Contact: joanna.jia@wecistanche.com / WhatsApp: 008618081934791

Pls click here to Part 1

Pls click here to Part 3

Cistanche can improve memory

Given these findings, we next examined if upregulated hippocampal miR-466f-3p expression in mice upon spatial learning and memory formation also affected neuronal morphology in vivo. Significantly, we found that average spine density of the hippocampal neurons of GLN mice was higher than that of PLN mice, as revealed by Golgi impregnation of the pyramidal

Figure 3. Effects of alteration of the hippocampal miR-466f-3p expression levels on mouse MWM performance

(A) Experimental timeline of the MWM task, expression analysis, or electrophysiology measurement of the mouse brains after lentivirus injection.

(B) Left panels: expression of dsRed in mouse hippocampus. Representative low- and high- magnification IF images of hippocampus from mice infected with recombinant lentiviruses ex- pressing dsRed-only as a control (left column of images) or miR-466f-3p + dsRed (right column of images). The boxed areas of the top two images are magnified and shown below. Scale bars, 100 mm. The dotted lines indicate the boundaries of DG. Nuclei were stained with DAPI. GCL, granule cell layer; ML, molecular layer. Right histogram, relative expression levels of miR-466f-3p in mouse hippocampus infected with lentivirus expressing control or miR-466f-3p, as assayed by RT-qPCR (n = 16 per group).

(C) MWM performances of mice infected with different recombinant lentiviruses, packaged with the use of five different plasmids expressing dsRed-only, miR-466f-3p, mut-miR-466f-3p, and miR/SCR-sponges, respectively, into their hippocampus (n = 13, 26, 16, 19, and 11 per group, respectively). Left panel: escape latency during the training. Right panel: individual escape latency on the 6th session.

Data shown in (B) are presented as mean ± SD, and data shown in (C) are presented as mean ± SEM. Statistical significance was assessed by unpaired t test (B), two-way ANOVA with Bonfer- roni post hoc comparison (C, left panel), or one-ANOVA with Tukey’s post hoc test (C, right panel). Statistical differences: #p < 0.05, **/##p < 0.01 and ****p < 0.0001; *miR-466f-3p versus others; #miR-sponge versus controls.

neurons in the GLN and PLN mouse hippocampus (Figure S3). These findings demonstrate that upregulation of hippocampal miR-466f-3p promotes neurite outgrowth and dendritic spine formation, which is similar to the induction of spine density observed in the hippocampus of GLN mice relative to PLN mice.

miR-466f-3p positively modulates mouse spatial learning and memory performance, as well as synaptic plasticity

To investigate whether miR-466f-3p positively regulates mouse spatial learning and memory formation, we used a recombinant lentivirus infection approach to overexpress the miRNA in the mouse hippocampus. The mice were analyzed 7 days after lentivirus injection into the DG (Figure 3A). Representative images of the hip-expression levels of miR-466f-3p were indeed elevated in the miR-466f-3p overexpression group compared to the control group (right histogram in Figure 3B). We found that mice with hippocampal overexpression of miR-466f- 3p and subjected to the MWM task exhibited escape latencies of 88 ± 5 s in the 1st session and 19 ± 1 s in the 6th session (red dot line, Figure 3C), which were better than for vector control mice (black line) or mutant control mice (red circle line) and similar to the escape latencies of GLN mice described in Figure 1A. In parallel, we examined the effect of miR-466f-3p loss of function on spatial learning and memory by inhibiting miR-466f-3p using miR-sponge. Notably, the MWM performance of mice in which hippocampal miR-466f-3p was trapped by the miR-sponge was similar to that of PLN mice, i.e., they did not learn to find the platform even by the last session (solid green square line,

Figure 3C). Furthermore, mice injected with lentivirus did not appear to have disrupted homeostatic plasticity, since some mice in each group had learned or at least were in the process of learning during the last session (Figure 3C, right histogram).

