Age-Related Onset of Obesity Corresponds with Metabolic Dysregulation and Altered Microglia Morphology in Mice Deficient for Ifitm Proteins

The IfitmDel mouse lacks all five of the Ifitm genes via LoxP deletion. This animal breeds normally with no obvious defect in development. The IfitmDel animals exhibit a steady and significantly enhanced weight gain relative to wild-type controls beginning about three months of age and under normal feeding conditions. The increased weight corresponds with elevated fat mass, and in tolerance tests they are hyporesponsive to insulin but respond normally to glucose. Both young (4 mo) and older (12 mo) IfitmDel mice have enhanced levels of serum leptin suggesting a defect in leptin/leptin receptor signaling. Analysis of the gene expression profiles in the hypothalamus of IfitmDel animals, compared to WT, demonstrated an altered ratio of Pomc and Npy neuropeptide expression, which likely impairs the satiation response of the IfitmDel animal leading to an increased eating behavior. Also elevated in hypothalamus of IfitmDel mice were pro-inflammatory cytokine expression and reduced IL-10. Anatomical analysis of the hypothalamus using immunohistochemistry revealed that microglia exhibit an abnormal morphology in IfitmDel animals and respond abnormally to Poly:IC challenge. These abnormalities extend the phenotype of the IfitmDel mouse beyond abnormal responses to viral challenge to include a metabolic phenotype and weight gain. Further, this novel phenotype for the IfitmDel mouse could be related to abnormal neuropeptide production, inflammatory status and microglia status in the hypothalamus.


Introduction
The interferon-induced transmembrane gene family (Ifitm) consists of four genes in humans and five in the mouse that encode very similar proteins of 40-60 residues. Each Ifitm protein consists of a unique extracellular N-terminus, a highly conserved transmembrane domain, and equally well-conserved hydrophilic (cytoplasmic) domain followed by a much more diverse transmembrane-like domain [1]. Although the Ifitm proteins can be detected with sequences on the outside of the plasma membrane, an alternative prediction is that they are embedded on the cytoplasmic side of the membrane [2]. The proteins possess a number of functional sites including cysteine residues that are palmitoylated, lysine residues that are ubiquitinylated and serine/threonine residues that are phosphorylated following cellular activation ( [1,2],unpublished data). Ifitm cell location includes on the surface as well as inside the cell where they are associated with endosomal and golgi membranes [3][4][5]. As a class, the Ifitm proteins have a proclivity to bind to multi-spanning/tetraspanin proteins such as CD81 and CD9 [6,7].
Of the Ifitm gene family members, Ifitm3 is the best characterized and it shows the greatest transcriptional response to type I and type II interferon induction [8]. Ifitm3 was shown in a broad siRNA screen to be essential for the interferon-induced cellular resistance to viruses that infect from the endosomal compartment to the cytoplasm such as influenza and dengue [4,5,9,10]. A defective human IFITM3 allele has been linked to increased severity of human infections to influenza virus [11] and we have recently shown this same allele is linked to coronary heart damage associated with Kawasaki Disease, an immune inflammation of unknown initiation [12]. A number of models have been proposed to describe the function of Ifitm3 in providing resistance to cellular infections including building a protein lattice in the membrane to block endosomal exit, blocking fusion pores during virus-endosome hemifusion, enhancing the deposition of cholesterol to also block virus exit or by blocking virus entry by enhancing the stability of the clathrin/vATPase complexes on the endosomal membrane [13][14][15].
The mouse Ifitm gene family encompasses about 65,000bp on mouse chromosome 7. This section of the chromosome has been removed by LoxP mediated deletion to create the IfitmDel animal which lacks all five of the Ifitm genes [16]. No other coding sequences or functional noncoding RNA's are included within this section of the genome. The IfitmDel animal was originally created to test the necessity of the Ifitm proteins for germinal cell speciation [17][18][19] and embryo generation [20]. IfitmDel animals are generated in normal Mendelian numbers and have few if any obvious defects in development and survival [16]. We have made extensive use of these animals to study the roles that the Ifitm proteins have in immune signaling pathways. As we maintain these animals as homozygous deletion lines, over time we have observed a pronounced enhanced weight gain and an obesity phenotype (e.g., [21][22][23]) in older IfitmDel mice compared to C57BL/6 controls. In this report we quantify the obesity phenotype and link this to altered leptin/neuropeptide signaling, and demonstrate abnormal microglia morphology in the IfitmDel animal.

