A bidirectional competitive interaction between circHomer1 and Homer1b within the orbitofrontal cortex regulates reversal learning

Alexander K Hafez; Amber J Zimmerman; Grigorios Papageorgiou; Jayapriya Chandrasekaran; Stephen K Amoah; Rixing Lin; Evelyn Lozano; Caroline Pierotti; Michela Dell’Orco; Brigham J Hartley; Begüm Alural; Jasmin Lalonde; John Matthew Esposito; Sabina Berretta; Alessio Squassina; Caterina Chillotti; Georgios Voloudakis; Zhiping Shao; John F Fullard; Kristen J Brennand; Gustavo Turecki; Panos Roussos; Roy H Perlis; Stephen J Haggarty; Nora Perrone-Bizzozero; Jonathan L Brigman; Nikolaos Mellios

doi:10.1016/j.celrep.2021.110282

. Author manuscript; available in PMC: 2022 Feb 2.

Published in final edited form as: Cell Rep. 2022 Jan 18;38(3):110282. doi: 10.1016/j.celrep.2021.110282

A bidirectional competitive interaction between circHomer1 and Homer1b within the orbitofrontal cortex regulates reversal learning

Alexander K Hafez ^1,^2,²¹, Amber J Zimmerman ^1,²¹, Grigorios Papageorgiou ^1,²¹, Jayapriya Chandrasekaran ¹, Stephen K Amoah ^1,^2,¹⁸, Rixing Lin ³, Evelyn Lozano ¹, Caroline Pierotti ¹, Michela Dell’Orco ¹, Brigham J Hartley ^4,¹⁹, Begüm Alural ⁵, Jasmin Lalonde ^5,²⁰, John Matthew Esposito ⁶, Sabina Berretta ^6,^7,⁸, Alessio Squassina ⁹, Caterina Chillotti ¹⁰, Georgios Voloudakis ^11,^12,^13,¹⁴, Zhiping Shao ^11,^12,¹³, John F Fullard ^11,^12,¹³, Kristen J Brennand ^4,¹⁹, Gustavo Turecki ³, Panos Roussos ^11,^12,^13,^14,¹⁵, Roy H Perlis ^16,¹⁷, Stephen J Haggarty ⁵, Nora Perrone-Bizzozero ¹, Jonathan L Brigman ¹, Nikolaos Mellios ^1,^2,^22,^*

¹Department of Neurosciences, University of New Mexico School of Medicine, Albuquerque, NM, USA

²Autophagy Inflammation and Metabolism Center of Biomedical Research Excellence, University of New Mexico Health Sciences Center, Albuquerque, NM, USA

³McGill Group for Suicide Studies, Douglas Mental Health University Institute, Department of Psychiatry, McGill University, Montreal, QC, Canada

⁴Pamela Sklar Division of Psychiatric Genomics, Friedman Brain Institute, Departments of Genetics and Genomic Sciences, Neuroscience, and Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY, USA

⁵Center for Genomic Medicine, Chemical Neurobiology Laboratory, Departments of Neurology and Psychiatry, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA

⁶Translational Neuroscience Laboratory, Mclean Hospital, Belmont, MA, USA

⁷Department of Psychiatry, Harvard Medical School, Boston, MA, USA

⁸Program in Neuroscience, Harvard Medical School, Boston, MA, USA

⁹Department of Biomedical Sciences, Section of Neuroscience and Clinical Pharmacology, University of Cagliari, Cagliari, Italy

¹⁰Unit of Clinical Pharmacology, University Hospital Agency of Cagliari, Cagliari, Italy

¹¹Pamela Sklar Division of Psychiatric Genomics, New York, NY, USA

¹²Department of Genetics and Genomic Sciences, New York, NY, USA

¹³Icahn Institute for Data Science and Genomic Technology, New York, NY, USA

¹⁴Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY, USA

¹⁵Mental Illness Research, Education and Clinical Centers, James J. Peters VA Medical Center, Bronx, NY, USA

¹⁶Harvard Medical School, Department of Psychiatry, Boston, MA, USA

¹⁷Center for Experimental Drugs and Diagnostics, Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, USA

¹⁸Present address: University of Cape Coast, Cape Coast, Ghana

¹⁹Present address: Departments of Psychiatry and Genetics, Division of Molecular Psychiatry, Yale University School of Medicine, New Haven, CT, USA

²⁰Present address: Department of Molecular and Cellular Biology, University of Guelph, Guelph, ON, Canada

²¹These authors contributed equally

²²Lead contact

AUTHOR CONTRIBUTIONS

N.M. conceived the hypothesis; designed, performed, and supervised experiments; analyzed data; and wrote the manuscript. J.L.B. and N.P.-B. designed and supervised experiments, analyzed data, and provided feedback on the hypothesis and manuscript. A.K.H., A.J.Z., and G.P. designed and performed experiments, analyzed data, provided feedback on the hypothesis, and helped in manuscript preparation. All other authors performed experiments and analyzed data. All authors reviewed the paper.

Correspondence: nmellios@salud.unm.edu

PMCID: PMC8809079 NIHMSID: NIHMS1772785 PMID: 35045295

SUMMARY

Although circular RNAs (circRNAs) are enriched in the brain, their relevance for brain function and psychiatric disorders is poorly understood. Here, we show that circHomer1 is inversely associated with relative HOMER1B mRNA isoform levels in both the orbitofrontal cortex (OFC) and stem-cell-derived neuronal cultures of subjects with psychiatric disorders. We further demonstrate that in vivo circHomer1 knockdown (KD) within the OFC can inhibit the synaptic expression of Homer1b mRNA. Furthermore, we show that circHomer1 directly binds to Homer1b mRNA and that Homer1b-specific KD increases synaptic circHomer1 levels and improves OFC-mediated behavioral flexibility. Importantly, double circHomer1 and Homer1b in vivo co-KD results in a complete rescue in circHomer1-associated alterations in both chance reversal learning and synaptic gene expression. Lastly, we uncover an RNA-binding protein that can directly bind to circHomer1 and promote its biogenesis. Taken together, our data provide mechanistic insights into the importance of circRNAs in brain function and disease.

In brief

Through in vivo circRNA and mRNA isoform-specific knockdown in mouse orbitofrontal cortex (OFC), Hafez et al. elucidate the antagonistic interaction between the psychiatric-disease-associated circHomer1 and Homer1b mRNA and their opposing effects on OFC-mediated reversal learning.

Graphical abstract

graphic file with name nihms-1772785-f0007.jpg

INTRODUCTION

Emerging data suggest the presence of significant alterations in cognitive flexibility, which is the ability to adjust one’s thoughts and behavior following changing circumstances, in patients with psychiatric disorders, such as bipolar disorder (BD) and schizophrenia (SCZ) (Leeson et al., 2009; O’Donnell et al., 2017; Waltz and Gold, 2007; Wegbreit et al., 2016). The orbitofrontal cortex (OFC) is among the brain regions strongly involved in behavioral flexibility, and its dysfunction can selectively impair reversal learning (Hamilton and Brigman, 2015; Izquierdo et al., 2017; Mainen and Kepecs, 2009). Multiple studies have suggested the presence of disturbances in glutamatergic neurotransmission and alterations in synaptic gene expression in the prefrontal cortex (PFC) of subjects with BD and SCZ (Crabtree and Gogos, 2014; Hashimoto et al., 2007; Schloesser et al., 2008). Furthermore, OFC lesions and either pharmacological inhibition or knockdown (KD) of N-methyl-D-aspartate (NMDA) receptors within the OFC have been shown to specifically disrupt reversal learning (Brigman et al., 2013; Rudebeck and Murray, 2008). However, very little is known about the molecular mechanisms that could underlie OFC-mediated behavioral flexibility and their relevance to psychiatric disorders.

Circular RNAs (circRNAs) are a novel subtype of brain-enriched non-coding RNAs (ncRNAs), which are abundantly expressed in the PFC and display strong developmental and activity-dependent expression within the mammalian brain (Gruner et al., 2016; Rybak-Wolf et al., 2014; You et al., 2015; Zimmerman et al., 2020). Although, circRNAs are predominantly synthesized following backsplicing and covalent joining of exons and/or introns derived from protein-coding genes, they are with very few exceptions incapable of being translated into proteins (Chen and Yang, 2015; Guo et al., 2014; Jeck et al., 2013; Liang and Wilusz, 2014; Salzman et al., 2012; Zhang et al., 2014). Instead, circRNAs appear to exert robust and diverse regulatory effects in protein-coding gene expression by either interacting with RNA-binding proteins (Abdelmohsen et al., 2017; Conn et al., 2015; Dell’Orco et al., 2020), sequestering microRNAs (miRNAs) (Hansen et al., 2013; Memczak et al., 2013), or influencing important transcriptional and post-transcriptional processes within a cell (Abdelmohsen et al., 2017; Li et al., 2015). Recent studies suggest that circRNAs could be regulated independently of their linear mRNA counterparts or compete with linear splicing of precursor mRNAs (pre-mRNAs) (Ashwal-Fluss et al., 2014; Jeck et al., 2013; Rybak-Wolf et al., 2014; Salzman et al., 2012). Moreover, manipulating the expression of specific circRNAs in the brain can result in significant changes in neuronal gene expression and impairments in neuronal activity and behavior (Piwecka et al., 2017; Zimmerman et al., 2020). Furthermore, recent studies have suggested that different subsets of circRNAs could be differentially expressed in the frontal cortex of subjects with psychiatric disorders, such as BD and SCZ (Liu et al., 2019; Luykx et al., 2019; Mahmoudi et al., 2019; Zimmerman et al., 2020), while alterations in circRNA expression have also been observed in the blood of patients with SCZ and unipolar depression (Yao et al., 2019; Zhang et al., 2020). However, little is known about the role of psychiatric-disorder-associated circRNAs in the control of cognitive flexibility, as well as their interplay with protein-coding gene expression.

We recently showed that circHomer1 (also known as circHomer1a) is a neuronal-enriched circRNA derived from the backsplicing of exons 2–5 of HOMER protein homolog 1 (HOMER1) (Zimmerman et al., 2020), a gene important for neuronal function and synaptic plasticity that is also implicated in psychiatric disorders (Bottai et al., 2002; Cao et al., 2015; Chen et al., 2012; Gimse et al., 2018; Hu et al., 2010; Klugmann et al., 2005; Lominac et al., 2005; Szumlinski et al., 2005, 2006). Importantly, both HOMER1 and HOMER2 (a homologue of HOMER1) are generated from genome-wide-significant (genome-wide association study [GWAS]) loci for BD (Mullins et al., 2021). Our initial study suggested that circHomer1 is significantly reduced in the PFC and induced pluripotent stem cell (iPSC)-derived neuronal cultures from subjects with BD and SCZ and that modest reductions in circHomer1 in mouse OFC are sufficient to disrupt specific stages of OFC-mediated reversal learning (Zimmerman et al., 2020).

Here, we show that deficits in circHomer1 levels within the OFC and iPSC-derived neuronal cultures of patients with psychiatric disorders are inversely associated with the relative expression of the long HOMER1B mRNA isoform. Using in vivo circHomer1- and Homer1b-specific KD in mouse OFC, we demonstrate that mature circHomer1 and Homer1b inhibit each other’s synaptic expression and that circHomer1 could bind to the 3′ UTR of Homer1b mRNA. We further provide evidence that Homer1b KD within the OFC results in improved chance reversal learning and that double Homer1b and circHomer1 co-KD in the OFC can rescue circHomer1-associated deficits in chance reversal learning and restore the alterations in synaptic gene expression. Lastly, we identify that the RNA-binding protein (RBP) eukaryotic initiation factor 4A-III (EIF4A3) displays a positive association with circHomer1 expression in human and mouse brain, can directly bind to circHomer1, and is able to positively regulate circHomer1 synthesis. Taken together, our data reveal an antagonistic interplay between a circRNA and its linear mRNA counterpart that is important for regulating synaptic gene expression and cognitive flexibility.

RESULTS

Changes in relative HOMER1 mRNA isoform levels are differentially associated with deficits in circHomer1 expression

In order to examine whether alterations in circHomer1 are associated with the expression of specific HOMER1 mRNA isoforms, we used isoform-specific quantitative real-time PCR in postmortem samples from the OFC of subjects with BD (n = 30), SCZ (n = 32), and unaffected controls (n = 33) derived from the Stanley Medical Research Institute (SMRI) (Tables S1 and S2; Torrey et al., 2000). We quantified the expression of HOMER1A, a short activity-dependent, dominant-negative isoform produced from exons 1–5 and parts of intron 5; HOMER1B, the longest detected in human brain mRNA isoform composed of all nine exons; and HOMER1H, an intermediate in size and moderately expressed transcript produced by exons 1–5 and exon 9 (Figures 1A, 1B, and S1A–S1C). Statistical analysis via a univariate general linear model with post hoc Bonferroni correction for multiple comparisons, as well as correction for multiple potential postmortem confounding factors, such as postmortem interval, RNA integrity number, brain pH, and age, showed a significant reduction in both BD and SCZ in HOMER1H, with no changes observed for HOMER1B and modest reductions in HOMER1A in SCZ that did not reach statistical significance (Figures 1A, 1B, and S1A–S1C). Using this more stringent statistical analysis in our previously published data on circHomer1 (Zimmerman et al., 2020) within the same RNA samples from the OFC, we found a significant reduction in BD (p < 0.01) with reductions in SCZ not reaching statistical significance (Figure 1B). In addition, we also validated a robust reduction in circHomer1 levels in the OFC of patients with BD, but not SCZ, from another postmortem brain cohort (n = 20–21 per group; Tables S3 and S4) from the Harvard Brain Tissue Resource Center (HBTRC) (Figure 1C). It is known that both circHomer1 and HOMER1B are derived from the long HOMER1B precursor mRNA (Dudekula et al., 2016; Glažar et al., 2014; Figure 1A). Moreover, neuronal activity has been shown to induce the expression of both Homer1a and circHomer1 in mouse neurons (Bottai et al., 2002; You et al., 2015). We therefore hypothesized that circHomer1 could be differentially associated with the relative expression of these two main HOMER1 mRNA isoforms. In order to correct for overall effects in total HOMER1 transcription, we calculated the ratios of HOMER1A, B, or H versus total HOMER1 mRNA and correlated them to the relative changes in circHomer1 in the larger SMRI OFC cohort (Figures 1D–1F). Our results showed a modest but significant positive correlation between circHomer1 and relative HOMER1A expression within the OFC but a robust negative correlation between circHomer1 and relative HOMER1B mRNA isoform levels (Figures 1D and 1E). However, no association was observed between circHomer1 and changes in relative HOMER1H expression (Figure 1F).

Figure 1. — (A) Schematic representation of isoform-specific linear splicing from the human *HOMER1* gene versus *circHomer1* backsplicing.

(B) Statistical analysis using a univariate general linear model correcting for *postmortem* interval, RNA integrity number, brain pH, and age and with post hoc Bonferroni correction for *circHomer1, HOMER1A, HOMER1B*, and *HOMER1H*. ns, not significant. # 0.10 < p < 0.05, *p < 0.05, **p < 0.01.

(C) Relative to control *circHomer1* levels (normalized to the geometric mean of *circTulp4* and *CDR1as*) from the OFC of BD, SCZ, and controls from the HBTRC cohort.

(D–F) Correlations between relative to control changes in *circHomer1* in the OFC of patients with SCZ and BD (SMRI cohort) and the relative to control ratios of *HOMER1A* (D), *HOMER1B* (E), and *HOMER1H* (F) mRNAs to *HOMER1all* levels. Spearman correlation coefficient and two-tailed p values are shown in the graph.

(G and H) Relative to control circHomer1 (G) and relative to control *HOMER1B* versus *HOMER1all* levels (H) from the OFC of BD patients from the SMRI cohort with and without psychosis at the time of death.

