Background
Gestational diabetes mellitus (GDM) is a state of glucose intolerance and hyperglycemia with first onset during pregnancy and is the most common complication of pregnancy, affecting up to 10% of expectant mothers [
1]. The prevalence of GDM is on the rise since more women are overweight or obese when they become pregnant. The combination of human cohort studies and research using rodent models with controlled pre- and postnatal conditions has demonstrated that diabetes during pregnancy during critical stages of development affects health outcomes in the offspring [
2]. Several human cohort studies have linked diabetes during pregnancy with impaired cognitive abilities [
3‐
6] and psychiatric disorders [
7,
8] in the offspring. Similar correlations exist between poor quality early childhood diets and cognitive performance [
9,
10]. Brain development is particularly sensitive to environmental influences during infancy [
11]. Animal studies utilizing controlled conditions confirmed that maternal high-sugar intake caused deficits in cognitive functions in the offspring [
12,
13]. However, the mechanisms that link GDM to the development of cognitive impairments in the offspring remain to be elucidated.
GDM is associated with mild but chronic systemic inflammation [
14] and elevated circulating free fatty acids [
15,
16]. Although the transfer of cytokines across the placenta from mother to fetus is restricted [
17], maternal hyperglycemia and hyperlipidemia directly impact the fetus through the placental overproduction of pro-inflammatory cytokines [
14,
18]. We have also recently shown that obesity-associated GDM promotes the production of the pro-inflammatory cytokine, interleukin (IL)-1β, and toll-like receptor activity, in the spleen cells of neonatal and 15-week-old offspring that may contribute to chronic low-grade systemic inflammation [
19]. Cytokines can induce pro-inflammatory responses in the central nervous system (CNS) resulting in neuroinflammation, which includes the release of cytokines, chemokines, adhesion molecules, free radicals, and destructive enzymes. Activation of microglia, resident brain immune cells, is a central event and driving force in neuroinflammation [
20]. Pro-inflammatory responses of activated microglia can jeopardize neurogenesis and promote neurodegeneration [
21]. Microglia have an important role in neuroplasticity as they assist in the development and maintenance of neuronal networks via trophic factor release and controlled phagocytosis, which allows microglia to prune synapses and remodel neuronal transmission [
22,
23]. Microglial pro-inflammatory responses [
20] can also indirectly affect the neuronal vitality by changing the function of astroglial cells, which have supportive role in neuronal metabolism and modulation of synaptic transmission [
24]. GDM-induced inflammation during embryogenesis could perturb microglial functions resulting in a limited ability to support the development of a healthy neuronal network that detrimentally influences brain development in the offspring.
We utilized a high-fat and sucrose (HFS) diet-induced GDM model [
19,
25] to evaluate the effects of GDM exposure and postnatal diets on cognitive abilities and neuroinflammation in the offspring. In this GDM model, hyperglycemia is observed in mid-gestation and resolves after the dam litters out, thus mimicking some features of the clinical presentation of GDM in overweight and obese women [
19,
25]. We determined that GDM exposure impaired recognition memory in the offspring and when combined with postnatal HFS diet, resulted in atypical inattentive behavior. These cognitive changes were associated with reduced density and derangement of the Cornu Ammonis 1 (CA1) pyramidal neuronal layer, decreased hippocampal synaptic integrity, and increased neuroinflammation. Exposure of primary cultures of microglia to elevated levels of glucose and/or fatty acids triggered pro-inflammatory responses capable of disturbing development of neuronal circuitry and functions, suggesting that gluco- and lipotoxicity contribute to microglial-mediated neuroinflammation.