Given that overexpression of miR-466f-3p in mouse hippocampus enhanced their learning and memory capability, we analyzed the comparative electrophysiology of cultured hippocampal neurons expressing different levels of miR-466f-3p. The miniature excitatory postsynaptic current (mEPSC) from DIV14 hippocampal neurons overexpressing miR-466f-3p or mut-miR-466f-3p, miR-sponge, or control was recorded using whole-cell patch clamps. Whereas there were no significant differences in the mEPSC amplitude, rise Tau, or decay Tau among the four sets of samples, the mEPSC frequency of neurons over-expressing miR-466f-3p was significantly higher compared to other groups (Figure 4A), indicating that postsynaptic glutaminergic receptors were more strongly activated upon miR-466f- 3p overexpression.

We then measured long-term potentiation (LTP) to directly determine the role of miR-466f-3p in synaptic plasticity in vivo. Weinjectedthehippocampusofmicewithrecombinantlentivirus asdescribedinFigure3andtheninducedLTPinhippocampalslices by tetanic stimulation (three trains of high-frequency stimulation [3xHFS]) of the Schaffer collateral pathway. We found that our protocol induced LTP in all groups, as evidenced by the persistent increase in field excitatory postsynaptic potentials (fEPSPs) in the cornu ammonis 1 (CA1) region (Figure 4B, left panel). The LTP was stronger in miR-466f-3p-overexpressing slices (188% ± 2% of baseline at 40–50 min after stimulation, mean ± SEM) compared to mutant (169% ± 1% of baseline), SCR- sponge (173% ± 3% of baseline), and control-virus-infected slices (159% ± 2% of baseline), as miR-466f-3p inhibition by miR-sponge reduced the LTP (128% ± 1% of baseline) relative to controls (Figure 4B, right panel). We also measured the LTP of the GLN and PLN groups after training. The respective data also revealed significant differences in fEPSP slope in the CA1 region between the GLN and PLN groups (Figure 4C). Thus, an elevated level of miR-466f-3p enhances LTP and synaptic plasticity that, in turn, can promote the learning and memory capability of mice. Together, the data presented in Figures 2, 3, and 4 demonstrate that miR-466f-3p plays a critical positive role in spatial learning and memory formation, likely by enhancing spine formation, LTP, and the strength of synaptic plasticity.

Mef2a mRNA is a regulatory target of miR-466f-3p

We conducted a bioinformatics analysis to identify potential target mRNAs regulated by binding of miR-466f-3p to their 30 UTRs. Among the candidates we identified was the Mef2a mRNA encoding MEF2A. Interestingly, MEF2A expression was previously found to be downregulated after MWM training, and overexpression of this factor exerted a negative effect on the performances of mice (Cole et al., 2012). Therefore, we used a luciferase reporter assay to examine if miR-466f-3p regulates the translation of Mef2a mRNA by binding to its 30 UTR. We inserted wild-type or mutant Mef2a 30 UTR sequences downstream of an SV40-promoter-driven luciferase cDNA (Luc), resulting in the reporter plasmid psiCHECK2-MEF2A 30 UTR or psiCHECK2-mut-MEF2A 30 UTR (Figure 5A). As shown in the left histogram of Figure 5B, co-expression of miR-466f-3p attenuated the expression of luciferase directed by Mef2a 30 UTR. This effect appeared to be dependent on the interaction between miR-466f-3p and Mef2a 30 UTR since mutation of either the predicted binding site of miR-466f-3p on the Mef2a 30 UTR (50-UGU- GUAU-30) or the seed region of miR-466f-3p (50-AUACACA-30) recognizing Mef2a 30 UTR abrogated the inhibitory effect of miR-466f-3p on luciferase activity (Figure 5B, middle histogram). Furthermore, co-expression of the miR-sponge, but not the CR sponge, also eliminated the repressive effect of miR-466f-3p on luciferase activity (Figure 5B, right histogram). Notably, neither overexpression of miR-466f-3p nor miR-sponge had any effect on the levels of Mef2a mRNA in primary hippocampal neurons (Figure 5C), but they did reduce or increase, respectively, the level of MEF2A protein (Figure 5D). We also found that overexpression of miR-466f-3p or miR-sponge, respectively, downregulated or upregulated the mRNA level of activity-regulated cytoskeletal associated protein (Arc) in primary hippocampal neurons (Figure 5C), which was a known downstream target positively regulated by MEF2A (Flavell et al., 2006)