(I) Relative to control circHomer1 levels (normalized circTulp4) from the DLPFC of BD, SCZ, and unaffected controls from the Mt. Sinai postmortem cohort.For (C)–(I): *p < 0.05, **p < 0.01, and ***p < 0.001 based on a univariate general linear model corrected for RNA integrity number (RIN), *postmortem* interval (PMI), brain pH, and age.

(J) Relative to control *circHomer1* levels (quantitative real-time PCR, normalized to 18S rRNA) in B lymphoblastoids derived from patients with BD and healthy controls. **p < 0.05, based on two-tailed one-sample t test.

(K and L) Relative to control *circHomer1* levels (normalized to 18S *rRNA*) from B lymphoblastoid cultures from healthy controls (K) and patients with BD (L) with and without treatment with lithium for 1 week. #p < 0.05, based on two-tailed paired t test.

All data are shown as mean ± SEM based on quantitative real-time PCR, and the individual biological replicates are shown within each graph.

For the patients with BD of the SMRI cohort, we had demographic information related to the presence of psychosis at the time of death. Interestingly, circHomer1 levels within the OFC were found to be significantly reduced relative to unaffected Controls only in patients with BD with presence of psychosis (Figure 1G). However, no changes in circHomer1 levels were seen in the OFC of the subset of BD patients with absence of psychosis at the time of death (Figure 1G). Given the inverse correlation between circHomer1 expression and relative HOMER1B mRNA levels in the OFC (see Figure 1E), we plotted the changes in relative HOMER1B mRNA levels (HOMER1B versus total HOMER1 expression) in the OFC of BD patients with and without psychosis at the time of death (Figure 1G). Our results showed a significant upregulation of relative HOMER1B expression in the OFC of patients with BD with psychosis at the time of death (Figure 1H).

We had previously shown that circHomer1 is significantly downregulated in the dorsolateral PFC (DLPFC), but not OFC, of subjects with SCZ (Zimmerman et al., 2020). To further investigate this, we measured circHomer1 expression in a larger DLPFC postmortem cohort from the Icahn School of Medicine at Mount Sinai (n = 54 for SCZ, n = 40 for BD, and n = 60 for controls; see Tables S5 and S6). Our data suggested a robust reduction in circHomer1 levels in the DLPFC of subjects with SCZ, but not BD (Figure 1I). We then cultured B lymphoblastoids (BLCLs) from the blood of 19 patients with BD and 12 unaffected controls (Squassina et al., 2013), split these BLCLs into two equal aliquots, and treated one of each aliquot with 1 mM of lithium chloride (LiCl) for 7 days. We found a significant reduction in circHomer1 expression in BLCLs of subjects with BD (Figure 1J). Moreover, lithium treatment resulted in no changes in circHomer1 expression in controls but caused a modest increase in circHomer1 expression in patients with BD (Figures 1K and 1L).

We had previously shown that circHomer1, but not total HOMER1, mRNA levels are significantly reduced in iPSC-derived neuronal cultures from subjects with BD and early onset SCZ (Zimmerman et al., 2020). Using the same approach described above, we quantified the expression of HOMER1A, HOMER1B, and HOMER1H mRNAs in iPSC-derived neuronal progenitors (NPs) and neuronal cultures differentiated for 2–6 weeks for BD and 6 weeks for early-onset SCZ (Figure 2A; Bavamian et al., 2015; Hoffman et al., 2017; Zimmerman et al., 2020). We found significant increases in HOMER1B in 6-week iPSC-derived neurons and modest increases in both HOMER1A and HOMER1H in iPSC-derived NPs, but not neurons from subjects with SCZ (Figures 2B–2D). No notable changes in relative HOMER1 mRNA isoform expression were observed in iPSC-derived NPs or neurons from subjects with BD (Figures 2E–2G). Comparison between circHomer1 and relative HOMER1B mRNA levels revealed a robust negative correlation (Figures 2H and 2I). Lastly, a modest negative correlation was observed between circHomer1 and relative HOMER1H, but not HOMER1A, mRNA levels (Figures 2H–2J). We conclude that, in both postmortem brain and stem-cell-derived neuronal samples, circHomer1 is robustly inversely associated with relative HOMER1B mRNA isoform expression, suggestive of a potential competitive interaction between circHomer1 and HOMER1B synthesis.

Figure 2. — (A) Schematic representation of the process for generating iPSC-derived neuronal progenitors (NPs) and neuronal cultures.

(B–D) Relative to the mean of control NPs *HOMER1A* (B), *HOMER1B* (C), and *HOMER1H* (D) mRNA levels in iPSC-derived, early-onset SCZ patients and control (n = 10 control; n = 9 SCZ subjects) NPs and 6-week differentiated neurons.

(E–G) Relative to the mean of control NPs *HOMER1A* (E), *HOMER1B* (F), and *HOMER1H* (G) mRNA levels in iPSC-derived BD patient and control (n = 3 control and n = 4 BD subjects) NPs and 2-, 4-, and 6-week differentiated neurons. (B–G) *p < 0.05, two-tailed one sample t test relative to the control of the same developmental time point.

(H–J) Correlation between relative to control changes in *circHomer1* levels in iPSC-derived NPs and neuronal cultures of patients with SCZ and BD and relative to control changes in the ratio of *HOMER1A* (H), *HOMER1B* (I), and *HOMER1H* (J) mRNAs to *HOMER1all* levels. Spearman correlation coefficient and two-tailed p values are shown in the graph.

All data are shown as mean ± SEM based on quantitative real-time PCR, and the individual biological replicates are shown within each graph.

circHomer1 and Homer1b mRNA inhibit each other’s synaptic expression within the OFC

We have validated that a short hairpin RNA (shRNA) designed asymmetrically against the unique mouse circHomer1 splice junction (Figure 3A) can achieve a significant and specific KD in mature circHomer1 levels in mouse OFC without resulting in significant off-target effects (Zimmerman et al., 2020). Such an approach did not result in any changes in overall Homer1 mRNA isoform levels, since shRNA KD reduces mature circHomer1 levels but does not affect its biogenesis. We then designed a shRNA against the unique mouse Homer1b mRNA linear splice junction between exons 5 and 7 using a similar asymmetric design to avoid disrupting the Homer1b precursor mRNA (Figure 3A). In vivo expression of this shRNA via lentiviral injection in the OFC resulted in a significant reduction in Homer1b mRNA levels versus a scrambled shRNA negative control (Figures 3B and 3C). Importantly, no changes were observed in the expression of circHomer1, as well as Homer1a and the long Homer1c mouse mRNA isoform (Figure 3C), thus validating the specificity of our shRNA design. We then injected equal amounts of circHomer1 (sh-circHomer1) and Homer1b (sh-Homer1b) shRNA-expressing lentiviruses within the OFC, as well as an equal volume of the same scrambled shRNA negative control (sh-Control). Using isoform-specific PCR primers, we found a significant double KD of both circHomer1 and Homer1b in the OFC of mice expressing both circHomer1 and Homer1b shRNAs (sh-Double-KD) versus sh-Control (Figure 3D). Notably, no changes were observed in Homer1a or Homer1c mRNA expression in sh-Double-KD OFC, validating once more the specificity of our KD approaches (Figure 3D).

We then extracted synaptosomes from the OFC of sh-circHomer1-, sh-Homer1b-, sh-Double-KD-, and sh-Control-injected mice and measured the expression of circHomer1, Homer1a, Homer1b, and Homer1c mRNAs. Our results showed that circHomer1 KD in the OFC resulted in significantly increased synaptic Homer1a and Homer1b mRNA expression with no changes observed in synaptic Homer1c mRNA levels (Figure 3E). Single Homer1b KD resulted in a significant upregulation in synaptic circHomer1 levels in the OFC (Figure 3F), along with a reduction in synaptic Homer1b mRNA levels, which was expected given the overall KD of this mRNA. Of note, no changes were seen in synaptic Homer1a and Homer1c mRNA levels in the OFC of sh-Homer1b mice (Figure 3F), suggesting that Homer1b could specifically inhibit circHomer1 synaptic expression. Looking at synaptosomes from the OFC of sh-Double-KD- and sh-Control-injected mice, we found that double circHomer1 and Homer1b KD in the OFC resulted in normal synaptic Homer1b expression (Figure 3G), which is to be expected given that we are reducing overall Homer1b expression via shRNA-mediated degradation, thus balancing any potential circHomer1 KD-mediated increases in synaptic Homer1b localization. Furthermore, given that Homer1b KD can also increase synaptic circHomer1 levels and thus compensate for the overall KD in circHomer1, we also found no significant changes in synaptic circHomer1 expression following double circHomer1 and Homer1b KD (Figure 3G). Of note, only a modest close to 35% increase in synaptic Homer1a expression was observed in the OFC of sh-Double-KD mice (versus the close to 3.7-fold increase detected in single circHomer1 KD OFC), suggesting a partial rescue in Homer1a synaptic levels following co-KD of Homer1b (Figure 3G). No changes in synaptic Homer1c mRNA levels were observed in the OFC of sh-Double-KD mice, similar to what was seen in the OFC of single circHomer1 KD mice (Figure 3G). We conclude that circHomer1 and Homer1b inhibit each other’s expression within synapses in the OFC and that co-KD of both circHomer1 and Homer1b can normalize their synaptic levels.

CircHomer1 directly binds to the 3′ UTR of Homer1b mRNA

We previously showed that, in addition to circHomer1, Homer1a and Homer1b (but not Homer1c) mRNAs are evolutionary-conserved isoforms that can directly bind to HuD (Zimmerman et al., 2020). We therefore hypothesized that circHomer1 could compete for binding to HuD with Homer1a and Homer1b mRNAs, which could account for increased synaptic expression of these mRNAs following reductions in circHomer1 expression. However, HuD is known to affect the stability and localization of numerous neuronal mRNAs (Deschênes-Furry et al., 2006; Tiruchinapalli et al., 2008). That means that the ability of circHomer1 to bind and potentially sequester HuD could only be specific to Homer1 mRNAs if additional interactions could allow for a direct proximity between circHomer1 and Homer1a or Homer1b mRNAs. Using a long non-coding RNA (lncRNA) and mRNA binding prediction software (IntaRNA) (Mann et al., 2017), we uncovered a very strong predicted binding site between circHomer1 and the 3′ UTR of Homer1b that allowed for 28 out of 30 nt within Homer1b 3′ UTR to directly interact with circHomer1 (Figures 4A and 4B) and a weaker predicted potential for binding to circHomer1 for the 3′ UTR of Homer1a (14 out 15 nt being complementary; Figure S2). No direct binding interactions (minimum of 10 consecutive base pairs was chosen) were predicted between Homer1c and circHomer1 (data not shown; Homer1c differs from Homer1b by having an additional exon 6 and a shorter 3′ UTR). Interestingly, the predicted interaction between circHomer1 and the 3′ UTR of Homer1b was in direct proximity to the strongest predicted HuD-binding site within the Homer1b 3′ UTR and very close to a predicted HuD-binding site close to the circHomer1 splice junction (Figures 4B and 4C). Importantly, the predicted binding interaction between the 3′ UTR of Homer1b mRNA and circHomer1 was made accessible due to the circHomer1 secondary structure created via putative interactions between exon 2 and the unique circHomer1 splice junction (Figure 4A).

Figure 4. — (A) Schematic representation of the predicted *circHomer1* secondary structure (RNAfold) and *Homer1b* mRNA 3′ UTR and *circHomer1* interaction site (IntaRNA; highlighted in blue). The *circHomer1* splice junction and adjacent 30 nt are highlighted in red.

(B) Sequence of the mouse *Homer1b* 3′ UTR with the predicted *circHomer1* binding site highlighted in blue and the predicted HuD-binding site highlighted in purple.

(C) Schematic showing the predicted interaction between *circHomer1* and the *Homer1b* 3′ UTR and the proximity to the predicted HuD-binding sites, highlighted in fuchsia.

(D) Schematic showing the process of *circHomer1* antisense purification.

(E) *CircHomer1* splice-junction probe sequence used for *circHomer1* antisense purification.

(F) Mean ± SEM relative to non-specific control probe mouse *circHomer1, circTulp4, Homer1b, Homer1a*, and *Homer1c* RNA levels after *circHomer1* antisense purification in Neuro-2a cell line. *p < 0.05, two-tailed, one-sample t test relative to non-specific control probe mean.

(G) Schematic of RNA antisense purification probe design for *Homer1b* 3′ UTR.

(H) Mean ± SEM relative to non-specific control probe mouse *circHomer1* and *Homer1b* mRNA levels after RNA antisense purification using biotinylated probes against the 3′ UTR of *Homer1b* mRNA in Neuro-2a cell line. **p < 0.01, two-tailed, one-sample t test relative to non-specific control probe mean. Individual biological replicates are shown within each graph.

We then used a modified RNA antisense modification technique (Engreitz et al., 2015; Torres et al., 2018) in mouse Neuro2A cultures that utilizes biotinylated probes against the unique circRNA splice junction (Figure 4D) and designed a splice-junction probe specific for mouse circHomer1 (Figure 4E). We found a significant close to 20-fold enrichment in circHomer1 levels in the RNA derived from circRNA antisense purification using the circHomer1-specific probe in comparison to the non-specific negative control probe (Figure 4F). Using Homer1 mRNA isoform-specific primers, we found that there was a significant enrichment for Homer1b, but not Homer1c, mRNA (Figure 4F). However, we also found a smaller but significant enrichment for Homer1a mRNA, suggesting that it could also bind to circHomer1 (Figure 4F). To further validate this direct interaction between circHomer1 and the 3′ UTR of Homer1b, we performed RNA antisense purification in mouse Neuro2A cultures using biotinylated probes against the 3′ UTR of Homer1b (Figure 4G). As expected, this experiment was able to significantly pull down Homer1b mRNA relative to a non-specific probe (Figure 4H). Importantly, a significant enrichment was found for circHomer1 (Figure 4H). Moreover, this effect was specific, since no enrichment was found for the brain-enriched circTulp4 (Figure S2B). We then utilized digital droplet PCR (ddPCR) to determine the absolute expression of circHomer1 and Homer1a and Homer1b mRNAs in adult mouse brain. Our results revealed that circHomer1 and Homer1a had an almost identical absolute expression in mouse OFC, while Homer1b showed the highest expression (Figure S2C). As a control, we found that Homer1d (a muscle-enriched Homer1 isoform) displayed the lowest expression in mouse OFC (Figure S2C). Taken together, our data suggest that circHomer1 can directly bind to the 3′ UTR of Homer1b mRNA with the putative binding sites being adjacent to predicted HuD-binding elements.

In vivo KD of Homer1b improves OFC-mediated chance reversal learning

Despite the well-documented importance of Homer1b in neuronal function and behavior (Cao et al., 2015; Chen et al., 2012; Gimse et al., 2018; Huang et al., 2012), no studies have examined its influence on OFC-mediated cognitive flexibility. We have previously shown that circHomer1 KD can significantly impair chance reversal learning in a touch-screen reversal learning paradigm, without influencing discrimination learning, as well as the early (perseverative) and late (criterion) stages of reversal learning (Zimmerman et al., 2020). To examine whether KD of Homer1b in the OFC could also affect reversal learning, we utilized the same touch-screen-based behavioral paradigm (Figure 5A) in mice injected with either sh-Homer1b- or sh-Control-expressing lentiviruses within their OFC. As described above, this resulted in specific KD in Homer1b within the OFC but also an increase in synaptic circHomer1 levels (see Figures 3C and 3F). Notably, we found that sh-Homer1b mice had significantly improved performance in OFC-dependent chance reversal learning (Figure 5B), suggesting that Homer1b KD might have opposing effects on OFC function relative to what is observed following circHomer1 KD. Moreover, no changes were found during early and late reversal learning (Figures S3D and S3E), suggesting that Homer1b KD could specifically promote chance reversal learning. No changes were also seen in overall motivation and motor function, but a modest improvement in discrimination learning was detected (Figures S3A–S3C). We had previously shown that the increase in the number of trials during chance reversal in mice with in vivo circHomer1 KD was observed in three out of four types of choice combinations (lose-shift, win-stay, and regressive; with smaller changes in perseverative) (Zimmerman et al., 2020). Interestingly, a similar effect was seen for the improvements in chance reversal learning in sh-Homer1b mice, with notable reduction in number of trials needed in all but perseverative trials (Figure 5B). These data suggest that Homer1b can inhibit OFC-mediated chance reversal learning.

