Distinct cognitive effects and underlying transcriptome changes upon inhibition of individual miRNAs in hippocampal neurons

Abstract

MicroRNAs (miRNA) are small, non-coding RNAs mediating post-transcriptional regulation of gene expression. miRNAs have recently been implicated in hippocampus-dependent functions such as learning and memory, although the roles of individual miRNAs in these processes remain largely unknown. Here, we achieved stable inhibition using AAV-delivered miRNA sponges of individual, highly expressed and brain-enriched miRNAs; miR-124, miR-9 and miR-34, in hippocampal neurons. Molecular and cognitive studies revealed a role for miR-124 in learning and memory. Inhibition of miR-124 resulted in an enhanced spatial learning and working memory capacity, potentially through altered levels of genes linked to synaptic plasticity and neuronal transmission. In contrast, inhibition of miR-9 or miR-34 led to a decreased capacity of spatial learning and of reference memory, respectively. On a molecular level, miR-9 inhibition resulted in altered expression of genes related to cell adhesion, endocytosis and cell death, while miR-34 inhibition caused transcriptome changes linked to neuroactive ligand-receptor transduction and cell communication. In summary, this study establishes distinct roles for individual miRNAs in hippocampal function.

Figure 1. Activity of three miRNAs in hippocampal neurons.

miRNA activity was visualised in transgenic mice using a negative GFP reporter system incorporating perfectly complementary binding sites for each respective miRNA. (A) Expression of the GFP reporter was achieved under the regulation of a PGK promoter with (upper panel) or without (lower panel; GFP Ctrl) the inclusion of four perfectly complementary target sites (miR.T) for a specific miRNA, separated by a WPRE element in a self-integrating (SIN) lentiviral vector. Lentiviral transgenesis resulted in integration of these vectors in all cell types. B-D) The presence of four perfectly complementary target sites for either miR-124 (B) miR-9 (C) or miR-34 (D), limited GFP expression to a subpopulation of cells or in no cells at all, indicating activity of each respective miRNA in GFP-negative cells. Hippocampal neurons were GFP-negative in miR-124.T.GFP, miR-9.T.GFP and miR-34.T.GFP mice, indicating the activity of each respective miRNA in these cells. miR-124.T.GFP mice displayed GFP expression, i.e. lack of miR-124 activity, in astrocytes and microglia (see Akerblom et al. 2012a), whereas only microglia had this characteristic in miR-9.T.GFP mice (see Akerblom et al. 2013). E) In GFP Ctrl transgenic mice, any cell type, including hippocampal neurons of the dentate gyrus, can express GFP. Scale bars in merged middle panels are 100 μm.

Figure 2. miRNA inhibition using AAV-miRNA sponges.

(A) AAV-miRNA sponges were expressed in order to inhibit individual miRNA families. These vectors express a GFP reporter under the influence of a synapsin promoter (p-syn) with (upper panel) or without (lower panel; GFP Ctrl) the presence of eight imperfectly complementary target sites for a specific miRNA, all within inverted terminal repeats (ITRs) of an AAV2/5 vector. (B) The design of the miR-124sp, miR-9sp and miR-34sp sequences in comparison to the endogenous miRNA. (C–G). (C) DAB immunohistochemistry, (D–G) fluorescence immunohistochemistry. Injection of AAV2/5 pseudotyped vectors into the hippocampus (three locations per hemisphere), resulted in the expression of the GFP reporter in a majority of hippocampal neurons of the dentate gyrus in the presence of a miR-124sp sequence (C–D), a miR-9sp sequence (E), a miR-34sp sequence (F), or without such a sequence (G). Scale bars 100 μm (D–G); (left merged panels) and 20 μm (D–G); (right merged panels).

Figure 3. Transcriptome changes of probable targets after miRNA inhibition.

(A) In the list of all detected genes using mRNA-seq after each respective miRNA inhibition (light grey), RISC genes (light purple; Malmevik et al. 2015) with at least one computationally predicted and evolutionary conserved target site for this miRNA (miR-target site(s); dark grey; TargetScanMouse 6.2) were identified as probable targets (overlapping area, left panel). Some probable targets were shared between the three analysed miRNAs (right panel). Each subset of probable targets was compared to all other genes in each respective miRNA sponge (miR-sp) experiment. (B) After miR-124 inhibition, this subset (n = 337, red/dotted, left panel) had a significantly different cumulative fraction distribution of its log2(fold change in miR-124sp/GFP Ctrl) in comparison to that of all other genes (n = 11852, black; ****p < 0.0001, Kolmogorov-Smirnov Z test), demonstrating up-regulated expression of miR-124 target genes. The genes of this subset (red bar, right panel) also had an overall higher average fold change after miR-124sp injection in comparison to all other genes (black bar; ****p < 0.0001, unpaired parametric two-tailed t-test). (C) Similarly, the equivalent subset of probable targets of miR-9 (n = 211; brown/dotted) had a significantly different fold change after miR-9sp inhibition in comparison to all other genes (n = 12115, black), as shown in a log2(fold change) cumulative fraction graph (C, left panel; ****p < 0.0001, Kolmogorov-Smirnov Z test) and in an overall fold change bar chart (C) right panel; ***p < 0.001, unpaired parametric two-tailed t-test). (D) The equivalent subset of probable targets of miR-34 (n = 114; blue/dotted) after miR-34sp inhibition exhibited the same trend as miR-124 and miR-9 in comparison to all other genes (n = 12091, black), however this did not reach significance. FC > 1.2 and FC < 1/1.2 are presented as shaded grey areas of each respective graph.

