FANCM regulates repair pathway choice at stalled replication forks.

Repair pathway "choice" at stalled mammalian replication forks is an important determinant of genome stability; however, the underlying mechanisms are poorly understood. FANCM encodes a multi-domain scaffolding and motor protein that interacts with several distinct repair protein complexes at stalled forks. Here, we use defined mutations engineered within endogenous Fancm in mouse embryonic stem cells to study how Fancm regulates stalled fork repair. We find that distinct FANCM repair functions are enacted by molecularly separable scaffolding domains. These findings define FANCM as a key mediator of repair pathway choice at stalled replication forks and reveal its molecular mechanism. Notably, mutations that inactivate FANCM ATPase function disable all its repair functions and "trap" FANCM at stalled forks. We find that Brca1 hypomorphic mutants are synthetic lethal with Fancm null or Fancm ATPase-defective mutants. The ATPase function of FANCM may therefore represent a promising "druggable" target for therapy of BRCA1-linked cancer.

Mechanism of tandem duplication formation in BRCA1-mutant cells.

Small, approximately 10-kilobase microhomology-mediated tandem duplications are abundant in the genomes of BRCA1-linked but not BRCA2-linked breast cancer. Here we define the mechanism underlying this rearrangement signature. We show that, in primary mammalian cells, BRCA1, but not BRCA2, suppresses the formation of tandem duplications at a site-specific chromosomal replication fork barrier imposed by the binding of Tus proteins to an array of Ter sites. BRCA1 has no equivalent role at chromosomal double-stranded DNA breaks, indicating that tandem duplications form specifically at stalled forks. Tandem duplications in BRCA1 mutant cells arise by a replication restart-bypass mechanism terminated by end joining or by microhomology-mediated template switching, the latter forming complex tandem duplication breakpoints. Solitary DNA ends form directly at Tus-Ter, implicating misrepair of these lesions in tandem duplication formation. Furthermore, BRCA1 inactivation is strongly associated with ~10 kilobase tandem duplications in ovarian cancer. This tandem duplicator phenotype may be a general signature of BRCA1-deficient cancer.

Picky comprehensively detects high-resolution structural variants in nanopore long reads.

Acquired genomic structural variants (SVs) are major hallmarks of cancer genomes, but they are challenging to reconstruct from short-read sequencing data. Here we exploited the long reads of the nanopore platform using our customized pipeline, Picky ( https://github.com/TheJacksonLaboratory/Picky ), to reveal SVs of diverse architecture in a breast cancer model. We identified the full spectrum of SVs with superior specificity and sensitivity relative to short-read analyses, and uncovered repetitive DNA as the major source of variation. Examination of genome-wide breakpoints at nucleotide resolution uncovered micro-insertions as the common structural features associated with SVs. Breakpoint density across the genome is associated with the propensity for interchromosomal connectivity and was found to be enriched in promoters and transcribed regions of the genome. Furthermore, we observed an over-representation of reciprocal translocations from chromosomal double-crossovers through phased SVs. We demonstrate that Picky analysis is an effective tool for comprehensive detection of SVs in cancer genomes from long-read data.

The tandem duplicator phenotype as a distinct genomic configuration in cancer.

Next-generation sequencing studies have revealed genome-wide structural variation patterns in cancer, such as chromothripsis and chromoplexy, that do not engage a single discernable driver mutation, and whose clinical relevance is unclear. We devised a robust genomic metric able to identify cancers with a chromotype called tandem duplicator phenotype (TDP) characterized by frequent and distributed tandem duplications (TDs). Enriched only in triple-negative breast cancer (TNBC) and in ovarian, endometrial, and liver cancers, TDP tumors conjointly exhibit tumor protein p53 (TP53) mutations, disruption of breast cancer 1 (BRCA1), and increased expression of DNA replication genes pointing at rereplication in a defective checkpoint environment as a plausible causal mechanism. The resultant TDs in TDP augment global oncogene expression and disrupt tumor suppressor genes. Importantly, the TDP strongly correlates with cisplatin sensitivity in both TNBC cell lines and primary patient-derived xenografts. We conclude that the TDP is a common cancer chromotype that coordinately alters oncogene/tumor suppressor expression with potential as a marker for chemotherapeutic response.

Systems consequences of amplicon formation in human breast cancer.

