Insulin signaling occurs when insulin activates the insulin receptor (IR), this ultimately results in glucose uptake by the cell. The process is orchestrated by intracellular signalling, including the phosphoinositide 3-kinase (PI3K) pathway and the phosphorylation of Akt/protein kinase B [1]. Insulin signaling within the brain plays important roles in cognition, learning and memory. When insulin signaling becomes dysregulated, such as in the case of insulin resistance and long-term compensatory elevations of circulating insulin, an increased risk for developing cognitive pathologies occur, such as Alzheimer’s disease [2]. The link between insulin resistance and cognition is shown in rodent studies which use the Morris Water Maze (MWM), a task used to evaluate spatial learning and memory [3]. Animals that present with insulin resistance have reduced performance in the MWM [4, 5, 6, 7, 8, 9, 10, 11, 12]. These deficits can be ameliorated by central insulin infusion, which can increase performance in the memory component of the MWM [13].

Within the central nervous system (CNS), the hippocampus, an area responsible for memory and learning, is a prominent target for insulin signaling. A high density of IRs within the dorsal hippocampus suggests a role for insulin in the formation of spatial memory [14, 15]. In support, CA1 hippocampal insulin administration enhanced memory consolidation and retrieval in the MWM, which is considered to specifically test for hippocampal-dependent cognitive deficits. This suggests that insulin directly regulates memory by acting on the hippocampus [14, 16]. Using a lentiviral vector to downregulate IRs in the rat hippocampus, Grillo and colleagues demonstrated that hippocampal insulin resistance results in deficits in hippocampal synaptic transmission and hippocampal-dependent learning [17]. Importantly, these deficits were seen to be independent of metabolic or endocrine imbalances. This study provides important evidence that hippocampal insulin signaling facilitates neuroplasticity and cognition [17].

Insulin signaling within the CNS declines with age and insulin resistance is a risk factor for age-related Alzheimer’s disease and cognitive decline [18]. With aging, a marked decrease in glucose transporters [19], insulin receptors [20], insulin levels [21] and insulin signaling [22] have all been observed, which has been linked to age related hippocampal memory impairments [23]. However, central insulin resistance and aging do not always go hand-in-hand, with factors such as body composition, lifestyle and exercise having important influences. Notably, central and peripheral insulin resistance do not always occur concurrently, as dysregulated CNS insulin signaling can precede, or help initiate, the onset of peripheral insulin resistance [24].

As well as aging, sex can also influence insulin signaling within the CNS. Male rats have increased anorexigenic sensitivity to insulin infusion compared to female rats [25], while acute intranasal insulin in female patients improved hippocampal-dependent memory, whereas males did not show an enhancement [26]. Another study examined individuals with mild cognitive impairments and demonstrated that men show an improvement in working memory after intranasal insulin, but only at a dose twice as high as shown to be effective in women [27]. Some of these sex differences appear to be the result of differential gonadal hormone levels [28]. Overall, a strong link between intra-hippocampal insulin signaling and cognitive performance has been identified, however, the specific neuronal population mediating these effects within the hippocampus has not yet been identified.

Neuropeptide-Y (NPY), a 36-amino acid peptide, is abundantly expressed throughout the brain, including the hippocampus [29]. NPY is involved in the regulation of biological and pathophysiological functions including feeding behaviours, neuroplasticity, memory and learning [30]. NPY has a modulatory role in spatial memory and learning as it appears to exercise both stimulatory and inhibitory effects on memory, contingent on the NPY receptor subtype manipulated, dose applied, neuroanatomical brain systems involved, temporal step (i.e., retrieval, acquisition, retention, consolidation) and memory type [31, 32, 33, 34]. Hippocampal NPY has been associated with spatial learning and memory during the MWM, as increased levels of NPY mRNA were observed in the dentate gyrus of the hippocampus following MWM exposure [35]. Further, brains from Alzheimer’s patients show loss of NPY-positive neurons in the hippocampus [36], while NPY injections into the dorsal hippocampus increases memory retention in mice [37]. Altogether, these data show that NPY and insulin signaling play important roles for learning and memory by signaling within the hippocampus.

