Deep learning is widely applicable to phenotyping embryonic development and disease.

Genome editing simplifies the generation of new animal models for congenital disorders. However, the detailed and unbiased phenotypic assessment of altered embryonic development remains a challenge. Here, we explore how deep learning (U-Net) can automate segmentation tasks in various imaging modalities, and we quantify phenotypes of altered renal, neural and craniofacial development in Xenopus embryos in comparison with normal variability. We demonstrate the utility of this approach in embryos with polycystic kidneys (pkd1 and pkd2) and craniofacial dysmorphia (six1). We highlight how in toto light-sheet microscopy facilitates accurate reconstruction of brain and craniofacial structures within X. tropicalis embryos upon dyrk1a and six1 loss of function or treatment with retinoic acid inhibitors. These tools increase the sensitivity and throughput of evaluating developmental malformations caused by chemical or genetic disruption. Furthermore, we provide a library of pre-trained networks and detailed instructions for applying deep learning to the reader's own datasets. We demonstrate the versatility, precision and scalability of deep neural network phenotyping on embryonic disease models. By combining light-sheet microscopy and deep learning, we provide a framework for higher-throughput characterization of embryonic model organisms. This article has an associated 'The people behind the papers' interview.

HNF1B Alters an Evolutionarily Conserved Nephrogenic Program of Target Genes.

SIGNIFICANCE STATEMENT: Mutations in hepatocyte nuclear factor-1 ß ( HNF1B ) are the most common monogenic causes of congenital renal malformations. HNF1B is necessary to directly reprogram fibroblasts to induced renal tubule epithelial cells (iRECs) and, as we demonstrate, can induce ectopic pronephric tissue in Xenopus ectodermal organoids. Using these two systems, we analyzed the effect of HNF1B mutations found in patients with cystic dysplastic kidney disease. We found cross-species conserved targets of HNF1B, identified transcripts that are differentially regulated by the patient-specific mutant protein, and functionally validated novel HNF1B targets in vivo . These results highlight evolutionarily conserved transcriptional mechanisms and provide insights into the genetic circuitry of nephrogenesis. BACKGROUND: Hepatocyte nuclear factor-1 ß (HNF1B) is an essential transcription factor during embryogenesis. Mutations in HNF1B are the most common monogenic causes of congenital cystic dysplastic renal malformations. The direct functional consequences of mutations in HNF1B on its transcriptional activity are unknown. METHODS: Direct reprogramming of mouse fibroblasts to induced renal tubular epithelial cells was conducted both with wild-type HNF1B and with patient mutations. HNF1B was expressed in Xenopus ectodermal explants. Transcriptomic analysis by bulk RNA-Seq identified conserved targets with differentially regulated expression by the wild-type or R295C mutant. CRISPR/Cas9 genome editing in Xenopus embryos evaluated transcriptional targets in vivo . RESULTS: HNF1B is essential for reprogramming mouse fibroblasts to induced renal tubular epithelial cells and induces development of ectopic renal organoids from pluripotent Xenopus cells. The mutation R295C retains reprogramming and inductive capacity but alters the expression of specific sets of downstream target genes instead of diminishing overall transcriptional activity of HNF1B. Surprisingly, targets associated with polycystic kidney disease were less affected than genes affected in congenital renal anomalies. Cross-species-conserved transcriptional targets were dysregulated in hnf1b CRISPR-depleted Xenopus embryos, confirming their dependence on hnf1b . CONCLUSIONS: HNF1B activates an evolutionarily conserved program of target genes that disease-causing mutations selectively disrupt. These findings provide insights into the renal transcriptional network that controls nephrogenesis.

Mutations in transcription factor CP2-like 1 may cause a novel syndrome with distal renal tubulopathy in humans.

BACKGROUND: An underlying monogenic cause of early-onset chronic kidney disease (CKD) can be detected in ∼20% of individuals. For many etiologies of CKD manifesting before 25 years of age, >200 monogenic causative genes have been identified to date, leading to the elucidation of mechanisms of renal pathogenesis. METHODS: In 51 families with echogenic kidneys and CKD, we performed whole-exome sequencing to identify novel monogenic causes of CKD. RESULTS: We discovered a homozygous truncating mutation in the transcription factor gene transcription factor CP2-like 1 (TFCP2L1) in an Arabic patient of consanguineous descent. The patient developed CKD by the age of 2 months and had episodes of severe hypochloremic, hyponatremic and hypokalemic alkalosis, seizures, developmental delay and hypotonia together with cataracts. We found that TFCP2L1 was localized throughout kidney development particularly in the distal nephron. Interestingly, TFCP2L1 induced the growth and development of renal tubules from rat mesenchymal cells. Conversely, the deletion of TFCP2L1 in mice was previously shown to lead to reduced expression of renal cell markers including ion transporters and cell identity proteins expressed in different segments of the distal nephron. TFCP2L1 localized to the nucleus in HEK293T cells only upon coexpression with its paralog upstream-binding protein 1 (UBP1). A TFCP2L1 mutant complementary DNA (cDNA) clone that represented the patient's mutation failed to form homo- and heterodimers with UBP1, an essential step for its transcriptional activity. CONCLUSION: Here, we identified a loss-of-function TFCP2L1 mutation as a potential novel cause of CKD in childhood accompanied by a salt-losing tubulopathy.