Figure 5. — (A) Stages of the touch-screen reversal learning paradigm and trial outcomes associated with performance.

(B and C) Number of trials during chance reversal following *in vivo* KD of *Homer1b* (*sh-Homer1b*) (B) and *circHomer1*/*Homer1b* (*sh-Double-KD*) (C) within the OFC. Trial numbers are separated as lose-shift, win-stay, perseverative, and regressive trials. *p < 0.05, two-way ANOVA.

(D) Relative to *sh-Control* changes in baseline *circHomer1* levels within the OFC are inversely correlated to the number of lose-shift trials during chance reversal. Spearman correlation coefficient and two-tailed p values are shown in the graph.

(E and F) Relative to D50 *circHomer1* (E) and *Homer1b* mRNA (F) levels in non-injected adult mouse OFC during different stages of the touch-screen reversal learning paradigm. *p < 0.05, two-tailed, one-sample t test relative to D50 levels.

(G) Mean relative to *sh-Control* mRNA isoform fold changes from mouse OFC RNA-seq analysis of *sh-circHomer1* versus *sh-Control* (orange) and *sh-Double-KD* versus *sh-Control* (purple). The 17 mRNA isoforms shown are the most significantly altered following single *circHomer1* KD (more than 1.5-fold change and q < 0 × 10¹).

(H) Volcano plot showing differential mRNA isoform expression in the OFC of *sh-Double-KD* versus *sh-Control* mice (x axis: relative to control log2 fold changes; y axis: negative log10 of the q values). Vertical lines correspond to >2-fold changes, and the horizontal lines represent q < 0.10.

All data are shown as mean ± SEM, and individual biological replicates are shown within each graph.

In vivo double KD of Homer1b and circHomer1 rescues OFC-mediated chance reversal learning

Given the above findings, we hypothesized that the observed deficits in chance reversal learning as a result of circHomer1 KD could be due to the presence of increased Homer1b synaptic levels. To address this hypothesis, we injected the OFC of adult mice with lentiviruses expressing both Homer1b shRNA and circHomer1 shRNA (sh-Double-KD; see also Figure 3A) and repeated the same touch-screen reversal learning paradigm. Intriguingly, our results showed that knocking down Homer1b in addition to circHomer1 was sufficient to restore normal chance reversal learning (Figure 5C). As expected, sh-Double-KD mice displayed no changes in early and late reversal learning and had normal overall motivation and motor function (Figures S3F and S3G). We did observe though a modest trend for improvement in discrimination learning similar to what was observed in sh-Homer1b mice. However, this was mainly driven by changes during win-stay trials and did not reach significance (Figure S3H). Since all the measurements in circHomer1 levels shown above (see Figure 3) were performed weeks after the termination of all behavioral experiments, we compared animal-to-animal differences in overall circHomer1 expression within the OFC to each animal’s performance in chance reversal learning. We included molecular and behavioral data from sh-circHomer1 (Zimmerman et al., 2020), sh-Homer1b, and sh-Double-KD mice. Interestingly, we noticed that baseline circHomer1 levels within the OFC were significantly inversely correlated to each animal’s individual performance during chance reversal learning (Figures 5D and S4A–S4C), suggesting that the extent of reduction in circHomer1 levels in the OFC could determine the magnitude of the disturbances in chance reversal. Of note, the observed association between baseline circHomer1 expression and chance reversal learning performance was also found in the three out of four types of choice combinations that showed alterations following single circHomer1 KD (lose-shift, win-stay, and regressive, but not perseverative; Figures 5D and S4A–S4C; Zimmerman et al., 2020). We conclude that co-KD of Homer1b is sufficient to rescue the alterations in chance reversal learning observed following single circHomer1 KD.

Experience-dependent and opposing changes in circHomer1 and Homer1b mRNA levels in the OFC during chance reversal learning

In order to determine whether circHomer1 could display experience-dependent changes within the OFC, we had adult, wildtype (WT), non-injected mice undergo different stages of the reversal learning behavioral paradigm. These mice were euthanized immediately after mid- and late discrimination learning (D50 and D85), as well as after early, chance, and late reversal learning (R01, R50, and R85, respectively; see also Figure 5A), and RNA was extracted from the OFC. Using circRNA-specific quantitative real-time PCR, we found a significant reduction in circHomer1 levels within the OFC during chance reversal learning (Figure 5E), suggesting that circHomer1 expression could be specifically modulated during this specific stage of reversal learning. The expression of circTulp4, another brain-enriched circRNA, showed no changes in the OFC during any of the behavioral stages examined (Figure S4D). Furthermore, levels of Homer1b were found to significantly increase during chance reversal learning (Figure 5F). On the other hand, no changes were observed in Homer1a and Homer1c mRNA expression in the same samples (Figures S4E and S4F). Taken together, our data uncover opposing changes in circHomer1 and Homer1b expression within the OFC during chance reversal learning.

In vivo double KD of Homer1b and circHomer1 normalizes gene expression within the OFC

Based on our previous study, circHomer1 KD within the OFC resulted in robust alterations in mRNA isoform expression, which were preferentially derived from genes important for synaptic function (Zimmerman et al., 2020). Focusing on the most significantly altered mRNA isoforms following single circHomer1 KD (more than 1.5-fold change and q < 0.10; Zimmerman et al., 2020), we found very few alterations via deep RNA sequencing (RNA-seq) in the OFC of sh-Double-KD versus their respective sh-Control mice (Figure 5G). This observed rescue in circHomer1-regulated mRNA isoforms was pretty robust, since we could see no changes in 16 out of 17 mRNA isoforms, even when we lowered the cutoff for the sh-Homer1b RNA-seq data (p < 0.05 instead of q < 0.10 used for sh-circHomer1 and fold-change cutoff of 1.25 instead of 1.5; Figure 5G). Of note, with the exception of mRNA isoforms derived from just three genes, we found no significant differences in the levels of any other mRNA isoforms in the OFC of sh-Double-KD mice (Figure 5H; 1.5-fold change and q<0.10). Taken together, these results suggest that double Homer1b and circHomer1 KD can rescue the circHomer1-mediated molecular abnormalities in mRNA isoform expression within the OFC.

EIF4A3 binds to circHomer1 and can positively regulate its expression

In order to uncover potential upstream regulators of circHomer1 biogenesis, we used the Circinteractome in silico bioinformatics tool (Dudekula et al., 2016) to search for RBPs that could bind to circHomer1. This analysis uncovered nine potential binding sites for EIF4A3 (Dudekula et al., 2016), an RBP part of the exon junction complex (EJC) that has been previously shown to regulate the biogenesis of a subset of circRNAs (Wang et al., 2018; Zheng et al., 2020). Interestingly, analysis of EIF4A3 mRNA expression in human postmortem brains revealed that EIF4A3 is significantly downregulated in the OFC of patients with BD and is positively correlated with the relative changes in OFC circHomer1 expression (Figures 6A and 6B). Moreover, measurements of mouse Eif4a3 mRNA in the OFC of WT mice from different stages of the reversal learning behavioral paradigm revealed that its expression changes during reversal learning mirrored those of circHomer1 (Figure 6C; see also Figure 5E). Using RNA immunoprecipitation (RIP) with an anti-EIF4A3 antibody and circHomer1 quantitative real-time PCR, we found that EIF4A3 can indeed directly bind to mouse circHomer1 (Figures 6D and 6E). Furthermore, pharmacological inhibition of EIF4A3 in human cell lines and mouse cortical cultures resulted in a significant downregulation of circHomer1 expression (Figures 6F and 6G). On a similar note, shRNA-mediated KD of EIF4A3 mRNA in two different human cell lines showed significant downregulation of circHomer1 expression (Figures 6H and 6I). Taken together, these data identify EIF4A3 as an upstream regulator of circHomer1 synthesis, which correlates with both the experience-dependent changes in circHomer1 expression in mouse OFC and the observed alterations in circHomer1 levels in human postmortem brains.

Figure 6. — (A) Relative to control *EIF4A3* mRNA levels from the OFC of BD patients. *p < 0.05, based on a univariate general linear model corrected for RIN, PMI, brain pH, and age.

(B) Correlations between relative to control changes in *circHomer1* in the OFC of patients with SCZ and BD (SMRI cohort, RNaseR-treated samples used for *circHomer1* detection) and the relative to control ratios of *EIF4A3* mRNA levels. Spearman correlation coefficient and two-tailed p values are shown in the graph.

(C) Relative to D50 *Eif4a3* mRNA levels in non-injected adult mouse OFC during different stages of the touch-screen reversal learning paradigm. **p < 0.01, two-tailed, one-sample t test relative to D50 levels.

(D and E) RIP using an antibody against EIFA3 protein. Western blotting data are shown in (D). Mean ± SEM relative to immunoglobulin G (IgG) control *circHomer1* and *circCrebbp* levels (based on quantitative real-time PCR and normalized to *18S rRNA*) after EIF4A3 RIP (E). *p < 0.05, two-tailed, one-sample t test relative to IgG control levels.

(F and G) Relative to vehicle *circHomer1* and *circTulp4* levels 24 h after treatment with EIF4A3 inhibitor in mouse primary cortical neurons (F) and human HEK293T cells (G). *p < 0.05 and **p < 0.01, two-tailed, one-sample t test relative to vehicle.

(H and I) Relative to vehicle *circHomer1*, *circCDR1as*, and *EIF4A3* mRNA levels 48 h after shRNA-mediated KD of *EIF4A3* in human HEK293T cells (H) and human differentiated SH-SY5Y cells (I). *p < 0.05 and ****p < 0.0001, two-tailed, one-sample t test relative to vehicle.

All data are shown as mean ± SEM based on quantitative real-time PCR and normalization to *18S rRNA*. The individual biological replicates are shown within each graph.

DISCUSSION

Here, we provide evidence of an antagonistic interaction between a circRNA and its linear mRNA counterpart in the OFC that can significantly modulate behavioral flexibility. Specifically, we show that circHomer1 and Homer1b can directly bind to each other next to putative regulatory RBP-binding sites and they can inhibit each other’s synaptic expression within the OFC. Intriguingly, we also provide evidence that Homer1b KD in the OFC results in improved chance reversal learning and that its co-KD with circHomer1 can rescue the observed circHomer1-mediated abnormalities in both chance reversal learning and synaptic gene mRNA isoform expression. Furthermore, we uncover that the expression of circHomer1 and Homer1b is differentially affected specifically during chance reversal learning and that baseline circHomer1 levels within the OFC inversely correlate with individual performances in chance reversal learning. We also show that robust deficits in circHomer1 are present in both the brain and peripheral cells of patients with BD and that changes in circHomer1 expression in both the OFC and in iPSC-derived neuronal cultures of patients with BD and SCZ are inversely associated with relative HOMER1B mRNA expression. Moreover, we validate that circHomer1 is significantly downregulated in the DLPFC, but not OFC, of subjects with SCZ and that the predominant changes in circHomer1 observed in the OFC of patients with BD are limited to subjects with a positive history of psychosis at the time of death. Lastly, we identify that the RBP EIF4A3 can bind to circHomer1 and positively regulate its biogenesis. Together, our data elucidate a molecular interplay between a brain-enriched circRNA and its linear mRNA counterpart that is important for the control of both synaptic gene expression and OFC-mediated cognitive flexibility.

Based on our data, there appear to be two distinct competitive interactions between circHomer1 and Homer1b with relevance to OFC function and/or psychiatric disorders. Firstly, overall circHomer1 OFC levels reduce while Homer1b mRNA levels increase during chance reversal learning in control (non-injected) adult mice. Such an opposing change between circHomer1 and Homer1b expression is expected to be a result of a competition between circRNA backsplicing and linear splicing, since both circHomer1 and Homer1b originate from Homer1b pre-mRNA. Such a competition between circHomer1 backsplicing and linear Homer1 splicing could also underlie the robust negative correlations between changes in circHomer1 and relative (to total HOMER1 levels) HOMER1B mRNA expression in both the OFC and iPSC-derived neuronal cultures of subjects with BD and SCZ. In addition to a potential inhibitory interaction between circHomer1 and Homer1b biogenesis, our data support another layer of antagonistic interplay between circHomer1 and Homer1b that is related to a mutual inhibition of their synaptic expression. Given that shRNA-mediated KD of circHomer1 and Homer1b in the OFC does not affect overall levels of Homer1b and circHomer1, respectively, it is likely that circHomer1 and Homer1b inhibit each other’s synaptic localization or stability within the synapses. Interestingly, our data revealed that circHomer1 and Homer1b directly bind to each other and the putative binding site for Homer1b is within its 3′ UTR, a known regulatory region that can influence mRNA stability and localization. Moreover, the predicted binding sites were in close proximity to highly predicted binding elements for HuD, a known regulator of Homer1 mRNA transport (Tiruchinapalli et al., 2008; Zimmerman et al., 2020). It is thus tempting to hypothesize that circHomer1 competes with Homer1b for binding to HuD and for HuD-mediated transport into the synapses. However, the synaptic expression of the activity-dependent Homer1a, whose synaptic localization is also known to be activity dependent, was also found in our study to be robustly increased during single circHomer1 KD by around 4-fold but was only upregulated by 37% in sh-Double-KD mice. Thus, additional research is needed to determine whether such a reduction in synaptic Homer1a levels could be associated with potential changes in neuronal activity or whether it could be related to its capacity to also bind to both circHomer1 and HuD. On a similar note, future studies using in vivo electrophysiological analysis during reversal learning behavior in conjunction with manipulation of either Homer1a or Homer1b levels would allow us to properly elucidate the overall effects of circHomer1 KD in OFC neuronal activity.

It has to be noted though that our shRNA-mediated circHomer1 KD works by specifically degrading the mature circHomer1 molecules through binding to their unique exon 2–5 backsplice junctions. As a result, circHomer1 KD via shRNA does not influence circHomer1 biogenesis or the competition between linear splicing and circRNA backsplicing within the Homer1 gene. We have previously shown our shRNA approach to be very specific and devoid of notable off-target effects (Zimmerman et al., 2020). Moreover, our data showing that double Homer1b and circHomer1 KD can rescue all the molecular and behavioral abnormalities observed following single circHomer1 KD cannot be explained by any potential off-target effects in another gene other than circHomer1. Specifically, a potential off-target gene targeting would not be rescued by just correcting Homer1b levels, nor would it explain why we see the opposite behavioral effects from single Homer1b KD and single circHomer1 KD in the OFC. Our data thus strongly support a very specific effect and a very efficient targeting of circHomer1 in the OFC.

Our findings that EIF4A3 can directly bind to circHomer1 and promote its biogenesis in conjunction with the fact that changes in circHomer1 levels in mouse and human OFC are associated with changes in EIF4A3 mRNA levels, allowing us to identify an up-stream regulator of circHomer1 synthesis. Interestingly, EIF4A3 has been previously shown to be linked to synaptic plasticity and learning and memory and, together with other components of the EJC, control the decay of the immediate-early-gene-activity-regulated cytoskeleton-associated protein (Arc/Arg3.1; Barker-Haliski et al., 2012; McMahon et al., 2016). It is therefore tempting to hypothesize that circHomer1 biogenesis is regulated by the same pathways that are responsible for the control of other important mediators of experience-dependent synaptic plasticity.