Chromosomal structural variations play an important role in determining the transcriptional landscape of human breast cancers. To assess the nature of these structural variations, we analyzed eight breast tumor samples with a focus on regions of gene amplification using mate-pair sequencing of long-insert genomic DNA with matched transcriptome profiling. We found that tandem duplications appear to be early events in tumor evolution, especially in the genesis of amplicons. In a detailed reconstruction of events on chromosome 17, we found large unpaired inversions and deletions connect a tandemly duplicated ERBB2 with neighboring 17q21.3 amplicons while simultaneously deleting the intervening BRCA1 tumor suppressor locus. This series of events appeared to be unusually common when examined in larger genomic data sets of breast cancers albeit using approaches with lesser resolution. Using siRNAs in breast cancer cell lines, we showed that the 17q21.3 amplicon harbored a significant number of weak oncogenes that appeared consistently coamplified in primary tumors. Down-regulation of BRCA1 expression augmented the cell proliferation in ERBB2-transfected human normal mammary epithelial cells. Coamplification of other functionally tested oncogenic elements in other breast tumors examined, such as RIPK2 and MYC on chromosome 8, also parallel these findings. Our analyses suggest that structural variations efficiently orchestrate the gain and loss of cancer gene cassettes that engage many oncogenic pathways simultaneously and that such oncogenic cassettes are favored during the evolution of a cancer.

QuIN: A Web Server for Querying and Visualizing Chromatin Interaction Networks.

UNLABELLED: Recent studies of the human genome have indicated that regulatory elements (e.g. promoters and enhancers) at distal genomic locations can interact with each other via chromatin folding and affect gene expression levels. Genomic technologies for mapping interactions between DNA regions, e.g., ChIA-PET and HiC, can generate genome-wide maps of interactions between regulatory elements. These interaction datasets are important resources to infer distal gene targets of non-coding regulatory elements and to facilitate prioritization of critical loci for important cellular functions. With the increasing diversity and complexity of genomic information and public ontologies, making sense of these datasets demands integrative and easy-to-use software tools. Moreover, network representation of chromatin interaction maps enables effective data visualization, integration, and mining. Currently, there is no software that can take full advantage of network theory approaches for the analysis of chromatin interaction datasets. To fill this gap, we developed a web-based application, QuIN, which enables: 1) building and visualizing chromatin interaction networks, 2) annotating networks with user-provided private and publicly available functional genomics and interaction datasets, 3) querying network components based on gene name or chromosome location, and 4) utilizing network based measures to identify and prioritize critical regulatory targets and their direct and indirect interactions. AVAILABILITY: QuIN's web server is available at http://quin.jax.org QuIN is developed in Java and JavaScript, utilizing an Apache Tomcat web server and MySQL database and the source code is available under the GPLV3 license available on GitHub: https://github.com/UcarLab/QuIN/.

Mapping the genetic landscape establishing a tumor immune microenvironment favorable for anti-PD-1 response in mice and humans.

Identifying host genetic factors modulating immune checkpoint inhibitor (ICI) efficacy has been experimentally challenging because of variations in both host and tumor genomes, differences in the microbiome, and patient life exposures. Utilizing the Collaborative Cross (CC) multi-parent mouse genetic resource population, we developed an approach that fixes the tumor genomic configuration while varying host genetics. With this approach, we discovered that response to anti-PD-1 (aPD1) immunotherapy was significantly heritable in four distinct murine tumor models (H2 between 0.18-0.40). For the MC38 colorectal carcinoma system (H2 = 0.40), we mapped four significant ICI response quantitative trait loci (QTL) localized to mouse chromosomes (mChr) 5, 9, 15 and 17, and identified significant epistatic interactions between specific QTL pairs. Differentially expressed genes within these QTL were highly enriched for immune genes and pathways mediating allograft rejection and graft vs host disease. Using a cross species analytical approach, we found a core network of 48 genes within the four QTLs that showed significant prognostic value for overall survival in aPD1 treated human cohorts that outperformed all other existing validated immunotherapy biomarkers, especially in human tumors of the previously defined immune subtype 4. Functional blockade of two top candidate immune targets within the 48 gene network, GM-CSF and high affinity IL-2/IL-15 signaling, completely abrogated the MC38 tumor transcriptional response to aPD1 therapy in vivo. Thus, we have established a powerful cross species in vivo platform capable of uncovering host genetic factors that establish the tumor immune microenvironment configuration propitious for ICI response.

