Topographically positioned MEC neurons have distinct axonal projection patterns.
To explore the differences in connectivity mediated by Lphn2 or Ten3 expressing neuron populations, we first looked to survey their expression profile across parahippocampal regions. To do so, we performed single molecule fluorescent in-situ hybridization (smFISH) for Ten3 and Lphn2 in horizontal brain sections containing high Lphn2 expression in the MEC (Fig. 1A). In addition to the adjacent Lphn2 and Ten3 expressing neuron segregation that is observed in the CA1 and subiculum that has been previously reported[8, 19], we found similar patterning in parahippocampal regions of the presubiculum (PrS), parasubiculum (PaS), and MEC (Fig. 1A). This expression patterning matched the connectivity patterning that is observed between these interconnected regions[9], which we hypothesize are molecular elements mediating proximal and distal connectivity segregation of the MEC and its inputs (summarized in Fig. 1B).
In addition to the Lphn2 topographical expression profile that is observed in the MEC, Lphn2 expression is confined to select neurons in superficial II/III layers[9]. To test cell-type specific Lphn2 dependent circuit assembly, we sought to selectively delete Lphn2 from layer III pyramidal MEC neurons (MECIII). To accomplish this, we utilized a transgenic mouse that expresses CRE-recombinase using the OXR1 promoter that is selectively expressed in MECIII neurons (pOXR1-Cre)[20]. To test for Lphn2 protein elimination with this selective Cre-recombinase mediate deletion, we crossed pOXR1-Cre mice with transgenic Lphn2-mVenusfl/fl mice[12]. This mouse strain permits for Lphn2 visualization, coupled with the ability of genetic deletion with the presence of Cre-recombinase. Using these mice, MEC Lphn2 is found to be topographically and superficial-layer enriched[9], as well as enriched in the hippocampal stratum lacunosum-moleculare (SLM)[12]. Comparing Lphn2-mVenus mice with or without the pOXR1-Cre allele (Fig. 1C), we measured Lphn2 immunohistochemical signal in the MEC (Fig. 1D) and hippocampal SLM (Fig. 1E). Doing so, we observe robust decrease in Lphn2 signal for MEC sections, with no signal loss observed in the hippocampal SLM (Fig. 1C-E). Thus, genetic deletion of Lphn2 expression from layer III neurons represents a large percentage of Lphn2 protein in the MEC, while having no observable impact on Lphn2 protein levels and localization in the hippocampal output region for these neurons.
To confirm the cell-type specific expression for Lphn2 deletion as well as testing the use of this genetic tool to investigate differential MEC topographical circuitry, we proceeded to test selective and topographical Cre-recombinase expression in this genetic line. Using pOXR1-Cre mice, we co-labelled proximal and distal discrete MEC populations with Cre-recombinase dependent GFP (DIO-GFP-AAV5) or tdTomato (DIO-tdTomato-AAV5) expression AAVs (Fig. 1F). To confirm precise regional boundaries and analyze layer specific expression, we utilized an immunohistochemical (IHC) marker for Wolframin Syndrome 1 (Wfs1)[21]. Enrichment in Wfs1 IHC labeling is observed in the parasubiculum (PaS) but not presubiculum (PrS), and strongly in layer II of the MEC adjacent to the rhinal fissure beyond the rhinal fissure (Fig. 1G). Using Wfs1 IHC to precisely label regional boundaries in pOXR1-Cre injected animals, we observe distinct proximal/distal MEC neuron labelling confined to MECIII neurons not observed in the adjacent regions of the PaS or perirhinal cortex (Fig. 1H-I). Secondly, when comparing MECIII projection patterns between Lphn2 enriched neurons in the distal MEC (dMEC) and Ten3 enriched neurons in the proximal MEC (pMEC) (Fig. 1A/H), we observe distinct topographical projections to the stratum lacunosum-moleculare regions of the hippocampus (Fig. 1H). Taking these experiments together, we validate the use of pOXR1-Cre mice as a genetic tool to evaluate the role of Lphn2 in MECIII neurons.