Animals
The mice were housed and used for this study in accordance with protocols approved in advance by the Institutional Animal Care and Use Committee at the University of Utah (Protocol Number (09-07003). In all cases animals were maintained in according to the Guide for the Care and use of Laboratory Animals of the National Institutes of Health. IfitmDel mice were backcrossed for greater than 10 generations to the C57BL/6 strain. Background-and age-matched littermates were used as WT controls. These mice were fed ad libitum with normal chow. For food intake studies, mice were kept individually and a similar amount of normal chow was given to each mouse. Average three-day consumption of food was measured for 21 days.

Metabolic studies
For insulin and glucose tolerance tests, mice fasted 5 hours followed by intraperitoneal injection of human recombinant insulin (1U/kg, Novalin R) or glucose (1.5g/kg, Sigma) respectively. Blood levels were measured at indicated time points by FreeStyle Lite blood glucose monitoring system. Blood samples were obtained from tail bleeds.
For blood tests of fasting and fed animals, including blood glucose and leptin, blood was obtained at 10 am and 10 pm (12h) for analysis.
Energy expenditure and locomotor activity were analyzed by indirect colorimetry using CLAMS metabolic cages, HSC Cores Research Facility at University of Utah. Body composition was determined by nuclear magnetic resonance (NMR Bruker Minispec).

Histology, immunohistochemistry and microscopy
The epididymal fat pads were dissected from animal and fixed immediately in 4% paraformaldeyde in PBS overnight at 4°C. Next day, tissues were dehydrated with serial alcohol solutions (50%, 70% and 100%) at room temperature for 2 hours each. Tissues were then infiltrated in Immuno-bed resin (Polyscience, Inc.) at room temperature overnight. Tissues were moved to fresh Immuno-bed resin with catalyst (Polyscience, Inc.) at 1:25 ratio and polymerized in the mold at room temperature overnight. Tissues were carefully removed from the mold and sectioned to 3 microns with a rotary microtome (Thermo Scientific, Microm HM310). Tissue sections were adhered to microscope slides and proceed to H&E staining according to manufacturer's instructions (VWR). Stained slides were mounted with cytoseal mounting medium (Fisher Scientific) before imaging. Pictures were acquired with Zeiss Axiovert 100 microscope equipped with a Microfire CCD camera (Optimetrix). Brain sections were prepared as before [28][29][30]. Briefly, mice were lethally anesthetized with tribromoethanol (TBE) by intraperitoneal injection. For perfusion, the right ventricle of the heart was punctured and 10mL of saline was perfused into the left ventricle, followed by 20mL of 3% paraformaldehyde (PFA, EMS) plus 5% sucrose in PBS. The brains were post-fixed with 3% paraformaldehyde plus 5% sucrose solution and subsequently infiltrated with 15% and then 30% sucrose. Brain tissues were embedded in 2% gelatin (Sigma-Aldrich) and sectioned at 10μm with the Thermo Scientific HM 550 Cryostat (ThermoFisher Scientific). Sections were permeabilized with 0.02% Triton (Sigma) and stained with Iba1 (Abcam), F4/80 (eBioscience), CD11b (eBioscience), GFAP (Abcam), O2A (Abcam) and MAP-2(Abcam) followed by secondary antibody FITC-or PE-conjugated anti-rabbit antibodies or Alexa Fluor 546 Goat Anti-Rat IgG (invitrogen) as appropriate.
BMDM were isolated and cultured using standard protocols (14). Cells were fixed with 4% paraformaldehyde and permeabilized with 0.02% Triton (Sigma) and stained with Iba1.

