We recently wrote and submitted a Letter to the Editor of Neuron in response to the recent article by Bundo et al. (2014) reporting increased LINE1 retrotransposition in schizophrenia genomes. Unfortunately, although the editors were interested in the piece, they no longer accept Letters to the Editor and our brief piece fit neither the formatting of a research article nor a viewpoint piece. So, alas, it was ultimately rejected. But so as not to let work go to waste, I’ve elected to publish it here, lacking in peer review though it may be. Hopefully it might be of interest to a few readers.
Here, also, is the original abstract from Bundo et al.’s article for reference:
Recent studies indicate that long interspersed nuclear element-1 (L1) are mobilized in the genome of human neural progenitor cells and enhanced in Rett syndrome and ataxia telangiectasia. However, whether aberrant L1 retrotransposition occurs in mental disorders is unknown. Here, we report high L1 copy number in schizophrenia. Increased L1 was demonstrated in neurons from prefrontal cortex of patients and in induced pluripotent stem (iPS) cell-derived neurons containing 22q11 deletions. Whole-genome sequencing revealed brain-specific L1 insertion in patients localized preferentially to synapse- and schizophrenia-related genes. To study the mechanism of L1 transposition, we examined perinatal environmental risk factors for schizophrenia in animal models and observed an increased L1 copy number after immune activation by poly-I:C or epidermal growth factor. These findings suggest that hyperactive retrotransposition of L1 in neurons triggered by environmental and/or genetic risk factors may contribute to the susceptibility and pathophysiology of schizophrenia.
And our letter, reporting some of its own preliminary findings on increased transposable element content in schizophrenia and central nervous system-related genes:
Recent research has highlighted the potential role of transposable elements (TE) in the genetic instability of certain psychiatric and neurodevelopmental conditions, such as schizophrenia and autism. For instance, the recent study published by Bundo et al. (2014) reported an increase in copy number of LINE1 (L1) retrotransposable elements in postmortem tissues and induced pluripotent stem cells (iPSC) from schizophrenic patients. A similar study reported higher percentages of transcripts from subfamilies of long terminal repeat (LTR) elements in cases of autism as compared to controls (Balestrieri et al., 2012). In addition, our own recent work has highlighted the potential importance of transposable content, including that of extinct sequences, in the stability of autism candidate genes, many of which are shared risk genes in schizophrenia (Williams et al., 2013).
Interpretation of this type of data has generally focused on the roles which transpositionally active or “hot” TEs may play in genome stability. Bundo et al. (2014) state that their “findings suggest that hyperactive retrotransposition of L1 in neurons triggered by environmental and/or genetic risk factors may contribute to the susceptibility and pathophysiology of schizophrenia” (p. 1). However, the circumstances that increase transcription and insertion of active mobile elements likewise destabilize their extinct predecessors and other repetitive sequences within the genome, leading to a variety of mutational events, such as nonallelic homologous recombination of low copy repeats (LCR). Therefore, gene-associated forms of schizophrenia and autism are more likely the results of a variety recombinatory (inversion, duplication, deletion) and transposition events arising from gene destabilization such as occurs during transcription (Gottipati et al., 2008). And as has been shown in a number of studies, both hot and cold or extinct TEs provide an important reservoir for such rearrangements.
To illustrate this point, we have utilized a core set of schizophrenia candidate genes compiled by Allen et al. (2008), made available through the Schizophrenia Research Forum database, and have compared their TE content to non-overlapping controls (see Williams et al., 2013 for methods of control selection). We collected TE data on our genes of interest utilizing the database, TranspoGene (Levy et al., 2008). Using a Kuiper nonparametric analysis due to the nature of the gene groups’ distributions, we find that schizophrenia genes display significantly greater total TE content than control genes [V = 0.4363, p = 0.0005203, Bonferroni-corrected; schizophrenia: N = 38, μ = 220.67 ± 317.69; control: N = 429, μ = 44.25 ± 106.06]. The majority of these elements would be considered transpositionally extinct; however, it has been shown that many TEs maintain the capacity to form alternate conformations such as hairpin structures and other random coils, either within themselves or paired with homologous sequences within neighboring elements, occasionally leading to recombination events. Therefore, schizophrenia-related genes may be prone to a variety of mutational events, including retrotransposition by L1 elements.
