A Revolution in Germline Toxicology: Dr. Toshi Shioda’s “PGC-LCs”
"If epigenetic errors occur in the germline genome during the reprogramming process
Erasure of DNA methylation, genomic imprints and epimutations during primordial germ cell differentiation from mouse pluripotent stem cells
I was excited to learn about your work developing a new model for addressing epigenomic vulnerabilities of germ cells—this is an area badly in need of innovation and solutions. How did you come to work in this field?
About 15 years ago I was a breast cancer researcher focusing on genetics and epigenetics, seeking diagnostic markers and drug targets. However, after I recognized the importance of breast cancer prevention, my research focus gradually shifted to epigenetic effects of fetal exposure to hormonally active toxicants and drugs, which are among the major risk factors for breast cancer.
Then I was exposed to an array of the emerging concepts including DOHaD, or Developmental Origins of Health and Disease, and transgenerational epigenetic effects, which led me to the research field of stem cell epigenetics. After realizing that even deficiency of vitamin C strongly affects the epigenetic health of stem cells, I concluded that I must look at the epigenetic status of germline cells to better understand the epigenetic effects of fetal exposure on human health.
So, I no longer limit myself to breast cancer research. Anything that may be linked to unhealthy epigenetic status of germline cells caused by any types of harms (e.g., exposure to drugs or chemicals, malnutrition, extreme stresses) is my interest.
So you looked for a way to test these phenomena?
As an expert in genetic and epigenetic analyses, I have been emphasizing the importance of reproducible, specific, and experimentally testable epigenetic phenomena as a model to understand mechanisms of the harmful effects. While I have been collaborating with a number of animal model experts who are exposing pregnant animals to drugs and toxicants, a couple years ago I decided to develop a cell culture model of human and mouse germline cells derived from the embryonic stem cells and the iPS cells for my germline toxico-epigenomics research.
It had been believed that germline epimutations were always corrected during a second wave reprogramming process, restoring normal gene expression, but now we know that is not always true. Can you summarize how some epimutations can persist in the germline?
We still don’t have a confirmed example of an epimutation that survives the epigenetic reprogramming in the genome of mammalian germline and causes aberrant gene expression. However, recent studies, including ours, suggest that there are exceptional genomic elements that do not lose DNA methylation during the germline epigenetic reprogramming. Some of them are retrotransposons, which move around the genomic DNA if their gene expression is de-repressed. The germline cells seem to have a special but uncharacterized mechanism to suppress such “toxic” genes even during the short period of germline epigenetic reprogramming. Other exceptional genomic elements include sequences important for maintaining the normal structure of chromosomes, and the gamma-satellite repeat seems like an example that has to maintain DNA methylation to prevent chromosomal damages. Again, we don’t know why such sequences are untouched by the genome-wide epigenetic erasure in the germline. Once an epimutation occurs somewhere within these exceptional genomic elements, it might be inherited to the subsequence generations. We need to know if this actually occurs and, if so, whether it significantly affects health of the offspring.
The epigenetic reprogramming in the mammalian germline cells is a highly complex process. It is amazing to see how the epigenetic marks are thoroughly erased from the entire genome and how a new, correct marks are established during the germline development. The existence of the exceptional genomic elements that escape this process — which makes perfect sense due to biological necessities — adds one more layer of challenge to understand the mechanisms of this entire process. However, if you think about how accurately this process has to be to prevent epigenetic errors that cause disorders, I can easily assume that the germline epigenetic reprogramming is highly vulnerable to a lot of factors, including exposure to chemicals or malnutrition. If epigenetic errors occur in the germline genome during the reprogramming process during pregnancy of a woman, her sons or daughters appear normal, but their germline cells carry a potential bomb. However, if the effects of exposure during pregnancy are detectable only in the grandchildren - not in the direct sons or daughters, the present “commonsense” of cause-and-effect relationship would tell you that there is no connection. We need reliable epigenetic biomarkers that detect such direct children who carry germline epimutations that may affect health of their own children.
