Genetic inactivation of Translin/Trax RNase disrupts miRNAs, hippocampal synaptic plasticity, and memory

Summary

Neurons utilize RNA interference in the reversible translational repression of synaptically localized mRNAs, enabling rapid translation in response to synaptic activity. Two evolutionarily conserved proteins, Translin and Trax, form an RNase complex that processes miRNAs, tRNAs, and siRNAs. To determine the specific role of the RNase activity of this complex in brain function, we employed a mouse line harboring a point mutation in Trax (E126A) that renders the Translin/Trax RNase inactive. At the molecular level, we found alterations in the levels of multiple small RNAs, including miRNAs, tsRNAs, and substantial downregulation of gene expression at the mRNA level in the hippocampus of TraxE126A mice. At the synaptic level, TraxE126A mice exhibit deficits in specific forms of long-term hippocampal synaptic plasticity. At the behavioral level, TraxE126A mice display impaired long-term spatial memory and altered open-field behavior. These studies reveal the functional role of Translin/Trax RNase in the mammalian brain.

Graphical abstract

Highlights

• •Evolutionarily conserved proteins, Translin and Trax, form an RNase complex
• •Mutant mice with RNase-dead complex show altered hippocampal miRNA levels
• •Mutant mice show impaired hippocampal gene expression, plasticity, and behavior
• •Translin/Trax RNase is an important regulator of miRNA-mediated gene expression

Genetics; Molecular biology; Neuroscience; Omics

Introduction

Gene regulation plays a critical role in long-lasting forms of synaptic plasticity and long-term memory. Gene regulation occurs at multiple levels, including transcriptional, post-transcriptional, translational, post-translational, and epigenetic regulation. Identifying the molecular players orchestrating these processes will help us better understand the mechanisms and develop strategies to target them. RNA interference (RNAi) is one such mechanism used by cells for post-transcriptional gene silencing. In RNAi, small interfering RNAs (siRNAs) produced by processing double-stranded RNA (dsRNA) are loaded into an effector complex called RNA-induced silencing complex (RISC) that contains target messenger RNA (mRNA), an RNase argonaute-2 (Ago2), and accessory proteins, which act to degrade or silence target mRNAs.^1,2,3 Similarly, microRNAs (miRNAs), upon cleavage from their precursor forms, also take part in post-transcriptional gene regulation by binding to the 3′-untranslated regions (3′-UTRs) of the target mRNAs, thereby leading to their translational repression or degradation.^4,5,6,7,8 Neurons utilize this mechanism of miRNA-mediated translational repression to traffic and localize mRNAs in the vicinity of synapses, where, in response to specific synaptic activity, miRNAs are rapidly degraded to allow the translation of a select set of mRNAs.^{9,10,11,12,13,14,15} The presence of numerous miRNAs and translationally silent mRNAs near the synapses supports this.^9,16,17,18 Several proteins involved in RNA trafficking, stabilization, compartmentalization, activity-mediated signaling, and translational activation have been shown to regulate this process.^{15,18,19,20,21}

The Translin/Trax complex (TN/TX), formed by Translin (also called TBRBP; testis, brain RNA-binding protein) and Trax (Translin-associated factor X) proteins, is an emerging player in the regulation of post-transcriptional gene expression.^14,22,23,24 Both Translin and Trax are evolutionarily conserved, possess nucleic acid binding capability, and are implicated in several biological processes such as DNA damage repair,^25,26 homologous recombination,^27,28 cell division and proliferation,^29,30 genome stability,³¹ spermatogenesis,³² RNA trafficking and nucleic acid metabolism.^33,34,35 Apart from each other, they also have other interaction partners such as adenosine A2A receptor,³⁶ phospholipase C beta (PLCβ),³⁷ and glycogen synthase kinase 3 beta (GSK3β).²⁶

A key feature of the TN/TX complex is that it possesses endonuclease activity against single-stranded RNA (ssRNA), circular RNA (circRNA), and DNA.^23,27,38,39 The complex is also called C3PO (component 3 promoter of RISC) because in Drosophila and human systems, it was found to enhance the activity of RISC by facilitating endoRNase cleavage of the siRNA passenger strand.^23,24 However, it was subsequently found that the RNase activity of the TN/TX complex can cleave many precursor and mature miRNAs by targeting mismatches in the stem region.^14,15,39,40 Studies from us and others have shown that the TN/TX complex indeed degrades a select set of miRNAs to reverse the translational silencing of specific mRNA transcripts in hippocampal neurons,⁴¹ in cerebellum,⁴² in striatum,^43,44 and in vascular smooth muscle cells.^45,46 Thus, these findings suggest that the TN/TX complex appears to have opposite effects on silencing mediated by miRNAs and siRNAs. Furthermore, TN/TX RNase also plays a role in tRNA processing, where it removes 5′ pre-tRNA fragments after the processing of pre-tRNAs by RNase P.³⁴ Growing evidence seems to suggest that small RNA fragments derived from tRNAs could have a role in translational regulation.^47,48

Given the prominent expression of both Translin and Trax in the mammalian brain,^40,49 several studies have investigated their role in the nervous system and revealed that they are involved in mechanisms underlying synaptic plasticity and memory,^14,41,50,51 normal as well as addictive behavior.^44,51,52 Trax has also been proposed as a candidate susceptibility gene for psychiatric disorders^53,54,55,56 and to provide neuroprotection in Huntington’s disease.⁴³ Important clues to the functional role of Translin and Trax proteins have been uncovered from studies using a mouse model that constitutively lacks Translin.^{41,50,51,52,57} Interestingly, in the brain, constitutive deletion of Translin led to the degradation of Trax at the protein level,⁵² highlighting a crucial role of the TN/TX complex. Therefore, it is important to assess the specific role of TN/TX RNase activity in neuronal function without affecting other functions of Translin and Trax. Although the complex is heterooligomeric, comprising different ratios of Translin and Trax subunits,³⁸ the RNase activity of the complex is exclusively conferred by the Trax subunits and studies have identified three key acidic residues E126, E123 and D204 within Trax, each of which disrupts the catalytic activity when mutated.²³ Introduction of the E126A point mutation in Trax abolishes the RNase activity of the TN/TX complex against synthetic miRNA, miR-409, in an in vitro cleavage assay.⁴² We have previously generated a mouse line that constitutively harbors the TraxE126A point mutation, which abolishes the RNase activity of the TN/TX complex without affecting protein expression, stability or the ability of Translin and Trax to interact with each other.⁴² In this study, we utilized this point mutant mouse line to characterize the impact of impairing TN/TX RNase activity on various classes of small RNAs in the hippocampus and the subsequent alterations in synaptic plasticity, behavior, and long-term memory.

Results

Expression of multiple small RNA species, including miRNAs, is altered in the hippocampus of mice carrying RNase-dead Trax

We first assessed the impact of abolishing the RNase activity of the TN/TX complex on the expression of small RNAs in the hippocampus of TraxE126A point mutant mice.⁴² Based on our previous work with TN/TX,⁴¹ our primary interest was in examining the role of TN/TX RNase on the expression of miRNAs in the hippocampus. We performed miRNA sequencing using total RNA extracted from the hippocampus of naive adult male mice and found that the expression of multiple mature miRNAs was significantly altered in the TraxE126A mutants compared to their wildtype littermates (Figure 1A and File S1). Previous studies have shown that the TN/TX complex is involved in the degradation of precursor miRNAs (pre-miRNAs).^15,39,46,58 However, we were not able to assess the alterations in pre-miRNA levels from our small RNA sequencing data because the kit used for cDNA library preparation was based on Dicer recognition sequences. Therefore, we analyzed the same total RNA samples using a small RNA microarray, which utilizes direct end labeling and allows us to determine the expression of multiple small RNA species, including mature miRNAs, pre-miRNAs, small nucleolar RNAs (snoRNAs), mature transfer RNAs (tRNAs), and tRNA-derived small RNAs (tsRNAs). Microarray analysis also revealed that the expression of multiple mature miRNAs was altered in the hippocampus of TraxE126A mutants (Figure 1B and File S2). Twelve miRNAs (10 upregulated: miR-412-5p, miR-496a-3p, miR-409-3p, miR-409-5p, miR-330-3p, miR-330-5p, miR-128-2-5p, miR-667-3p, miR-378a-5p, miR-125b-2-3p; 2 downregulated: miR-3068-3p and miR-143-3p) were identified both in the miRNA sequencing and in the microarray analysis. Surprisingly, microarray analysis identified no pre-miRNA to be significantly upregulated and only one pre-miRNA (pre-miR-378b) to be significantly downregulated in the hippocampus of TraxE126A mutants (Figure 1C and File S3). Among snoRNAs, only two (Mir1949–201 and Gm24598-201) were found to be significantly upregulated in the TraxE126A mutants (Figure 1D and File S3). Microarray analysis of mature tRNAs identified five tRNAs to be significantly upregulated in the TraxE126A mutants, and none were found to be downregulated (Figure 1E and File S3). The largest alteration among the small RNA species investigated using microarray analysis was found in the tsRNA species with 91 tsRNAs significantly upregulated and 22 tsRNAs significantly downregulated in the hippocampus of TraxE126A mutants (Figure 1F and File S3). Interestingly, almost all the significantly altered tsRNAs were 5′-fragments or halves.

Figure 1.