The fact that circHomer1 is an experience-dependent circRNA within the OFC that can affect OFC-mediated reversal learning suggests that circHomer1 could be a major regulator of cortical plasticity. Furthermore, the recent findings that circHomer1 is significantly altered in old adult brains of patients with mild cognitive decline and Alzheimer’s disease (AD) and is robustly associated with AD-related neuropathology (Dube et al., 2019) also suggest that circHomer1 could have diverse physiological roles during different developmental time points. Such developmental-stage- and brain-region-specific effects in cortical function and cognition have been previously observed for other experience-dependent ncRNAs, such as miR-132 (Mellios et al., 2011; Chen et al., 2018; Xu et al., 2019). Furthermore, our findings of significantly reduced circHomer1 levels in the blood of patients with BD suggest the possibility for circHomer1 to be used as a biomarker for BD. Lastly, given that our work has revealed robust alterations in circHomer1 expression not only in the OFC of patients with BD but also the DLPFC of patients with SCZ, it would be of great interest to investigate the role of circHomer1 in additional brain regions and determine whether this circRNA could have widespread importance for PFC function and psychiatric disease.

Limitations of the study

As discussed above, one of the major technical limitations of this study is that it was not feasible to utilize multiple shRNA sequences against the circHomer1 splice junction that could specifically affect circHomer1 levels without influencing linear circHomer1 expression. Moreover, given the very high expression of this circRNA in the brain, we could not achieve a multifold reduction in its levels following shRNA-mediated KD. However, we did achieve significant and very specific reductions in circHomer1 expression, which allowed us to study its physiological role in OFC function and its interactions with Homer1 mRNAs. Another limitation of our study is that we were not able to show any HOMER1B protein isoform changes following circHomer1 KD, due to the lack of availability of HOMER1B-specific antibodies.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for reagents should be directed to and will be fulfilled by lead contact Nikolaos Mellios (nmellios@salud.unm.edu).

Materials availability

ShRNA Vectors can be made available upon request. Restrictions in availability could apply since due to the fact that it is a custom built by a commercial company.

Data and code availability

All RNA-seq data associated with this project have been submitted to GEO (Accession #GSE160874).
This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available form the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

Eight-week-old male C57BL/6J mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Mice were housed in pairs initially, and separated following surgery (please see below for detailas on post-surgical animal housing). Housing took place in a temperature- and humidity-controlled vivarium under a 12 hour reverse light/dark cycle (lights off at 0800 h) and behavioral testing occurred during the dark phase in a red-light room. Animal care was performed in accordance with the National Institutes of Health Guide for Care and Use of Laboratory Animals and approved by the University of New Mexico Health Sciences Center Institutional Animal Care and Use Committee.

Postmortem samples

Human postmortem brain total RNA samples from the OFC and DLPFC of subjects with BD (N=30), SCZ (n=32), or unaffected Control subjects (n=33) were obtained from SMRI (Torrey et al., 2000). An additional smaller cohort of postmortem brain total RNA samples from OFC of subjects with BD (n=21), SCZ (n=20) or unaffected Control subjects (n=20) was obtained from the Harvard Brain Tissue Resource Center (HBTRC). A larger postmortem brain total RNA cohort (DLPFC) was obtained from the Icahn School of Medicine at Mount Sinai (N=54 for SCZ, N=40 for BD, and N=60 for Controls). Details on demographics are shown in Tables S1–S6. Details on all primers used in these study are found on Table S7.

METHOD DETAILS

RNA extraction and mRNA/circRNA quantification

RNA extraction was done as previously shown (Mellios et al., 2011, 2018; Zimmerman et al., 2020). Briefly, RNA was isolated using the miRNeasy RNA isolation kit (Qiagen, Hilden, Germany) following the manufacturer’s supplied protocol. RNA quality as well as concentration of isolated and purchased total RNA was assayed through Nanodrop 2000c spectrophotometer and Qubit 3 (Thermo Fisher Scientific, Waltham, Massachusetts). RNA Integrity Number of total RNA samples from HBTRC was determined using the Agilent 2100 Bioanalyzer System (Agilent, Santa Clara, California). Reverse transcription of total RNA was carried out using the SuperScript IV First-Strand Synthesis System (Thermo Fisher Scientific) with oligo-dT for linear mRNAs and random hexamers for circRNA detection. qRT-PCR was done using either PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) along with custom designed, validated, and sequence-verified circRNA and mRNA primers or TaqMan Gene Expression Assays (Thermo Fisher Scientific) for mRNA detection. All circRNA qRT-PCR products were run on an agarose gel and sequence validated. 18S rRNA was used as a normalizer for mRNA detection, whereas circTulp4 and circCDR1as were used for circRNA normalization. For mRNA qRT-PCR quantification the following formula was used: Relative value = ÂCt^{18S rRNA}/ÂCt^mRNA, where A = 10^(−1/primer slope). For circRNA qRT-PCR quantification the following formula was used: Relative value = ÂCt^{2circRNA normalizers} (geometric mean of Ct^circTulp4 and Ct^CDR1as)/ÂCt^circRNA, where A = 10^(−1/primer slope). In the cases where one circRNA normalizer (circTulp4) was used the formula changes to: Relative value = ÂCt^circTulp4/ÂCt^circRNA, where A = 10^(−1/primer slope). Lastly when no normalizers were used circRNA relative expression was calculated as: Relative value = [Â – Ct^circRNA)] × 10^⁶, where A = 10^(−1/primer slope). Detailed information regarding the primers used in our study are included in Table S7.

In silico circRNA RNA predictions and circRNA-Antisense purification

CircHomer1 sequence was previously obtained from circbase and Homer1b 3’ UTR sequence was obtained from Ensemble genome database (Yates et al., 2020). IntaRNA, a program used to provide fast and accurate predictions of interactions between two RNA molecules, was used to predict binding sites between circHomer1 and the 3’ UTR or Homer1b 3’ UTR (Mann et al., 2017). RNAfold, part of the ViennaRNA Package (Lorenz et al., 2011), was used to predict the least free energy secondary structure of circHomer1 and examine potential interaction sites between the two transcripts.

CircRNA-Antisense Purification was carried out as previously published (Engreitz et al., 2015; Torres et al., 2018) with modifications for circRNA specificity. The secondary structure of circHomer1 was used to examine predicted binding and probe sites in the sequence. Due to the shared sequence between circHomer1 and exons 2, 3, 4, and 5 of Homer1, a single 30 nucleotide probe was designed to span the backsplice junction of circHomer1, which was not predicted to interact with any Homer1 isoform. A second 30 nt probe lacking specificity to any sequence in the mouse genome was designed as well (non-specific probe). The probes were ordered from Integrative DNA Technologies Inc (Coralville, IA) with a 3’ biotin label and triethyleneglycerol spacer to reduce steric hinderance. Mouse Neuro-2a (ATCC CCL-131) cells were grown to confluency and crosslinked with fresh 1% paraformaldehyde solution in PBS, washed twice with PBS, and collected in fresh lysis buffer (50 mM Tris-HCl pH 7.0, 10 mM EDTA, 1% SDS supplemented with 200 U/mL of RNAseOUT Ribonuclease Inhibitor (Invitrogen) and a cocktail of protease inhibitor 5 uL/mL (Roche)). The lysate was briefly sonicated to fragment the RNA between 200 and 800 bp. The lysed samples were divided equally and 2 volumes of hybridization buffer (50 mM Tris-HCl pH 7.0, 750 mM NaCl, 1 mM EDTA, 1% SDS, 15% Formamide added extemporaneously) were added to the supernatants. 100 pmol of biotinylated oligonucleotide probes were added to the respective circHomer1-specific or non-specific samples and incubated 6 hours on a tube rotator at room temperature. 50 μl of magnetic streptavidin beads (ThermoFisher Scientific) supplemented with 200 U/mL RNAseOUT Ribonuclease Inhibitor (Invitrogen) were added and incubated for 16 hours on a tube rotator at room temperature. A magnetic tube rack was used to separate the beads from the lysate and the supernatant was discarded. The beads were washed 5 times with wash buffer (SDS 0.5%, SSC 2x) interspersed with 5 minutes of agitation on a tube rotator. After the final wash, 95 μl of Proteinase K buffer (10 mM Tris-HCl pH 7.0, 100 mM NaCl, 1 mM EDTA, 0.5% SDS) and 5 μl of proteinase-K (20 mg/mL) were added to the samples. All samples were incubated for 45 minutes at 50°C followed by 10 minutes at 95°C. The samples were placed on the magnetic rack and the supernatant was collected. The same procedure was utilized to perform RNA antisense purification for the Homer1b 3’ UTR. In this case we used a number of overlapping biotinylated probes against Homer1b 3’ UTR. Probe 1 sequence: GAAACAAACGATTTATTTTCATTCTAATAT, Probe 2 sequence: ATTCTAATATAAATT AGTATGTATGTATTG, Probe 3 sequence: GTATGTATTGACATACATACACATAAATAT, Probe 4 sequence ACATAAATATAAAA CA TACTAAAGAATCTG, Probe 5 sequence: TAAAGAATCTGTTTGAGTGAGAAAGATGGG. The isolated RNA following these procedures was purified using a NucleoSpin RNA Mini kit (Machery-Nagel) according to the manufacturers supplied protocol. Purified RNA was quantified using a Nanodrop 2000 spectrophotometer (ThermoFisher Scientific). Reverse transcription was performed using the SuperScript IV First-Strand Synthesis System (ThermoFisher) with random hexamers. The cDNA was then used for qRT-PCR as previously described.

EIF4a3 KD and pharmacological inhibition

A shRNA-mediated KD approach was conducted by using lentiviral shRNA vectors (pLKO.1), already predesigned for EIF4A3 (Ref Seq Number: NM_014740) and purchased from (Millipore Sigma-Aldrich - MISSION® shRNA Products library). Multiple shRNA clones were purchased targeting the CSD region and after a round of experiment testing their efficiency, the clone that led to the highest % of KD of EIF4A3, was further used to determine changes in circHomer1 expression levels. EIF4a3 shRNA clone information: Target sequence: CGCATCTTGGTGAAACGTGAT, Product number: TRCN0000061854. Non-Target shRNA Control clone information: MISSION® pLKO.1-puro, shRNA insert that does not target any known genes from any species (Product number: SHC016). For EIF4A3 inhibition, we chose eIF4A3-IN-2, which is a highly selective and noncompetitive EIF4A3 inhibitor with an IC50 of 110 nM that binds to the allosteric region in EIF4A3 and inhibits in vitro ATPase, helicase, and cellular nonsense-mediated RNA decay (NMD) activities. The inhibitor was purchased by MedChemExpress (Cat. No.: HY-101785) in a solution form of 10 mM * 1 mL in DMSO. After a set of experiments trying to determine the optimal concentration and cytotoxicity, 10μM of inhibitor was selected for the experiments described below.

HEK293 embryonic kidney human cell line experiments

HEK293 epithelial embryonic kidney cells were purchased from ATCC CRL-1573™. Cells were plated in a 48 well plate at passage #9 at a concentration of 30,000 cells per well. Cells were fed with DMEM, high glucose, HEPES Catalog number: 12430112, Fetal Bovine Serum, Premium, Catalog number: A4766801, Penicillin-Streptomycin-Neomycin (PSN) Antibiotic Mixture, Catalog number: 15640055 ∼ ThermoFisher Scientific. 48 hours later were transfected with the purchased with various EIF4A3 clones targeting the CDS region to test their efficiency and non-target shRNA clone mentioned using LipofectamineTM 3000 / P3000 reagent (200 ng DNA, 0.5 μl Lipofectamine and 0.5 μl P3000 reagent per well). 48hrs later, the HEK293 cells were subjected to RNA extraction and then qRT-PCR to assay overall changes in circHomer1 expression and the efficiency of EIF4A3 clones. Pharmacological inhibition of EIF4A3 activity was performed using the optimal concentration after cell viability assays from a series of experiments. Thus, 72 hours after plating the HEK293, EIF4A3 pharmacological inhibitor was added at 10μM concentration. 24hours after the addition of the inhibitor, HEK293 cells were subjected to RNA extraction and then qRT-PCR to assay overall changes in circHomer1 expression.

SH-SY5y human glioblastoma cell line and EIF4A3 KD

SH-SY5Y epithelial human neuroblastoma cell line was purchased from ATCC CRL-2266™. SY-5 can differentiate under certain circumstances. To that end, cells were fed for 5 days using Neurobasal Plus, 1XB27 Plus, 5% Pen/Strep ∼ ThermoFisher Scientific. The neuronal-like morphology was observed under the microscope. After differentiation, SY-5 cells were plated in a 24 well plate at passage #7 at a concentration of 100,000 cells per well. 48 hours later they were transfected with the EIF4A3 shRNA clone and non-target shRNA clone mentioned above using LipofectamineTM 3000 / P3000 reagent (500 ng DNA, 1 μl Lipofectamine and 1 μl P3000 reagent per well). 48hrs following the transfection, the differentiated SH-SY5Y cells were subjected to RNA extraction and then qRT-PCR to assay overall changes in circHomer1 expression.

Neuro2a mouse neuroblastoma cell line and EIF4A3 inhibition

Neuro2a mouse cell line was purchased by ATCC (CCL-131™). Cells were plated in a 48 well plate at passage #8 and with a concentration of seeding density 30,000 cells per well using as a feeding medium: MEM Catalog number: 11095080, Fetal Bovine Serum, Premium, Catalog number: A4766801, Penicillin-Streptomycin-Neomycin (PSN) Antibiotic Mixture, Catalog number: 15640055 ∼ ThermoFisher Scientific. Neuro2A cells were then treated with 10μM of retinoic acid RA (R2625 Sigma-Aldrich) and 4 days later, their neuronal-like morphology was observed under the microscope. At that point Neuro2a cells had also reached 70% of confluency and therefore EIF4A3 inhibitor was added at a concentration of 10μM (determined by cell viability assay and from the HEK293 successful experiment). 24 hours later following treatment, cells were subjected to RNA extraction and then qRT-PCR to assay overall changes in circHomer1 expression.

Primary mouse cortical neurons and EIF4A3 inhibition

Mouse cortical neuronal cultures were purchased from ThermoFisher Scientific - Catalog number: A15585 (C57BL/6 mice). Neurons were plated at a density of 4–5×10^4 cells/12-mm coverslip coated with poly-Ornithine on a 24-well plate. Primary cortical neuronal cells were allowed to adhere for 20 min before addition of 500ul plating neuronal media. Neurons were fed by replacing half the volume with fresh media every third day (Neurobasal Plus, 1XB27 Plus, 2 mM Glutamax, 5% Pen/Strep ∼ ThermoFisher Scientific). At DIV13, 10μM of the EIF4A3 inhibitor were added. Of note, the inhibitor was dissolved in DMSO and to compensate the effects of DMSO on cell viability, a negative control / vehicle in the experiments mentioned above (HEK293-Neuro2A-primary neurons) was a treatment group of culture medium along with DMSO at the same concentration ∼ 0.1%. 24 hours later following treatment, neurons were subjected to RNA extraction and then qRT-PCR to assay overall changes in circHomer1 expression.

OFC crude synaptosome extractions

Crude synaptosomes were extracted from the OFC of mice as previously described (Boese et al., 2016; Rao and Steward, 1991). Briefly, the OFC tissue was immersed in 20% w/v of homogenization buffer (0.32 M Sucrose, 0.1 mM EDTA, 0.25 mM DTT, 2 mM HEPES, pH 7.4) supplemented with 200 U/mL RNAseOut (Invitrogen) and homogenized by mechanical disruption with pestles. Nuclei and cell debris were pelleted by centrifugation for 2 minutes at 2000 x g and the supernatant was collected. The resulting pellet was washed once again with homogenization and was pelleted again by centrifugation with the resulting supernatant combined. The supernatant was centrifuged an additional 10 minutes at 14,000 x g and the resulting pellet, representing the crude synaptosome fraction, was collected. Total RNA was extracted from the resulting pellet and supernatant using the miRNeasy Mini RNA Extraction Kit (Qiagen) as described above.