A way to test germ cell toxicity is your PGC-LCs, or primordial germ cell-like cells— can you briefly describe how you create the PGC-LCs?
We generate PGC-LCs from pluripotent stem cells. We use embryonic stem cells or iPS cells to generate mouse PGC-LCs. We use only iPS cells, generated in our own lab from foreskin cells of newborn boys, to generate human PGC-LCs. These pluripotent stem cells are exposed to cocktails of various growth factors and chemicals to mimic the normal process of germline differentiation. The first half of PGC-LC induction is performed in the standard plastic dishes, but the latter half occurs only in the embryoid bodies, which are cell aggregates that partly mimic that three-dimensional body structures of the human or mouse embryos. The entire process takes 8-10 days.
In what respects are the PGC-LCs different from natural germ cells, both male and female?
My lab has just completed thorough epigenetic characterization of a mouse primordial germ cell-like cell culture model, demonstrating striking similarities of our model with the natural primordial germ cells isolated from mouse embryos. This is reported in my newest manuscript. Deep sequencing analyses of mouse PGC-LCs and gonadal natural PGCs for mRNA expression, DNA methylation, DNA hydroxymethylation, and histone modiﬁcations have demonstrated signiﬁcant epigenomic and transcriptomal similarities between them, supporting the usefulness of PGC-LCs as a model for epigenomic research on germline cells.
Although we observed significant erasure of DNA methylation at the imprinted genes, some laboratories argue that PGC-LCs do not affect epigenetic regulation of the imprinted genes. Epigenetic marks of the imprinted genes are completely erased in normal mouse or human primordial germ cells in vivo. It appears that the exact characteristics of PGC-LCs differ among laboratories although most of the gene expression marks agree well.
We are also generating a human primordial germ cell-like cell culture model routinely in the lab, and we have finished their gene expression profiling, which also showed great similarities with human primordial germ cells reported by Amander Clark of UCLA. However, whether the epigenetic profile of human PGC-LCs faithfully reflects that of primordial germ cells in vivo remains to be established.
How do you anticipate the PCG-LCs will ultimately be utilized?
I think there are three major applications of PGC-LCs.
Firstly, PGC-LCs can be used to develop testable hypotheses about mechanisms through which exposures or harms induce epimutations in the genome of germline cells to cause diseases. This avenue of research may be important to develop knowledge about how to prevent diseases mediated by germline epimutations.
Secondly, PGC-LCs may be useful for safety screenings of drugs or potentially toxic chemicals. The germline toxicity screenings of drugs or chemicals typically detect reduction, malfunction, or genomic DNA mutations of germline cells after exposure of adult animals. Such tests do not examine whether apparently normal germline cells could convey diseases through epigenetic aberrations. PGC-LCs may introduce an additional layer of safety screening criteria that are important to prevent human diseases that are inherited through cryptic epimutations.
Thirdly, PGC-LCs may provide unique opportunities to develop epigenetic biomarkers that reflect epigenetic health or damages of the germline genome, especially in humans. Reliable epigenetic biomarkers that reflect exposure to epimutagens will be useful for not only clinical applications for individual persons but also epidemiological investigations that attempt to determine what factors (exposure to chemicals, nutrition, lifestyle including smoking or drinking, sex, age, ethnicity, other diseases, drugs, etc) affect health of the human epigenome. This approach may also identify “hot spots” in which epimutations occur more frequently than other regions and/or persist over generations.
What types of compounds could you anticipate testing?
Epidemiological and animal experiment information may help us to select drugs and toxicants to be tested in such screening experiments. After completion of more thorough evaluations of the epigenetic status of the human and mouse germline cell culture models, we will initiate exposure experiments to observe induced epigenetic abnormalities.
I think it is important to point that some compounds may affect the germline genome indirectly through disrupting the body functions of their hosts. If compounds damage the host’s endocrine system, blood flow, metabolism, etc, such compounds may significantly affect the epigenome of the germline cells even when the germline cells themselves have never been exposed to the compounds. Results of culture-based exposure assessments have to be interpreted very carefully; apparently negative results are far from approving the compounds as germline-safe.