Lphn2 deletion in MEC layer III neurons does not affect hippocampal projection topology.
After confirming selective labeling and topographical projection patterning that exits between the MECIII neurons and the hippocampus, we next set out to survey Lphn2 function in controlling this circuit. To do so, we crossed pOXR1-Cre mice with Lphn2 conditional knock-out mice that have loxP sites flanking the first coding exon of Lphn2 (Lphn2fl/fl)[12]. The resulting transgenic mouse line lead to selective deletion of Lphn2 in layer III neurons of the medial entorhinal cortex (Lphn2fl/fl;pOXR1-Cre). As a control, we used animals with native Lphn2 alleles not impacted by Cre-recombinase expression (Lphn2wt/wt;pOXR1-Cre). Using these animals, we performed proximal/distal MECIII neuron labeling with Cre-recombinase dependent tdTomato and GFP expression AAVs (Fig. 2A). We first quantified injection localization between Lphn2wt/wt;pOXR1-Cre (Control) and Lphn2fl/fl;pOXR1-Cre (MECIII-KO) mice by measuring relative fluorescence levels across layer III from the beginning of proximal MEC (pMEC; marked at the PaS border with immunolabelled Wfs1), through the distal MEC (dMEC; marked by the rhinal fissure and decrease in prominent Wfs1 labeling). While quantifying viral localization across the MEC in this fashion, we compared samples with similar localization and infectivity between genotypes for both tdTomato labeled pMEC (Fig. 2B-C) and GFP labelled dMEC (Fig. 2D-E).
With similar populations of neurons labeled in control and MECIII KO mice, we then proceeded to compare MECIII axonal projections into the hippocampal stratum lacunosum-moleculare (SLM) (Fig. 2F-K). In this analysis the genetic manipulation is purely presynaptic as hippocampal cells do not express Cre-recombinase in pOXR1 mice[20]. Axonal topographical patterning was quantified within the SLM layer from the CA1 region proximal to CA3 (pCA1) to the distal subiculum (dSUB) region ending at the presubiculum (PrS) border. By comparing relative fluorescent intensity measurements in this fashion, we observe no difference between control and MECIII-KO mice for axon topographical organization in the SLM (Fig. 2F-G). Additionally, we measured SLM layer specific targeting in CA1 by measuring fluorescence intensity across hippocampal layers from the alveus through the SLM. Comparing CA1 layer specificity of MECIII axon targeting between genotypes, no alterations in SLM targeting was observed (Fig. 2H). This analysis was repeated for dMEC injections, with similar findings observed with SLM projection topology (Fig. 2I-J) and layer specificity (Fig. 2K). In addition to these MECIII projections to the ipsilateral SLM, we observed strong projections as well to the contralateral SLM as similarly described[9, 22]. Performing similar analysis on these same labeled MECIII inputs for contralateral SLM projections, we again observed no differences between genotypes (data not shown). Collectively, it appears that neither proximal nor distal MECIII neurons require Lphn2 for their spatial axon targeting to the hippocampus.
Lphn2 deletion from MEC layer III neurons alters contralateral MEC projection topology.