RNA sequencing in OFC samples with circHomer1/Homer1b double-KD

Total RNA extracted as previously described from OFC of sh-Double-KD and sh-Scramble mice was used for RNA sequencing (RNA-seq). Library preparation, RNA-seq, and analysis was carried out by Arraystar, Inc. First, Libraries were constructed using Kapa Stranded RNA-seq library kit (Illumina, San Diego, CA). Paired end RNA-seq was done on an Illumina HiSeq 4000 with a read length of 150 bps. Image analysis and base calling were performed using Solexa Platform v1.8 and sequence quality was assessed using FastQC software. Adapter trimming and filtering was performed by cut adapt software and the reads were then aligned to mouse genome GRCm28 using HiSat 2 software. The transcript abundance for each sample was estimated using StrongTie, and the GPKM value for the gene and transcript levels were calculated using the R package Ballgown (Frazee et al., 2014). All RNA-seq data associated with this project have been submitted to GEO (Accession #GSE160874).

Generation of shRNAs and lentiviruses in mouse neuronal cultures

ShRNA sequences against the unique exon 5/ exon 2 backsplice junction of mouse circHomer1 (GCCATTTCCACATAGGGAGCA), the unique exon 5 / exon 7 linear splice junction of mouse Homer1b mRNA (GCCATTTCCACATAGGGAGCA), and a scrambled control shRNA sequence were inserted in pLV-mU6-SYN-GFP lenti-vectors by Biosettia (San Diego, CA). These vectors can drive the expression of shRNAs via the mouse U6 promoter and the co-expression of GFP via the human Synapsin (SYN) promoter (Zimmerman et al., 2020). Ultra-high titer lentiviral particles (LVPs) carrying shRNAs were purchased from System Biosciences (Palo Alto, CA).

Absolute quantification of gene expression via digital droplet PCR

Digital droplet PCR (dd-PCR) was used to determine the exact number of copies per ul of DNA. The setup reaction per 21 μl for each sample was as following: 11 μl of QX200™ ddPCR™ EvaGreen Supermix #1864033, 2 μl of validated sequence verified circRNA or mRNA primers (forward and reverse respectively) and 8 μl of RNAase/DNAase free water. Supermix for each sample was vortexed vigorously and centrifuged at 12,000 rpm. Of note, for each sample 1 μl of cDNA was used. Using DG8™ Cartridges for QX200™/QX100™ Droplet Generator #1864008, 20 ml of each sample was loaded in the central row, while in the row below, 70 μl of the X200™ Droplet Generation Oil for EvaGreen #1864005 was loaded to generate the oil droplets. The cartridge was covered by DG8™ Gaskets for QX200™/QX100™ Droplet Generator #1863009 and it was loaded in the Digital Droplet generator QX200 for 2 minutes. The generated droplets (now located at the top row of the cartridge) were then transferred using a Rainin P50 multichannel pipettor. 40 μl of generated dropletswere carefully and slowly transferred in a ddPCR™ 96-Well Plates#12001925, specialized for the ddPCR assay. The PCR Plate Heat Seal, foil, pierceable #1814040 was used for sealing the plate with the help of the PX1 PCR Plate Sealer (ddPCR protocol sealing: 180C for 5 seconds). Immediately after sealing, the plate was moved to the C1000 Touch Thermal Cycler, since at that point the droplets are fragile. After the thermocycling protocol is complete, the plate was loaded to the QX200 Droplet Reader #1864003 for the droplet reading process (about 1 ½ minute per well of the 96 well plate). For data analysis of the Direct Quantification DQ, the QX Manager 1.2.345 Standard Edition program as used. Briefly, 10,000 generated droplets per well (event count) were the limit for data analysis. For the sake of consistency and liability, the thresholds were set manually for each well to eliminate the negative droplets of low fluorescence and collect only the positive ones. Finally, the total number of copies, copies per 20 μl and copies per ml were calculated by the algorithm of the software.

In vivo KD of circHomer1/Homer1b in mouse OFC and touch-based reversal learning

Mice underwent stereotaxic surgery as previously described (Zimmerman et al., 2020) in which they received bilateral injections into the OFC (AP: + 2.60, ML: ± 1.35, DV: - 2.70 to Bregma). Animals were match-pair randomly assigned to receive 0.5 μl of ultra-high titer LVPs expressing either the non-specific scrambled negative Control viral vector (sh-Control) or the Homer1b shRNA (sh-Homer1b) lentiviruses. In the case of double circHomer1 and Homer1b double KD (sh-Double-KD), animals were injected with 1 μl of ultra-high titer LVPs expressing either the sh-Control or a combination of 0.5 μl of the circHomer1 shRNA (sh-circHomer1) and 0.5 μl sh-Homer1b shRNA lentiviruses. Following surgery, animals were given two weeks to allow for recovery and stable lentiviral expression. Total trials, correct, incorrect, and correction trials were analyzed and further divided into win-stay, lose-shift, regressive, and perseverative trials for statistical comparison between groups. Animals who did not reach criterion for a behavioral stage were excluded, and those considered to be outliers by Rout test were excluded from analysis. Eighteen male mice (n = 9 per group) were allowed to acclimate for one week upon arrival to the facility, and were slowly weight-reduced to 85% of their free-feeding weight to ensure appropriate motivation for reward consumption. Following weight reduction, animals underwent operant training as described previously (Zimmerman et al., 2020). Briefly, animals were acclimated to the liquid reward (strawberry Nesquik) through presentation of 5–6 drops of strawberry Nesquik mixed with low fat milk in a plastic weigh boat placed in the animal’s homecage for 15 minutes per day for 3 consecutive days. The following day, animals were habituated to reward retrieval in the operant box for 10 minutes or 10 retrievals, whichever came first. All animals retrieved 10 rewards (10 μl strawberry Nesquik) in under 10 minutes and moved to pretraining, during which animals learned to bar press and screen touch to obtain a 10 μl liquid reward dispensed into a magazine within the operant box (model # ENV-307W, Med Associates, St. Albans, VT) housed within a sound- and light-attenuating box (Med Associates, St. Albans, VT). Animals continued through “punish” training, in which they learned to discriminate between a correct (white shape) or incorrect (black screen) stimulus. Correct responses were rewarded with 10 μl liquid reward accompanied by secondary reinforcers (1sec tone and magazine illumination) upon choice and a five second intertrial interval. Incorrect choices resulted in a white houselight presented during a 10 second timeout in which the animal could not initiate another trial. Incorrect trials were followed by correction trials where the stimulus was presented in the same location until the animal performed a correct trial. Once an animal reached criterion of 85% (>25/30) correct over two consecutive days, it was given food ad libitum for two days and then underwent surgery. Following surgery, animals were then given a pretraining reminder and moved to the initial discrimination where they learned to discriminate two novel equi-luminescent, counter-balanced stimuli until they reached criterion of 85% correct over two consecutive days. Mice then entered the reversal phase wherein the choice contingencies were switched. Previously incorrect choice was now the correct choice which resulted in reward and vice versa and were trained to 85% correct over two consecutive days. Following all behavioral testing, the majority of animals were sacrificed by cervical dislocation at the same time of day to account for any potential differences in circadian rhythms on gene expression. Two-millimeter micropunches were taken from bilateral OFC and frozen on dry ice, then stored at −80°C until processing.

OFC extraction across discrimination and reversal stages in wild-type C57BL6/J mice

Sixty (N = 12 mice/ stage) C57BL6/J mice were trained on pairwise discrimination as described above. In order to capture the possible dynamic regulation of Homer1 isoforms, mice were trained to five distinct behavioral stages as previously described (Brigman et al., 2013; Marquardt et al., 2019). D^early where the mice were performing approximately 50% correct over two consecutive days. D^late where mice had attained >85% correct over two consecutive days. R^early which is the first day of reversal where performance was approximately 20% correct, R^mid where during the reversal phase the mice were again at 50% correct over two consecutive days, R^late were they had attained >85% over two consecutive days. Upon reaching the specific stage, mice were sacrificed by cervical dislocation two hours following the start of the behavioral session. Brains were extracted and snap frozen using 2-methyl isobutane and then 2 mm micropunches (1 mm thickness) were taken of bilateral OFC.

Lymphoblastoid cell lines

B lymphoblastoids (BLCL) were collected from blood of BD patients and healthy Controls as described by (Squassina et al., 2013). In short, each 31 BLCL sample (19 cases and 12 Controls) was split into two equal aliquots and cultured in separate flasks; one with culture media containing 1 mM of lithium chloride (LiCl) and one without for seven days. BLCLs were pelleted via centrifugation for further total RNA extraction using miRNeasy Mini Kit (Qiagen) following manufactures protocol. Total RNA was reverse transcribed using M-MLV reverse transcriptase (Gibco) following manufactures protocol with random hexamer priming. To measure RNA expression, we used custom designed probes (IDT) with PowerUp SYBR Green Master Mix (ThermoFisher). Real-time PCR reactions were run in triplicates using the QuantStudio 6 Flex Real-Time PCR System and data was collected using the QuantStudio Real-Time PCR Software v1.1 (Applied Biosystems).

Induced pluripotent stem cell (iPSC)-derived neuronal cultures

Fibroblasts were derived from dermal skin punches extracted from 4 BD patients and three unaffected Controls enrolled in research studies in the Center for Genomic Medicine at the Department of Psychiatry of Massachusetts General Hospital. Derivation of iPSCs from fibroblasts and neuronal progenitors (NPs) and 2w, 4w, and 5w differentiated neurons from iPSCs was performed as shown before (Zimmerman et al., 2020). Skin biopsies of 9 childhood-onset SCZ and 10 healthy Controls were used to derive fibroblasts, which were reprogrammed to iPSCs and differentiated to NPs and 6w neuronal cultures in Mount Sinai Icahn School of Medicine as shown before (Hoffman et al., 2017; Zimmerman et al., 2020).

QUANTIFICATION AND STATISTICAL ANALYSIS

For postmortem measurements, a Univariate General Linear Model with post-hoc Bonferroni correction for multiple comparisons, which corrects for RIN, Brain pH, PMI, and age, was used (IBM SPSS Statistics 24 – IBM, Armonk, New York – (IBM, 2016)). Normalized values were divided to the mean of each Control group and the relative to control ratios were plotted as means ± S.E.M. using GraphPad Prism and after removing up to 2 outliers using Roots test (Graphpad Software, La Jolla, CA). A two-tailed one sample t-test was used for comparisons between two groups or more than 2 groups where all were compared to the same control group. A two-tailed paired t-test was used to compare circRNA levels in the same samples before and after treatment. In other cases of more than 2 groups where all groups were compared to each other, a one-way Analysis of Variance (ANOVA) with correction for multiple comparisons was used. Spearman correlation coefficients and two-tailed p-values were calculated. The significant level for each test was set as p < 0.05.

Supplementary Material

NIHMS1772785-supplement-1.pdf^{(921.8KB, pdf)}

NIHMS1772785-supplement-2.xlsx^{(8.9KB, xlsx)}

NIHMS1772785-supplement-3.xlsx^{(18.7KB, xlsx)}

NIHMS1772785-supplement-4.xlsx^{(8.8KB, xlsx)}

NIHMS1772785-supplement-5.xlsx^{(14.6KB, xlsx)}

NIHMS1772785-supplement-6.xlsx^{(8.8KB, xlsx)}

NIHMS1772785-supplement-7.xlsx^{(14.6KB, xlsx)}

NIHMS1772785-supplement-8.xlsx^{(9.8KB, xlsx)}

NIHMS1772785-supplement-9.pdf^{(5.8MB, pdf)}

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies

rabbit polyclonal anti-eIF4A3	Bethyl Laboratories	A302-981A

Bacterial and virus strains

pLV-mU6-SYN-GFP high titer lentiviruses	Systems Biosciences	N/A

Biological samples

Human postmortem brain total RNA samples from the OFC and DLPFC of subjects with BD N=30), SCZ (n=32), or unaffected Control subjects (n=33)	SMRI, Torrey et al. (2000)	N/A
An additional smaller cohort of postmortem brain total RNA samples from OFC of subjects with BD (n=21), SCZ (n=20) or unaffected Control subjects (n=20)	Harvard Brain Tissue Resource Center (HBTRC)	N/A
A larger postmortem brain total RNA cohort (DLPFC) (N=54 for SCZ, N=40 for BD, and N=60 for Controls)	Icahn School of Medicine at Mount Sinai	N/A
Crude synaptosomes were extracted from the OFC of mice	Boese et al. (2016); Rao and Steward (1991)	N/A
iPSC-derived neuronal cultures extracted from 4 BD patients and 3 unaffected Controls	Center for Genomic Medicine at the Department of Psychiatry of Massachusetts General Hospital (Zimmerman et al., 2020)	N/A
iPSC derived neuronal cultures derived from 9 childhood-onset SCZ and 10 healthy Controls	Mount Sinai Icahn School of Medicine (Hoffman et al., 2017; Zimmerman et al., 2020)	N/A
B lymphoblastoids (BLCL) were collected from blood of BD patients and healthy Controls	(Squassina et al., 2013)	N/A

Chemicals, peptides, and recombinant proteins

eIF4A3-IN-2 (EIF4A3 inhibitor)	MedChemExpress	Cat#: HY-101785
DMEM, MEM, Fetal Bovine Serum, Penicillin-Streptomycin-Neomycin (PSN) Antibiotic Mixture	ThermoFisher Scientific	Cat#: 12430112, 11095080, A4766801, 15640055
Retinoic acid RA	Sigma-Aldrich	R2625
Neurobasal™ Plus Medium, B-27™ Supplement (50X)	ThermoFisher Scientific	Cat#: A3582901, 17504044
LipofectamineTM 3000 / P3000 reagent	ThermoFisher Scientific	Cat#: 15338030
RNAseOUT Ribonuclease Inhibitor	Invitrogen	Cat#: 10777019
Lithium Chloride Solution, 1M	Fisher Science Education™	Cat#: S25390

Critical commercial assays

miRNeasy RNA isolation kit	Qiagen, Hilden, Germany	Cat#: 217004
SuperScript IV First-Strand Synthesis System	Thermo Fisher Scientific	Cat#: 18091200
PowerUp SYBR Green Master Mix	Thermo Fisher Scientific	Cat#: A25471
TaqMan Gene Expression Assays	Thermo Fisher Scientific	Cat#: 4331182
Pierce Magnetic streptavidin beads	Thermo Fisher Scientific	Cat#: 88816
NucleoSpin RNA Mini kit	Machery-Nagel	REF 740984.50
Kapa Stranded RNA-seq Library kit	Illumina, San Diego, CA	N/A
QX200™ ddPCR™ EvaGreen Supermix, DG8™ Cartridges for QX200™/QX100™ Droplet Generator, X200™Droplet Generation Oil for EvaGreen	Bio-Rad Laboratories \| Life Science Group	Cat#: 1864033, 1864008, 1864005
M-MLV Reverse Transcriptase (200 U/μL)	Invitrogen™	Cat#: 28025013

Deposited data

Ensemble genome database	Yates et al. (2020)	N/A
RNA-seq data associated with this project have been submitted to GEO	GEO	Accession #GSE160874

Experimental models: Cell lines

HEK293 epithelial embryonic kidney cells	ATCC	CRL-1573™
Neuro2a mouse cell line	ATCC	CCL-131™
SH-SY5Y epithelial human neuroblastoma cell line	ATCC	CRL-2266™
Mouse cortical neuronal cultures	ThermoFisher Scientific	Cat# A15585

Experimental models: Organisms/strains

Eight-week-old male C57BL/6J mice	The Jackson Laboratory	Jax: 000664

Oligonucleotides

Probe 1 sequence: GAAACAAACGATT TATTTTCATTCTAATAT	Integrative DNA Technologies Inc	N/A
Probe 2 sequence: ATTCTAATATAAAT TAGTATGTATGTATTG	Integrative DNA Technologies Inc	N/A
Probe 3 sequence: GTATGTATTGACAT ACATACACATAAATAT	Integrative DNA Technologies Inc	N/A
Probe 4 sequence: ACATAAATATAAAA ATACTAAAGAATCTG	Integrative DNA Technologies Inc	N/A
Probe 5 sequence: TAAAGAATCTGTTT GAGTGAGAAAGATGGG	Integrative DNA Technologies Inc	N/A
ShRNA sequences against the unique exon 5/ exon 2 backsplice junction of mouse circHomer1 (GCCATTTCCACATAGGGAGCA)	Biosettia	Zimmerman et al. (2020)
ShRNA sequences against the unique exon 5 / exon 7 linear splice junction of mouse Homer1b mRNA (GCCATTTCCACATAGGGAGCA)	Biosettia	Zimmerman et al. (2020)

Recombinant DNA

shRNA EIF4A3 (lentiviral vector (pLKO.1) Target sequence:CGCATCTTGGTGAAACGTGAT	Millipore Sigma-Aldrich - MISSION® shRNA Products	TRCN0000061854
Non-Target shRNA Control Target sequence: GAGAAATTATTAGCG CTATCGCGCTTTTT CCGGGCGCGAT AGCGCTAATAATTTCTC	Millipore Sigma-Aldrich	MISSION® pLKO.1-puro: SHC016
pLV-mU6-SYN-GFP lenti-vectors	Biosettia	N/A

Software and algorithms

IntaRNA	Mann et al. (2017)	N/A
RNAfold, part of the ViennaRNA Package	Lorenz et al. (2011)	N/A
Library preparation, RNA-seq, and analysis	Arraystar, Inc.	N/A
Image analysis and base calling	Solexa Platform v1.8	N/A
Sequence quality	FastQC software	N/A
Read alignment	HiSat 2 software	N/A
For data analysis of the Direct Quantification DQ	QX Manager 1.2.345 Standard Edition	N/A

Other

Bar press and screen touch dispensed into a magazine within the operant box	model # ENV-307W, Med Associates, St. Albans, VT	N/A
Sound- and light-attenuating box	Med Associates, St. Albans, VT	N/A

Open in a new tab

Highlights.