We also performed fluorescence in situ hybridization (FISH) to detect miR-466f-3p and combined it with IF labeling of MEF2A in DIV14 primary hippocampal neurons without or with forskolin treatment. Forskolin is known to induce chemical LTP and activate adenylyl cyclase, thus raising intracellular cAMP levels. We observed nuclear colocalization of MEF2A with miR-466f- 3p, as illustrated by the representative images in Figure 5E. After forskolin stimulation, however, the miR-466f-3p signal increased in both soma and dendrites, whereas the MEF2A signal in the

nucleus diminished, as shown by the reciprocal changes of the intensities of the MEF2A and Fast Red signals, respectively, of the individual neuronal cells (Figure 5E). In sum, the data in Figures 5A–5E indicates that miR-466f-3p negatively regulates the expression of MEF2A protein by binding to the 30 UTR of Mef2a mRNA and consequently repressing its translation.

Consistent with the above-described results, we found that hippocampal levels of MEF2A protein were downregulated in GLN mice after MWM training, but not in PLN mice (Figure 5F). Furthermore, the relative levels of MEF2A in individual GLN mice (dots) and PLN mice (squares) were inversely correlated with the levels of miR-466f-3p (R = 0.60, Figure 5G). Thus, heterogeneous patterns of spatial learning and memory capability are modulated by stochastic increases of hippocampal miR- 466f-3p in individual mice and consequent decreases of MEF2A upon stimulation of neuronal activity.

Stochastic activation of hippocampal CREB and consequent transcriptional upregulation of the miR- 466-669 cluster during MWM training

WeinvestigatedthemechanismsbywhichmiR-466f-3p in mouse hippocampus could be stochastically upregulated by the MWM task. There are three miRNA precursors encoding miR-466f-3p, all of which belong to the rodent-specific miR-466-669 cluster located in intron 10 of its host gene, mSfmbt2 (Inoue et al., 2017) (Figure 6A). Interestingly, in contrast to miR-466f-3p, there was no significant difference in the expression levels of mSfmbt2 betweentheGLNandPLNgroups(Figure6B), suggesting that the miR-466-669 cluster might encode a separate transcript instead of being part of the primary transcript of mSfmbt2. Accordingly, we designed several sets of primers to identify the putative primary transcript of miR-466-669 cluster by RT-qPCR. As depicted in figure 6A, a long PCR fragment (+282 to 2,381), together with a series of overlapping qPCR bands, namely A (+282 to 115), B ( 131 to 406), C ( 388 to 686), D ( 662 to 1,028), E ( 1,004 to 1,299), F ( 1,242 to 1,629), G ( 1,605 to 2,029), and H( 2,005 to 2,381), could be detected in mouse hippocampus, but not part I ( 2,306 to 2,776) (data not shown). These data support that the miR-466-669 cluster encoded a long transcript near approximately 2,388 bp upstream of the first miRNA precursor(i.e.,pre-mir-466m,denoted+1inFigure6A). Remarkably, similar to what we found for miR-466f-3p, the average level of this transcript in the hippocampus of GLN mice was higher than for PLN mice (Figure 6C). Thus, enhanced learning and memory capability of GLN mice is due to transcriptional activation of the miR-466-669 cluster.