Because the Bundo et al. (2014) study noted that L1 insertion was not only increased in schizophrenia but was over-represented specifically in neuronal genes within schizophrenic genomes, we have hypothesized that TE content is likewise greater in central nervous system-related genes, which may explain their propensity for recombination and transposition events. We have therefore used a list of central nervous system-related genes from the Neurogenesis and Neural Stem Cell (PAMM-404) PCR array from Qiagen, containing a total of 85 genes across functional areas of cell proliferation, cell cycle, cell motility and migration, differentiation, growth factors and cytokines, synaptogenesis and synapse function, apoptosis, cell adhesion, neurogenesis, and transcriptional regulation. This particular array was selected to form our gene list because it is a representative sample of important CNS gene expression markers commonly used in research today. We find that, as with the schizophrenia candidate sample, this panel of CNS genes exhibits significantly greater TE content than our controls, suggesting that a gene’s tendency towards recombination and transposition shares a positive relationship with its overall TE content [V = 0.2728, p = 0.003226, Bonferroni-corrected; CNS: N = 85, μ = 106.69 ± 202.00; control: N = 441, μ = 44.25 ± 106.06].
Bundo et al. (2014) performed whole-genome sequencing on iPSC-derived neurons from two female schizophrenic patients diagnosed with 22q11-deletion syndrome. While the authors report increased L1 copy numbers in the genomes of these two patients, researchers have found that most cases of 22q11 deletion syndrome likely stem from Alu-mediated recombination, not from L1 insertion (Babcock et al., 2003). Because hot TEs are dramatically outweighed in number by their cold counterparts, this suggests that active retrotransposition is not the major cause of gene-associated schizophrenia but may only be one of a number of related causes.
A given gene’s vulnerability towards mutation is highly dynamic and is not only subject to its primary sequence conformation but to its secondary and tertiary structures which are constantly changing with epigenetic interaction. When a gene is held in the open conformation to allow for transcription or replication it is made more vulnerable to mutation due to the torsional strain placed upon its strands (Gottipati et al., 2008). With this strain can arise alternate conformations such as hairpins, cruciforms, random coils, and other slippage events that can subsequently disrupt transcription or replication, promoting stalled forks, or cause duplications or deletions of repetitive elements. Alternative conformations such as the slippage events that can likewise lead to “alternative transposition” also allow repetitive elements such as TEs to form bonds with neighboring or even more distant homologous sequences, occasionally resulting in inversions, deletions, and duplications.
Perhaps because of variations in gene expression over the different developmental time periods of a cell and variable instability across different genes, rates of transposition are not equal across all cell stages. For instance, Muotri et al. (2005) reported that occurrences of L1 retrotransposition increased substantially in neural progenitors during early stages of differentiation. Of interest to schizophrenia research, the Bundo et al. (2014) study may indirectly suggest that differentiation is somehow disturbed in some cases of schizophrenia and that certain target genes are either overexpressed or their expression prolonged, increasing the opportunity for transposition into these genes.
The Bundo et al. (2014) study has highlighted the potential importance of genetic instability in schizophrenia’s etiology. However, while some gene mutations may arise due to retrotransposon insertion, this is probably only one of a number of destabilizing events that target schizophrenia-risk genes. Due in part to their high transposable element content, the repetitive content of many schizophrenia candidate genes places them at higher risk for transposition and recombination in general.
References
Allen, N. C., Bagade, S., McQueen, M. B., Ioannidis, J. P. A., Kavvoura, F. K., Khoury, M. J., et al. (2008). Systematic meta-analyses and field synopsis of genetic association studies in schizophrenia: The SzGene Database. Nature Genetics, 40(7), 827-834.
Babcock, M., Pavlicek, A., Spiteri, E., Kashork, C. D., Ioshikhes, I., Shaffer, L. G., et al. (2003). Shuffling of genes within low-copy repeats on 22q11 (LCR22) by Alu-mediated recombination events during evolution. Genome Res. 13, 2519-2532.