In addition to the well-known MEC projections into the hippocampus, we also observed prominent MECIII axons that project to the contralateral MEC as previously described[9, 23]. Similarly as with MECIII axon projections to the hippocampus, these MECIII projections to contralateral MEC appear to be both topographically organized and layer targeted (Fig. 3A). Neurons from pMEC target the contralateral pMEC, and dMEC targeting contralateral dMEC. For both of these projections, MECIII axons appear to be targeted exclusively to contralateral MEC layer I. With this clear topographical and layer organization, we next explored the impact Lphn2 MECIII neuron deletion would have on the specificity of this patterning. With this genetic manipulation, both presynaptic and postsynaptic Lphn2 expression is deleted. Comparing topographical projection patterns from pMEC tdTomato labelled MECIII neurons axons into MEC layer I, line scan quantification was performed traversing the proximal to distal axis across MEC layer I as described previously (Fig. 2). In doing so, a slight shift is observed in the axonal targeting for Lphn2 deletion in MECIII neurons, with a small increase in axonal targeting observed in the dMEC compartment (Fig. 3B-D). Lastly, we analyzed the layer specific targeting for these pMECIII axon projections. Performing a line scan analysis of the contralateral MEC through layers I-VI, we observed layer I specific targeting that remains intact regardless of Lphn2 MECIII status (Fig. 3E). We then repeated this topographical and layer targeting analysis for dMECIII neurons and their projections to contralateral MEC (Fig. 3F-I). For these projections, we found a more robust shift in the axon topological targeting to Lphn2 MECIII deletion (Fig. 3F-H). For these Lphn2 deleted dMECIII projections, rather than projecting strongly to the contralateral dMEC as normally observed, these axons appear to shift their targeting more proximally within the MEC (Fig. 3G-H). Layer I targeting however, appears to be Lphn2 independent (Fig. 3I). Altogether, we found that Lphn2 is essential for dMECIII axon targeting to contralateral dMEC topographical patterning, while simultaneously serving as a repulsive signal for pMEC projections to contralateral dMEC.
Lphn2 deletion from MEC layer III neurons alters PrS to MEC topology.
Previously, we found that by using retrograde neuron labeling, deletion of Lphn2 in one MEC hemisphere leads to a selective impairment in connections between the MEC and the ipsilateral PrS[9]. With the majority of Lphn2 protein in the MEC being genetically removed in pOXR1-Cre mice (Fig. 1D), we hypothesized that this cell type is critical for the topographical organization of PrS to MEC connections. To test this hypothesis, we utilized Lphn2wt/wt;pOXR1-Cre (Control) and Lphn2fl/fl;pOXR1-Cre (MECIII-KO) mice as before. We then labelled proximal PrS neurons and their projections in these animals with stereotaxic targeted viral injects of EYFP expression AAVs (CaMKIIa-EYFP-AAV5). In this experiment, genetic deletion of Lphn2 is purely in postsynaptic MECIII neurons, as presubiculum neurons do not express Cre-recombinase in pOXR1-Cre mice[20]. Comparing the injection localization/expression of EYFP, we analyzed animals with similar PrS targeting (Fig. 4B). We then proceeded to compare the projection patterning of their outputs into the MEC layer III for both brain hemispheres (Fig. 4C-H). Analyzing the PrS axon organization in the MEC, we find that proximal PrS targeted the dMEC layer III on both hemispheres as previously reported[4, 9]. When comparing control and MECIII-KO mice however, there is a clear shift in this PrS topographical targeting to the ipsilateral MEC with MECIII Lphn2 deletion (Fig. 4C-E). Interestingly, when similarly analyzing the PrS projections to MEC on the opposite hemisphere, this long-range topographical targeting appears intact (Fig. 4F-H). Lastly, when comparing the layer specificity of PrS axons to the MEC, both ipsilateral and contralateral projections exhibited similar layer III targeting specificity (Fig. 4I-J). Together, this implicates Lphn2-MECIII neurons to be selectively controlling short-rage ipsilateral PrS to MEC connections for their topographical, but not layer specified targeting. Long-range contralateral PrS to MEC input targeting on the other hand, appears to be Lphn2 independent.
MECIII neuron Lphn2 deletion leads to compartment specific dendritic spine deficits.