Your model to test genetic/epigenetic toxicity of PGC-like cells in culture is fantastic and very badly needed. If I were a researcher I would try to ascertain the germline effects of cigarette smoke, general anesthetic agents, and synthetic steroid hormones, including those I had been exposed to prenatally. But there are many others.
Those potential germline epimutagens seem reasonable. I think there can be two categories among them — namely, specific strong epimutagens (perhaps hormonal drugs) and relatively nonspecific and weak epimutagens. For example, DES (diethylstilbestrol) is a very strong estrogen—I don’t know why it was even approved for human use—but other drugs may be also significant owing to their chronic exposure.
What germline effects have you discovered so far with respect to different toxicants?
Exposure of pregnant mice to the environmental toxic chemicals such as Bisphenol A or tributyltin causes transgenerationally transmittable disorders including breast hyperplasia or obesity. Collaborating with multiple extramural laboratories that perform animal exposure studies, our laboratory has been searching for possible epigenetic changes in germline cells as well as tissues showing the adult-onset phenotypes (e.g., mammary glands, adipocytes, and mesenchymal stem cells). The goal of these collaborative projects is to identify toxicant-induced “epigenetic lesions” that are responsible for the late-onset and/or transgenerational disease phenotypes in the genomes of the exposed fetuses and their progenies.
What specific molecular artifacts can you detect in your analyses of the affected PGC-LCs? Methylation? Histone modifications? ncRNAs?
Well, the term “molecular artifacts” means experimental by-products that do not reflect the biological processes in the real world…. If you mean epigenetic marks, we can detect genomic DNA methylation and hydroxymethylation, histone tail modifications, long- and short ncRNAs. We can also detect the “openness” or “accessibility” of chromosomes.
With respect to autism, there’s no known “epigenetic” signature — and I’m not sure how a germline epimutation could even be detected in subjects with autism. If you find epimutation in PGC-LCs induced by, say, cigarette smoke or synthetic steroids, how could researchers then possibly connect those molecular errors to a possible phenotypes such as abnormal neurodevelopment?
First of all, we need to know what exactly autism is. Is it a single disease, or multiple different diseases that show similar symptoms recognized or diagnosed as autistic? Transcriptomal and epigenome analyses should be performed for such “subtypes” of autism, and we need to identify candidates of subtype-specific and/or common autism-causing genes (possibly important for neuro-development). Knowledge on autism biomarkers that may not “cause” autism but reflect the risk would also help. Then PGC-LCs will be useful to examine if any conditions in cell culture can cause transcriptional or epigenetic changes similar to changes observed in autism. In my opinion, we do not need to be thorough or comprehensive to make a breakthrough. As long as observations are reproducible, an example in which a specific epimutation linked to even a small subset of autism is generated in PGC-LCs by exposure to certain chemical would suffice. Once we penetrate the wall, even through a very small hall, we often can see what’s going on beyond the wall.
The FDA at this time requires no ascertainment of fetal germline effects of pregnancy drugs, an omission I consider unconscionable considering both the vulnerability and the importance of those cells. Do you think this lack of germ cell safety testing will change with your new model?
I am not very optimistic to assume that the US FDA is willing to change their procedure or policy only because a new single assay for a newly proposed type of health risk is made available, no matter how important the consequence of the risk can be. We cannot, or should not, ask FDA to screen drugs without knowing what exactly we are asking this agency to do. I rather worry that a premature FDA action forced to be initiated by political or public pressure without establishing definitive positive control cases or scientific bases might incorrectly result in negative conclusions, misjudging many compounds as germline-safe.
If a pregnant woman is exposed to a compound and her direct babies show immediately discernible health problems, it is easy to convince politicians and the public about the threat. But if the consequence skips one generation between the exposed pregnant woman and the affected grandchildren, some people may feel that it is a too long shot or a sophistry. But the more I learn biological details of how germline cells develop and reprogram their genomes, the more I am getting convinced that environmental factors would readily damage such extremely delicate process and cause health problems in very elusive but significant manners.