Expression profile of frizzled receptors in human medulloblastomas.

Secreted WNT proteins signal through ten receptors of the frizzled (FZD) family. Because of the relevance of the WNT/ß-catenin (CTNNB1) signaling pathway in medulloblastomas (MBs), we investigated the expression of all ten members of the FZD gene family (FZD1-10) in 17 human MBs, four MB cell lines and in normal human cerebellum, using real-time PCR. We found that FZD2 transcript was over-expressed in all MBs and MB cell lines. Western blot analysis confirmed the expression of FZD2 at the protein level. Moreover, the levels of FZD2 transcript were found to correlate with those of ASPM transcript, a marker of mitosis essential for mitotic spindle function. Accordingly, ASPM mRNA was expressed at a very low level in the adult, post-mitotic, human cerebellum, at higher levels in fetal cerebellum and at highest levels in MB tissues and cell lines. Unlike FZD2, the other FZDs were overexpressed (e.g., FZD1, FZD3 and FZD8) or underexpressed (e.g., FZD7, FZD9 and FZD10) in a case-restricted manner. Interestingly, we did not find any nuclear immuno-reactivity to CTNNB1 in four MBs over-expressing both FZD2 and other FZD receptors, confirming the lack of nuclear CTNNB1 staining in the presence of increased FZD expression, as in other tumor types. Overall, our results indicate that altered expression of FZD2 might be associated with a proliferative status, thus playing a role in the biology of human MBs, and possibly of cerebellar progenitors from which these malignancies arise.

Genomic and epigenomic BRCA alterations predict adaptive resistance and response to platinum-based therapy in patients with triple-negative breast and ovarian carcinomas.

Triple-negative breast cancer (TNBC) and ovarian carcinomas (OvCas) with BRCA1 promoter methylation (BRCA1meth) respond more poorly to alkylating agents compared to those bearing mutations in BRCA1 and BRCA2 (BRCAmut). This is a conundrum given the biologically equivalent homologous recombination deficiency (HRD) induced by these genetic and epigenetic BRCA perturbations. We dissected this problem through detailed genomic analyses of TNBC and OvCa cohorts and experimentation with patient-derived xenografts and genetically engineered cell lines. We found that despite identical downstream genomic mutational signatures associated with BRCA1meth and BRCAmut states, BRCA1meth uniformly associates with poor outcomes. Exposure of BRCA1meth TNBCs to platinum chemotherapy, either as clinical treatment of a patient or as experimental in vivo exposure of preclinical patient derived xenografts, resulted in allelic loss of BRCA1 methylation and increased BRCA1 expression and platinum resistance. These data suggest that, unlike BRCAmut cancers, where BRCA loss is a genetically "fixed" deficiency state, BRCA1meth cancers are highly adaptive to genotoxin exposure and, through reversal of promoter methylation, recover BRCA1 expression and become resistant to therapy. We further found a specific augmented immune transcriptional signal associated with enhanced response to platinum chemotherapy but only in patients with BRCA-proficient cancers. We showed how integrating both this cancer immune signature and the presence of BRCA mutations results in more accurate predictions of patient response when compared to either HRD status or BRCA status alone. This underscores the importance of defining BRCA heterogeneity in optimizing the predictive precision of assigning response probabilities in TNBC and OvCa.

Global analysis of gene expression in NGF-deprived sympathetic neurons identifies molecular pathways associated with cell death.