Discussion
Neurons in the CNS require a healthy microenvironment to survive, which is a central function of microglia, the resident immune cells of the brain. In the adult brain, microglia phagocytose invading pathogens and clear away debris to maintain a healthy CNS. During brain development in utero, microglia control the wiring of the brain by performing synaptic pruning and assisting in synapse maturation [
23,
46]. Microglial phagocytosis and cytokine release profile also affect neurogenesis and neuronal viability [
20,
46]. In obese adult humans and rodents, a prolonged hypercaloric challenge coincides with activation of microglia, which is initially transient but, when sustained, leads to detrimental effects on the surrounding neurons [
47].
Our data provides the first evidence that in the rat, maternal obesity-associated GDM influences microglial activation and neuroinflammation in newborn offspring. Moreover, we show for the first time that microglial activation in GDM offspring is sustained into young adulthood and is associated with astrogliosis and derangement of the hippocampal CA1 layer. Interestingly, these alterations in neuronal arrangement in the young adult GDM offspring were associated with impaired recognition memory, reduced anxiety level, and inattentive behavior. Therefore, offspring of GDM dams exhibit many of the cognitive changes reported in some clinical cohort studies of populations of children from mothers diagnosed with diabetes during pregnancy [
3‐
6].
Our study used obesity-associated GDM model, in which dams are fed a diet enrich in high levels of fats and simple sugars. The diet increased the weight of the dams and caused a pregnancy-driven hyperglycemia and hyperinsulinemia, as the glucose intolerance starts after mid-pregnancy [
19,
25]. The offspring of our GDM mice have elevated weight gain and, at 15 weeks, show mild insulin resistance, but no glucose intolerance or hyperglycemia [
2,
25]. While these symptoms in both dams and offspring are consistent with the clinical presentation of obesity-associated GDM, our GDM rat model does not recapitulate all of the aspects of non-obese GDM that has different etiology [
48].
At present, little is known about how excess saturated fats and simple sugars affect microglial activation. Recent research has shown that high-glucose levels increase microglial activation [
49] and cytokine production [
50,
51]. Similarly, a diet rich in saturated fats [
52] as well as intracerebral ventricular injections of saturated fatty acids [
53] stimulated microglial activation and inflammatory responses. Previous clinical research [
54,
55] and results from experimental mouse models [
56] showed that diets high in saturated fat and simple sugars lead to cognitive impairments. Obesity has also reported to induce microglial activation and IL-1R-mediated deficits in hippocampal synaptic plasticity in mice [
57]. In agreement with these findings, we observed elevated neuroinflammatory status (microglial morphological activation and astrogliosis) in lean-HFS offspring. We extend these findings to show for the first time that GDM induced by a diet high in saturated fat and simple sugars during pregnancy caused reactive gliosis and elevated pro-inflammatory cytokine (IFN-γ, IL-1α, MCP-1, TNFα) levels in the neonate offspring. Reactive gliosis was maintained until early adulthood; however, inflammatory cytokine levels in the brain were not. Together, this data suggested that fetal exposure to maternal obesity-associated GDM induced robust neuroinflammation in the offspring brain during the critical period of hippocampal synaptic development (up to 3–4 weeks of age). Interestingly, derangement of the CA1 pyramidal neurons and behavioral changes in the GDM offspring were observed only when a postnatal HFS diet was consumed, suggesting that GDM conditions microglia to be in an activated state and the addition of the HFS diet further damages the hippocampal CA1 synaptic organization and maintenance. Previous research using primary microglial cultures showed that raising the glucose concentration from 25 to 50 mM or from 10 to 25 mM increased TNFα secretion two- or [
51] fourfold [
50], respectively. However, 25 mM glucose is rarely observed in the brain [
44]. The glucose levels in adult are about 70–80% lower in CSF than in plasma and do not generally exceed 16.7 mM (300 mg/dl) [
58]. However, in neonates and infants, CSF/plasma ratio is higher than in adults 0.88–1.1 [
