Department News



(l to r) Olivia Ho-Shing, Leigh Needleman, Nimrod Rubinstein, Julio Perez, and Catherine Dulac

Most cells in the mammalian body contain two copies of each chromosome, one passed from each of the parents. While most of the genes are equally active from both of the parental chromosomes, in a limited set of genes, a copy from only one of the parents is inactivated in a process known as genomic imprinting. The process of inactivation, or imprinting, generally takes place in the oocyte and sperm cells, and is achieved through addition of chemical marks on the DNA, which in the offspring, ensures inactivation of the genes. As a result of this phenomenon the maternal and paternal genomes contribute differently to the tissues in which they are expressed. While it is known that many imprinted genes are expressed in the embryo and placenta, and hence play key roles in early development, it is becoming evident that many imprinted genes are also expressed in the developing and adult brain, potentially regulating functions such as feeding, as well as social and motivated behaviors. Mapping imprinted genes in the brain can therefore shed significant light on the genetic basis of these functions.

Despite the advent of RNA sequencing technologies, which provide sensitive measurements of gene expression, several previous efforts to map imprinting in the mouse brain have yielded inconsistent results, most likely due to inadequate experimental design, data analysis, and lack of secondary validation. People from the laboratory of Catherine Dulac, in collaboration with the group of Jun Liu (Department of Statistics), have addressed these issues by applying a combined approach that includes developing a powerful statistical model that accounts for many of the sources of noise and variation inherent to RNA sequencing data, and extensively validating each imprinted gene candidate using an independent experimental technique. Using this approach for mapping imprinted genes in the developing and adult mouse cerebellum, they detected and independently validated 41 novel imprinted genes, in addition to 74 previously known imprinted genes.  This is a staggering increase of the total number of genes known to be imprinted in the mouse.

Their analysis confirmed that imprinting for many genes is manifested as lack of expression from one of the parental copies. However, it also highlighted that many other imprinted genes show a moderate deviation from equal expression of both copies (parental bias). Among many exciting findings of this work, the comparison between the two animal age groups revealed that the parental bias is not constant through life but generally its magnitude tends to decrease with age. Furthermore, quantifying the parental bias for a subset of imprinted genes across the brain and in tissues outside the brain showed a striking preference for these genes to be parentally biased only in the brain.  In addition, they observed a remarkable variability in the degree of the parental bias across the different brain regions (Figure 1).

While it was known that many imprinted genes are involved in tissue growth processes, programmed cell death (apoptosis) emerged as another biological pathway in which imprinted genes are significantly involved. Specifically, the long isoform of the Bcl-x gene (Bcl-xL), whose function is to prevent apoptosis, was found to have a 60:40 paternal-to-maternal expression level ratio. Despite this moderate parental bias, mice in which the paternal copy of the Bcl-x gene was deleted from all brain cells showed a significantly smaller brain size compared to either mice in which the maternal copy of Bcl-x was deleted or to control mice. Studying the brain histology of the paternal Bcl-x deleted mice showed that this loss of brain volume is largely due to a specific loss of excitatory neurons within the brain. This suggests that genomic imprinting of Bcl-xL is specific to certain cell types in the brain. It also clearly demonstrates the biological significance of a key finding of this work – that imprinted genes with moderate parental biases have a functional biological significance.

Read more in eLife or download PDF

Read more in the Harvard Gazette