ureter. Without such a drainage tube, any urine generated by the implant would simply diffuse back into the host. Efforts are underway to generate ureteric tissues from PSCs, with reports of success in driving PSCs to become epithelia with a urothelial phenotype [24][25][26]. These models, however, do not appear to generate ureteric smooth muscle cells.
Without proximal to distal directional peristalsis, that initiates before birth in native ureters, there would be the risk of 'functional obstruction to urine flow' as occurs in Teashirt-3 mutant mice that lack ureteric smooth muscle yet have an anatomically patent urothelial tube [27]. In normal organogenesis, immature urothelia act as a paracrine signalling centre, secreting the sonic hedgehog growth factor that induces nearby mesenchymal cells to form smooth muscle [28,29].
Of note, when PSC-derived urothelial were implanted next to similar native mesenchymal cells in explant culture, the latter were induced to differentiate into contractile muscle [26]. As for kidney organoids generated from PSCs, ureteric tissues derived from PSCs are smaller than their in vivo native counterparts. It is established that the linear growth of explanted mouse embryonic ureters can be inhibited by application of exogenous transforming growth factor  (TGF) or accelerated by blockade of endogenous TGF signalling ( Figure 5) [30]. Moreover, application of exogenous fibroblast growth factor 10 (FGF10) stimulated linear growth [30]. Such observations may inform manipulations of PSC-derived urothelial tissues to increase their size.
Indeed, it has been reported that application of FGF10 to PSC-derived bladder-like urothelia enhanced their molecular maturity and stratification [25].

Modelling human congenital kidney disease with human PSC technology
As discussed in the Introduction to review, congenital disorders of development and differentiation of the kidney and urinary tract can cause ESKD. In the last two decades, it has become apparent that at least a small subset of individuals both with kidney [7,31] or urinary tract [32] malformations carry mutations of genes active during normal renal tract development.
At the time of writing, the commonest mutated gene associated with cystic dysplastic kidneys is hepatocyte nuclear factor 1B (HNF1B); for example, pathogenic heterozygous variants of HNF1B were reported in 52 of 199 children with kidney malformations [33]. This gene codes for a transcription factor expressed in tubules derived from both the UB and the MM [34]. It is rare to obtain tissue samples of these native kidneys and researchers are turning PSC organoid technology to try to model the disease. One research group [35] used CRISPR-Cas9 gene editing to mutate one allele of HNF1B in PSCs. After driving these to become UB-like tissues, they recorded impaired branching into collecting duct precursors. Moreover, these primitive ducts showed aberrant expression of several molecular markers of the UC/collecting duct lineage [35]. Another research group mutated both alleles of HNF1B in human PSC and then drove them to become nephron organoids [36]. They reported that the mutant organoids contained glomeruli but appeared to lack proximal and distal tubules. A caveat with this PSC study is that their model had both copies of the gene mutated whereas humans have only been reported with one copy mutated, with dominant inheritance of the disease through several generations.
The interested reader is directed to a recent review that addresses the use of human PSCs to study a wide range of inherited human kidney diseases including the polycystic kidney diseases and congenital nephrotic syndrome [7]. In the longer term it is hoped that mechanistic insights 11 from such studies will suggest druggable targets to enhance differentiation and preclude or at least delay the onset of ERKF.

Conclusions
Human PSC technology has made striking advances in generating kidney-like tissues. The potential use of organoids for regenerative medicine therapies is currently precluded by their small size, the lack of large vessels feeding blood to the organoid, and the lack of a urinary tract to drain urine made by the the newly grown kidney. The ability to model kidney diseases starting with human PSC should allow insights into the pathobiology of congenital renal diseases, especially those caused by mutations.

Disclosure
The author declared no competing interests.