Expression of many small RNA species, including mature microRNAs, is altered in the hippocampus of TraxE126A mutant mice

Volcano plots show the mature miRNAs differentially expressed in the TraxE126A mutants compared to wildtype littermates as identified by (A) miRNA sequencing and (B) miRNA microarray. The 12 common miRNAs identified in both approaches to be significantly altered are highlighted. Full lists of miRNAs are provided in File S1 and S2. Volcano plots show the differentially expressed small RNA species identified using microarray analysis, including (C) precursor miRNAs (pre-miRNAs), (D) small nucleolar RNAs (snoRNAs), (E) mature tRNAs, and (F) tRNA-derived small RNAs (tsRNAs). In all the plots, the upregulated and downregulated miRNAs (false discovery rate, FDR <0.050 and log₂fold change ≥0.200) are highlighted in red and blue, respectively. (TraxE126A, n = 4; WT, n = 5, all males). Largest changes were seen in tsRNA levels (majority are 5′-fragments) and mature miRNAs. Full lists of all the pre-miRNAs, snoRNAs, tRNAs, and tsRNAs are provided in File S3.

These results suggest that the RNase activity of the TN/TX complex could be involved in the regulation of multiple small RNA species, especially miRNAs and tsRNAs.

MicroRNAs regulated by TN/TX RNase activity target transcripts predicted to be involved in synaptic plasticity, neuronal function, and cell adhesion

Previous studies have provided evidence to show that the TN/TX complex is involved in the degradation of miRNAs and thereby relieves the translational repression of the mRNA targets of those miRNAs in multiple tissues, including brain.^{15,39,42,46,58} Our own work has shown that the global deletion of Translin and Trax leads to impairments in hippocampus-dependent memory and synaptic plasticity,^41,50 at least in part due to accumulation of a set of miRNAs which repress the activity-induced translation of activin receptor type 1C (Acvr1c) mRNA.⁴¹ Therefore, we focused our further investigation in this study on miRNAs.

We compared the mature miRNAs identified to be significantly altered in our small RNA sequencing analysis (Figure 1A) and the microarray analysis (Figure 1B). This revealed 12 common miRNAs (10 upregulated and 2 downregulated) that were identified in both the analyses (Figure 2A). Because of the greater confidence in these 12 common miRNAs as being regulated by TN/TX RNase activity, in this study, we focused on them for downstream target prediction. We used the public miRNA target prediction database miRDB⁵⁹ to identify mRNAs predicted to be targeted by the 12 common miRNAs. We only included targets with miRDB Target Score ≥60, indicating higher confidence. Although there were some shared targets between different miRNAs, a significant proportion of the targets were distinct for each miRNA (Figure 2B and File S4). We then used the combined list of predicted targets (miRDB Target Score ≥60; total 2921 targets) of all the 12 common miRNAs to perform functional enrichment and annotation analysis using the DAVID database.^60,61 Top KEGG pathways that showed significant enrichment (ranked by adjusted P value) included MAPK signaling, axon guidance, long-term depression, PI3K-Akt signaling, focal adhesion, Ras signaling pathway, among others (Figure 2C and File S4). Top Gene Ontology (GO) Biological Process terms that showed significant enrichment (ranked by adjusted P value) included the regulation of transcription by RNA polymerase II, nervous system development, and cell adhesion, among others (Figure 2D and File S4). We validated the expression levels of some of the predicted targets with prior evidence for involvement in neuronal function and plasticity using qRT-PCR and found that only some of these selected targets were differentially expressed in the hippocampus of TraxE126A mice compared to WT (Figure S1). We then looked at a public resource webpage “CNS MicroRNA Profiles,” which contains data for the expression of miRNAs in different neuronal cell types in adult mouse neocortex^62,63 extracted expression data for these 12 miRNAs. These data show that among the 10 common miRNAs, 4 miRNAs (409-3p, 128-2-5p, 667-3p, 378a-5p) are expressed in excitatory neurons (CamkIIa+); 5 miRNAs (409-3p, 409-5p, 330-3p, 330-5p, 667-3p) are expressed in Gad2+ neurons which are predominantly GABAergic inhibitory neurons; 5 miRNAs (496a-3p, 330-3p, 330-5p, 667-3p, 378a-5p) are expressed in parvalbumin (Pv+) interneurons and 6 miRNAs (412-5p, 409-5p, 330-3p, 330-5p, 667-3p, 125b-2-5p) are expressed in somatostatin (Som+) interneurons (Figures S2A–S2J). Out of the 2 common downregulated miRNAs, only 3068-3p was expressed in excitatory neurons (CamkIIa+) and parvalbumin (Pv+) interneurons, whereas the expression of 143-3p was very low in all the cell types (Figure S2K–S2L).

Figure 2.

The predicted targets of a subset of mature miRNAs differentially expressed in the hippocampus of TraxE126A mutants show functional enrichment for pathways involved in neuronal function and plasticity

(A) Venn diagram shows the overlap between mature miRNAs identified using miRNA sequencing and microarray analysis (with FDR<0.050 and log₂fold change ≥0.200). A total of 12 miRNAs (10 upregulated and 2 downregulated) were found to be common and were used for target prediction using the miRDB database. A full list of miRDB targets of the 12 miRNAs is provided in File S4. Also see Figure S2 for neuronal cell type-specific expression data for these miRNAs.

(B) An upset plot shows the shared and unique predicted mRNA target profiles in the miRDB database for the 12 common miRNAs. Only targets with miRDB Target Score ≥60 are included (Also see Figure S1).

(C) Top 15 KEGG pathways and (D) Gene Ontology (GO) Biological Process terms from the functional enrichment analysis of the predicted targets of the 12 common miRNAs performed using the DAVID database (File S4).

mRNAs regulated by TN/TX RNase activity show enrichment for pathways involved in immune response, extracellular matrix signaling, and cell adhesion

Because it is difficult to individually validate the huge number of predicted targets of the selected miRNAs, next, we performed mRNA sequencing from the same total RNA samples that were used for small RNA sequencing and microarray analysis. We reasoned that this would help us identify the overlap between the predicted targets of the miRNAs and changes in mRNA expression levels. The mRNA sequencing analysis revealed alterations in the expression of a significant number of genes, with a large proportion of them being downregulated (Figure 3A and File S5). We then performed functional pathway enrichment analysis on the differentially expressed genes using the DAVID database. Top KEGG pathways that showed significant enrichment (ranked by adjusted P value) included ECM-receptor interaction, response to viral infection, focal adhesion, PI3K-Akt signaling pathway, complement and coagulation cascades, and cell adhesion, among others (Figure 3B and File S5). Top GO-Biological Process terms that showed significant enrichment (ranked by adjusted P value) included immune response and related signaling, cell adhesion, integrin-mediated signaling, and extracellular matrix organization, among others (Figure 3C and File S5). Top GO-Molecular Function terms that showed significant enrichment (ranked by adjusted P value) included protein binding, integrin binding, extracellular matrix components, nucleotide and ATP-binding, and kinase activity, among others (Figure 3D and File S5).

Figure 3.

Abolishing the RNase activity of Trax leads to significant alterations in the expression of genes involved in immune function, extracellular matrix, and integrin-mediated cell signaling in the hippocampus

(A) Volcano plot shows the differentially expressed mRNAs in the hippocampus of TraxE126A mutants (n = 4) compared to wildtype littermates (n = 5). The upregulated and downregulated mRNAs (FDR<0.050 and log₂fold change ≥0.263) are highlighted in red and blue, respectively. A full list of mRNAs analyzed is provided in File S5.

(B) Top 15 KEGG pathways from the functional enrichment analysis of the differentially expressed mRNAs performed using the DAVID database.

(C) Top 15 Gene Ontology (GO) Biological Process and (D) Molecular Function terms from the functional enrichment analysis of the differentially expressed mRNAs performed using the DAVID database. Also see File S5 for details of the DAVID functional enrichment analysis.

Together, these results indicate that abolishing the RNase activity of the TN/TX complex leads to significant alterations in the expression levels of genes involved in many signaling pathways, such as extracellular matrix-mediated signaling, focal adhesion, cell adhesion, PI3K-Akt signaling, and immune response. Some of these pathways were also annotated to be regulated by the predicted gene targets of 12 miRNAs altered in the TraxE126A mice (Figure 2C and File S4).

miRNAs targeted by TN/TX RNase regulate the expression of genes involved in focal adhesion, integrin signaling, and extracellular matrix in the hippocampus

To identify the extent of overlap between the predicted targets (2921) of the 12 common differentially expressed miRNAs and the mRNAs (525) that we identified as significantly altered in the TraxE126A mutants, we compared these lists (Figure 4A). Surprisingly, only a small number of the predicted targets (63 out of 2921) were identified as significantly altered in the mRNA sequencing analysis. The primary focus of our study was to understand the impact of the abolished RNase activity of the TN/TX complex, which is known to degrade miRNAs. Therefore, we hypothesized that the miRNAs targeted by TN/TX RNase would be upregulated in the TraxE126A mutants, which in turn, would bind to 3′-UTRs of their target mRNAs, leading to translational repression or degradation of those targets.^4,5,6,7,8 Accordingly, we were specifically interested in the predicted targets of the 10 upregulated miRNAs, which were also found to be downregulated at the mRNA level. We found 43 of the predicted targets of the upregulated miRNAs to be significantly downregulated at mRNA levels (Figure 4A and File S5). We focused on these 43 genes (Figure 4A inset and Table S2) and performed functional enrichment analysis using DAVID (File S5). Focal adhesion was the only pathway that showed significant enrichment in KEGG and WikiPathways databases (Figure 4B). GO term analysis identified integrin binding and protein binding as the enriched Molecular Function terms, and focal adhesion, cell surface, and collagen-containing extracellular matrix as the enriched Cellular Component terms. The GO-Biological Process terms that showed highest enrichment included cell adhesion, integrin-mediated signaling, immune response pathways and receptor tyrosine kinase signaling among others (Figure 4B). Given that the TraxE126A knock-in is global and not cell type-specific, we checked the Allen Brain Cell Atlas database to extract mRNA expression data for these genes in excitatory neurons, inhibitory neurons, astrocytes, and vascular cell types within mouse hippocampus. Out of the 43 target genes, we could find expression data for 24, along with expression data for Translin, but not Trax (Figures S3A–S3E). These data revealed that Translin (Tsn) shows the highest expression in excitatory and inhibitory neurons but is also expressed in astrocytes and vascular cells. Among the 24 target genes, Mtmr10, Lyn, Myrf, Pax, Hpgds, Pcolce2, Dram1 and Bach1 showed expression in excitatory neurons; Mtmr10, Lyn, Pxn, Pcolce2, Dram1 and Bach1 showed expression in inhibitory neurons; Mtmr10, Sfrp1, Lyn, Pxn, Tnc, Slc13a3, Col4a5 and Bach1 showed expression in astrocytes; Sulf1, Ctsc, Mtmr10, Lyn, Pxn, Itga5, Ret, Gpsm3, Col4a5, Gbp6, Ifih1, Bach1, Emp1 and Trpv4 showed expression in vascular cells. We then performed clustering analysis on these 43 targets based on protein-protein interaction data using the STRING database, which identified significantly enriched clustering of genes involved in focal adhesion, extracellular matrix, and integrin signaling (Figure 4C and File S5). Next, we validated the expression of a selected few genes involved in these signaling pathways at the mRNA level using qRT-PCR and confirmed that they are indeed downregulated in the hippocampus of TraxE126A mutants (Figure 4D).