Activation of nuclear CREB by phosphorylation during MWM tasks or in other forms of learning is a critical step for converting short-term into long-term memory (Lisman et al., 2018; Roger- son et al., 2014). Therefore, we investigated if the varying levels of miR-466f-3p among individual mice that underwent the MWM task might be correlated with the activation status of CREB. To explore this idea, first we examined the phosphorylation levels of hippocampal CREB in individual mice. As shown in Figure 6D, CREB activation (by phosphorylation at residue S-133) was enhanced in GLN mice relative to PLN and HC mice. We also confirmed that miR-466f-3p was indeed co-expressed with pCREB in neurons by performing miRNA ISH combined with IF staining of pCREB and MAP2 in primary hippocampal neurons (Figure S2A). Furthermore, the primary transcript levels of miR- 466-669 cluster were positively correlated with pCREB in GLN mice (R = 0.52), whereas it was negatively correlated with pCREB in PLN mice (R = 0.71) (Figure 6E). We checked for a correlation between the levels of another active factor, phos- pho-extracellular signal-regulated kinase (pERK), and the primary transcript levels of the miR-466-669 cluster. We found that both types of MWM-trained mice presented positive correlations between miR-466-669 cluster signal and pERK (R = 0.59 and 0.48, respectively; Figure S4A), and there was no difference in pERK/tERK expression between those two groups (Figure S4B). To further clarify the impact of CREB activation on transcriptionalupregulationofthemiR-466-669clusterand, consequently,

inductionofmiR-466f-3p,weusedaspecificCREBinhibitor,666- 15 (Xie et al., 2015), in combination with forskolin treatment of DIV14 primary hippocampal neurons. Forskolin is known to activate CREB phosphorylation (Malhotra et al., 2015). We found that pretreatment of primary hippocampal neurons with 666-15 blocked forskolin-induced CREB activation (Figure 6F) and reduced the levels of miR-466f-3p and miR-466-669 cluster transcript (Figures 6G and 6H). Concomitantly, mRNA levels of the known pCREB target genes nurr1 and homer1a (Bridi et al., 2017; Jensen et al., 2017) were also reduced upon 666-15 treatment (Figure 6H). Taken together, Figure 6 shows that stochastic activation of CREB in the hippocampus of a subpopulation of inbred mice undergoing MWM task induces transcription of the miR-466-669 cluster and, consequently, induction of miR-466f- 3p, thereby enhancing their spatial learning and memory capability to perform better in the task.

DISCUSSION

We have explored the possible role of miRNAs in modulating the varying capability of spatial learning and memory among inbred C57BL/6J mice. We suggest that stochastic activation of CREB and consequent upregulation of hippocampal miR-466f-3p ac- counts for the greater spatial learning and memory capability of a subgroup of our test mice, likely due to miR-466f-3p mediating translational inhibition of Mef2a mRNA that encodes the memory negative regulator MEF2A. Our findings provide a scenario demonstrating the functional and evolutionary role of a specific miRNA in regulating cognition through stochastic induction of a CREB-pCREB-miR-466f-3p-MEF2A axis by environmental stimuli.

The MWM task has been widely used to examine the varying capability of spatial learning and memory for rodents. Under normal conditions, rodents use distal cues to orient themselves, allowing them to learn and memorize the location of the hidden platform in the task. Most of our test inbred C57BL/6J mice(62%) exhibited a normal pattern of escape latency, finding the platform within 30 s by sessions 3rd-6th (Figure 1A, GLN). In contrast, a group of mice (38%) did not learn the task even by the 6th session (PLN). Notably, we have also subjected the mice to two other recognition tests. The first was the novel object recognition (NOR) test, which is a single trial based on innate exploration without reinforcement or stress to motivate behavior (Leger et al., 2013). The correlation between the NOR and MWM tasks in terms of mouse performances is poor (R = 0.13), which is similar to the correlation between NOR performances and the hippocampal expression levels of miR-466f-3p (R = 0.01; data not shown). The NOR test also revealed no differences in the discrimination index between the MWM GLN and PLN groups (0.36 ± 0.25 versus 0.29 ± 0.18; Figure S1B). The other spatial learning and memory-dependent task we have applied is the Barnes maze (BM), which is similar to the MWM. It is based on the assumption that rodents placed into an aversive environment should learn and remember the location of an escape box located below the surface of the platform. As shown in Figure S1C, we have found that mouse performances in the BM probe trials are also positively correlated with the hippocampal expression levels of miR-466f-3p. Thus, induction of the CREB-pCREB-miR-466f-3p-MEF2A axis in the hippocampus during learning and memory formation is spatially context-dependent.