Results
The IfitmDel animals demonstrate enhanced adiposity The IfitmDel strain possesses a defined and engineered (via Cre-Lox) chromosomal deletion of all five of the Ifitm genes on a mouse genetic background of C57BL/6 [16]. No coding sequences or regulatory sequences other than those associated with the Ifitm genes are known to be lost in this deletion. Mice lacking these Ifitm genes are fertile and thrive in colonies except for the welldescribed sensitivity to viral infections. In maintaining such mice in our colony we noted that the older IfitmDel animals were generally larger than their WT (C57BL/6) age and sex matched counterparts. To quantify these differences, body weights were taken from male mice maintained in the colony on regular mouse chow. Fig 1A, shows the slow but significant enhanced weight gain associated with the IfitmDel animals compared to WT. This body weight increase corresponded with an increased total fat mass as the animals aged ( Fig 1B). Measurement of 3 day food intake averages for a period of 21 days shows that the IfitmDel mice consume more chow than the WT age-matched controls ( Fig 1C). Also increased levels of epididymal fat deposits were present in these ( Fig 1D). Histological examination suggested the adipocyte size was increased compared to WT adipose tissue ( Fig 1E) and this was confirmed by quantitation as shown in Fig 1F. The same trends were observed for female mice (not shown).

The IfitmDel animals demonstrate metabolic alterations
We next determined if the enhanced body weight of the IfitmDel animals was associated with altered physiological stasis. Blood glucose levels of fasting 2 month and 9-month-old male IfitmDel animals were measured and observed to be consistently higher than WT (Fig 2A). However, this difference was not observed under normal feeding (fed) schedules. This difference in fasting blood glucose was clear in the 2 month old animals.
Two standard tests to measure blood glucose levels are the insulin tolerance test (ITT) ( Fig  2B) where blood glucose levels are quantified in fasted mice following a single injection of insulin and the glucose tolerance test (GTT; Fig 2C) in which blood glucose levels are quantified following a single injection glucose [31]. As shown, the IfitmDel animals responded poorly in the ITT, maintaining higher levels of blood glucose than the WT animals. The IfitmDel and WT animals were indistinguishable in their response in the GTT. The same trends were observed with female IfitmDel animals (not shown). These results suggest that as the IfitmDel animals age, they enter into a metabolic syndrome with a more moderate phenotype than animals displaying morbid obesity [23].
The IfitmDel animals were also analyzed using metabolic chambers that quantify oxygen consumption (V02), heat production, and activity (movement). As shown in Fig 2D the Ifitm-Del animals have reduced metabolism compared to the WT animals as reflected by lower oxygen consumption and decreased heat production in both the light and dark cycles. The activity of the IfitmDel animal in the cage trended towards less than the WT in the 12 month age group, but these differences were not statistically significant.

The IfitmDel animals possess elevated levels of serum leptin
Mammals with a metabolic syndrome phenotype often have altered levels of leptin in the blood stream [32]. The leptin cytokine is critical in regulating appropriate food uptake [22,33,34] and we have shown (Fig 1C) that the IfitmDel mice have increased food intake. This level of food intake, however, is significantly less than observed with the ob/ob or db/db mice. To assess leptin levels in the IfitmDel and control mice, two sets of male animals (4 and 12 months) were analyzed ( Fig 3A). In order to correct for the differences in body weight, serum leptin levels are shown as ng per μl per gram of lean body mass. As shown in Fig 3A, the blood leptin levels were dramatically elevated in the IfitmDel animal compared to WT. When blood leptin levels in mice of the same weight (e.g., IfitmDel vs WT weighing 35 g), but differing age, were compared the IfitmDel animals still consistently have higher levels of blood leptin compared to WT controls (not shown). Leptin is primarily produced by fat cells and serves, via the leptin receptor expressed in the hypothalamus, to regulate food intake [23,34,35]. Animals lacking leptin (ob/ob) or a functional leptin receptor (db/db) become morbidly obese due to their lack of control over food intake [34].
Leptin's actions are largely through interaction with leptin receptors expressed in the hypothalamus, which makes this brain region a target for the interaction between leptin and Ifitm. To confirm that Ifitm genes are expressed by cells of the hypothalamus, the expression of Ifitm1, Ifitm2, Ifitm3 and Ifitm6 was quantified in the WT hypothalamus samples (Ifitm5 is only expressed by osteoblasts). As shown in Fig 3B, Ifitm2, Ifitm3 and Ifitm6 are all expressed in WT tissue, with Ifitm2 and Ifitm3 showing dramatically elevated expression in older animals. Therefore, the correlation between the deletion of Ifitm genes and leptin levels occurs in both young and older mice, the older IfitmDel mice have the added impact of not having the normal increase in these proteins seen during the aging process.