Our previous work examined pyramidal neurons in the MEC and spine density for their local processes, finding deep layer III pyramidal neurons had fewer spines on their secondary apical dendrites[9]. To get a more complete understanding, we wanted to compare neuron morphology and spine density across the entire dendritic tree of these distal MEC layer III pyramidal cells enriched in Lphn2 expression. To accomplish this, we injected Lphn2fl/fl littermate paired neonatal mice with either a control tdTomato AAV (CAG-tdTomato-AAV5) for neuronal labelling, or a cocktail of Cre-recombinase expressing and Cre-recombinase dependent tdTomato expressing AAVs (hSyn-CRE-AAV5; CAG-FLEX-tdTomato-AAV5) to elicit Lphn2 deletion in sparse neurons of distal MEC (Fig. 5A, B). We confocal imaged MEC layer III pyramidal neurons and their dendrites in different layers, then digitally reconstructed 3-dimensional images of secondary dendrites identified by basal/apical type and by their layer positioning. Using these reconstructions, we compared MEC layer III control and Lphn2 deleted pyramidal neurons for their secondary basal and apical dendrites in deep MEC layer III and superficial I (Fig. 5C) where Lphn2 protein is most enriched[9]. With this analysis, we find that spine density on secondary basal dendrites to be unaffected with Lphn2 deletion, but secondary apical dendrites in both deep layer III and superficial layer I exhibited a decrease of spine density (Fig. 5D-E). Together this shows that Lphn2 expression in layer III MEC pyramidal neurons is important for post-synaptic spine development in deep layer III and for apical tuft dendrites that are positioned in layer I for terminal dendritic branches on the apical shaft.
Lphn2 deletion in MECIII neurons impairs spatial-sequence learning.
To examine whether Lphn2 function in MECIII neurons affects animal behavior, we next set out to test if this genetic manipulation impacts entorhinal cortex dependent episodic learning and memory. Previous work that targeted Lphn2 deletion to the CA1 hippocampal region, was found to result in the learning deficits for temporally separated sequence spatial tasks (i.e. water T-maze), but no effect on spatial learning and memory (i.e. novel object recognition, novel area exploration, Barnes maze). We then asked, is MECIII Lphn2 expression similarly required for temporal/spatial sequence learning? To test this hypothesis, we set out to perform a list entorhinal cortex dependent behaviors in sex/gender matched littermate pairs of control and MECIII Lphn2 deleted mice (Fig. 6A). First, we measured novel area exploration with a 5-minute open field test, dividing the arena into central and peripheral zones with behavior tracking software. We analyzed movement speed, distance, and periphery distance to determine if there were differences in locomotor activity in these mice, or anxiety like behavior indicated by an increase of time spent in the periphery of the arena[24]. We did not observe any significant differences in distance traversed, distanced traversed in the periphery, or movement speed of the mice due to genotype (Fig. 6B). Next, we performed a novel object recognition task, implicated as dependent on entorhinal cortex connections with the hippocampus[25]. Using a discrimination ratio to compare time spent with novel and familiar objects 1 hour after exposure to familiar objects, we did not find significant differences between groups due to genotype (Fig. 6C). Third, we proceeded to test spatial learning and memory, by performing a Barnes maze behavior analysis[26]. We subjected these same mice to the Barnes Maze where they are placed in a brightly lit platform containing 20 evenly spaced closed holes around the perimeter (Fig. 6D). Visual cues were placed outside of the maze to help orient the mice, and one of the platform holes (target) was opened to provide a darkened space under the platform. For 4 days, mice were tested 5 times each day and measured for their ability to find the target hole at a fixed location. On day 5, a single trial (test 1) was performed to examine short term memory. Mice were then home cage housed for 2 weeks, after which another single trial was performed to later to assess long term spatial memory (test 2). With these experiments, we found no changes in learning the task (Fig. 6E), nor in the short-term or long-term memory recall test trials (Fig. 6F). Lastly, we then performed a spatial task that assesses animal ability to be cognitively flexible and learn sequences of spatial events, the water t-maze test (Fig. 6G). This assay incorporates an alternating spatial task with two swim trials separated by a 30 second time delay, in which in which the mice must learn the escape platform on the second trial is always on the opposite side of the first randomized trial (Fig. 6G). In performing this assay, we find that Lphn2 MECIII deleted mice exhibited a longer latency to find the escape platform upon reversal in the 2-arm maze during the first day of training but not subsequent training days (Fig. 6H-I). This suggests that MECIII deletion of Lphn2 impairs cognitive flexibility learning rates, but animals are able to retain these spatial-sequence memories once the task is acquired. Altogether, these behavioral assays indicate Lphn2 MECII deletion results in a selective impairment in spatial/temporal sequence learning tasks that requires cognitive flexibility, while other entorhinal cortex dependent tasks are unaffected.