BACKGROUND: Developing sympathetic neurons depend on nerve growth factor (NGF) for survival and die by apoptosis after NGF withdrawal. This process requires de novo gene expression but only a small number of genes induced by NGF deprivation have been identified so far, either by a candidate gene approach or in mRNA differential display experiments. This is partly because it is difficult to obtain large numbers of sympathetic neurons for in vitro studies. Here, we describe for the first time, how advances in gene microarray technology have allowed us to investigate the expression of all known genes in sympathetic neurons cultured in the presence and absence of NGF. RESULTS: We have used Affymetrix Exon arrays to study the pattern of expression of all known genes in NGF-deprived sympathetic neurons. We identified 415 up- and 813 down-regulated genes, including most of the genes previously known to be regulated in this system. NGF withdrawal activates the mixed lineage kinase (MLK)-c-Jun N-terminal kinase (JNK)-c-Jun pathway which is required for NGF deprivation-induced death. By including a mixed lineage kinase (MLK) inhibitor, CEP-11004, in our experimental design we identified which of the genes induced after NGF withdrawal are potential targets of the MLK-JNK-c-Jun pathway. A detailed Gene Ontology and functional enrichment analysis also identified genetic pathways that are highly enriched and overrepresented amongst the genes expressed after NGF withdrawal. Five genes not previously studied in sympathetic neurons - trib3, ddit3, txnip, ndrg1 and mxi1 - were validated by real time-PCR. The proteins encoded by these genes also increased in level after NGF withdrawal and this increase was prevented by CEP-11004, suggesting that these genes are potential targets of the MLK-JNK-c-Jun pathway. CONCLUSIONS: The sympathetic neuron model is one of the best studied models of neuronal apoptosis. Overall, our microarray data gives a comprehensive overview of, and provides new information about, signalling pathways and transcription factors that are regulated by NGF withdrawal.

DNA microarray analysis identifies CKS2 and LEPR as potential markers of meningioma recurrence.

Meningiomas are the most frequent intracranial tumors. Surgery can be curative, but recurrences are possible. We performed gene expression analyses and loss of heterozygosity (LOH) studies looking for new markers predicting the recurrence risk. We analyzed expression profiles of 23 meningiomas (10 grade I, 10 grade II, and 3 grade III) and validated the data using quantitative polymerase chain reaction (qPCR). We performed LOH analysis on 40 meningiomas, investigating chromosomal regions on 1p, 9p, 10q, 14q, and 22q. We found 233 and 268 probe sets to be significantly down- and upregulated, respectively, in grade II or III meningiomas. Genes downregulated in high-grade meningiomas were overrepresented on chromosomes 1, 6, 9, 10, and 14. Based on functional enrichment analysis, we selected LIM domain and actin binding 1 (LIMA1), tissue inhibitor of metalloproteinases 3 (TIMP3), cyclin-dependent kinases regulatory subunit 2 (CKS2), leptin receptor (LEPR), and baculoviral inhibitor of apoptosis repeat-containing 5 (BIRC5) for validation using qPCR and confirmed their differential expression in the two groups of tumors. We calculated ΔCt values of CKS2 and LEPR and found that their differential expression (C-L index) was significantly higher in grade I than in grade II or III meningiomas (p < .0001). Interestingly, the C-L index of nine grade I meningiomas from patients who relapsed in <5 years was significantly lower than in grade I meningiomas from patients who did not relapse. These findings indicate that the C-L index may be relevant to define the progression risk in meningioma patients, helping guide their clinical management. A prospective analysis on a larger number of cases is warranted.

Extrachromosomal DNA is associated with oncogene amplification and poor outcome across multiple cancers.

Extrachromosomal DNA (ecDNA) amplification promotes intratumoral genetic heterogeneity and accelerated tumor evolution1-3; however, its frequency and clinical impact are unclear. Using computational analysis of whole-genome sequencing data from 3,212 cancer patients, we show that ecDNA amplification frequently occurs in most cancer types but not in blood or normal tissue. Oncogenes were highly enriched on amplified ecDNA, and the most common recurrent oncogene amplifications arose on ecDNA. EcDNA amplifications resulted in higher levels of oncogene transcription compared to copy number-matched linear DNA, coupled with enhanced chromatin accessibility, and more frequently resulted in transcript fusions. Patients whose cancers carried ecDNA had significantly shorter survival, even when controlled for tissue type, than patients whose cancers were not driven by ecDNA-based oncogene amplification. The results presented here demonstrate that ecDNA-based oncogene amplification is common in cancer, is different from chromosomal amplification and drives poor outcome for patients across many cancer types.

Methylation of O6-methylguanine DNA methyltransferase and loss of heterozygosity on 19q and/or 17p are overlapping features of secondary glioblastomas with prolonged survival.