Expression of circHomer1 is inversely associated with relative HOMER1B mRNA levels
CircHomer1 and Homer1b bind to each other and inhibit each other’s synaptic expression
In vivo KD of Homer1b improves OFC-mediated chance reversal learning
Co-KD of Homer1b and circHomer1 restores chance reversal learning

ACKNOWLEDGMENTS

This work was supported by the NIGMS P20 grant (1P20-GM121176; N.M., A.K.H., and S.K.A.), a Young Investigator Grant (FP00000839; N.M.) by the Brain & Behavior Research Foundation, and a high-priority, short-term R56 award from the NIMH (1R56-MH119150; N.M.). This work was partially supported by NIH grants 5R01-AA026583, R01-MH101454, R01-MH106056, R33-MH087896, R01-AG067025, R01-AG065582, R01-AG050986, R01-MH110921, U01-MH116442, R01-MH125246, R01-MH106056, and R01-MH109897. Further support was available by the Veterans Affairs Merit grant BX002395 (to P.R.), a BBRF Young Investigator Grant (K.J.B.), the MGH Research Scholars Program (S.J.H.), and the New York Stem Cell Foundation (K.J.B.). B.A. was supported by the Scientific and Technological Council of Turkey (2214/A) International Research Fellowship Program. We would like to thank Dr. Maree Webster and the SMRI brain bank for providing us with postmortem brain specimen. We would also like to thank Brian Rodriguez, Cole Bird, and Mate Fisher for technical assistance; Dr. Fernando Valenzuela for advice on the manuscript; and the UNM Clinical & Translational Science Center (CTSC) for allowing us access to ddPCR equipment. The authors gratefully acknowledge use of the Center of Biomedical Research Excellence (COBRE) data (P20-GM103472) and the Function Biomedical Informatics Research Network (FBIRN) data.

Footnotes

DECLARATION OF INTERESTS

A.K.H. and N.M. have a financial interest as co-founders of Circular Genomics Inc. and are inventors of patents related to the use of circRNAs for brain disease diagnostics (N.M.) or therapeutics (N.M. and A.K.H.).

SUPPLEMENTAL INFORMATION

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2021.110282.

REFERENCES

Abdelmohsen K, Panda AC, Munk R, Grammatikakis I, Dudekula DB, De S, Kim J, Noh JH, Kim KM, Martindale JL, et al. (2017). Identification of HuR target circular RNAs uncovers suppression of PABPN1 translation by CircPABPN1. RNA Biol. 14, 361–369. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ashwal-Fluss R, Meyer M, Pamudurti NR, Ivanov A, Bartok O, Hanan M, Evantal N, Memczak S, Rajewsky N, and Kadener S. (2014). CircRNA biogenesis competes with pre-mRNA splicing. Mol. Cell 56, 55–66. [DOI] [PubMed] [Google Scholar]
Barker-Haliski ML, Pastuzyn ED, and Keefe KA (2012). Expression of the core exon-junction complex factor eukaryotic initiation factor 4A3 is increased during spatial exploration and striatally-mediated learning. Neuroscience 226, 51–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
Bavamian S, Mellios N, Lalonde J, Fass DM, Wang J, Sheridan SD, Madison JM, Zhou F, Rueckert EH, Barker D, et al. (2015). Dysregulation of miR-34a links neuronal development to genetic risk factors for bipolar disorder. Mol. Psychiatr. 20, 573–584. [DOI] [PMC free article] [PubMed] [Google Scholar]
Boese AS, Saba R, Campbell K, Majer A, Medina S, Burton L, Booth TF, Chong P, Westmacott G, Dutta SM, et al. (2016). MicroRNA abundance is altered in synaptoneurosomes during prion disease. Mol. Cell. Neurosci. 71, 13–24. [DOI] [PubMed] [Google Scholar]
Bottai D, Guzowski JF, Schwarz MK, Kang SH, Xiao B, Lanahan A, Worley PF, and Seeburg PH (2002). Synaptic activity-induced conversion of intronic to exonic sequence in Homer 1 immediate early gene expression. J. Neurosci. 22, 167–175. [DOI] [PMC free article] [PubMed] [Google Scholar]
Brigman JL, Daut RA, Wright T, Gunduz-Cinar O, Graybeal C, Davis MI, Jiang Z, Saksida LM, Jinde S, Pease M, et al. (2013). GluN2B in corticostriatal circuits governs choice learning and choice shifting. Nat. Neurosci. 16, 1101–1110. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cao L, Tian Y, Jiang Y, Zhang GJ, Lei H, and Di ZL (2015). Down-regulation of homer1b/c protects against chemically induced seizures through inhibition of mTOR signaling. Cell. Physiol. Biochem. 35, 1633–1642. [DOI] [PubMed] [Google Scholar]
Chen LL, and Yang L. (2015). Regulation of circRNA biogenesis. RNA Biol. 12, 381–388. [DOI] [PMC free article] [PubMed] [Google Scholar]
Chen T, Fei F, Jiang XF, Zhang L, Qu Y, Huo K, and Fei Z. (2012). Down-regulation of Homer1b/c attenuates glutamate-mediated excitotoxicity through endoplasmic reticulum and mitochondria pathways in rat cortical neurons. Free Radic. Biol. Med. 52, 208–217. [DOI] [PubMed] [Google Scholar]
Chen D, Hu S, Wu Z, Liu J, and Li S. (2018). The role of MiR-132 in regulating neural stem cell proliferation, differentiation and neuronal maturation. Cell. Physiol. Biochem. 47, 2319–2330. [DOI] [PubMed] [Google Scholar]
Conn SJ, Pillman KA, Toubia J, Conn VM, Salmanidis M, Phillips CA, Roslan S, Schreiber AW, Gregory PA, and Goodall GJ (2015). The RNA binding protein quaking regulates formation of circRNAs. Cell 160, 1125–1134. [DOI] [PubMed] [Google Scholar]
Crabtree G, and Gogos JA (2014). Synaptic plasticity, neural circuits and the emerging role of altered short-term information processing in schizophrenia. Front. Synaptic Neurosci. 6, 28. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dell’Orco M, Oliver RJ, and Perrone-Bizzozero N. (2020). HuD binds to and regulates circular RNAs derived from neuronal development- and synaptic plasticity-associated genes. Front. Genet. 11, 790. [DOI] [PMC free article] [PubMed] [Google Scholar]
Deschênes-Furry J, Perrone-Bizzozero N, and Jasmin BJ (2006). The RNA-binding protein HuD: a regulator of neuronal differentiation, maintenance and plasticity. BioEssays 28, 822–833. [DOI] [PubMed] [Google Scholar]
Dube U, Del-Aguila JL, Li Z, Budde JP, Jiang S, Hsu S, Ibanez L, Fernandez MV, Farias F, Norton J, et al. (2019). An atlas of cortical circular RNA expression in Alzheimer disease brains demonstrates clinical and pathological associations. Nat. Neurosci. 22, 1903–1912. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dudekula DB, Panda AC, Grammatikakis I, De S, Abdelmohsen K, and Gorospe M. (2016). Circinteractome: a web tool for exploring circular RNAs and their interacting proteins and microRNAs. RNA Biol. 13, 34–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
Engreitz J, Lander ES, and Guttman M. (2015). RNA antisense purification (RAP) for mapping RNA interactions with chromatin. Methods Mol. Biol. 1262, 183–197. [DOI] [PubMed] [Google Scholar]
Frazee a.C., Pertea G, Jaffe a.E., Langmead B, Salzberg SL, and Leek JT (2014). Flexible isoform-level differential expression analysis with Ballgown. BioRxiv, 003665. [Google Scholar]
Gimse K, Gorzek RC, Olin A, Osting S, and Burger C. (2018). Hippocampal Homer1b/c is necessary for contextual fear conditioning and group I metabotropic glutamate receptor mediated long-term depression. Neurobiol. Learn. Mem. 156, 17–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
Glažar P, Papavasileiou P, and Rajewsky N. (2014). CircBase: a database for circular RNAs. RNA20, 1666–1670. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gruner H, Cortés-López M, Cooper DA, Bauer M, and Miura P. (2016). CircRNA accumulation in the aging mouse brain. Sci. Rep. 6, 38907. [DOI] [PMC free article] [PubMed] [Google Scholar]
Guo JU, Agarwal V, Guo H, and Bartel DP (2014). Expanded identification and characterization of mammalian circular RNAs. Genome Biol. 15, 409. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hamilton DA, and Brigman JL (2015). Behavioral flexibility in rats and mice: contributions of distinct frontocortical regions. Gene Brain Behav. 14, 4–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hansen TB, Jensen TI, Clausen BH, Bramsen JB, Finsen B, Damgaard CK, and Kjems J. (2013). Natural RNA circles function as efficient microRNA sponges. Nature 495, 384–388. [DOI] [PubMed] [Google Scholar]
Hashimoto R, Tankou S, Takeda M, and Sawa A. (2007). Postsynaptic density: a key convergent site for schizophrenia susceptibility factors and possible target for drug development. Drugs Today 43, 645–654. [DOI] [PubMed] [Google Scholar]
Hoffman GE, Hartley BJ, Flaherty E, Ladran I, Gochman P, Ruderfer DM, Stahl EA, Rapoport J, Sklar P, and Brennand KJ (2017). Transcriptional signatures of schizophrenia in hiPSC-derived NPCs and neurons are concordant with post-mortem adult brains. Nat. Commun. 8, 2225. 10.1038/s41467-017-02330-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hu JH, Park JM, Park S, Xiao B, Dehoff MH, Kim S, Hayashi T, Schwarz MK, Huganir RL, Seeburg PH, et al. (2010). Homeostatic scaling requires group I mGluR activation mediated by Homer1a. Neuron 68, 1128–1142. [DOI] [PMC free article] [PubMed] [Google Scholar]
Huang WD, Liu XB, Fei Z, Zhang YL, and Yang J. (2012). Down-regulation of Homer1b/c expression protects cultured neurons after traumatic injury. Neural Regen. Res. 7, 2176–2181. [DOI] [PMC free article] [PubMed] [Google Scholar]
IBM. (2016). IBM SPSS Advanced Statistics 24 (IBM; ). [Google Scholar]
Izquierdo A, Brigman JL, Radke AK, Rudebeck PH, and Holmes A. (2017). The neural basis of reversal learning: an updated perspective. Neuroscience 345, 12–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
Jeck WR, Sorrentino JA, Wang K, Slevin MK, Burd CE, Liu J, Marzluff WF, and Sharpless NE (2013). Circular RNAs are abundant, conserved, and associated with ALU repeats. RNA 19, 141–157. [DOI] [PMC free article] [PubMed] [Google Scholar]
Klugmann M, Symes CW, Leichtlein CB, Klaussner BK, Dunning J, Fong D, Young D, and During MJ (2005). AAV-mediated hippocampal expression of short and long Homer 1 proteins differentially affect cognition and seizure activity in adult rats. Mol. Cell. Neurosci. 28, 347–360. [DOI] [PubMed] [Google Scholar]
Leeson VC, Robbins TW, Matheson E, Hutton SB, Ron MA, Barnes TRE, and Joyce EM (2009). Discrimination learning, reversal, and set-shifting in first-episode schizophrenia: stability over six years and specific associations with medication type and disorganization syndrome. Biol. Psychiatr. 66, 586–593. [DOI] [PMC free article] [PubMed] [Google Scholar]
Li Z, Huang C, Bao C, Chen L, Lin M, Wang X, Zhong G, Yu B, Hu W, Dai L, et al. (2015). Exon-intron circular RNAs regulate transcription in the nucleus. Nat. Struct. Mol. Biol. 22, 256–264. [DOI] [PubMed] [Google Scholar]
Liang D, and Wilusz JE (2014). Short intronic repeat sequences facilitate circular RNA production. Genes Dev. 28, 2233–2247. [DOI] [PMC free article] [PubMed] [Google Scholar]
Liu Z, Ran Y, Tao C, Li S, Chen J, and Yang E. (2019). Detection of circular RNA expression and related quantitative trait loci in the human dorsolateral prefrontal cortex. Genome Biol. 20, 99. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lominac KD, Oleson EB, Pava M, Klugmann M, Schwarz MK, Seeburg PH, During MJ, Worley PF, Kalivas PW, and Szumlinski KK (2005). Distinct roles for different Homer1 isoforms in behaviors and associated prefrontal cortex function. J. Neurosci. 25, 11586–11594. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lorenz R, Bernhart SH, Höner zu Siederdissen C, Tafer H, Flamm C, Stadler PF, and Hofacker IL (2011). ViennaRNA package 2.0. Algorithms Mol. Biol. 6, 26. [DOI] [PMC free article] [PubMed] [Google Scholar]
Luykx JJ, Giuliani F, Giuliani G, and Veldink J. (2019). Coding and noncoding RNA abnormalities in bipolar disorder. Genes (Basel) 10, 946. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mahmoudi E, Fitzsimmons C, Geaghan MP, Shannon Weickert C, Atkins JR, Wang X, and Cairns MJ (2019). Circular RNA biogenesis is decreased in postmortem cortical gray matter in schizophrenia and may alter the bioavailability of associated miRNA. Neuropsychopharmacology 44, 1043–1054. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mainen ZF, and Kepecs A. (2009). Neural representation of behavioral outcomes in the orbitofrontal cortex. Curr. Opin. Neurobiol. 19, 84–91. [DOI] [PubMed] [Google Scholar]
Mann M, Wright PR, and Backofen R. (2017). IntaRNA 2.0: enhanced and customizable prediction of RNA-RNA interactions. Nucleic Acids Res. 45, W435–W439. [DOI] [PMC free article] [PubMed] [Google Scholar]
Marquardt K, Josey M, Kenton JA, Cavanagh JF, Holmes A, and Brigman JL (2019). Impaired cognitive flexibility following NMDAR-GluN2B deletion is associated with altered orbitofrontal-striatal function. Neuroscience 404, 338–352. [DOI] [PMC free article] [PubMed] [Google Scholar]
McMahon JJ, Miller EE, and Silver DL (2016). The exon junction complex in neural development and neurodevelopmental disease. Int. J. Dev. Neurosci.: Off. J. Int. Soc. Dev. Neurosci. 55, 117–123. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mellios N, Sugihara H, Castro J, Banerjee A, Le C, Kumar A, Crawford B, Strathmann J, Tropea D, Levine SS, et al. (2011). MiR-132, an experience-dependent microRNA, is essential for visual cortex plasticity. Nat. Neurosci. 14, 1240–1242. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mellios N, Feldman DA, Sheridan SD, Ip JPK, Kwok S, Amoah SK, Rosen B, Rodriguez BA, Crawford B, Swaminathan R, et al. (2018). MeCP2-regulated miRNAs control early human neurogenesis through differential effects on ERK and AKT signaling. Mol. Psychiatr. 23, 1051–1065. [DOI] [PMC free article] [PubMed] [Google Scholar]
Memczak S, Jens M, Elefsinioti A, Torti F, Krueger J, Rybak A, Maier L, Mackowiak SD, Gregersen LH, Munschauer M, et al. (2013). Circular RNAs are a large class of animal RNAs with regulatory potency. Nature 495, 333–338. [DOI] [PubMed] [Google Scholar]
Mullins N, Forstner AJ, O’Connell KS, Coombes B, Coleman JRI, Qiao Z, Als TD, Bigdeli TB, Børte S, Bryois J, and Charney AW (2021). Genome-wide association study of more than 40,000 bipolar disorder cases provides new insights into the underlying biology. Nat. Genet. 53, 817–829. 10.1038/s41588-021-00857-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
O’Donnell LA, Deldin PJ, Pester B, McInnis MG, Langenecker SA, and Ryan KA (2017). Cognitive flexibility: a trait of bipolar disorder that worsens with length of illness. J. Clin. Exp. Neuropsychol. 39, 979–987. [DOI] [PubMed] [Google Scholar]
Piwecka M, Glazar P, Hernandez-Miranda LR, Memczak S, Wolf SA, Rybak-Wolf A, Filipchyk A, Klironomos F, Cerda Jara CA, Fenske P, et al. (2017). Loss of a mammalian circular RNA locus causes miRNA deregulation and affects brain function. Science 357, eaam8526. [DOI] [PubMed] [Google Scholar]
Rao A, and Steward O. (1991). Evidence that protein constituents of postsynaptic membrane specializations are locally synthesized: analysis of proteins synthesized within synaptosomes. J. Neurosci. 11, 2881–2895. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rudebeck PH, and Murray EA (2008). Amygdala and orbitofrontal cortex lesions differentially influence choices during object reversal learning. J. Neurosci. 28, 8338–8343. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rybak-Wolf A, Stottmeister C, Glažar P, Jens M, Pino N, Giusti S, Behm M, Bartok O, Ashwal-Fluss R, Herzog M, et al. (2014). Circular RNAs in the mammalian brain are highly abundant, conserved, and dynamically expressed. Mol. Cell 58, 870–885. [DOI] [PubMed] [Google Scholar]
Salzman J, Gawad C, Wang PL, Lacayo N, and Brown PO (2012). Circular RNAs are the predominant transcript isoform from hundreds of human genes in diverse cell types. PLoS One 7, e30733. [DOI] [PMC free article] [PubMed] [Google Scholar]
Schloesser RJ, Huang J, Klein PS, and Manji HK (2008). Cellular plasticity cascades in the pathophysiology and treatment of bipolar disorder. Neuropsychopharmacology 33, 110–133. [DOI] [PubMed] [Google Scholar]
Squassina A, Manchia M, Chillotti C, Deiana V, Congiu D, Paribello F, Roncada P, Soggiu A, Piras C, Urbani A, et al. (2013). Differential effect of lithium on spermidine/spermine N1-acetyltransferase expression in suicidal behaviour. Int. J. Neuropsychopharmacol. 16, 2209–2218. [DOI] [PubMed] [Google Scholar]
Szumlinski KK, Lominac KD, Kleschen MJ, Oleson EB, Dehoff MH, Schwartz MK, Seeberg PH, Worley PF, and Kalivas PW (2005). Behavioral and neurochemical phenotyping of Homer1 mutant mice: possible relevance to schizophrenia. In Genes, Brain and Behavior, (Genes Brain Behav), pp. 273–288. [DOI] [PubMed] [Google Scholar]
Szumlinski KK, Kalivas PW, and Worley PF (2006). Homer proteins: implications for neuropsychiatric disorders. Curr. Opin. Neurobiol. 16, 251–257. [DOI] [PubMed] [Google Scholar]
Tiruchinapalli DM, Caron MG, and Keene JD (2008). Activity-dependent expression of ELAV/Hu RBPs and neuronal mRNAs in seizure and cocaine brain. J. Neurochem. 107, 1529–1543. [DOI] [PubMed] [Google Scholar]
Torres M, Becquet D, Guillen S, Boyer B, Moreno M, Blanchard MP, Franc JL, and François-Bellan AM (2018). RNA pull-down procedure to identify RNA targets of a long non-coding RNA. J. Vis. Exp. 2018, 134. [DOI] [PMC free article] [PubMed] [Google Scholar]
Torrey EF, Webster M, Knable M, Johnston N, and Yolken RH (2000). The Stanley foundation brain collection and neuropathology consortium. Schizophr. Res. 44, 151–155. [DOI] [PubMed] [Google Scholar]
Waltz JA, and Gold JM (2007). Probabilistic reversal learning impairments in schizophrenia: further evidence of orbitofrontal dysfunction. Schizophr. Res. 93, 296–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wang R, Li J, Jia R, Pan Y, and Liang H. (2018). EIF4A3-induced circular RNA MMP9 (circMMP9) acts as a sponge of miR-124 and promotes glioblastoma multiforme cell tumorigenesis. Mol. Cancer 17, 166. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wegbreit E, Cushman GK, Weissman AB, Bojanek E, Kim KL, Leibenluft E, and Dickstein DP (2016). Reversal-learning deficits in childhood-onset bipolar disorder across the transition from childhood to young adulthood. J. Affect. Disord. 203, 46–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
Xu N, Li AD, Ji LL, Ye Y, Wang ZY, and Tong L. (2019). MiR-132 regulates the expression of synaptic proteins in app/ps1 transgenic mice through C1q. Eur. J. Histochem. 63, 69–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
Yao G, Niu W, Zhu X, He M, Kong L, Chen S, Zhang L, and Cheng Z. (2019). Hsa-circRNA-104597: a novel potential diagnostic and therapeutic biomarker for schizophrenia. Biomark. Med. 13, 331–340. [DOI] [PubMed] [Google Scholar]
Yates AD, Achuthan P, Akanni W, Allen J, Allen J, Alvarez-Jarreta J, Amode MR, Armean IM, Azov AG, Bennett R, et al. (2020). Ensembl 2020. Nucleic Acids Res. 48, D682–D688. [DOI] [PMC free article] [PubMed] [Google Scholar]
You X, Vlatkovic I, Babic A, Will T, Epstein I, Tushev G, Akbalik G, Wang M, Glock C, Quedenau C, et al. (2015). Neural circular RNAs are derived from synaptic genes and regulated by development and plasticity. Nat. Neurosci. 18, 603–610. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang XO, Wang HB, Zhang Y, Lu X, Chen LL, and Yang L. (2014). Complementary sequence-mediated exon circularization. Cell 159, 134–147. [DOI] [PubMed] [Google Scholar]
Zhang Y, Du L, Bai Y, Han B, He C, Gong L, Huang R, Shen L, Chao J, Liu P, et al. (2020). CircDYM ameliorates depressive-like behavior by targeting miR-9 to regulate microglial activation via HSP90 ubiquitination. Mol. Psychiatr. 25, 1175–1190. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zheng X, Huang M, Xing L, and Yang R. (2020). The circRNA circSEPT9 mediated by E2F1 and EIF4A3 facilitates the carcinogenesis and development of triple-negative breast cancer. Mol. Cancer 19, 73. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zimmerman AJ, Hafez AK, Amoah SK, Rodriguez BA, Dell’Orco M, Lozano E, Hartley BJ, Alural B, Lalonde J, Chander P, et al. (2020).A psychiatric disease-related circular RNA controls synaptic gene expression and cognition. Mol. Psychiatr. 24, 2712–2727. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS1772785-supplement-1.pdf^{(921.8KB, pdf)}