Figure 4.

Components of focal adhesion, integrin signaling, and extracellular matrix are significantly downregulated in the hippocampus of TraxE126A mutant mice

(A) Venn diagram shows the overlap between the predicted targets (miRDB Target Score ≥60) of the 12 common miRNAs upregulated (UP miRNAs) or downregulated (DOWN miRNAs) (FDR<0.050 and log₂fold change ≥0.200) and mRNAs upregulated (mRNAseq UP) or downregulated (mRNAseq DOWN) (FDR<0.050 and log₂fold change ≥0.263) in the TraxE126A mutants (see File S5 for lists of genes shown in the Venn diagram). Inset shows the 43 genes that are targets of upregulated miRNAs and are also found to be downregulated in mRNA-seq. No genes were found that were targets of downregulated miRNAs and were upregulated in the mutants (Also see Figure S3 for cell type-specific expression data for 24 of the 43 genes and Table S2 for information on which of the 10 upregulated miRNAs target these 43 genes).

(B) Significantly enriched (adjusted P value <0.050) pathways (KEGG and WikiPathways) and Gene Ontology (GO) terms from the functional enrichment analysis of the 43 downregulated genes that are also targets of the upregulated miRNAs performed using the DAVID database (File S5).

(C) Clustering analysis of these 43 genes using the STRING protein-protein interaction database, showing clustering of genes involved in focal adhesion, integrin signaling, and extracellular matrix. Genes within the cluster are also colored for significant functional enrichment of these terms in GO, KEGG, and WikiPathways databases (File S5).

(D) qPCR validation of the expression of selected genes (Integrin alpha 5, Talin1, Paxilin, Tenascin C, and Plasminogen activator urokinase receptor) involved in focal adhesion, integrin signaling, and extracellular matrix showed significant downregulation in the hippocampus of TraxE126A mice (n = 4, males) compared to WT mice (n = 5, males).

Together, these results indicate that miRNAs targeted for degradation by TN/TX RNase regulate the expression of genes involved in focal adhesion, integrin signaling, and extracellular matrix in the hippocampus.

Abolishing the RNase activity of the TN/TX complex leads to impairments in specific forms of hippocampal synaptic plasticity and synaptic tagging/capture

Previously, we have shown that the global deletion of Translin and Trax proteins leads to impairments in distinct forms of synaptic plasticity in the hippocampal CA1 region.^41,50 However, both Translin and Trax have been shown to have independent functions in addition to their role as a proposed miRNA-degrading complex.^{14,23,25,26,28,33,34,37,40,49} Because the TraxE126A point mutation only abolishes the RNase activity of the complex but affects neither the protein levels of Translin and Trax nor their interaction in the hippocampus,⁴² we assessed its impact on hippocampal synaptic plasticity to determine the specific role of the RNase activity of the TN/TX complex. Using acute hippocampal slices from WT and TraxE126A mutant mice (both males and females), we performed field-EPSP recordings in the CA1 stratum radiatum region by stimulating the Schaffer collateral pathway from CA3. We first assessed basal synaptic transmission using input-output responses and found no changes in the presynaptic fiber volley or fEPSP amplitudes (Figures S4A and S4B). We also measured paired-pulse facilitation (PPF), a very short-term plasticity, and found no significant alterations over a range of inter-stimulus intervals (Figure S4C). We then tested synaptic plasticity in response to distinct stimulation paradigms that induce short-term or long-term LTP and LTD. We found that the slices from TraxE126A mutant mice showed deficits only in the long-lasting LTP induced by the spaced 4-train stimulation paradigm (Figures 5A and 5B) but not by the massed 4-train stimulation paradigm (Figures 5C and 5D). TraxE126A mutant mice showed no impairments either in a short-lasting form of LTP induced by 1-train stimulation (Figures 5E and 5F) or in LTD induced by prolonged low-frequency stimulation (Figures 5G and 5H).

Figure 5.

TraxE126A mutant mice show deficits in distinct forms of hippocampal synaptic plasticity in the hippocampal CA1 region

(A and B) Long-lasting LTP induced by spaced-4-train stimulation protocol is impaired in slices from TraxE126A mutant mice (n = 13 mice, 20 slices; 7 males, 6 females) compared to WT littermates (n = 13 mice, 17 slices; 6 males, 7 females). The mean fEPSP slope over the last 20 min recordings is significantly lower in TraxE126A mutants compared to WT (Welch’s t test, two-tailed, t = 3.307, df = 17.76, p = 0.0040, eta squared = 0.381).

(C and D) Long-lasting LTP induced by massed-4-train stimulation protocol is not impaired in slices from TraxE126A mutant mice (n = 12 mice, 15 slices; 6 males, 6 females) compared to WT littermates (n = 13 mice, 16 slices; 7 males, 6 females). The mean fEPSP slope over the last 20 min recordings is not significantly different in TraxE126A mutants compared to WT (Unpaired t test, two-tailed, t = 0.820, df = 23, p = 0.4205, eta squared = 0.028).

(E and F) Short-lasting LTP induced by 1-train stimulation protocol is identical in slices from TraxE126A mutant mice (n = 12 mice, 14 slices; 6 males, 6 females) compared to WT littermates (n = 13 mice, 14 slices; 6 males, 7 females). The mean fEPSP slope over the last 20 min recordings is not significantly different in TraxE126A mutants compared to WT (Unpaired t test, two-tailed, t = 0.152, df = 23, p = 0.8808, eta squared = 0.0009).

(G and H) LTD induced by 1 Hz, 15 min stimulation protocol shows no significant alterations in slices from TraxE126A mutant mice (n = 15 mice, 24 slices; 6 males, 9 females) compared to WT littermates (n = 14 mice, 24 slices; 7 males, 7 females). The mean fEPSP slope over the last 20 min recordings is not significantly different in TraxE126A mutants compared to WT (Unpaired t test, two-tailed, t = 0.782, df = 27, p = 0.4408, eta squared = 0.022). In all bar graphs, open symbols represent females and closed symbols represent males. Representative fEPSP traces are shown in the insets for each genotype, one sampled at baseline (black trace) and another at the end of the recording (red trace). Calibration bars for fEPSP traces: 2 mV vertical, 5 ms horizontal. The error bars represent SEM in all graphs. Also see Figure S4 for basal transmission and paired-pulse facilitation data.

We next investigated synaptic tagging and capture (STC), a form of associative heterosynaptic plasticity.^64,65,66 STC involves the interaction between local synaptic events (the “tag”) and broader dendritic or somatic events, such as protein synthesis, that supply the plasticity-related products (PRPs) for synapse-strengthening (“capture”).^65,66 This enables the heterosynaptic association between two synaptic plasticity events (e.g., LTP or LTD) that occur within a specific temporal window in synaptic inputs onto a common population of neurons. We have previously found that STC is impaired in the hippocampal slices from mice globally lacking both Translin and Trax proteins.⁴¹ To know whether this impairment is specifically due to the lack of the RNase activity of TN/TX complex, we investigated STC in the hippocampal slices from TraxE126A mutant and WT mice using a similar two-pathway design to stimulate two independent synaptic inputs and recorded fEPSPs in the CA1 stratum radiatum (Figure 6A). In each experiment, we confirmed that the two pathways were independent using the lack of PPF (50 ms ISI) between the pathways as a measure. After 20 min of baseline recording, massed 4-train stimulation was delivered to one pathway (S1) followed 30 min later by 1-train stimulation in the second pathway (S2) (Figure 6B). In the slices from WT mice, 1-train stimulation in S2 results in long-lasting LTP thereby demonstrating intact STC (Figure 6C) whereas in slices from TraxE126A mutant mice, this heterosynaptic facilitation of 1-train LTP into a long-lasting LTP is impaired (Figure 6D). Upon comparison of LTP maintenance between genotypes, massed 4-train LTP in pathway S1 showed no significant difference (Figure 6E) whereas 1-train LTP in pathway S2 was significantly lower in mutants than WT (Figure 6F).

Figure 6.

TraxE126A mutant mice show deficits in synaptic tagging and capture, a form of associative synaptic plasticity, in the hippocampal CA1 region

(A) A schematic representation of a hippocampal slice shows the location of electrodes in the hippocampal CA1 stratum radiatum. Two electrodes, S1 and S2, are used to stimulate two independent Schaffer collateral inputs onto a population of neurons, and the fEPSP responses are recorded with a common recording electrode.

(B) Timeline of the synaptic tagging and capture (STC) experiments. After a stable baseline recording of 20 min in both the inputs, massed 4-train stimulation is delivered to input S1, followed 30 min later by 1-train stimulation in input S2. fEPSPs in both inputs are recorded for 2 h after the 1-train stimulation.