Mutations in TBX18 Cause Dominant Urinary Tract Malformations via Transcriptional Dysregulation of Ureter Development.

Congenital anomalies of the kidneys and urinary tract (CAKUT) are the most common cause of chronic kidney disease in the first three decades of life. Identification of single-gene mutations that cause CAKUT permits the first insights into related disease mechanisms. However, for most cases the underlying defect remains elusive. We identified a kindred with an autosomal-dominant form of CAKUT with predominant ureteropelvic junction obstruction. By whole exome sequencing, we identified a heterozygous truncating mutation (c.1010delG) of T-Box transcription factor 18 (TBX18) in seven affected members of the large kindred. A screen of additional families with CAKUT identified three families harboring two heterozygous TBX18 mutations (c.1570C>T and c.487A>G). TBX18 is essential for developmental specification of the ureteric mesenchyme and ureteric smooth muscle cells. We found that all three TBX18 altered proteins still dimerized with the wild-type protein but had prolonged protein half life and exhibited reduced transcriptional repression activity compared to wild-type TBX18. The p.Lys163Glu substitution altered an amino acid residue critical for TBX18-DNA interaction, resulting in impaired TBX18-DNA binding. These data indicate that dominant-negative TBX18 mutations cause human CAKUT by interference with TBX18 transcriptional repression, thus implicating ureter smooth muscle cell development in the pathogenesis of human CAKUT.

A Dominant Mutation in Nuclear Receptor Interacting Protein 1 Causes Urinary Tract Malformations via Dysregulation of Retinoic Acid Signaling.

Congenital anomalies of the kidney and urinary tract (CAKUT) are the most common cause of CKD in the first three decades of life. However, for most patients with CAKUT, the causative mutation remains unknown. We identified a kindred with an autosomal dominant form of CAKUT. By whole-exome sequencing, we identified a heterozygous truncating mutation (c.279delG, p.Trp93fs*) of the nuclear receptor interacting protein 1 gene (NRIP1) in all seven affected members. NRIP1 encodes a nuclear receptor transcriptional cofactor that directly interacts with the retinoic acid receptors (RARs) to modulate retinoic acid transcriptional activity. Unlike wild-type NRIP1, the altered NRIP1 protein did not translocate to the nucleus, did not interact with RARα, and failed to inhibit retinoic acid-dependent transcriptional activity upon expression in HEK293 cells. Notably, we also showed that treatment with retinoic acid enhanced NRIP1 binding to RARα RNA in situ hybridization confirmed Nrip1 expression in the developing urogenital system of the mouse. In explant cultures of embryonic kidney rudiments, retinoic acid stimulated Nrip1 expression, whereas a pan-RAR antagonist strongly reduced it. Furthermore, mice heterozygous for a null allele of Nrip1 showed a CAKUT-spectrum phenotype. Finally, expression and knockdown experiments in Xenopus laevis confirmed an evolutionarily conserved role for NRIP1 in renal development. These data indicate that dominant NRIP1 mutations can cause CAKUT by interference with retinoic acid transcriptional signaling, shedding light on the well documented association between abnormal vitamin A levels and renal malformations in humans, and suggest a possible gene-environment pathomechanism in this disease.

Engineering kidney cells: reprogramming and directed differentiation to renal tissues.