However, I do not think that the classic cause-and-effect judgment model used by the FDA for decades can effectively handle this new type of health risk. It could be easier to launch a new movement to assess the germline safety of drugs outside the FDA rather than to ask the FDA for changing their established standard. I have no intention to criticize the FDA because the health threats we are investigating are not what the FDA is meant to go after with their present resources. We scientists have to develop effective resources to deal with threats to the epigenetic health of the germline cells, and the public have to support the importance to manage such threats. Then the opinion will be formed as to whether the FDA is the best regulatory agency to handle the new resources or it could be time to launch a new one with new goals and expertise.
Explain to me your interest in monoallelic expression. Is this just regarding imprinted genes, or other genes as well?
To examine effects of toxic agents on monoallelic gene expression in germline cells, we have generated mouse iPS cells whose paternal and maternal chromosomes are derived from Mus spretus and Mus musculus, respectively, by interspeciﬁc in vitro fertilization. Taking advantage of the rich SNPs between these two distant species of Mus, which appear at approximately every 100 bp in their nucleotide base sequences, we are presently developing a deep sequencing pipeline for sensitive and quantitative determination of monoallelic gene expression in these iPS cells, their differentiated products such as PGC-LCs, and various tissues of animals generated by tetraploid complementation. Monoallelic gene expression will be further examined in the context of single cell analysis, which may reveal signiﬁcant intercellular heterogeneity among normal cells and cellular responses to epimutagens.
What are the limitations of using human PGC-LCs as a surrogate model of human natural PGCs?
The unique problem of this surrogate model is that we actually have very limited information on the human natural PGCs. The PGCs are present only in early stages of embryos, and their number in each embryo is extremely small and typically insufficient for genomic analyses. We cannot ask pregnant women to take risky compounds and then give us their embryos growing in their wombs….
So we cannot rigorously assess the advantages and limitations of human PGC-LCs. This is why my laboratory started a mouse PGC-LC research project ahead of the human PGC-LC project. Although the biology of mouse germline is significantly different from the biology of human germline, we can compare mouse PGC-LCs with mouse natural PGCs, through which we can assess the pros and cons of this surrogate model. Then we will attempt to extrapolate the limitations of the human PGC-LC models based no the knowledge accumulated from the mouse model. It is my gut feeling that we need to establish a cell culture system in which human PGC-LCs can differentiate into more advanced stages of germline cells, probably involving the Sertoli cells or follicular cells.
Thank you so much for your time and your fascinating, important work, Dr. Shioda.
"I concluded that I must look at the epigenetic status of germline cells to better understand the epigenetic effects of fetal exposure on human health."
"Recent studies, including ours, suggest that there are exceptional genomic elements that do not lose DNA methylation during the germline epigenetic reprogramming."
"We need reliable epigenetic biomarkers that detect such direct children who carry germline epimutations that may affect health of their own children."
"PGC-LCs can be used to develop testable hypotheses about mechanisms through which exposures or harms induce epimutations in the genome of germline cells to cause diseases."
"It is important to point that some compounds may affect the germline genome indirectly through disrupting the body functions of their hosts."
"The goal of these collaborative projects is to identify toxicant-induced 'epigenetic lesions' that are responsible for the late-onset and/or transgenerational disease phenotypes in the genomes of the exposed fetuses and their progenies."
"PGC-LCs will be useful to examine if any conditions in cell culture can cause transcriptional or epigenetic changes similar to changes observed in autism."
"If a pregnant woman is exposed to a compound and her direct babies show immediately discernible health problems, it is easy to convince politicians and the public about the threat. But if the consequence skips one generation between the exposed pregnant woman and the affected grandchildren, some people may feel that it is a too long shot or a sophistry."
"The more I learn biological details of how germline cells develop and reprogram their genomes, the more I am getting convinced that environmental factors would readily damage such extremely delicate process and cause health problems in very elusive but significant manners."