(C) STC in the slices of WT mice (n = 14 mice, 16 slices; 8 males, 6 females) as shown by the successful facilitation of short-lasting LTP to long-lasting LTP in input S2 in response to weak 1-train stimulation.

(D) In the slices of TraxE126A mutant mice (n = 12 mice, 13 slices; 6 males, 6 females), 1-train stimulation in S2 only results in short-lasting LTP, demonstrating impaired STC.

(E) A bar graph of the mean fEPSP slope over the last 20 min recordings in input S1 (massed 4-train) shows no significant difference in potentiation between TraxE126A mutants and WT (Unpaired t test, two-tailed, t = 0.741, df = 24, p = 0.4657, eta squared = 0.022).

(F) A bar graph of the mean fEPSP slope over the last 20 min recordings in input S2 (1-train) shows a significantly lower potentiation in TraxE126A mutant slices compared to WT; (Unpaired t test, two-tailed, t = 3.080, df = 24, p = 0.0051, eta squared = 0.283). In all bar graphs, open symbols represent females and closed symbols represent males. Representative fEPSP traces are shown in the insets for each input at baseline (black trace) and at the end of the recording (red trace). Calibration bars for traces: 2 mV vertical, 5 ms horizontal. The error bars represent SEM in all graphs.

Together, these results show that TraxE126A mice, which lack the RNase activity of the TN/TX complex, exhibit deficits only in specific forms of hippocampal synaptic plasticity, such as spaced 4-train LTP and STC, which phenocopy the synaptic plasticity deficits in global Translin KO mice.^41,50

Mice lacking the RNase activity of the TN/TX complex show altered open-field behavior and impaired long-term spatial memory, but normal contextual fear memory

We next tested the impact of abolishing the RNase activity of the TN/TX complex on certain behavioral measures and long-term memory. We first checked open-field behavior because alterations in this could confound the use of spatial object exploration tasks for long-term memory assessment. To assess open-field behavior, WT and TraxE126A mice (both males and females) were allowed to individually explore an open arena for 6 min, and the time spent in the central and peripheral areas, and total distance traveled were calculated (Figure 7A). TraxE126A mutant mice spent significantly less time in the central area of the arena compared to WT mice (Figure 7A1). We found a sex-specific effect wherein overall, female mice spent significantly less time in the central area of the arena compared to males (Figure 7A2), however, no significant interaction was found between genotype and sex effects. TraxE126A mutant mice also traveled significantly less distance compared to WT mice in the open-field task (Figure 7A3) but neither a significant effect of sex was found (Figure 7A4) nor an interaction between sex and genotype. We then assessed long-term contextual fear memory in these mice where they were allowed to explore a grid-floor context for 3 min during which they received a single foot shock (1.5 mA, 2s). After 24 h, mice were re-introduced to the same context for 5 min, and their freezing levels were measured. Subsequently, freezing levels were also tested in an altered context. Increased freezing in the shocked context compared to pre-shock levels was considered as a proxy for long-term fear memory. The data revealed no significant alteration in contextual fear memory in TraxE126A mice compared to WT mice, no significant sex-specific effects, or an interaction between genotype and sex. The fear conditioning training led to significantly higher freezing levels in both WT and TraxE126A mice, in the trained as well as the altered context, compared to pre-shock freezing levels (Figure 7B). Finally, we performed a spatial version of the Y-maze task involving a single training session and a test session (2-trial Y-maze task). During the training session (10 min), mice were allowed to explore the Y-maze with one of its arms blocked. After 24 h, mice were re-introduced to the maze and allowed to explore all the three arms (5 min test session) during which the number of entries into the novel and familiar arms as well as the time spent exploring each of the arms were measured (Figure 7C). The WT mice made significantly more entries into the novel arm during the test session (Figures 7C1 and S5) whereas TraxE126A mice did not show this preference for the novel arm demonstrating a deficit in long-term spatial memory (Figures 7C1 and S6). We found no significant sex-specific effect in the discrimination index for the novel arm entry (Figure 7C2), as well as no significant interaction between the genotype and sex effects. Intriguingly, the discrimination index for the time spent exploring the novel arm was not significantly different between WT and TraxE126A mice (Figure S7), which could potentially be due to the criteria used for scoring exploration time (Figure S8).

Figure 7.

TraxE126A mutant mice show altered open-field behavior and impaired long-term memory in the spatial Y-maze task, but show no deficits in contextual fear memory

(A) A schematic representation of the Open-Field Task Arena with central and peripheral areas. (A1) TraxE126A mutant mice (n = 22; 12 males, 10 females) show altered open-field behavior as they spend significantly less time in the central area of the arena [two-way ANOVA, F (1,52) = 6.539, p = 0.0135] compared to WT littermate mice (n = 33; 18 males, 15 females). (A2) A significant main effect of sex is also observed in the open-field behavior wherein female mice (n = 25; 15 WT, 10 TraxE126A) spend less time in the central area of the arena [two-way ANOVA, F(1,52) = 6.195, p = 0.0161] compared to male mice (n = 30; 18 WT, 12 TraxE126A). However, there is no significant interaction between sex and genotype (F (1,51) = 0.8903, p = 0.3498. (A3) TraxE126A mutant mice travel significantly less distance [two-way ANOVA, F(1,52) = 16.84, p = 0.0001] compared to WT littermate mice in the open-field task. (A4) There is neither a main effect of sex [two-way ANOVA F(1,52) = 0.0823, p = 0.7753] nor an interaction between sex and genotype [two-way ANOVA, F(1,51) = 0.0792, p = 0.7795] in the total distance traveled in the open-field task.

(B) TraxE126A mutant mice (n = 18; 10 males, 8 females) show no deficits in long-term contextual fear memory compared to WT littermate mice (n = 21; 11 males, 10 females) Three-way repeated measures ANOVA, no main effect of genotype [F(1,35) = 3.254, p = 0.0799]; no main effect of sex [F(1,35) = 1.276, p = 0.2664], main effect of sessions [F(1.852,64.83) = 148.5, p < 0.0001] and no significant interactions among genotype, sessions, and sex [F(1.852,64.83) = 0.5641, p = 0.5586].

(C) A schematic representation of the spatial 2-trial Y-maze Task showing the maze configuration during training and test sessions. (C1) TraxE126A mutant mice show deficits in long-term memory in a spatial 2-trial Y-maze task compared to WT littermates. During the 24h test, the WT mice (n = 24; 17 males, 7 females) preferentially make more entries into the novel arm as indicated by their strong positive discrimination index (DI) for the novel arm over familiar arm, whereas TraxE126A mutant mice (n = 12; 8 males, 4 females) show significantly lower discrimination index for the novel arm entry [two-way ANOVA, F(1,33) = 5.368, p = 0.0269]. (C2) For the novel arm entry DI, there is neither a main effect of sex [F(1,33) = 0.8049, p = 0.3761] nor an interaction between sex and genotype [F(1,32) = 0.8512, p = 0.3631]. (Also see Figures S5–S8 for additional data and details related to the Y-maze task). In all bar graphs, open symbols represent females and closed symbols represent males. The error bars represent SEM in all graphs.

Discussion

The TN/TX complex has been implicated in multiple cellular processes including the processing of miRNAs and pre-miRNAs during RNAi,²³ hippocampal synaptic plasticity,^41,50 vascular stiffness^45,46 and adiposity.⁴² Additionally, Translin and Trax have been shown to have independent functions in homologous recombination and DNA damage repair,^25,67 cell division and proliferation,^29,30 DNA/RNA metabolism,^33,34,35 mesenchymal cell differentiation,^67,68 sperm maturation.³² In this study, we investigated the specific role of the RNase activity of the TN/TX complex in the adult hippocampus using the TraxE126A point mutant mice in which the RNase activity of this complex has been abolished. At the molecular level, we found alterations in the levels of miRNAs in the TraxE126A mice. Forty-three of the predicted targets of these upregulated miRNAs were found to be significantly downregulated at the mRNA level. These genes are involved in focal adhesion, integrin signaling, and extracellular matrix-mediated signaling in the hippocampus. At the synaptic level, we found that TraxE126A mice, which lack the RNase activity of the TN/TX complex, exhibit deficits only in specific forms of hippocampal synaptic plasticity, such as spaced 4-train LTP and STC. At the behavioral level, we observed deficits in long-term spatial memory and behavioral alterations in the TraxE126A mice. These findings show for the first time the functional role of TN/TX RNase activity in the brain.

Our data reveal that the TN/TX RNase activity is involved in the regulation of multiple small RNA species, including miRNAs, tRNAs, and tsRNAs. Inactivating the RNase activity of this complex led to alterations in the levels of many of the small RNAs in the hippocampus of naive mice. Although our hypothesis was that the TN/TX RNase complex acts to degrade precursor and mature miRNAs, we found miRNA levels being both upregulated and downregulated. This could be due to multiple reasons. First, the mouse model used in this study is a constitutive mutant, and it is possible that the alterations in some miRNAs could be a compensatory mechanism during development. Second, changes in some miRNAs could be induced secondary to changes in miRNAs directly targeted by the RNase. Third, TN/TX RNase potentially plays a role in the biogenesis of miRNAs. A direct validation of TN/TX RNase targeting sequences on the miRNAs found to be altered in our study and the use of conditional deletion of the RNase activity would be helpful in this regard, and this can be explored in future experiments. Nevertheless, among the common miRNAs we identified using both small RNA sequencing and microarray, we found a majority (10 out of 12) to be upregulated. Interestingly, some of these upregulated miRNAs (miR-412, miR-496a, miR-330) were also found to be upregulated in the mouse striatum in a previous study following shRNA-mediated downregulation of Trax.⁴³ MiR-409, another miRNA found to be upregulated in the TraxE126A mutants, was validated as a target of the TN/TX RNase in an in vitro cleavage assay.⁴² Additionally, several miRNAs (miR-412, miR-93, miR-7b, miR-1946b) identified to be upregulated in the hippocampus of TraxE126A mice in this study either by sequencing or by microarray were also found to be upregulated in the nucleus accumbens of Translin KO mice in our recent study⁴⁴ while some other miRNAs (miR-409, miR-412, miR-330, miR-370) were found to be upregulated in the cerebellum of Translin KO mice.³⁹ Given that the RNase activity of the complex is conferred by Trax, and that Translin KO mice lack Trax protein, this provides further support that these miRNAs are targeted for degradation by the TN/TX RNase activity.