Growing knowledge of how cell identity is determined at the molecular level has enabled the generation of diverse tissue types, including renal cells from pluripotent or somatic cells. Recently, several in vitro protocols involving either directed differentiation or transcription-factor-based reprogramming to kidney cells have been established. Embryonic stem cells or induced pluripotent stem cells can be guided towards a kidney fate by exposing them to combinations of growth factors or small molecules. Here, renal development is recapitulated in vitro resulting in kidney cells or organoids that show striking similarities to mammalian embryonic nephrons. In addition, culture conditions are also defined that allow the expansion of renal progenitor cells in vitro. Another route towards the generation of kidney cells is direct reprogramming. Key transcription factors are used to directly impose renal cell identity on somatic cells, thus circumventing the pluripotent stage. This complementary approach to stem-cell-based differentiation has been demonstrated to generate renal tubule cells and nephron progenitors. In-vitro-generated renal cells offer new opportunities for modelling inherited and acquired renal diseases on a patient-specific genetic background. These cells represent a potential source for developing novel models for kidney diseases, drug screening and nephrotoxicity testing and might represent the first steps towards kidney cell replacement therapies. In this review, we summarize current approaches for the generation of renal cells in vitro and discuss the advantages of each approach and their potential applications.

Pandemic 2009 H1N1 influenza A virus carrying a Q136K mutation in the neuraminidase gene is resistant to zanamivir but exhibits reduced fitness in the guinea pig transmission model.

Resistance of influenza A viruses to neuraminidase inhibitors can arise through mutations in the neuraminidase (NA) gene. We show here that a Q136K mutation in the NA of the 2009 pandemic H1N1 virus confers a high degree of resistance to zanamivir. Resistance is accompanied by reduced numbers of NA molecules in viral particles and reduced intrinsic enzymatic activity of mutant NA. Interestingly, the Q136K mutation strongly impairs viral fitness in the guinea pig transmission model.

Synthetic genetic circuits to uncover and enforce the OCT4 trajectories of successful reprogramming of human fibroblasts.

Reprogramming human fibroblasts to induced pluripotent stem cells (iPSCs) is inefficient, with heterogeneity among transcription factor (TF) trajectories driving divergent cell states. Nevertheless, the impact of TF dynamics on reprogramming efficiency remains uncharted. Here, we identify the successful reprogramming trajectories of the core pluripotency TF, OCT4, and design a genetic controller that enforces such trajectories with high precision. By combining a genetic circuit that generates a wide range of OCT4 trajectories with live-cell imaging, we track OCT4 trajectories with clonal resolution and find that a distinct constant OCT4 trajectory is required for colony formation. We then develop a synthetic genetic circuit that yields a tight OCT4 distribution around the identified trajectory and outperforms in terms of reprogramming efficiency other circuits that less accurately regulate OCT4. Our synthetic biology approach is generalizable for identifying and enforcing TF dynamics for cell fate programming applications.

Synthetic genetic circuits to uncover the OCT4 trajectories of successful reprogramming of human fibroblasts.

Reprogramming human fibroblasts to induced pluripotent stem cells (iPSCs) is inefficient, with heterogeneity among transcription factor (TF) trajectories driving divergent cell states. Nevertheless, the impact of TF dynamics on reprogramming efficiency remains uncharted. We develop a system that accurately reports OCT4 protein levels in live cells and use it to reveal the trajectories of OCT4 in successful reprogramming. Our system comprises a synthetic genetic circuit that leverages noise to generate a wide range of OCT4 trajectories and a microRNA targeting endogenous OCT4 to set total cellular OCT4 protein levels. By fusing OCT4 to a fluorescent protein, we are able to track OCT4 trajectories with clonal resolution via live-cell imaging. We discover that a supraphysiological, stable OCT4 level is required, but not sufficient, for efficient iPSC colony formation. Our synthetic genetic circuit design and high-throughput live-imaging pipeline are generalizable for investigating TF dynamics for other cell fate programming applications.

Plasmacytoid dendritic cells and Toll-like receptor 7-dependent signalling promote efficient protection of mice against highly virulent influenza A virus.

Types I and III interferons (IFNs) elicit protective antiviral immune responses during influenza virus infection. Although many cell types can synthesize IFN in response to virus infection, it remains unclear which IFN sources contribute to antiviral protection in vivo. We found that mice carrying functional alleles of the Mx1 influenza virus resistance gene partially lost resistance to infection with a highly pathogenic H7N7 influenza A virus strain if Toll-like receptor 7 (TLR7) signalling was compromised. This effect was achieved by deleting either the TLR7 gene or the gene encoding the TLR7 adaptor molecule MyD88. A similar decrease of influenza virus resistance was observed when animals were deprived of plasmacytoid dendritic cells (pDCs) at day 1 post-infection. Our results provide in vivo proof that pDCs contribute to the protection of the lung against influenza A virus infections, presumably via signals from TLR7.

Tuning the 3D microenvironment of reprogrammed tubule cells enhances biomimetic modeling of polycystic kidney disease.