NIHMS1772785-supplement-2.xlsx^{(8.9KB, xlsx)}

NIHMS1772785-supplement-3.xlsx^{(18.7KB, xlsx)}

NIHMS1772785-supplement-4.xlsx^{(8.8KB, xlsx)}

NIHMS1772785-supplement-5.xlsx^{(14.6KB, xlsx)}

NIHMS1772785-supplement-6.xlsx^{(8.8KB, xlsx)}

NIHMS1772785-supplement-7.xlsx^{(14.6KB, xlsx)}

NIHMS1772785-supplement-8.xlsx^{(9.8KB, xlsx)}

NIHMS1772785-supplement-9.pdf^{(5.8MB, pdf)}

Data Availability Statement

All RNA-seq data associated with this project have been submitted to GEO (Accession #GSE160874).
This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available form the lead contact upon request.

[R1] Abdelmohsen K, Panda AC, Munk R, Grammatikakis I, Dudekula DB, De S, Kim J, Noh JH, Kim KM, Martindale JL, et al. (2017). Identification of HuR target circular RNAs uncovers suppression of PABPN1 translation by CircPABPN1. RNA Biol. 14, 361–369. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] Ashwal-Fluss R, Meyer M, Pamudurti NR, Ivanov A, Bartok O, Hanan M, Evantal N, Memczak S, Rajewsky N, and Kadener S. (2014). CircRNA biogenesis competes with pre-mRNA splicing. Mol. Cell 56, 55–66. [DOI] [PubMed] [Google Scholar]

[R3] Barker-Haliski ML, Pastuzyn ED, and Keefe KA (2012). Expression of the core exon-junction complex factor eukaryotic initiation factor 4A3 is increased during spatial exploration and striatally-mediated learning. Neuroscience 226, 51–61. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] Bavamian S, Mellios N, Lalonde J, Fass DM, Wang J, Sheridan SD, Madison JM, Zhou F, Rueckert EH, Barker D, et al. (2015). Dysregulation of miR-34a links neuronal development to genetic risk factors for bipolar disorder. Mol. Psychiatr. 20, 573–584. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] Boese AS, Saba R, Campbell K, Majer A, Medina S, Burton L, Booth TF, Chong P, Westmacott G, Dutta SM, et al. (2016). MicroRNA abundance is altered in synaptoneurosomes during prion disease. Mol. Cell. Neurosci. 71, 13–24. [DOI] [PubMed] [Google Scholar]

[R6] Bottai D, Guzowski JF, Schwarz MK, Kang SH, Xiao B, Lanahan A, Worley PF, and Seeburg PH (2002). Synaptic activity-induced conversion of intronic to exonic sequence in Homer 1 immediate early gene expression. J. Neurosci. 22, 167–175. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R7] Brigman JL, Daut RA, Wright T, Gunduz-Cinar O, Graybeal C, Davis MI, Jiang Z, Saksida LM, Jinde S, Pease M, et al. (2013). GluN2B in corticostriatal circuits governs choice learning and choice shifting. Nat. Neurosci. 16, 1101–1110. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R8] Cao L, Tian Y, Jiang Y, Zhang GJ, Lei H, and Di ZL (2015). Down-regulation of homer1b/c protects against chemically induced seizures through inhibition of mTOR signaling. Cell. Physiol. Biochem. 35, 1633–1642. [DOI] [PubMed] [Google Scholar]

[R9] Chen LL, and Yang L. (2015). Regulation of circRNA biogenesis. RNA Biol. 12, 381–388. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] Chen T, Fei F, Jiang XF, Zhang L, Qu Y, Huo K, and Fei Z. (2012). Down-regulation of Homer1b/c attenuates glutamate-mediated excitotoxicity through endoplasmic reticulum and mitochondria pathways in rat cortical neurons. Free Radic. Biol. Med. 52, 208–217. [DOI] [PubMed] [Google Scholar]

[R11] Chen D, Hu S, Wu Z, Liu J, and Li S. (2018). The role of MiR-132 in regulating neural stem cell proliferation, differentiation and neuronal maturation. Cell. Physiol. Biochem. 47, 2319–2330. [DOI] [PubMed] [Google Scholar]

[R12] Conn SJ, Pillman KA, Toubia J, Conn VM, Salmanidis M, Phillips CA, Roslan S, Schreiber AW, Gregory PA, and Goodall GJ (2015). The RNA binding protein quaking regulates formation of circRNAs. Cell 160, 1125–1134. [DOI] [PubMed] [Google Scholar]

[R13] Crabtree G, and Gogos JA (2014). Synaptic plasticity, neural circuits and the emerging role of altered short-term information processing in schizophrenia. Front. Synaptic Neurosci. 6, 28. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] Dell’Orco M, Oliver RJ, and Perrone-Bizzozero N. (2020). HuD binds to and regulates circular RNAs derived from neuronal development- and synaptic plasticity-associated genes. Front. Genet. 11, 790. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] Deschênes-Furry J, Perrone-Bizzozero N, and Jasmin BJ (2006). The RNA-binding protein HuD: a regulator of neuronal differentiation, maintenance and plasticity. BioEssays 28, 822–833. [DOI] [PubMed] [Google Scholar]

[R16] Dube U, Del-Aguila JL, Li Z, Budde JP, Jiang S, Hsu S, Ibanez L, Fernandez MV, Farias F, Norton J, et al. (2019). An atlas of cortical circular RNA expression in Alzheimer disease brains demonstrates clinical and pathological associations. Nat. Neurosci. 22, 1903–1912. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] Dudekula DB, Panda AC, Grammatikakis I, De S, Abdelmohsen K, and Gorospe M. (2016). Circinteractome: a web tool for exploring circular RNAs and their interacting proteins and microRNAs. RNA Biol. 13, 34–42. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] Engreitz J, Lander ES, and Guttman M. (2015). RNA antisense purification (RAP) for mapping RNA interactions with chromatin. Methods Mol. Biol. 1262, 183–197. [DOI] [PubMed] [Google Scholar]

[R19] Frazee a.C., Pertea G, Jaffe a.E., Langmead B, Salzberg SL, and Leek JT (2014). Flexible isoform-level differential expression analysis with Ballgown. BioRxiv, 003665. [Google Scholar]

[R20] Gimse K, Gorzek RC, Olin A, Osting S, and Burger C. (2018). Hippocampal Homer1b/c is necessary for contextual fear conditioning and group I metabotropic glutamate receptor mediated long-term depression. Neurobiol. Learn. Mem. 156, 17–23. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] Glažar P, Papavasileiou P, and Rajewsky N. (2014). CircBase: a database for circular RNAs. RNA20, 1666–1670. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] Gruner H, Cortés-López M, Cooper DA, Bauer M, and Miura P. (2016). CircRNA accumulation in the aging mouse brain. Sci. Rep. 6, 38907. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] Guo JU, Agarwal V, Guo H, and Bartel DP (2014). Expanded identification and characterization of mammalian circular RNAs. Genome Biol. 15, 409. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] Hamilton DA, and Brigman JL (2015). Behavioral flexibility in rats and mice: contributions of distinct frontocortical regions. Gene Brain Behav. 14, 4–21. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] Hansen TB, Jensen TI, Clausen BH, Bramsen JB, Finsen B, Damgaard CK, and Kjems J. (2013). Natural RNA circles function as efficient microRNA sponges. Nature 495, 384–388. [DOI] [PubMed] [Google Scholar]

[R26] Hashimoto R, Tankou S, Takeda M, and Sawa A. (2007). Postsynaptic density: a key convergent site for schizophrenia susceptibility factors and possible target for drug development. Drugs Today 43, 645–654. [DOI] [PubMed] [Google Scholar]