A surprising finding from our small RNA microarray analysis is the extensive alteration in the levels of tsRNAs in the hippocampus of TraxE126A mutant mice. Strikingly, almost all the significantly altered tsRNAs were derived from the 5′-end of tRNAs. It is important to note that the TN/TX complex (C3PO) has been previously shown to be involved in tRNA maturation, acting as an RNase to remove the 5′ pre-tRNA fragments after the processing of pre-tRNAs by RNase P.³⁴ Therefore, it is possible that the lack of the RNase activity of the TN/TX complex in the E126A mutants leads to the accumulation of the 5′ pre-tRNA-derived fragments. An alternative possibility is that in the absence of the TN/TX RNase activity, other endonucleases cleave the accumulating pre-tRNAs, resulting in increased levels of tsRNAs. Significant upregulation of some mature tRNAs was found in the hippocampus of TraxE126A mutant mice, which was also reported in mouse embryonic fibroblasts derived from Translin KO mice³⁴ (which also lack Trax protein⁵²). Given the evidence that increased levels of 5′-tsRNA fragments can inhibit translational machinery,^47,69 it will be interesting to investigate whether this could contribute to the synaptic plasticity and memory deficits observed in TraxE126A mutants. Evidence is emerging implicating tsRNAs in neuronal physiology and in neurodevelopmental, neurodegenerative, neurobehavioral disorders⁷⁰ as well as metabolic disorders⁷¹ and uncovering the role of TN/TX RNase in tsRNA biology could potentially open up new therapeutic avenues. Furthermore, the TN/TX complex might also target other types of noncoding RNAs not investigated in our study, such as circRNAs, which have been shown to directly interact with 3′-untranslated regions (3′-UTRs) of mRNAs and trigger nonsense-mediated decay.⁷²

Our results show that abolishing the RNase activity of the TN/TX complex leads to impairments in only selected forms of hippocampal synaptic plasticity, such as spaced-tetanic train-induced LTP and STC, whereas a few other forms of LTP and LTD are not affected. This highlights a specific role of the TN/TX RNase in regulating synaptic plasticity. Impairments in spaced-tetanic train-induced LTP and STC are also reported in the CA1 region of Translin KO mice,^41,50 lending support to the role of TN/TX RNase. Some of the mechanisms shown to support heterosynaptic facilitation of synaptic plasticity in STC include remodeling of the actin cytoskeleton and synapse,^73,74 activity-dependent transport and/or translation of local PRPs,^65,75,76 activity-dependent secretion of signaling molecules or cleavage of precursor forms of signaling molecules in the extracellular matrix into their active forms,^77,78,79 and activity-dependent trafficking of signaling molecules, receptors, and ion channels into the synaptic membrane.^80,81,82,83 Impairments in LTP and STC observed in the TraxE126A mutants could be due to alterations in one or more of these mechanisms. Although our current study does not delineate the exact mechanism, it offers hints at several possibilities, including removal of miRNA-mediated translational repression on key target mRNAs, regulation of tRNA maturation and tsRNA-mediated translational control, regulation of ECM components and signaling through integrin and focal adhesions.

Inactivation of the TN/TX RNase activity also led to alterations in open-field behavior. The point mutant mice spent less time in the central area of the open field and also moved less. This could indicate more anxiety or decreased motivation for exploration or possibly deficits in motor function. There was also a sex-specific effect wherein overall, female mice tended to spend less time in the central area of the arena compared to male mice; however, there was no significant interaction of this effect with the genotype. In our previous study, we have shown prominent expression of Translin and Trax in striatal dopaminergic neurons.⁸⁴ It is possible that the loss of TN/TX RNase activity in these neurons might impact motor or motivational phenotypes. Future studies aimed at investigating striatal and cerebellar function in these mutants would prove useful in this regard. Translin-null mice, which lack both Translin and Trax proteins, were reported not to show significant alterations in locomotor activity in the open field,⁵¹ although the trend was toward decreased activity. In the contextual fear conditioning task, Trax point-mutant mice showed a trend for marginally higher freezing levels compared to WT mice, further suggesting an increased anxiety-like phenotype. Interestingly, enhanced contextual freezing was also reported in female Translin KO mice.⁵¹ However, Stein et al.⁵¹ found reduced anxiety-related behavior in Translin KO mice in the elevated zero maze test and light-dark box exploration test. In future experiments, it will be interesting to see if the Trax point mutants show deficits in acoustic startle responses and prepulse inhibition, which are known to be disrupted in several neuropsychiatric disorders.⁸⁵ The Trax global point mutant mice used in this study show reduced muscle mass,⁴² and therefore, are not ideal for examining acoustic startle responses. In future studies, this can be investigated using conditional deletion models with the point mutation restricted to the brain. In the two-trial Y-maze task, we found that the Trax point mutant mice showed weaker discrimination for the novel arm, indicating deficits in long-term spatial memory. Although this task has been mostly used to test working memory at shorter intervals, our study shows the utility of the two-trial Y-maze task as a long-term memory task. In future studies, we will further characterize this task and utilize in the functional target validation studies. Overall, the behavioral alterations in the TraxE126A mutant mice seem to phenocopy Translin KO mice only in specific phenotypes, whereas in some others, they differ. There also seem to be distinct sex-specific differences. One explanation for this discrepancy would be that the complete loss of both Trax and Translin proteins in the Translin KO mice leads to the loss of their other functions as well, whereas in the TraxE126A mice, only the RNase activity of the TN/TX complex is specifically abolished.

For the downstream target prediction and functional enrichment analysis, in this study, we focused only on the 12 miRNAs identified to be significantly altered both in miRNA sequencing and in microarray analysis because of the higher confidence level in these. But it is certainly possible that additional significantly altered miRNAs identified only in the miRNA-sequencing approach (Figure 1A) or in the microarray approach (Figure 1B) are also important, and they will be explored in future studies. To gain insight into the mRNA targets and signaling pathways impacted by the inactivation of the TN/TX RNase, we performed mRNA sequencing from the same samples, which revealed a vast majority of mRNAs to be downregulated, indicating the overall impact of abolished TN/TX RNase activity on gene expression. Functional enrichment analysis of the significantly altered genes showed extracellular matrix-mediated signaling, integrin signaling, focal adhesion, and immune system-related pathways to be predominantly impacted. When we compared the overlap between the high-confidence predicted target mRNAs of the 12 common miRNAs and the significantly altered mRNAs identified in the sequencing approach, we found only 63 target genes, which is a small portion of the total predicted targets (2921) (See Figure 4A). This discrepancy could be due to the following possibilities: (1) majority of these are the targets of other significantly altered miRNAs identified in this study (Figures 1A, 1B, and 2A) but not among the common 12 miRNAs selected for further validation here; (2) a large proportion of the “predicted” targets are not biologically relevant in the naive hippocampus; and (3) alterations in the levels of many of the target mRNAs is an indirect, secondary effect via the regulation of transcription factors, RNA-binding proteins and/or translational regulators. Because the primary focus of this study was to understand the impact of abolishing the RNase activity of the TN/TX complex, for further analysis we only focused on 43 predicted target genes of the 10 common upregulated miRNAs and were also found to be downregulated in in Trax mutants (Figure 4A). Functional annotation of these genes revealed that they also show significant enrichment for focal adhesion, ECM-mediated signaling and integrin signaling pathways. The modest overlap between predicted targets and differentially expressed transcripts could also indicate TN/TX RNase-mediated regulation at the translation level rather than transcriptional. It is also possible that levels of many of these targets are significantly altered only following synaptic activation or learning, which we have not examined in this study. Although we have not directly tested whether abolishing the TN/TX RNase activity impairs global translation, the fact that some forms of long-term synaptic plasticity and long-term memory are not affected in the TraxE126A mutant mice is not consistent with impaired global translation. Our guiding hypothesis is that TN/TX RNase mediates the ability of plasticity stimuli to rapidly reverse constraints on local synaptic translation. Accordingly, novel approaches are needed to determine if predicted targets of miRNAs elevated by TN/TX inhibition are, in fact, translated during or following plasticity events, and if blocking TN/TX impairs their local translation.