59]. As glucose levels fluctuate due to feeding patterns, we decided to use a 5.5 mM (100 mg/dl) glucose as a normoglycemic condition and significantly higher, 16.7 mM glucose (300 mg/dl) as a high-glucose condition in order to capture the fluctuation maximum in these 24-hour microglial culture experiments in order to assess the effects of gestational glucose exposure. This glucose increase (from 5.5 to 16.7 mM) stimulated morphological activation of microglia without increasing TNFα production, nor other cytokines in general. On the other hand, we observed that elevated palmitate markedly stimulated microglial activation including pro-inflammatory cytokine production, in agreement with previous findings [
52]. While the combination of HP and HG triggered superoxide production and morphological activation, it only increased IL-1α cytokine release, whereas HP alone elevated most of the measured pro-inflammatory cytokines. The combination of HP and HG maintained anti-inflammatory cytokines at basal levels (IL-4 and IL-10) or below (IL-13), while HP alone increased IL-4 and 10 levels. While the fold changes in cytokine levels in our study were not profound, the singular cytokine levels may not be as crucial as the sum of the changes in the whole cytokine profile. Based on these findings, elevated saturated fatty acid concentrations appeared to contribute a larger effect on microglial activation and neuroinflammation than elevated glucose levels. This is not surprising given that microglia express the toll-like receptor [
53] as well as lipoprotein lipase and acyl-CoA synthases that facilitate fatty acid uptake by cells [
51]. Nonetheless, additional studies are required to elucidate the effects of glucose and fatty acid supplementation on microglial cytokine production.
While our study provides strong evidence that microglial actions promote the pathological changes in the hippocampus of GDM offspring, it is important to keep in mind that these changes are likely multifactorial and represent a combination of elevated maternal glucose, fatty acids, and inflammation/cytokines. In fact, a HFS diet has been reported to reduce hippocampal neuronal viability via affecting microvascular insulin sensitivity [
60] and fatty acids can be directly neurotoxic [
61,
62].
Hippocampal microglial numbers were increased in lean-HFS as well as GDM-LF offspring, while the combination of GDM exposure and a HFS postnatal diet reduced microglial cell numbers. One explanation for this observation is that the HFS diet promotes GDM induced chronic microglial activation to a level where chronic pro-inflammatory activation jeopardized microglial viability. It has been postulated that microglial senescence and lack of microglial supportive functions could lead to neurodegeneration associated with various neurodegenerative diseases [
63]. Cytokine analysis of the young adults partially supports this notion, since IL-10, 4, and 13 levels were reduced in the GDM-HFS offspring compared to the GDM-LF offspring.
An additional explanation for the cognitive deficits seen in young adult GDM offspring was reduced levels of synaptic proteins and a drastic decrease in microglial expression of the fractalkine receptor, CX3CR1. The fractalkine (CX3CL1)-CX3CR1 signaling axis promotes microglial remodeling of synaptic circuits by regulating microglial motility, trophic factor release, and phagocytosis activity [
43]. Microglial CX3CR1 expression in the developing CNS is essential for hippocampal synapse development [
23], synaptic plasticity, and cognitive functions [
64]. In neurodegenerative disease models, CX3CR1 depletion was reported to sustain microglial inflammatory responses and degrade synaptic circuits, thus promoting neurodegeneration [
65]. Our in vivo findings are in line with these findings and reduced CX3CR1 expression provides a logical mechanistic explanation linking microglial changes to hippocampal synaptic degradation, neuronal derangement, and cognitive changes in GDM offspring. A recent report further supports the view that prenatal stress can program microglial CX3CR1 expression and promote microglial pro-inflammatory responses [
66]. Given that neuroinflammation is largely driven by microglia, and our data shows that GDM causes microglial activation that is associated with hippocampal neuronal weakening and cognitive deficits, a reduction in microglial CX3CR1 expression could be central in the sustained microglial activation observed in the 15-week-old offspring of GDM dams.
Acknowledgements
We thank Drs. Marc Del Bigio and Gilbert Kirouac for the assistance in the behavioral tests and Domenico Di Curzio and Mario Fonseca for the expert technical assistance.