Renal tubular cells frequently lose differentiation markers and physiological properties when propagated in conventional cell culture conditions. Embedding cells in 3D microenvironments or controlling their 3D assembly by bioprinting can enhance their physiological properties, which is beneficial for modeling diseases in vitro. A potential cellular source for modeling renal tubular physiology and kidney diseases in vitro are directly reprogrammed induced renal tubular epithelial cells (iRECs). iRECs were cultured in various biomaterials and as bioprinted tubular structures. They showed high compatibility with the embedding substrates and dispensing methods. The morphology of multicellular aggregates was substantially influenced by the 3D microenvironment. Transcriptomic analyses revealed signatures of differentially expressed genes specific to each of the selected biomaterials. Using a new cellular model for autosomal-dominant polycystic kidney disease, Pkd1-/- iRECs showed disrupted morphology in bioprinted tubules and a marked upregulation of the Aldehyde dehydrogenase 1a1 (Aldh1a1). In conclusion, 3D microenvironments strongly influence the morphology and expression profiles of iRECs, help to unmask disease phenotypes, and can be adapted to experimental demands. Combining a direct reprogramming approach with appropriate biomaterials will facilitate construction of biomimetic kidney tubules and disease models at the microscale.

Nanozyme-catalysed CRISPR assay for preamplification-free detection of non-coding RNAs.

CRISPR-based diagnostics enable specific sensing of DNA and RNA biomarkers associated with human diseases. This is achieved through the binding of guide RNAs to a complementary sequence that activates Cas enzymes to cleave reporter molecules. Currently, most CRISPR-based diagnostics rely on target preamplification to reach sufficient sensitivity for clinical applications. This limits quantification capability and adds complexity to the reaction chemistry. Here we show the combination of a CRISPR-Cas-based reaction with a nanozyme-linked immunosorbent assay, which allows for the quantitative and colorimetric readout of Cas13-mediated RNA detection through catalytic metallic nanoparticles at room temperature (CrisprZyme). We demonstrate that CrisprZyme is easily adaptable to a lateral-flow-based readout and different Cas enzymes and enables the sensing of non-coding RNAs including microRNAs, long non-coding RNAs and circular RNAs. We utilize this platform to identify patients with acute myocardial infarction and to monitor cellular differentiation in vitro and in tissue biopsies from prostate cancer patients. We anticipate that CrisprZyme will serve as a universally applicable signal catalyst for CRISPR-based diagnostics, which will expand the spectrum of targets for preamplification-free, quantitative detection.

CRISPR-based diagnostics.

The accurate and timely diagnosis of disease is a prerequisite for efficient therapeutic intervention and epidemiological surveillance. Diagnostics based on the detection of nucleic acids are among the most sensitive and specific, yet most such assays require costly equipment and trained personnel. Recent developments in diagnostic technologies, in particular those leveraging clustered regularly interspaced short palindromic repeats (CRISPR), aim to enable accurate testing at home, at the point of care and in the field. In this Review, we provide a rundown of the rapidly expanding toolbox for CRISPR-based diagnostics, in particular the various assays, preamplification strategies and readouts, and highlight their main applications in the sensing of a wide range of molecular targets relevant to human health.

A CRISPR-based assay for the detection of opportunistic infections post-transplantation and for the monitoring of transplant rejection.

In organ transplantation, infection and rejection are major causes of graft loss. They are linked by the net state of immunosuppression. To diagnose and treat these conditions earlier, and to improve long-term patient outcomes, refined strategies for the monitoring of patients after graft transplantation are needed. Here, we show that a fast and inexpensive assay based on CRISPR-Cas13 accurately detects BK polyomavirus DNA and cytomegalovirus DNA from patient-derived blood and urine samples, as well as CXCL9 messenger RNA (a marker of graft rejection) at elevated levels in urine samples from patients experiencing acute kidney transplant rejection. The assay, which we adapted for lateral-flow readout, enables-via simple visualization-the post-transplantation monitoring of common opportunistic viral infections and of graft rejection, and should facilitate point-of-care post-transplantation monitoring.

Impact of Diabetic Stress Conditions on Renal Cell Metabolome.