Our results show significant downregulation of several ECM components, including many collagens, multiple integrin subunits, tenascin-C (Tnc), and many ECM/integrin/focal adhesion signaling components, such as Lyn, Paxillin (Pxn), Talin1, at the mRNA level in the hippocampus of TraxE126A mutants. Components of ECM modulate the function of local receptors or ion channels and send diffuse molecular signals using products of activity-dependent proteolytic cleavage. Because of their role in the regulation of synaptic plasticity and remodeling involving both neurons and glia, they have been suggested as components of the “tetrapartite synapse.”^86,87 ECM and integrins exhibit extensive crosstalk and are important regulators of several intracellular signaling pathways underlying synaptic plasticity and memory.⁸⁸ Tnc, one of the target genes downregulated in the hippocampus of TraxE126A mutants, is an ECM glycoprotein expressed during the development of the nervous system and its expression is upregulated following synaptic stimulation.^89,90 Mice lacking Tnc show deficits in TBS-LTP and fear extinction but normal spaced-4-train LTP and normal long-term memory in the Morris Water Maze (MWM) task as well as the contextual fear memory task.^91,92,93 Several integrin family members (Itgax, Itgb2, Itgal, Itga5, Itga10, Itga2b, Itgb1) were found to be downregulated in the hippocampus of TraxE126A mutants. Integrin alpha 5 (Itga5) regulates spine and synapse formation in hippocampal neurons via Src kinase and Rac signaling.⁹⁴ Studies have implicated multiple integrins in regulating distinct forms of hippocampal synaptic plasticity and memory.^{95,96,97,98,99,100} Although reduced expression of Itga5 alone does not alter spaced-HFS LTP, mice with the reduced expression of Itga3, Itga5, and Itga8 show impaired LTP and spatial memory in the water maze but normal contextual fear memory.⁹⁸ Additionally, mice with the forebrain-specific deletion of β1-integrin (Itgb1) show normal spatial and contextual fear memory, but impaired working memory in nonmatching-to-place T-maze assay and deficits in spaced 2-train HFS-LTP that was less sensitive to a much stronger, 8-train HFS saturating protocol.⁹⁷ These results seem to closely resemble our findings in the Trax point mutant mice. Abnormal structure of perineuronal nets resulting from deficiency of ECM components could lead to subregion-specific changes in perisomatic GABAergic inhibition, resulting in metaplastic changes to synaptic plasticity.^101,102 Focal adhesions are transmembrane structures that connect the ECM and the actin cytoskeleton to regulate various cellular processes. Integrins are an important part of the signaling through focal adhesions. The ECM-integrin-focal adhesion signaling is capable of relaying both “outside-in” and “inside-out” signaling, thereby acting as a conduit between the interior of the cell and the extracellular environment,¹⁰³ which could be affected in the TraxE126A mutants. Interestingly, STC, a process impaired in Trax mutants, is shown to be regulated by extracellular processing of precursor forms of neurotrophic molecules (such as BDNF) by ECM proteases¹⁰⁴ and activity-dependent remodeling of the actin cytoskeleton, which is closely linked to integrin/focal adhesion signaling.^73,105,106 Our results also reveal the alteration in genes involved in innate immunity and inflammation response. Recent studies have uncovered mechanisms linking innate immunity and long-term memory.^107,108 The study by Jovasevic et al. showed that memory formation is accompanied by increased DNA damage and TLR9-mediated inflammatory signaling.¹⁰⁷ Given the known function of Trax in DNA repair,^26,49 and our results implicating TN/TX RNase in immune response, it will be interesting to investigate whether a combination of these mechanisms might underlie the plasticity and memory deficits in TraxE126A mutant mice. Together, our analyses suggest that alterations in these signaling pathways could contribute to the impairments in hippocampal synaptic plasticity and memory in the TraxE126A mutant mice, and we will focus our investigation on these pathways in future studies. Current methodological limitations do not allow us to simultaneously measure the expression of miRNAs and their target mRNAs in a cell type-specific manner, although recently developed techniques such as CIMERA-Seq¹⁰⁹ might help address this in the future.

In summary, our study offers a comprehensive analysis of the consequences of abolishing the TN/TX RNase activity on various classes of small RNAs and gene expression in the hippocampus, which consequently leads to impairments in synaptic plasticity, behavior, and long-term memory.

Limitations of the study

The TraxE126A point-mutant mouse model used in our study is a constitutive one, and therefore, we cannot rule out the impact of any developmental compensatory mechanisms in the observed effects. Additionally, because the mutant allele is not restricted to any specific cell type or to the hippocampal region in our study, the behavioral phenotypes observed may not be attributed to any specific cell type or to the hippocampal region alone. The identified miRNAs remain to be directly validated as targets of TN/TX RNase, and the alterations in specific target gene expression and implicated pathways need to be validated at the protein level in functional experiments.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Ted Abel (ted-abel@uiowa.edu).

Materials availability

• •This study did not generate new unique reagents.
• •The Trax point-mutant mouse line used in this study (TraxE126A) is cryopreserved at the Jackson Laboratory (C57BL/6-Tsnax^em1Jbar/J; Strain #035903) and is available by cryorecovery.
• •Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Ted Abel, Ph.D (ted-abel@uiowa.edu).

Data and code availability

• •Data: Source data for the main figures are provided in the manuscript or in the supplemental information. The miRNA and mRNA sequencing datasets generated in this study have been deposited at the Gene Expression Omnibus (GEO) database with the accession number GSE300964. The accession number is also provided in the key resources table.
• •Code: The code used for generating the figures representing the cell type-specific target mRNA expression data extracted from Allen Brain Cell Atlas is available on GitHub as “createCellTypeGeneFigure.py” under https://github.com/zjpeters/traxTranslinABCAnalysis. The link to this page is also listed in the key resources table.
• •Additional information: Any additional information required to reanalyze the data reported in this paper is available from the lead contact (Ted Abel, Ph.D., at ted-abel@uiowa.edu) upon request.

Acknowledgments

We thank members of the Abel Lab and Baraban Lab for their help with the project and for helpful discussions. We thank Jia Ern Ong, Achala Thippeswamy, Deaven Denis, Jacob Belardo, and Quinn Truax for their help with colony maintenance and genotyping in the Abel lab. We thank the Iowa Institute of Human Genomics (IIHG) Core facility for help with miRNA and mRNA sequencing. We thank the Neural Circuits and Behavioral Core (NCBC) at the University of Iowa for the use of their facilities for behavioral experiments. Funding sources: This work was supported by funding from the National Institutes of Health R01 MH 087463 to T.A. T.A. is also supported by the Roy J. Carver Charitable Trust.

Author contributions

M.S.S., J.M.B., and T.A. designed the study. M.S.S. performed the electrophysiology experiments and hippocampal sample collection, behavioral experiments, and data analysis. J.K. performed RNA isolation and qRT-PCR experiments. M.S.S. and J.K. performed RNA sequencing data analysis and functional annotation. X.F. and J.M.B. generated the TraxE126A mouse model. M.C.L. performed some of the behavioral experiments and helped analyze the open-field data. S.M.T. performed the preliminary western blotting experiments and helped with manuscript revision. T.C.C. helped with some behavioral experiments and colony maintenance. Z.P. extracted the single-cell target gene expression data from the Allen Brain Single-Cell Atlas. K.C. performed statistical analysis of the behavioral data. M.S.S. wrote the manuscript with inputs from other authors. M.S.S., J.K., and S.M.T. prepared the graphical abstract with input from others. All the listed authors have read and approved the final version of the manuscript.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
miRVana miRNA Isolation Kit with phenol	Thermo Fisher Scientific	Cat # AM1560
DNase I	Zymo	Cat # E1010
Illumina TruSeq Small RNA Library Preparation Kit	Illumina	Cat # RS-200
Illumina Stranded mRNA Library Preparation Kit	Illumina	Cat # 20040532
iScript cDNA Synthesis Kit	Bio-Rad	Cat # 1708890
SYBR™ Green Universal Master Mix	Thermo Fisher	Cat # 4309155
Deposited data
Trax miRNA sequencing data	This study	[Gene Expression Omnibus (GEO)]: [GSE300964]
Trax mRNA sequencing data	This study	[Gene Expression Omnibus (GEO)]: [GSE300964]
Experimental models: Organisms/strains
C57BL/6-Tsnaxem1Jbar/J	Jackson Lab	strain # 035903
Oligonucleotides
See Table S1 for the qPCR primers	Integrated DNA Technology (IDT), Coralville, IA, USA	N/A
Software and algorithms
GraphPad Prism	GraphPad	N/A
BioRender	BioRender	N/A
Adobe Illustrator 2024	Adobe	N/A
Axon pCLAMP 10 Suite (Clampex and Clampfit)	Molecular Devices, CA	Axon pCLAMP
Partek Flow	Partek, Illumina, CA	Partek Flow
ExpressAnalyst110	Express Analyst	ExpressAnalyst
Venn Diagram Maker	Good Calculators	Venn Diagram Maker | Good Calculators
STRING111	STRING Consortium 2025	STRING: functional protein association networks
Ethovision XT	Noldus, Leesburg, VA, USA	EthoVision XT | Video tracking software | Noldus
BORIS Behavioral Analysis Software	Olivier Friard, University of Torino, Italy.	BORIS
miRDB	miRNA target prediction and functional annotations	miRDB - MicroRNA Target Prediction Database
FreezeScan System	CleverSys Inc., Reston, VA, USA	FreezeScan - CleverSys Inc
DAVID Bioinformatics Database	NIH, USA	DAVID
Custom Code for gene expression analysis and visualization	Zeru Peterson’s Github page	Github-ZP-ABC
Allen Brain Cell (ABC) Atlas- Mouse	Allen Institute	ABC Atlas
CNS MicroRNA Profiles	Washington University School of Medicine, St. Louis	CNS MicroRNA Profiles
Other
Teklad irradiated S-2335 mouse breeder diet #7904	Inotiv	Teklad Diet, Bedding and Enrichment
Teklad NIH-31 irradiated modified mouse diet #7913	Inotiv	Teklad Diet, Bedding and Enrichment
Arraystar Mouse Small RNA Expression Microarray V1.0 (8x15K format)	Arraystar Inc. Rockville, Maryland	Arraystar Mouse small RNA Array

Experimental model and study participant details

All experiments were performed according to the National Institutes of Health guidelines and were fully approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Iowa (The University of Iowa - PHS Assurance No. D16-00009 A3021-01).