The leptin receptor, expressed by neurons in the hypothalamus, signals through a Stat3 dependent pathway, controlling the expression of two contrasting neuropeptides [34]. These are the proopiomelanocortin (Pomc) gene whose expression is increased upon leptin signaling while the transcription of the neuropeptide y (Npy) gene is depressed [36][37][38][39]. Bio-active peptides from the Pomc protein serve to dampen eating while those of Npy have the opposite effect and promote food intake. The expression of Pomc and Npy gene transcripts were analyzed from hypothalami obtained from WT and IfitmDel animals of various ages from 1 to 12 months of age. As shown in Fig 3C, IfitmDel animals have significantly reduced Pomc transcripts than WT suggesting that the IfitmDel animals lack appropriate signals to depress eating. Further, the IfitmDel samples have trending higher levels of Npy transcripts compared to WT samples suggesting elevated Npy peptides may contribute to the positive signal to maintain feeding.
IfitmDel mice demonstrated altered responses to chronic Type I interferon induction Chronic Poly I:C treatment, via activation of Tlr3 and the RIG-I-like receptors, results in the production of IL-6 and IFNγ that can, when provided chronically as a model for cachexia, lead to a progressive weight loss [40]. When administered in vivo, the major cell types responding to Poly I:C include macrophages, dendritic cells (DC) and microglia. Additionally responding DC's can undergo necroptosis that can exacerbate the inflammatory response. Metabolic dysfunction and neurological disorders have also been linked to cachexia, especially as the outcome of chronic infections and cancer metastasis that can lead to the chronic release of inflammatory cytokines. Based upon the previously described findings that the absence of the Ifitm proteins can alter cellular induction pathways following type I interferon treatment, we tested whether or not the IfitmDel animals would have an altered response, compared to WT, to chronic Poly I:C treatment. As shown in Fig 4A, 8 week old WT animals treated with Poly I: C over a time course of 28 days demonstrated the expected progressive weight loss, but age and sex matched IfitmDel animals were much more resistant to the cachexic effects of the Poly I:C treatment. While the IfitmDel animals treated with Poly I:C did lose weight, especially compared to their PBS-treated counterparts, the degree of weight loss was much less dramatic. To determine if the altered response to Poly I:C in the IfitmDel animals was also mirrored in altered cytokine responses, animals were treated with Poly I:C for 10 days (treatment every two days) weighed and analyzed. The hypothalamus was removed by dissection, total RNA isolated and cytokine expression by quantitative RT-PCR was measured. As a positive control, the IFNinduced protein with tetratricopeptide repeats (Ifit1), which is genetically and functionally distinct from the Ifitm proteins, was measured in response to type I interferon stimulated by Poly I:C injection. As shown in Fig 4B, WT and IfitmDel mice have equivalent levels of Ifit1 RNA following Poly I:C treatment indicating these two strains were equally responsive. However, the IfitmDel and WT animals displayed significant differences in TNFα and IL-1β mRNA levels with the IfitmDel animal showing significantly elevated levels of these inflammatory cytokines (Fig 4C and 4D). PBS mock activated mice of the WT and IfitmDel did not induce elevated cytokine levels in the hypothalamus (not shown). The expression of the IFNγ and iNOS genes was also elevated in the IfitmDel animals (Fig 4E and 4F) with differences trending toward significance. Interestingly the expression of the anti-inflammatory cytokine IL-10 showed lower levels of expression in the IfitmDel animal samples (Fig 4G). Finally the expression of F4/80 (Emr1), a macrophage/microglial marker, was compared between the WT and IfitmDel Poly I: C treated animals (Fig 4H). The increased level of F4/80 RNA in the hypothalamus of the Poly I:C treated IfitmDel animals could be for a number of reasons including enhanced recruitment of blood monocytes/macrophages to that anatomic site compared to similarly activated WT animals. Similar analyses were performed with RNA obtained from abdominal fat pads (not shown) which showed the same trend of inflammatory cytokine signature as that obtained from the hypothalamus samples. In total these cytokine profiles suggest the response to interferon induced by the Poly I:C treatment is intact but much more polarized towards a proinflammatory signature in the IfitmDel animal than WT. This is the case even though the weight loss in these animals upon Poly I:C treatment is less severe.