[R27] Hoffman GE, Hartley BJ, Flaherty E, Ladran I, Gochman P, Ruderfer DM, Stahl EA, Rapoport J, Sklar P, and Brennand KJ (2017). Transcriptional signatures of schizophrenia in hiPSC-derived NPCs and neurons are concordant with post-mortem adult brains. Nat. Commun. 8, 2225. 10.1038/s41467-017-02330-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] Hu JH, Park JM, Park S, Xiao B, Dehoff MH, Kim S, Hayashi T, Schwarz MK, Huganir RL, Seeburg PH, et al. (2010). Homeostatic scaling requires group I mGluR activation mediated by Homer1a. Neuron 68, 1128–1142. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] Huang WD, Liu XB, Fei Z, Zhang YL, and Yang J. (2012). Down-regulation of Homer1b/c expression protects cultured neurons after traumatic injury. Neural Regen. Res. 7, 2176–2181. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] IBM. (2016). IBM SPSS Advanced Statistics 24 (IBM; ). [Google Scholar]

[R31] Izquierdo A, Brigman JL, Radke AK, Rudebeck PH, and Holmes A. (2017). The neural basis of reversal learning: an updated perspective. Neuroscience 345, 12–26. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] Jeck WR, Sorrentino JA, Wang K, Slevin MK, Burd CE, Liu J, Marzluff WF, and Sharpless NE (2013). Circular RNAs are abundant, conserved, and associated with ALU repeats. RNA 19, 141–157. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] Klugmann M, Symes CW, Leichtlein CB, Klaussner BK, Dunning J, Fong D, Young D, and During MJ (2005). AAV-mediated hippocampal expression of short and long Homer 1 proteins differentially affect cognition and seizure activity in adult rats. Mol. Cell. Neurosci. 28, 347–360. [DOI] [PubMed] [Google Scholar]

[R34] Leeson VC, Robbins TW, Matheson E, Hutton SB, Ron MA, Barnes TRE, and Joyce EM (2009). Discrimination learning, reversal, and set-shifting in first-episode schizophrenia: stability over six years and specific associations with medication type and disorganization syndrome. Biol. Psychiatr. 66, 586–593. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] Li Z, Huang C, Bao C, Chen L, Lin M, Wang X, Zhong G, Yu B, Hu W, Dai L, et al. (2015). Exon-intron circular RNAs regulate transcription in the nucleus. Nat. Struct. Mol. Biol. 22, 256–264. [DOI] [PubMed] [Google Scholar]

[R36] Liang D, and Wilusz JE (2014). Short intronic repeat sequences facilitate circular RNA production. Genes Dev. 28, 2233–2247. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R37] Liu Z, Ran Y, Tao C, Li S, Chen J, and Yang E. (2019). Detection of circular RNA expression and related quantitative trait loci in the human dorsolateral prefrontal cortex. Genome Biol. 20, 99. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] Lominac KD, Oleson EB, Pava M, Klugmann M, Schwarz MK, Seeburg PH, During MJ, Worley PF, Kalivas PW, and Szumlinski KK (2005). Distinct roles for different Homer1 isoforms in behaviors and associated prefrontal cortex function. J. Neurosci. 25, 11586–11594. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] Lorenz R, Bernhart SH, Höner zu Siederdissen C, Tafer H, Flamm C, Stadler PF, and Hofacker IL (2011). ViennaRNA package 2.0. Algorithms Mol. Biol. 6, 26. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R40] Luykx JJ, Giuliani F, Giuliani G, and Veldink J. (2019). Coding and noncoding RNA abnormalities in bipolar disorder. Genes (Basel) 10, 946. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] Mahmoudi E, Fitzsimmons C, Geaghan MP, Shannon Weickert C, Atkins JR, Wang X, and Cairns MJ (2019). Circular RNA biogenesis is decreased in postmortem cortical gray matter in schizophrenia and may alter the bioavailability of associated miRNA. Neuropsychopharmacology 44, 1043–1054. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] Mainen ZF, and Kepecs A. (2009). Neural representation of behavioral outcomes in the orbitofrontal cortex. Curr. Opin. Neurobiol. 19, 84–91. [DOI] [PubMed] [Google Scholar]

[R43] Mann M, Wright PR, and Backofen R. (2017). IntaRNA 2.0: enhanced and customizable prediction of RNA-RNA interactions. Nucleic Acids Res. 45, W435–W439. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] Marquardt K, Josey M, Kenton JA, Cavanagh JF, Holmes A, and Brigman JL (2019). Impaired cognitive flexibility following NMDAR-GluN2B deletion is associated with altered orbitofrontal-striatal function. Neuroscience 404, 338–352. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] McMahon JJ, Miller EE, and Silver DL (2016). The exon junction complex in neural development and neurodevelopmental disease. Int. J. Dev. Neurosci.: Off. J. Int. Soc. Dev. Neurosci. 55, 117–123. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] Mellios N, Sugihara H, Castro J, Banerjee A, Le C, Kumar A, Crawford B, Strathmann J, Tropea D, Levine SS, et al. (2011). MiR-132, an experience-dependent microRNA, is essential for visual cortex plasticity. Nat. Neurosci. 14, 1240–1242. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] Mellios N, Feldman DA, Sheridan SD, Ip JPK, Kwok S, Amoah SK, Rosen B, Rodriguez BA, Crawford B, Swaminathan R, et al. (2018). MeCP2-regulated miRNAs control early human neurogenesis through differential effects on ERK and AKT signaling. Mol. Psychiatr. 23, 1051–1065. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] Memczak S, Jens M, Elefsinioti A, Torti F, Krueger J, Rybak A, Maier L, Mackowiak SD, Gregersen LH, Munschauer M, et al. (2013). Circular RNAs are a large class of animal RNAs with regulatory potency. Nature 495, 333–338. [DOI] [PubMed] [Google Scholar]

[R49] Mullins N, Forstner AJ, O’Connell KS, Coombes B, Coleman JRI, Qiao Z, Als TD, Bigdeli TB, Børte S, Bryois J, and Charney AW (2021). Genome-wide association study of more than 40,000 bipolar disorder cases provides new insights into the underlying biology. Nat. Genet. 53, 817–829. 10.1038/s41588-021-00857-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] O’Donnell LA, Deldin PJ, Pester B, McInnis MG, Langenecker SA, and Ryan KA (2017). Cognitive flexibility: a trait of bipolar disorder that worsens with length of illness. J. Clin. Exp. Neuropsychol. 39, 979–987. [DOI] [PubMed] [Google Scholar]

[R51] Piwecka M, Glazar P, Hernandez-Miranda LR, Memczak S, Wolf SA, Rybak-Wolf A, Filipchyk A, Klironomos F, Cerda Jara CA, Fenske P, et al. (2017). Loss of a mammalian circular RNA locus causes miRNA deregulation and affects brain function. Science 357, eaam8526. [DOI] [PubMed] [Google Scholar]

[R52] Rao A, and Steward O. (1991). Evidence that protein constituents of postsynaptic membrane specializations are locally synthesized: analysis of proteins synthesized within synaptosomes. J. Neurosci. 11, 2881–2895. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R53] Rudebeck PH, and Murray EA (2008). Amygdala and orbitofrontal cortex lesions differentially influence choices during object reversal learning. J. Neurosci. 28, 8338–8343. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] Rybak-Wolf A, Stottmeister C, Glažar P, Jens M, Pino N, Giusti S, Behm M, Bartok O, Ashwal-Fluss R, Herzog M, et al. (2014). Circular RNAs in the mammalian brain are highly abundant, conserved, and dynamically expressed. Mol. Cell 58, 870–885. [DOI] [PubMed] [Google Scholar]

[R55] Salzman J, Gawad C, Wang PL, Lacayo N, and Brown PO (2012). Circular RNAs are the predominant transcript isoform from hundreds of human genes in diverse cell types. PLoS One 7, e30733. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R56] Schloesser RJ, Huang J, Klein PS, and Manji HK (2008). Cellular plasticity cascades in the pathophysiology and treatment of bipolar disorder. Neuropsychopharmacology 33, 110–133. [DOI] [PubMed] [Google Scholar]

[R57] Squassina A, Manchia M, Chillotti C, Deiana V, Congiu D, Paribello F, Roncada P, Soggiu A, Piras C, Urbani A, et al. (2013). Differential effect of lithium on spermidine/spermine N1-acetyltransferase expression in suicidal behaviour. Int. J. Neuropsychopharmacol. 16, 2209–2218. [DOI] [PubMed] [Google Scholar]

[R58] Szumlinski KK, Lominac KD, Kleschen MJ, Oleson EB, Dehoff MH, Schwartz MK, Seeberg PH, Worley PF, and Kalivas PW (2005). Behavioral and neurochemical phenotyping of Homer1 mutant mice: possible relevance to schizophrenia. In Genes, Brain and Behavior, (Genes Brain Behav), pp. 273–288. [DOI] [PubMed] [Google Scholar]

[R59] Szumlinski KK, Kalivas PW, and Worley PF (2006). Homer proteins: implications for neuropsychiatric disorders. Curr. Opin. Neurobiol. 16, 251–257. [DOI] [PubMed] [Google Scholar]

[R60] Tiruchinapalli DM, Caron MG, and Keene JD (2008). Activity-dependent expression of ELAV/Hu RBPs and neuronal mRNAs in seizure and cocaine brain. J. Neurochem. 107, 1529–1543. [DOI] [PubMed] [Google Scholar]

[R61] Torres M, Becquet D, Guillen S, Boyer B, Moreno M, Blanchard MP, Franc JL, and François-Bellan AM (2018). RNA pull-down procedure to identify RNA targets of a long non-coding RNA. J. Vis. Exp. 2018, 134. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] Torrey EF, Webster M, Knable M, Johnston N, and Yolken RH (2000). The Stanley foundation brain collection and neuropathology consortium. Schizophr. Res. 44, 151–155. [DOI] [PubMed] [Google Scholar]

[R63] Waltz JA, and Gold JM (2007). Probabilistic reversal learning impairments in schizophrenia: further evidence of orbitofrontal dysfunction. Schizophr. Res. 93, 296–303. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] Wang R, Li J, Jia R, Pan Y, and Liang H. (2018). EIF4A3-induced circular RNA MMP9 (circMMP9) acts as a sponge of miR-124 and promotes glioblastoma multiforme cell tumorigenesis. Mol. Cancer 17, 166. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R65] Wegbreit E, Cushman GK, Weissman AB, Bojanek E, Kim KL, Leibenluft E, and Dickstein DP (2016). Reversal-learning deficits in childhood-onset bipolar disorder across the transition from childhood to young adulthood. J. Affect. Disord. 203, 46–54. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R66] Xu N, Li AD, Ji LL, Ye Y, Wang ZY, and Tong L. (2019). MiR-132 regulates the expression of synaptic proteins in app/ps1 transgenic mice through C1q. Eur. J. Histochem. 63, 69–76. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R67] Yao G, Niu W, Zhu X, He M, Kong L, Chen S, Zhang L, and Cheng Z. (2019). Hsa-circRNA-104597: a novel potential diagnostic and therapeutic biomarker for schizophrenia. Biomark. Med. 13, 331–340. [DOI] [PubMed] [Google Scholar]

[R68] Yates AD, Achuthan P, Akanni W, Allen J, Allen J, Alvarez-Jarreta J, Amode MR, Armean IM, Azov AG, Bennett R, et al. (2020). Ensembl 2020. Nucleic Acids Res. 48, D682–D688. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R69] You X, Vlatkovic I, Babic A, Will T, Epstein I, Tushev G, Akbalik G, Wang M, Glock C, Quedenau C, et al. (2015). Neural circular RNAs are derived from synaptic genes and regulated by development and plasticity. Nat. Neurosci. 18, 603–610. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R70] Zhang XO, Wang HB, Zhang Y, Lu X, Chen LL, and Yang L. (2014). Complementary sequence-mediated exon circularization. Cell 159, 134–147. [DOI] [PubMed] [Google Scholar]

[R71] Zhang Y, Du L, Bai Y, Han B, He C, Gong L, Huang R, Shen L, Chao J, Liu P, et al. (2020). CircDYM ameliorates depressive-like behavior by targeting miR-9 to regulate microglial activation via HSP90 ubiquitination. Mol. Psychiatr. 25, 1175–1190. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R72] Zheng X, Huang M, Xing L, and Yang R. (2020). The circRNA circSEPT9 mediated by E2F1 and EIF4A3 facilitates the carcinogenesis and development of triple-negative breast cancer. Mol. Cancer 19, 73. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R73] Zimmerman AJ, Hafez AK, Amoah SK, Rodriguez BA, Dell’Orco M, Lozano E, Hartley BJ, Alural B, Lalonde J, Chander P, et al. (2020).A psychiatric disease-related circular RNA controls synaptic gene expression and cognition. Mol. Psychiatr. 24, 2712–2727. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

A bidirectional competitive interaction between circHomer1 and Homer1b within the orbitofrontal cortex regulates reversal learning

Alexander K Hafez

Amber J Zimmerman

Grigorios Papageorgiou

Jayapriya Chandrasekaran

Stephen K Amoah

Rixing Lin

Evelyn Lozano

Caroline Pierotti

Michela Dell’Orco

Brigham J Hartley

Begüm Alural

Jasmin Lalonde

John Matthew Esposito

Sabina Berretta

Alessio Squassina

Caterina Chillotti

Georgios Voloudakis

Zhiping Shao

John F Fullard

Kristen J Brennand

Gustavo Turecki

Panos Roussos

Roy H Perlis

Stephen J Haggarty

Nora Perrone-Bizzozero

Jonathan L Brigman

Nikolaos Mellios

SUMMARY

In brief

Graphical abstract

INTRODUCTION

RESULTS

Changes in relative HOMER1 mRNA isoform levels are differentially associated with deficits in circHomer1 expression

Figure 1. Inverse correlation between changes in circHomer1 and relative HOMER1B expression in the OFC of subjects with BD and SCZ OFC.

Figure 2. Inverse correlation between changes in circHomer1 and relative HOMER1B expression in iPSC-derived neuronal cultures from subjects with BD and SCZ.

circHomer1 and Homer1b mRNA inhibit each other’s synaptic expression within the OFC

Figure 3. HOMER1B and circHomer1 inhibit each other’s synaptic localization in mouse OFC.

CircHomer1 directly binds to the 3′ UTR of Homer1b mRNA

Figure 4. HOMER1B directly binds to circHomer1.

In vivo KD of Homer1b improves OFC-mediated chance reversal learning

Figure 5. HOMER1B and circHomer1 co-KD results in a rescue of circHomer1-mediated alterations in synaptic gene expression and chance reversal learning.

In vivo double KD of Homer1b and circHomer1 rescues OFC-mediated chance reversal learning

Experience-dependent and opposing changes in circHomer1 and Homer1b mRNA levels in the OFC during chance reversal learning

In vivo double KD of Homer1b and circHomer1 normalizes gene expression within the OFC

EIF4A3 binds to circHomer1 and can positively regulate its expression

Figure 6. EIF4A3 binds to circHomer1 and upregulates its expression.

DISCUSSION

Limitations of the study

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Materials availability

Data and code availability

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

Postmortem samples

METHOD DETAILS

RNA extraction and mRNA/circRNA quantification

In silico circRNA RNA predictions and circRNA-Antisense purification

EIF4a3 KD and pharmacological inhibition

HEK293 embryonic kidney human cell line experiments

SH-SY5y human glioblastoma cell line and EIF4A3 KD

Neuro2a mouse neuroblastoma cell line and EIF4A3 inhibition

Primary mouse cortical neurons and EIF4A3 inhibition

OFC crude synaptosome extractions

RNA sequencing in OFC samples with circHomer1/Homer1b double-KD

Generation of shRNAs and lentiviruses in mouse neuronal cultures

Absolute quantification of gene expression via digital droplet PCR

In vivo KD of circHomer1/Homer1b in mouse OFC and touch-based reversal learning

OFC extraction across discrimination and reversal stages in wild-type C57BL6/J mice

Lymphoblastoid cell lines

Induced pluripotent stem cell (iPSC)-derived neuronal cultures

QUANTIFICATION AND STATISTICAL ANALYSIS

Supplementary Material

Highlights.

ACKNOWLEDGMENTS

Footnotes

REFERENCES