PURPOSE: Recent data suggest that methylation of the DNA repair gene O(6)-methylguanine DNA methyltransferase (MGMT), by increasing the chemosensitivity of glioblastoma multiforme, is significantly associated with improved prognosis. Results in contradiction with these findings, however, are present in the literature and the clinical and genetic context framing MGMT methylation is poorly characterized. EXPERIMENTAL DESIGN: To address these issues, we have investigated the MGMT methylation status, clinical and magnetic resonance imaging characteristics, and relevant genetic features (loss of heterozygosity on 17p and 19q, EGFR amplification, and p53 mutations) in a retrospective study on 86 patients affected by glioblastoma multiforme: 72 patients had a clinical history indicating de novo insurgence of the tumor and the remaining 14 were secondary glioblastoma multiforme. RESULTS: MGMT methylation was detected by methylation-specific PCR in 41 of 86 cases (47.7%; Meth+). Progression-free survival and overall survival were significantly longer in Meth+ than in Meth- patients [10 versus 7 months (P=0.003, log-rank test) and 18 versus 14 months (P=0.0003, log-rank test), respectively]. Mixed-nodular enhancement at magnetic resonance imaging was significantly more frequent in Meth+ and secondary glioblastoma multiforme and ring enhancement in Meth- and primary glioblastoma multiforme (P<0.005). MGMT methylation was more present in secondary glioblastoma multiforme (P=0.006) and associated with loss of heterozygosity on 17p and/or 19q (P=0.005). CONCLUSIONS: These observations suggest that MGMT methylation is part of a genetic signature of glioblastomas that developed from lower-grade gliomas.

Neurospheres enriched in cancer stem-like cells are highly effective in eliciting a dendritic cell-mediated immune response against malignant gliomas.

Cancer stem-like cells (CSC) could be a novel target for cancer therapy, including dendritic cell (DC) immunotherapy. To address this, we developed experiments aimed at DC targeting of neurospheres (NS) from GL261 glioma cells because neurospheres can be enriched in CSC. We obtained murine neurospheres by growing GL261 cells in epidermal growth factor/basic fibroblast growth factor without serum. GL261-NS recapitulated important features of glioblastoma CSC and expressed higher levels of radial glia stem cell markers than GL261 cells growing under standard conditions (GL261 adherent cells, GL261-AC), as assessed by DNA microarray and real-time PCR. GL261-NS brain gliomas were highly infiltrating and more rapidly lethal than GL261-AC, as evidenced by survival analysis (P < 0.0001), magnetic resonance imaging and histology. DC from the bone marrow of syngeneic mice were then used for immunotherapy of GL261-NS and GL261-AC tumors. Strikingly, DC loaded with GL261-NS (DC-NS) cured 80% and 60% of GL261-AC and GL261-NS tumors, respectively (P < 0.0001), whereas DC-AC cured only 50% of GL261-AC tumors (P = 0.0022) and none of the GL261-NS tumors. GL261-NS expressed higher levels of MHC and costimulatory molecules (CD80 and CD86) than GL261-AC; the JAM assay indicated that DC-NS splenocytes had higher lytic activity than DC-AC splenocytes on both GL261-NS and GL261-AC, and immunohistochemistry showed that DC-NS vaccination was associated with robust tumor infiltration by CD8+ and CD4+ T lymphocytes. These findings suggest that DC targeting of CSC provides a higher level of protection against GL261 gliomas, a finding with potential implications for the design of clinical trials based on DC vaccination.

The Tandem Duplicator Phenotype Is a Prevalent Genome-Wide Cancer Configuration Driven by Distinct Gene Mutations.

The tandem duplicator phenotype (TDP) is a genome-wide instability configuration primarily observed in breast, ovarian, and endometrial carcinomas. Here, we stratify TDP tumors by classifying their tandem duplications (TDs) into three span intervals, with modal values of 11 kb, 231 kb, and 1.7 Mb, respectively. TDPs with ∼11 kb TDs feature loss of TP53 and BRCA1. TDPs with ∼231 kb and ∼1.7 Mb TDs associate with CCNE1 pathway activation and CDK12 disruptions, respectively. We demonstrate that p53 and BRCA1 conjoint abrogation drives TDP induction by generating short-span TDP mammary tumors in genetically modified mice lacking them. Lastly, we show how TDs in TDP tumors disrupt heterogeneous combinations of tumor suppressors and chromatin topologically associating domains while duplicating oncogenes and super-enhancers.

High-resolution deconstruction of evolution induced by chemotherapy treatments in breast cancer xenografts.