Mice strains

The generation and maintenance of TraxE126A point mutant mice were described previously.⁴² Mice were on C57BL/6J background. Heterozygous male and female mice were mated to produce homozygous point mutant mice and wildtype littermates. Mice were maintained on a 12h light/12h dark cycle. Food and water were available ad libitum. Breeder mice were maintained on a diet with higher fat content (Teklad irradiated S-2335 mouse breeder diet #7904; 11.4% fat) to enhance breeding efficiency. Weaned mice were maintained on a standard mouse diet (Teklad NIH-31 irradiated modified mouse diet #7913; 6.2% fat). All experiments were performed during the light cycle using homozygous Trax point mutant mice and wildtype littermates as controls. Genotyping was performed either in-house using PCR or through service offered by Transnetyx (Cordova, TN).

Method details

Small RNA sequencing

3-4 months old naïve male TraxE126A point mutant mice (n = 4) and wildtype littermates (n = 5) were sacrificed by cervical dislocation and hippocampi were quickly dissected in ice-cold, oxygenated aCSF. The hippocampi were collected in sterile, DNase and RNase-free tubes chilled on dry ice and immediately stored at -80°C. Total RNA (including small RNAs) was isolated from these samples using Invitrogen miRVana miRNA Isolation Kit with phenol (Thermo Fisher Scientific, #AM1560) according to manufacturer’s instructions. The samples were treated with DNase I (Zymo, #E1010). The quality of the RNA samples was confirmed using NanoDrop 2000 Spectrophotometer and Agilent Bioanalyzer 2100. Small RNA cDNA libraries were prepared from these samples using Illumina TruSeq Small RNA Library Preparation Kit and were sequenced on one lane of the Illumina NovaSeq 6000 SP flowcell (50bp paired-end sequencing, ∼50 million pairs of reads) at the Iowa Institute of Human Genomics (IIHG) Core Facility at the University of Iowa.

Small RNA microarray

3-4 months old naïve male TraxE126A point mutant mice (n = 4) and wildtype littermates (n = 5) were sacrificed by cervical dislocation and hippocampi were quickly dissected in ice-cold, oxygenated aCSF. The hippocampi were collected in sterile, DNase and RNase-free tubes chilled on dry ice and immediately stored at -80°C. Total RNA (including small RNAs) was isolated from these samples using Invitrogen miRVana miRNA Isolation Kit with phenol (Thermo Fisher Scientific, #AM1560) according to manufacturer’s instructions. The samples were treated with DNase I (Zymo, #E1010). The quality of the RNA samples was confirmed using NanoDrop 2000 Spectrophotometer and Agilent Bioanalyzer 2100. The samples were then shipped on dry ice to Arraystar Inc. (Rockville, Maryland) for simultaneously profiling major small RNA classes including miRNA, pre-miRNA, tsRNA, tRNA and snoRNA using Arraystar Mouse Small RNA Array V1.0. RNA sample quantity and quality control was performed again prior to microarray analysis using NanoDrop ND-1000 spectrophotometer and RNA integrity by Bioanalyzer 2100 or gel electrophoresis. The microarray workflow involves 3’ terminal dephosphorylation of total RNA (100 ng for each sample) with T4 polynucleotide kinase followed by DMSO denaturation and direct enzymatic 3’ end labeling with Cy3. The labeled RNA species are then hybridized onto Arraystar Mouse Small RNA Expression Microarray V1.0 (8x15K format), scanned by an Agilent G2505C scanner followed by data processing and analysis. The Arraystar Mouse Small RNA Array V1.0 comprises a total of 14,192 distinct probes and includes the following coverage for different small RNA species: miRNAs – 1,949 (966 5-p and 983 3-p); pre-miRNAs – 1,122; tsRNAs – 1,767; mature tRNAs – 270; snoRNAs – 1,324. Following databases were used as sources for small RNA nomenclature and sequence: miRNA -miRBase(v22); tsRNA - tRFdb, GtRNAdb (Updated to 18.1 2019.08); pre-miRNA -miRBase(v22); tRNA - GtRNAdb(Updated to 18.1 2019.08), ENSEMBL(v99); snoRNA - ENSEMBL(v99).

mRNA sequencing

3-4 months old naïve male TraxE126A point mutant mice (n = 4) and wildtype littermates (n = 5) were sacrificed by cervical dislocation and hippocampi were quickly dissected in ice-cold, oxygenated aCSF. The hippocampi were collected in sterile, DNase and RNase-free tubes chilled on dry ice and immediately stored at -80°C. Total RNA (including small RNAs) was isolated from these samples using Invitrogen miRVana miRNA Isolation Kit with phenol (Thermo Fisher Scientific, #AM1560). The samples were treated with DNase I (Zymo, #E1010). The quality of the RNA samples was confirmed using NanoDrop 2000 Spectrophotometer and Agilent Bioanalyzer 2100. cDNA libraries were prepared from these samples using Illumina Stranded mRNA Library Preparation Kit with oligo-dT purification of polyadenylated RNA and were sequenced on one lane of the Illumina NovaSeq 6000 SP flowcell (50bp paired-end sequencing, ∼50 million pairs of reads) at the Iowa Institute of Human Genomics (IIHG) Core Facility at the University of Iowa.

qRT-PCR

Total RNA was isolated from the hippocampi of WT and TraxE126A mutant mice (3-4 months old naïve males) using Invitrogen miRVana miRNA Isolation Kit with phenol (Thermo Fisher Scientific, #AM1560) according to manufacturer’s instructions. The samples were treated with DNase I (#E1010, Zymo Research, Irvine, CA) further purified using the Zymo RNA Clean & Concentrator-25 Kit (Cat. # R1017, Zymo Research). The quality of the RNA samples was confirmed using NanoDrop 2000 Spectrophotometer and Agilent Bioanalyzer 2100. Next, 1 μg of DNase-treated RNA was reverse-transcribed into cDNA using the iScript cDNA Synthesis Kit (Cat. # 1708890, Bio-Rad, Hercules, CA). Quantitative PCR (qPCR) was performed using SYBR™ Green Universal Master Mix (Cat. # 4309155, Thermo Fisher, Waltham, MA) in four technical replicates in three independent experiments across four and five biological replicates for each genotype. Primer sequences used for qRT-PCR are provided in Table S1. The Δ threshold cycle (ΔCT) values were calculated by normalizing the CT values of genes to those of Hprt. The relative fold change in the gene expression was determined using the formula: Fold Change = 2^−ΔΔCT against average value of WT samples.

Electrophysiology

Experiments were performed as previously described.^112,113 Male and female mice (2.5- to 6-month-old for LTP experiments; 23- to 33-days-old for LTD experiments) were sacrificed by cervical dislocation and hippocampi were quickly collected in chilled, oxygenated artificial cerebrospinal fluid (aCSF; 124 mM NaCl, 4.4 mM KCl, 1.3 mM MgSO₄·7H₂O, 1 mM NaH₂PO₄·H₂O, 26.2 mM NaHCO₃, 2.5 mM CaCl₂·2H₂O and 10 mM D-glucose) bubbled with 95% O₂ / 5% CO₂. 400 μm-thick transverse slices were prepared from dorsal two-thirds of the hippocampus using a manual McIlwain slicer (Stoelting) and placed in an interface recording chamber at 29°C (Fine Science Tools, Foster City, CA). The slices were constantly perfused with aCSF at ∼1 ml/min. Slices were incubated for at least 2 hours in aCSF before starting the recordings. Field-EPSPs were recorded using aCSF-filled glass electrodes (resistance 2-5 MΩ) in the hippocampal area CA1 stratum radiatum by stimulating Schaffer collateral pathway with monopolar lacquer-coated stainless-steel electrodes (#571000; A-M Systems, Sequim, WA). Test stimulation was a biphasic, constant current pulse (0.1 ms duration per phase) delivered every minute at a stimulation intensity that evoked ∼40% (or ∼50% in LTD experiments) of the maximal fEPSP amplitude as determined by an input-output curve (stimulation intensity v/s fEPSP amplitude) in each experiment (stimulation intensity range 5 μA – 70 μA). Paired-pulse facilitation (PPF) was assessed by delivering a pair of stimuli at five different inter-stimulus intervals (ISI; 300 ms – 5 ms). A stable baseline response was recorded for at least 20 min before delivering LTP or LTD-inducing stimulus. LTP was induced by the following stimulation paradigms: spaced 4-train LTP paradigm comprised of four 100 Hz, 1 s trains (pulse width 0.1 ms per phase) delivered with an intertrain interval of 5 minutes; massed 4 train LTP paradigm comprised of four 100 Hz, 1 s trains (pulse width 0.1 ms per phase) delivered with an intertrain interval of 5 seconds; 1-train LTP stimulation comprised of a single 100 Hz train (1 s duration; pulse width 0.1 ms per phase). LTD was induced by 1 Hz stimulation for 15 minutes (900 pulses; pulse width 0.1 ms per phase). All stimulation protocols were delivered at the baseline stimulation intensity. In synaptic tagging and capture (STC) experiments, two stimulation electrodes were positioned in CA1 stratum radiatum on either side of a recording electrode to record fEPSPs evoked by the stimulation of two independent pathways (inputs). Pathway-independence was ensured by the lack of paired-pulse facilitation (at 50 ms ISI) between the two. Stimulus intensity was set to elicit ∼40% of the maximum field-EPSP amplitude determined by an input-output curve for each pathway. After a stable 20-min baseline response, massed 4-train stimulation was delivered to one pathway (S1) followed 30 min later by 1-train stimulation in the second pathway (S2).

Behavioral assays

Behavioral assays were performed using the facilities provided by the Neural Circuits and behavioral Core (NCBC) at the University of Iowa. Whenever possible we tried to minimize the number of mice used, by utilizing the mice for a maximum of two behavioral paradigms. The order of the behavioral tests was chosen carefully so that tests with stressful stimuli (such as contextual fear conditioning) were always performed last. Naïve mice were used for open-field test as well as for Y-maze experiments. When these mice were also used for contextual fear conditioning, at least a week elapsed between the tests.