Immunohistochemistry of the WT and IfitmDel mice hypothalamus
The RNA signaling results (Fig 4) led us to examine the morphology of the hypothalamus which to our knowledge has not been examined in depth in the IfitmDel mice. We first assessed the hypothalamus in unstimulated mice using markers for microglia/macrophages (Iba1), astrocytes (GFAP), oligodendrocytes (O2A) and neurons (MAP2). Coronal serial sections from the brains of saline perfused and paraformaldehyde-fixed tissue of IfitmDel and control animals (Methods) were prepared from equivalent anatomical locations, stained for cell specific markers and photographed for visual analysis. Overall no gross abnormalities in brain anatomy were detected (data not shown), which is consistent with earlier reports [16] that the IfitmDel animals exhibited no developmental defects (although CNS morphology was not specifically characterized). The regional distribution, appearance and staining of neurons, astrocytes or oligodendrocytes was also equivalent between animals of both genotypes (Methods). However, microglia revealed by immunostaining for the expression of the Iba1 antigen exhibited a markedly altered morphology between WT and IfitmDel mice (Fig 5). Among the most notable morphology differences was the occurrence of dramatically elongated processes that were often observed to extend well over 150 microns (Fig 5B, dotted line). These processes were almost exclusively unipolar, and they exhibited infrequent branching and few varicosities. The cell bodies of these microglia also tended to appear as smaller and more poorly defined relative to controls in the hypothalamus (Fig 5A-5D). The altered microglia morphology of the IfitmDel mice was evident in other brain regions including the cortex and hippocampus (data not shown). Microglia are mesoderm/mesenchymal derivatives that share a bone-marrow macrophage lineage [41]. To determine if the morphological differences observed in the IfitmDel microglia are intrinsic to this cell type, bone marrow macrophages (BMDM) from controls and IfitmDel mice were prepared from culture. After 10 days, cells were gently rinsed with PBS, cells fixed with paraformaldehyde and immunostained to reveal Iba1 expression. As shown in Fig 5E and 5F, the morphology of The double arrow identifies a cell with an elongated cell morphology that also common to cultures prepared from either genotype. Both are typical to both preparations. The arrow head identifies a cell with morphology unique to the IfitmDel culture that is characterized by exceptionally long thin processes (dots) that resemble the microglia morphology seen in the brain. Bar = 20 microns. doi:10.1371/journal.pone.0123218.g005 a subpopulation of IfitmDel BMDMs exhibited strikingly similar features to those seen in the microglia. In particular, these BMDMs from IfitmDel BM produced very thin extensions that extended for long distances in the cultures. These processes were rare or not detected with the control cells (Fig 5E). Also evident is the impression that IfitmDel BMDMs were more compact and in general looked smaller than cells of similar morphology in the WT preparation. This is also evident in cells producing fan-shaped lamella which are more compact in IfitmDel BMDMs. Thus the curious morphology of the IfitmDel microglia was recapitulated in cultured bone marrow derived macrophages possessing the same genetic defect suggesting that common cellular abnormalities may be associated with the IfitmDel in vulnerable cells.