The processes by which tumors evolve are essential to the efficacy of treatment, but quantitative understanding of intratumoral dynamics has been limited. Although intratumoral heterogeneity is common, quantification of evolution is difficult from clinical samples because treatment replicates cannot be performed and because matched serial samples are infrequently available. To circumvent these problems we derived and assayed large sets of human triple-negative breast cancer xenografts and cell cultures from two patients, including 86 xenografts from cyclophosphamide, doxorubicin, cisplatin, docetaxel, or vehicle treatment cohorts as well as 45 related cell cultures. We assayed these samples via exome-seq and/or high-resolution droplet digital PCR, allowing us to distinguish complex therapy-induced selection and drift processes among endogenous cancer subclones with cellularity uncertainty <3%. For one patient, we discovered two predominant subclones that were granularly intermixed in all 48 co-derived xenograft samples. These two subclones exhibited differential chemotherapy sensitivity-when xenografts were treated with cisplatin for 3 weeks, the post-treatment volume change was proportional to the post-treatment ratio of subclones on a xenograft-to-xenograft basis. A subsequent cohort in which xenografts were treated with cisplatin, allowed a drug holiday, then treated a second time continued to exhibit this proportionality. In contrast, xenografts from other treatment cohorts, spatially dissected xenograft fragments, and cell cultures evolved in diverse ways but with substantial population bottlenecks. These results show that ecosystems susceptible to successive retreatment can arise spontaneously in breast cancer in spite of a background of irregular subclonal bottlenecks, and our work provides to our knowledge the first quantification of the population genetics of such a system. Intriguingly, in such an ecosystem the ratio of common subclones is predictive of the state of treatment susceptibility, showing how measurements of subclonal heterogeneity could guide treatment for some patients.

Large-scale functional organization of long-range chromatin interaction networks.

Chromatin interactions play important roles in transcription regulation. To better understand the underlying evolutionary and functional constraints of these interactions, we implemented a systems approach to examine RNA polymerase-II-associated chromatin interactions in human cells. We found that 40% of the total genomic elements involved in chromatin interactions converged to a giant, scale-free-like, hierarchical network organized into chromatin communities. The communities were enriched in specific functions and were syntenic through evolution. Disease-associated SNPs from genome-wide association studies were enriched among the nodes with fewer interactions, implying their selection against deleterious interactions by limiting the total number of interactions, a model that we further reconciled using somatic and germline cancer mutation data. The hubs lacked disease-associated SNPs, constituted a nonrandomly interconnected core of key cellular functions, and exhibited lethality in mouse mutants, supporting an evolutionary selection that favored the nonrandom spatial clustering of the least-evolving key genomic domains against random genetic or transcriptional errors in the genome. Altogether, our analyses reveal a systems-level evolutionary framework that shapes functionally compartmentalized and error-tolerant transcriptional regulation of human genome in three dimensions.

Genome-wide analysis of alternative splicing in medulloblastoma identifies splicing patterns characteristic of normal cerebellar development.

Alternative splicing is an important mechanism for the generation of protein diversity at a post-transcriptional level. Modifications in the splicing patterns of several genes have been shown to contribute to the malignant transformation of different tissue types. In this study, we used the Affymetrix Exon arrays to investigate patterns of differential splicing between pediatric medulloblastomas and normal cerebellum on a genome-wide scale. Of the 1,262 genes identified as potentially generating tumor-associated splice forms, we selected 14 examples of differential splicing of known cassette exons and successfully validated 11 of them by reverse transcriptase PCR. The pattern of differential splicing of three validated events was characteristic for the molecular subset of sonic hedgehog (Shh)-driven medulloblastomas, suggesting that their unique gene signature includes the expression of distinctive transcript variants. Generally, we observed that tumor and normal fetal cerebellar samples shared significantly lower exon inclusion rates than normal adult cerebellum. We investigated whether tumor-associated splice forms were expressed in primary cultures of Shh-dependent mouse cerebellar granule cell precursors (GCP) and found that Shh caused a decrease in the cassette exon inclusion rate of five of the seven tested genes. Furthermore, we observed a significant increase in exon inclusion between postnatal days 7 and 14 of mouse cerebellar development, at the time when GCPs mature into postmitotic neurons. We conclude that inappropriate splicing frequently occurs in human medulloblastomas and may be linked to the activation of developmental signaling pathways and a failure of cerebellar precursor cells to differentiate.