To date, it is unknown how disruption of insulin signaling in NPY expressing neurons affects performance in hippocampal dependent cognitive tasks, such as the MWM. To address this, we utilised the Cre-lox recombination technique in mice to selectively knock out IRs in NPY expressing neurons (IR${}^{lox/lox}$;NPY${}^{Cre/+}$). Mice were tested in the MWM at 6, 12 and 24 months of age to assess how aging influences behavioural deficits induced by ablated IRs on NPY neurons.

In the present study, we aimed to determine if the loss of IRs in NPY-expressing neurons negatively affected cognitive performance in the MWM, a robust measure of hippocampal-dependent memory performance. Because cognitive performance can be affected by aging, amongst other factors, we tested mice at 6, 12 and 24 months of age. We then examined p-Akt as a measure of insulin signaling in our knockout IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mouse model. This mouse model has been validated previously and is a robust measure of IR deletion in NPY neurons [38]. This was demonstrated by the authors upon ICV infusion of insulin which caused downstream activation of p-AKT in IR${}^{lox/lox}$ mice but not IR${}^{lox/lox}$;NPY${}^{Cre/+}$mice. NPY cells were genetically tagged with a mCherry marker which showed extensive overlap with p-Akt upon ICV insulin infusion in IR${}^{lox/lox}$ mice [38].

Consistent with previous research, male IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice had significantly increased body weight compared with IR${}^{lox/lox}$ control mice [38]. This increased body weight may be due to two reasons; a loss of insulin signaling leading to an upregulation of the orexogenic NPY [45] and reduced energy expenditure in IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice [38]. In contrast to previous work, no difference in energy intake was observed between genotypes, however, this could be due to differences in study design [38]. Interestingly, differences in body weight by genotype were only observed in male mice at 6 months of age, suggesting that disrupted homeostatic signalling in IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice is reduced in later life. It is unclear why female mice did not show similar body weight differences between genotypes, however it is known that sex plays a role in insulin-dependent regulation of energy homeostasis [25]. Despite sex-differences in body weight at 6 months old, there were no sex differences observed in either genotype during MWM performance, hence the data for male and female were combined. This is in contrast to both rodent and human studies showing that sensitivity to ICV insulin infusion [25] or intranasal insulin [27] affects males and females behavior differently. While this is interesting, our data clearly show that disruption to endogenous insulin signaling is not significantly affected by sex, at least during hippocampal-dependent memory tasks.

In the MWM, IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice spent less time in the target quadrant compared to IR${}^{lox/lox}$ control mice during the probe trial at 6 and 12 months of age. Importantly, these differences were not a consequence of reduced motivation or locomotor abilities of mice, as all mice showed similar escape latencies during learning of the MWM task. Knockout of IRs on NPY neurons did not affect spatial learning during the hidden platform days, therefore this mouse model has specific deficits of spatial memory retrieval during a hippocampal-dependent memory task. These data are consistent with a previous study which found that insulin-resistant mice present with hippocampal memory impairments but intact spatial learning on the MWM [13]. Moreover, intrahippocampal insulin administration improves spatial memory performance in the probe test [13, 46]. Importantly, while these studies show that exogenously administered intra-hippocampal insulin significantly improves cognitive functioning, our results demonstrate that endogenous IR signaling has similarly vital roles in maintaining appropriate cognitive functioning. In line with our work, disrupted or ablated IR signaling is correlated with neuroinflammation and cognitive deficits, including reduced spatial memory acquisition [47, 48], while administration of intra-hippocampal PI3K inhibitors impaired memory retrieval [49].

Through the use of our conditional IR${}^{lox/lox}$;NPY${}^{Cre/+}$ knockout mouse model, we conclusively determined that insulin’s actions in promoting hippocampal function are, at least in part, mediated through NPY cells. The hippocampus is rich in NPY neurons expressing IRs [50, 51, 52, 53] and is also crucial in spatial navigation and memory formation [54, 55]. NPY signalling can enhance stem cell proliferation and neurogenesis via the Y1 receptor within the dentate gyrus [56] and it is possible that insulin signaling may potentiate NPY neuronal activity within the hippocampus to support learning and memory. While these findings do not exclude the contribution of other cell types in mediating hippocampal-dependent cognition, they do clearly indicate that IRs on NPY-expressing cells are an instrumental part of a spatial memory circuit.