Open field

Male and female mice (2.5- to 5-month-old) were single-housed 4-5 days before the day of experiment and handled for 3-4 days (∼1-2 min every day) by the experimenter. On the day of the test, mice were introduced into the arenas (Dimensions: 38L x 32W x 30H) and allowed to explore the arenas freely for 6 minutes. The sessions were recorded by an overhead camera. The arenas were cleaned with 70% ethanol and allowed to dry prior to and between sessions. The videos were then analyzed using Ethovision XT program (Noldus) to score the time spent in the inner and outer areas (of equal area) of the arena and total distance travelled. Data are reported as mean ± SEM. Statistical analysis was performed on GraphPad Prism 10 using two-way ANOVA with genotype and sex as factors. When no significant interaction between the two factors was observed, a simpler model without interaction term was used for the analysis of main effects, from which the F and P-values are reported. Data are graphed separately for the main effects of genotype and sex.

Contextual fear conditioning (CFC)

Male and female mice (2.5- to 5-month-old) were single-housed 4-5 days before the day of experiment and handled for 3-4 days (∼1-2 min every day) by the experimenter. On the day of the training, each mouse was introduced to a context with a grid floor (FreezeScan System, CleverSys Inc. VA) and allowed to explore for 180 seconds during which a single electrical shock of 1.5 mA intensity, 2 s duration was delivered through the grid after 148 s. At the end of 180 s, mice were returned to their home cage and to the housing room. 24 hours after the training session, mice were reintroduced to the trained context for 5 minutes. Freezing behavior of the mice during the training and test sessions was scored using the FreezeScan software. ∼30-40 min after testing in the trained context, freezing was also assessed in an altered context with solid plastic floors and walls in the presence of a lemon soap odor. Average freezing levels before the shock (pre-shock) in the training session and during the 5 min test session in the trained and altered contexts were compared to measure context-specific freezing as a proxy for contextual fear memory. Data are reported as mean ± SEM. Statistical analysis was performed on GraphPad Prism 10 using Three-way repeated measures ANOVA with genotype, sex and sessions as factors. When no significant interaction between the factors was observed, a simpler model without interaction term was used for the analysis of main effects, from which the F and P-values are reported.

Two-trial Y-maze

Group-housed male and female mice (2.5- to 5-month-old) were handled for 3-4 days (∼1-2 min every day) by the experimenter. On the day of the training, mouse was introduced to a Y-maze at the entry arm and allowed to explore the maze for 10 mins with one of the arms of the maze blocked. At the end of 10 min, mouse was returned to its home cage and the housing room. 24 hours after the training session, mouse was reintroduced to the Y-maze at the entry arm and allowed to explore the maze for 5 mins with none of the arms of the maze blocked. Distinct visual cues were present in each arm of the maze during both training and test sessions. The sessions were recorded using an overhead camera and the test session videos were manually scored using BORIS Software¹¹⁴ to assess the number of entries into the familiar and novel arms. Any time both the front paws of the mouse were past the location of the blocking slit in each arm, it was considered as an “entry” into that arm. The time spent exploring the familiar and novel arms was also scored. For this, time spent exploring the novel and familiar arm including the respective halves of the central triangular area was included in the “exploration time” for that arm (because the mice often tended to sit in the triangular area and sniff/explore the arm without fully entering the arm). Time spent grooming and time spent in the entry arm were not included. Data are presented as the discrimination index for the novel arm [(novel arm entry - familiar arm entry)/ total arm entries]. Data are reported as mean ± SEM. Statistical analysis was performed on GraphPad Prism 10 using two-way ANOVA with genotype and sex as factors. When no significant interaction between the two factors was observed, a simpler model without interaction term was used for the analysis of main effects, from which the F and P-values are reported. Data are graphed separately for the main effects of genotype and sex.

Quantification and statistical analysis

Small RNA sequencing data analyses

Data analysis was performed using Partek Flow interface (Partek, Illumina, CA). Following QA/QC, 5’-TGGAATTCTCGGGTGCCAAGG-3’ adaptor sequence was trimmed from the 3’ end of read 1 and 5’-GATCGTCGGACTGTAGAACTCTGAAC-3’ adaptor sequence was removed from the 3’ end of read 2 using Cutadapt 4.2,¹¹⁵ followed by base trimming. Trimmed reads were aligned using Bowtie¹¹⁶ and annotated using mm10 miRBase v22 mature miRNA database. The resulting miRNA counts were filtered to exclude miRNAs with minimum count <= 1 and normalized using median ratio method. The differential miRNA expression analysis was performed with DESeq2.¹¹⁷ Differentially expressed miRNAs with P-values (adjusted for multiple comparison using Benjamini-Hochberg method) less than 0.050 and fold change of 1.150 were considered for functional enrichment analysis.

Small RNA microarray data analyses

The intensity data were extracted from the acquired array images using Agilent Feature Extraction software (version 11.0.1.1). Quantile normalization and subsequent data processing were performed using GeneSpring GX v12.1 software package (Agilent Technologies). After normalization, the probe signals having Present (P) or Marginal (M) QC flags in at least 4 out of 9 samples were retained. Multiple probes from the same small RNA (miRNA/tsRNA(tRF&tiRNA)/pre-miRNA/tRNA/snoRNA) were combined into one RNA level. Differentially expressed small RNAs between two comparison groups were identified by fold change (FC; ≥1.150) and statistical significance (unpaired t-test P-value and False Discovery Rate <0.050) thresholds.

mRNA sequencing data analyses

Data analysis was performed using Partek Flow interface (Partek, Illumina, CA). Following QA/QC, raw reads were subjected to base trimming and aligned using STAR aligner version 2.7.8a.¹¹⁸ Aligned reads were quantified by Partek E/M model and annotated to mouse reference genome mm39 Ensemble release 110. The resulting mRNA counts were filtered to exclude mRNAs with minimum count <= 1 and normalized using median ratio method. Differential gene expression analysis was performed with DESeq2.¹¹⁷ Differentially expressed mRNAs with P-values (adjusted for multiple comparisons using Benjamini-Hochberg method) less than 0.050 and fold change of 1.200 were considered for functional enrichment analysis.

Electrophysiological data analyses

Data were acquired using Clampex 10 and Axon Digidata 1440/1550 digitizer (Molecular Devices, Union City, CA) at 20 kHz and were low-pass filtered at 2 kHz with a four-pole Bessel filter. Data analysis was performed using Clampfit 10 (Molecular Devices, Union City, CA). Data plots and statistical analyses were performed using GraphPad Prism10. Input-output and PPF data were analyzed using repeated measures ANOVA. The maintenance of LTP or LTD was analyzed by comparing the average of the normalized slopes over the last 20-min between two groups using unpaired t-test. The “n” used in all the experiments represents the number of mice and replicate slices from the same mouse were averaged. Differences were considered statistically significant when P values were < 0.05. Data are plotted as mean ± S.E.M.

miRNA target prediction

The predicted mRNA targets of the selected miRNAs were acquired from the miRDB database.⁵⁹ Only the targets with Target Score ≥60 were included in the analysis.

Functional annotation and enrichment

Functional annotation and pathway enrichment analysis was performed using DAVID platform^60,61 with the EASE score threshold set at 0.05 and minimum gene count threshold set at 5. Kyoto Encyclopedia of Genes and Genomes (KEGG), Reactome and WikiPathways databases were used for functional enrichment of pathways. Gene Ontology (GO) and UniProt databases were used for enrichment analysis of GO terms. Pathways or GO terms with the Benjamini-Hochberg adjusted P-value threshold <0.050 were considered to be significantly enriched. Protein-Protein interaction clustering was performed using STRING database¹¹¹ using k-means clustering method.

Neuronal cell type-specific miRNA expression data

Neuronal cell type-specific expression data in adult mouse neocortex for the 12 common miRNAs (10 upregulated and 2 downregulated) are directly downloaded from the “CNS microRNA Profiles” webpage^62,63 maintained by the Joseph Dougherty and Timothy Miller labs at Washington University, Saint Louis (https://mirna.wustl.edu/).

Cell type-specific gene expression data from Allen brain cell atlas

We used the Allen Brain Cell (ABC) Atlas to visualize the expression of the 43 genes listed found to be targets of upregulated miRNAs that are also downregulated in mRNA-seq experiments within the hippocampus. Of these 43 genes, 24 were found within the MERFISH data along with imputed 10Xv3 single cell data.¹¹⁹ Using the parcellation information we first limited our search space to only the Hippocampal formation in brain section C57BL6J-638850.38, we then displayed output data across all cell types of interest within that region. For our cell types of interest we used “Excitatory” (a combination of the following classes: “01 IT-ET Glut”, “02 NP-CT-L6b Glut”, “03 OB-CR Glut”, “04 DG-IMN Glut”, “13 CNU-HYa Glut”, “14 HY Glut”, “15 HY Gnrh1 Glut”, “16 HY MM Glut”, “17 MH-LH Glut”, “18 TH Glut”, “19 MB Glut”, “23 P Glut”, “24 MY Glut”), “Inhibitory” (combination of the following classes: “05 OB-IMN GABA”, “06 CTX-CGE GABA”, “07 CTX-MGE GABA”, “08 CNU-MGE GABA”, “09 CNU-LGE GABA”, “11 CNU-HYa GABA”, “12 HY GABA”, “20 MB GABA”, “27 MY GABA”), “Astrocytes”, and “Vascular” cell types. We then plotted the log2 gene expression of each of the 24 genes along with the expression of Translin (Tsn) for each of the four cell types. The code used for generating the figures is available on Github as “createCellTypeGeneFigure.py” under https://github.com/zjpeters/traxTranslinABCAnalysis.

Statistics

For the behavior experiments, the statistical details are described in the method details section above and the figures and in their corresponding legends for individual experiments. Significance was set at P < 0.05. GraphPad Prism 10 (GraphPad Software, Boston, MA, USA) was used for statistical analysis.