Figure 4. Specific transcriptome changes after inhibition of each miRNA.

(A) Out of all detected genes in the total mRNA-seq, 1435, 69 and 75 genes were significantly down-regulated after miR-124, miR-9 and miR-34 inhibition, respectively. (B) For each of these three miRNA families, 815, 31 and 82 genes were significantly up-regulated. Very little overlap existed between the different miRNA inhibitions. (C) DAVID gene ontology analysis was performed on genes that were significantly down-regulated (left side) and up-regulated (right side) in each miRNA sponge experiment, showing the -log10(p-value) significance of each GO term (GO_BP, or KEGG pathway in bold) together with the total number of members enriched in this GO term (white text on each bar). All, or the top 10, GO terms are included and also any shared significant GO terms between the experiments (stacked bars).

Figure 5. Ingenuity Pathway Analysis of transcriptome changes after miR-124 inhibition.

miR-124 inhibition resulted in significantly altered levels of mRNAs enriched in (A) canonical pathways, (B) upstream regulators (C) and disease & biological functions. (A) miR-124sp injection led to differential expression of mRNAs involved in EIF2-, calcium- and mammalian target of Rapamycin (mTOR) signalling with each respective z-score, ratio and –log(p-value). The calculated z-score indicates a pathway with genes exhibiting overall increased mRNA levels (orange bars) or decreased mRNA levels (blue bars). The ratio (orange dots connected by a line) indicates the ratio of genes from the dataset that map to the pathway divided by the total number of genes that map to the same pathway, e.g. >40% for EIF2 signalling. (B) Potential upstream regulators identified in the miR-124sp dataset included mTOR (upper panel) and Fragile X mental retardation 1 (FMR1; lower panel). mTOR was predicted as an inhibited upstream regulator (blue; Fischer’s Exact test, *p = 0.0389), potentially causing indirect (dashed lines) activation of three mRNAs. The inhibition of mTOR was also predicted to have caused lowered activity of one mRNA. Genes were highlighted (yellow aura) when multiple transcripts were present in the dataset, upon which the maximum absolute fold change (FC) value of any one transcript was visualised. FMR1 was predicted to be an enhanced upstream regulator (lower panel; Fischer’s Exact test, *p = 0.047). The predicted activation of FMR1 was thought to result in indirect inhibition of five mRNAs. Four mRNAs were believed to be either inhibited or enhanced by FMR1 but displayed inconsistent findings in fold change after miR-124 inhibition (yellow dashed lines). (C) The IPA disease/function analysis confirmed that gene level alterations in miR-124sp-injected mice corresponded to neurological disease, nervous system development and function, and cell-to-cell signalling and interaction, among others. Threshold levels in (A,C) are based on significance (p = 0.05).

Figure 6. Specific cognitive effects after inhibition of each miRNA.

(A) In a Morris Water Maze (MWM) test, miR-124sp mice swam significantly shorter distances over the paired trials (block 1 through 6; B1–B6) in order to find the hidden platform (PF; *p < 0.05 repeated measures two-way ANOVA with Sidak’s multiple comparisons test; Interaction between vector type and block). This was particularly evident in B6 (**p < 0.01; unpaired parametric two-tailed t-test). (B) The two experimental groups performed equally well in a MWM probe test (GFP Ctrl ****p < 0.0001; miR-124sp ***p < 0.001; unpaired parametric two-tailed t-tests). (C) In a spontaneous alternation task (SAT) test, miR-124sp-injected mice alternated between the arms of a symmetric plus maze to a larger extent than GFP Ctrls (*p < 0.05, unpaired parametric two-tailed t-test). (D) miR-9 inhibition resulted in impaired learning in the MWM training phase, where miR-9sp-injected mice learnt the task of finding the hidden platform differently from GFP Ctrl (*p < 0.05, repeated-measures two-way ANOVA with Sidak’s multiple comparisons test; Interaction between vector type and block). (E) There was no difference between the experimental groups in the MWM probe test, where both groups learnt the task (GFP Ctrl **p < 0.01; miR-9sp ****p < 0.0001; unpaired parametric two-tailed t-tests). (F) No significant difference was found in the SAT test between the groups. (G) miR-34 inhibition had no effect upon the training phase of the MWM. (H) However, while the GFP Ctrl mice learnt the MWM probe test (****p < 0.0001, unpaired parametric two-tailed t-test), the miR-34sp-injected group did not (ns p = 0.0522, unpaired parametric two-tailed t-test), indicating an impaired reference memory. (I) No significant difference was detected between the two groups in the SAT test.