Diabetic kidney disease is a major complication in diabetes mellitus, and the most common reason for end-stage renal disease. Patients suffering from diabetes mellitus encounter glomerular damage by basement membrane thickening, and develop albuminuria. Subsequently, albuminuria can deteriorate the tubular function and impair the renal outcome. The impact of diabetic stress conditions on the metabolome was investigated by untargeted gas chromatography-mass spectrometry (GC-MS) analyses. The results were validated by qPCR analyses. In total, four cell lines were tested, representing the glomerulus, proximal nephron tubule, and collecting duct. Both murine and human cell lines were used. In podocytes, proximal tubular and collecting duct cells, high glucose concentrations led to global metabolic alterations in amino acid metabolism and the polyol pathway. Albumin overload led to the further activation of the latter pathway in human proximal tubular cells. In the proximal tubular cells, aldo-keto reductase was concordantly increased by glucose, and partially increased by albumin overload. Here, the combinatorial impact of two stressful agents in diabetes on the metabolome of kidney cells was investigated, revealing effects of glucose and albumin on polyol metabolism in human proximal tubular cells. This study shows the importance of including highly concentrated albumin in in vitro studies for mimicking diabetic kidney disease.

Fabrication of Kidney Proximal Tubule Grafts Using Biofunctionalized Electrospun Polymer Scaffolds.

The increasing prevalence of end-stage renal disease and persistent shortage of donor organs call for alternative therapies for kidney patients. Dialysis remains an inferior treatment as clearance of large and protein-bound waste products depends on active tubular secretion. Biofabricated tissues could make a valuable contribution, but kidneys are highly intricate and multifunctional organs. Depending on the therapeutic objective, suitable cell sources and scaffolds must be selected. This study provides a proof-of-concept for stand-alone kidney tubule grafts with suitable mechanical properties for future implantation purposes. Porous tubular nanofiber scaffolds are fabricated by electrospinning 12%, 16%, and 20% poly-ε-caprolactone (PCL) v/w (chloroform and dimethylformamide, 1:3) around 0.7 mm needle templates. The resulting scaffolds consist of 92%, 69%, and 54% nanofibers compared to microfibers, respectively. After biofunctionalization with L-3,4-dihydroxyphenylalanine and collagen IV, 10 × 106 proximal tubule cells per mL are injected and cultured until experimental readout. A human-derived cell model can bridge all fiber-to-fiber distances to form a monolayer, whereas small-sized murine cells form monolayers on dense nanofiber meshes only. Fabricated constructs remain viable for at least 3 weeks and maintain functionality as shown by inhibitor-sensitive transport activity, which suggests clearance capacity for both negatively and positively charged solutes.

Metabolic characterization of directly reprogrammed renal tubular epithelial cells (iRECs).

Fibroblasts can be directly reprogrammed to induced renal tubular epithelial cells (iRECs) using four transcription factors. These engineered cells may be used for disease modeling, cell replacement therapy or drug and toxicity testing. Direct reprogramming induces drastic changes in the transcriptional landscape, protein expression, morphological and functional properties of cells. However, how the metabolome is changed by reprogramming and to what degree it resembles the target cell type remains unknown. Using untargeted gas chromatography-mass spectrometry (GC-MS) and targeted liquid chromatography-MS, we characterized the metabolome of mouse embryonic fibroblasts (MEFs), iRECs, mIMCD-3 cells, and whole kidneys. Metabolic fingerprinting can distinguish each cell type reliably, revealing iRECs are most similar to mIMCD-3 cells and clearly separate from MEFs used for reprogramming. Treatment with the cytotoxic drug cisplatin induced typical changes in the metabolic profile of iRECs commonly occurring in acute renal injury. Interestingly, metabolites in the medium of iRECs, but not of mIMCD-3 cells or fibroblast could distinguish treated and non-treated cells by cluster analysis. In conclusion, direct reprogramming of fibroblasts into renal tubular epithelial cells strongly influences the metabolome of engineered cells, suggesting that metabolic profiling may aid in establishing iRECs as in vitro models for nephrotoxicity testing in the future.

Direct reprogramming of fibroblasts into renal tubular epithelial cells by defined transcription factors.

Direct reprogramming by forced expression of transcription factors can convert one cell type into another. Thus, desired cell types can be generated bypassing pluripotency. However, direct reprogramming towards renal cells remains an unmet challenge. Here, we identify renal cell fate-inducing factors on the basis of their tissue specificity and evolutionarily conserved expression, and demonstrate that combined expression of Emx2, Hnf1b, Hnf4a and Pax8 converts mouse and human fibroblasts into induced renal tubular epithelial cells (iRECs). iRECs exhibit epithelial features, a global gene expression profile resembling their native counterparts, functional properties of differentiated renal tubule cells and sensitivity to nephrotoxic substances. Furthermore, iRECs integrate into kidney organoids and form tubules in decellularized kidneys. Our approach demonstrates that reprogramming factors can be identified by targeted in silico analysis. Renal tubular epithelial cells generated ex vivo by forced expression of transcription factors may facilitate disease modelling, drug and nephrotoxicity testing, and regenerative approaches.