Balestrieri, E., Arpino, C., Matteucci, C., Sorrentino, R., Pica, F., Alessandrelli, R., et al. (2012). HERVs expression in autism spectrum disorders. PLoS One 7.11, e48831.
Bundo, M., Toyoshima, M., Okada, Y., Akamatsu, W., Ueda, J., Nemoto-Miyauchi, T., et al. (2014). Increased L1 retrotransposition in the neuronal genome in schizophrenia. Neuron, http://dx.doi.org/10.1016/j.neuron.2013.10.053.
Gottipati, P., Cassel, T. N., Savolainen, L., & Helleday, T. (2008). Transcription-associated recombination is dependent on replication in mammalian cells. Mol. Cell. Biol. 28, 154-164.
Levy, A., Sela, N., and Ast, G. (2008). TranspoGene and microTranspoGene: transposed elements’ influence on the transcriptome of seven vertebrates and invertebrates. Nucleic Acids Res. 36, D47-D52.
Muotri, A. R., Chu, V. T., Marchetto, M. C., Deng, W., Moran, J. V., and Gage, F. H. (2005). Somatic mosaicism in neuronal precursor cells mediated by L1 retrotransposition. Nature 435, 903-910.
Williams, E. L., Casanova, M. F., Switala, A. E., Hong, L., & Qiu, M. (2013). Transposable elements occur more frequently in autism-risk genes: Implications for the role of genomic instability in autism. Transl. Neurosci. 4, 172-202.
Note: Statistics performed by Andrew E. Switala.
So to summarize the point of the letter: schizophrenia and brain-related genes likely house higher-than-average repetitive DNA content, such as transposable elements, which leads to their relative instability and higher rates of recombination and mobile element insertion. While some schizophrenia-related genes may acquire mutations due to mobile element insertion as shown in the Neuron study, this is likely a minority given the rates of “successful” insertion of the few mobile elements that are still active in the human genome. More instances of non-point mutation are probably due to recombination events, which themselves may often be tightly linked with the extinct mobile element content within the gene or nearby sequences. The higher rates of copy number variation seen within schizophrenia, as well as the various recombination events linked with schizophrenia-related mutations, may in fact bear this out.
Science Over a Cuppa 
Hi! Have you thought of putting your work onto Figshare or Arxiv.org, rather than just your blog?
Hmm, that’s a good thought which I hadn’t had. I should look into that. Thanks for the suggestion. 🙂
I apologize if I missed this, but did you take into account the length of SZ-related introns in compared to the controls?
Hi, Sara. We did at first normalize by overall gene length, the bulk of which is of course intronic. We’re currently working on a more thorough and larger project with this data as well as our earlier autism TE data, expanding to look at differences in gene length, amino acid length, as well as variations across TE types. According to our prelims so far, sz and autism genes are very similar in their characteristics in terms of increased TE content, larger gene size, and even longer amino acid length. In answer to your question though, when the TE content is normalized by gene size, there’s still about a 20% increase in the number of TEs as compared to randomized controls. However, for the above “attempted” article, as well as our earlier autism TE article, we opted to just focus on absolute numbers per gene, since genes are not just strands of nucleotides but have very characteristic landmarks and features, such that the nature of an exon is quite different from an intron or intergenic sequences. So the idea that, even if the normalization by size had lost us our lovely p-value, it would have still been of considerable note that the TE content was grouping within target regions, e.g., introns, and that these have significance for the stability of those regions.
We’ll definitely be publishing a more thorough study on schizophrenia and autism in future, so check back in a few months. Thanks for asking. 🙂
Thanks for the great response! I completely agree that whether or not the length of the gene reduces the significance, it’s still interesting that TEs are enriched in these genes. For full disclosure I’m working on a similar project (which is how I happened onto your blog), and we see a similar effect. Can’t wait to see your paper when it comes up.
Hey, that’s great! The more attention paid to these underrated genetic elements, the better. Likewise, I’d love to see some of your work in future. Please drop me a line or a link. Always love to know what’s happening in different parts of the field.
I forgot to add… this is a great report, thanks for sharing it!