Interestingly, during the probe test, no difference between IR${}^{lox/lox}$;NPY${}^{Cre/+}$ and IR${}^{lox/lox}$ controls were observed in the time spent in the target quadrant at 24 months of age. Together, both groups also scored significantly lower when compared to their probe trial at 6 and 12 months of age. Similarly, both groups of mice swam similar path lengths at 6 and 12 months of age, however at 24 months both groups of aged mice exhibited decreased path length in the target quadrant, consistent with a floor in performance. These results may be explained due to age-related effects on insulin sensitivity and cognition. Consistent with these findings, other work has found that while young rats show improved performance on the MWM after insulin infusion, aged rats do not show sensitivity to the effects of insulin [46]. This could be due to an age-related decline in IR-$\beta$ immunoreactivity within the hippocampus [57]. Aged animals also exhibit upregulated pro-inflammatory cytokines such as IL-1$\beta$, TNF-$\alpha$ and IL-6 in the hippocampus, which have been shown to directly impair insulin receptor activity and signaling [58]. These pro-inflammatory cytokines may also contribute to age-related hippocampal insulin resistance [59]. Therefore, our aged IR${}^{lox/lox}$ mice may have cognitive decline equivalent to the effect seen with loss of IR on NPY cells due to aged-related hippocampal insulin insensitivity. To support our MWM data, we identified that IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice exhibit reduced p-Akt in the dentate gyrus of the hippocampus at 6 months of age, suggesting that these mice exhibit down-regulated intra-hippocampal IR signaling. Reduced p-Akt is also seen in states of central insulin resistance [60], which has been implicated with the cognitive decline and cognitive pathologies seen in Alzheimer’s disease [61].

There are some limitations in the present study. This work focuses on the role of IR on NPY cells within the hippocampus, however NPY is expressed elsewhere in the brain, such as the hypothalamus [62]. Nevertheless, the present study uses the MWM as a test model, which is known to specifically test hippocampal-based spatial learning and memory [3]. Another consideration is that p-Akt signaling immunohistochemical analysis between IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice and IR${}^{lox/lox}$ mice was only undertaken at 6 months of age. Considering both genotypes of mice exhibited decreased path length and time spent in the target quadrant at 24 months of age, it would be interesting to determine if this memory deficit correlates with impaired p-Akt signaling in both IR${}^{lox/lox}$;NPY${}^{Cre/+}$ mice and IR${}^{lox/lox}$ mice.

The present study has shown that mice with a tissue specific knockout of IRs in NPY expressing neurons (IR${}^{lox/lox}$;NPY${}^{Cre/+}$) demonstrated impaired performance in the probe trial of the MWM compared with control mice at both 6 and 12 months, supporting previous studies that have shown IRs play crucial roles in spatial memory. Importantly, we present novel evidence that these effects are mediated by NPY-expressing cells. Interestingly, no difference between genotypes was observed in aged (24 months old) mice. Further research is required to determine age-related physiological changes in IR signaling which modulate spatial memory and learning. Together, this data provides valuable insights into how IR signaling in discrete regions affects cognition, which could have important implications for pathologies related to insulin resistance, such as obesity and Alzheimer’s disease.

CNS, Central Nervous System; GTT, Glucose Tolerance Test; IR, Insulin Receptor; ITT, Insulin Tolerance Test; MWM, Morris Water Maze; NPY, Neuropeptide Y.

DB and HH designed the research study. EG, JT, JG and KA performed the research and analysed the data. EG, CM, DB, NR and LZ wrote and reviewed the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Animals were raised and handled at Australian BioResources Ltd, Moss Vale, NSW, Australia, with genotyping also being performed at this facility. Animals were transferred to University of New South Wales for approved experiments. Experimental procedures were approved by the University of New South Wales Animal Care and Ethics Committee in accordance with the Australian Code of Practice and Use of Animals for Scientific Purposes (ACEC 16/21A).

We thank the anonymous reviewers for the constructive criticism of this article.

This work was supported by the Australian Research Council (DE160100088 and DP170100063) and a Ramaciotti Foundation Establishment Grant.

The authors declare no conflict of interest.