Having found that microglia exhibit altered morphology in IfitmDel animal and the hypothalamic transcriptional response to Poly I:C is also altered in this mouse (Fig 4), we examined by immunohistochemistry changes in the CNS induced by Poly I:C. WT and IfitmDel animals were activated with Poly I:C as above, sacrificed and perfused. A notable difference in the response of the WT compared to the Poly I:C treated mice was evident in choroid plexus structures within ventricles. Macrophages mobilize to the CNS upon inflammation and this is often evident by increased cellularity of the choroid plexus. In Fig 6 we show that Poly I:C treatment of WT and IfitmDel animals stimulates an increase in the accumulation of Iba1/CD11b positive cells (Fig 6C and 6D; i.e., macrophages) in the choroid plexus. Further, these cells are also Iba1/ F4-80 positive (Fig 6G and 6H), indicative of activation of macrophage lineage cells. A striking difference between WT and IfitmDel animals is the appearance of masses of these cells (Fig 6D   Fig 6 and 6H) that are not evident in the PBS animals (Fig 6A and 6B), Poly I:C treated WT animals (Fig 6E and 6F) or the PBS-treated IfitmDel animal (Fig 6C and 6D). Similar data were also obtained from stained sections of the median eminence of the ventral hypothalamic region (Fig  7), again showing enhanced macrophage/microglia staining in the Poly I:C treated IfitmDel animal compared to controls or sham activated IfitmDel animals. Collectively these findings suggest that the enhanced numbers of transcripts encoding for F4/80 in the hypothalamus of Poly I:C treated IfitmDel animals (see Figs 6H and 7H) may be due to increased recruitment and/or accumulation of inflammatory macrophages to the brain. Further, the activated macrophages in these regions at the site of blood-brain interfaces suggests that perhaps the macrophages do not enter the brain but may become entangled in these regions during normal migration.

Discussion
This report describes novel peripheral and central alterations associated with the lack of the Ifitm proteins. The IfitmDel animals lack all five of the Ifitm genes via an engineered genetic deletion without impact on other coding or control sequences that reside within the gene family locus. When bred as heterozygotes, the homozygous IfitmDel progeny are produced in the vivarium at normal Mendelian ratios and exhibit normal losses of the adult animals compared to WT C57BL/6 animals. What is clear, however, is that the longer the IfitmDel animals are maintained on normal chow diet, the more obese they become. This obesity is due to an enhanced accumulation of white adipose mass. We have not examined changes in brown adipose tissue in these animals.
While there are a variety of pathways that can lead to obesity, perhaps the best characterized are the aberrant feeding behaviors associated with alterations in the leptin/leptin receptor pathway [34]. Animals deficient in either the ligand or receptor become morbidly obese due to the uncontrolled feeding behavior of the animals. Leptin, produced by fat cells, binds to leptin receptor-bearing cells in the hypothalamus and engages the Stat3 signaling pathway to influence the expression of genes encoding neuropeptides [42]. These include genes that impact upon feeding behavior such as an increase in Pomc that suppresses feeding and Npy that promotes this activity. Hence, normally low leptin levels allow Pomc expression to drop and NPY to increase which leads to enhanced feeding. However, IfitmDel animals, even at an early age (1 month) with normal weight, despite increased leptin levels, exhibit depressed expression of Pomc that corresponds to the increased feeding behavior of these animals. This reveals a novel disparity in leptin signaling through its receptor in the hypothalamus. Our preliminary analyses of the Stat3 signaling pathways in the IfitmDel animals revealed no difference with WT animals (data not shown). Further, other signaling systems that require Stat3 activation such as the IL-6 receptor is similar in IfitmDel splenocytes to that of WT-responses (data not shown). However, the elevated expression of certain inflammatory cytokines such as TNFα, which is often associated with cachexia and weight loss, require further exploration in the IfitmDel mice. Thus, a future line investigation will be to detail the mechanism(s) through which the Ifitm proteins contribute to this unique metabolic phenotype.
Because we are unaware of any reports of brain structural anomalies for the IfitmDel animal, we performed a survey of brain using a histological approach to evaluate whether the altered Pomc expression regulation could be due to structural deficiencies in the hypothalamus. Overall, we observed no abnormal anatomical defects in any brain regions when compared to the C57BL/6 controls. Additional analyses of these sections stained with cell type specific immunofluorescent markers also failed to reveal any overall gross inconsistencies in the overall distribution or numbers of neurons, oligodendrocytes and astrocytes. Microglia, however, were different. The overall numbers appeared to be similar between the IfitmDel and control mice. It is worth noting that in occasional IfitmDel animals there appeared to be a substantial decrease in microglia in the cortex (not shown). What did consistently differ was a striking and common microglial cell morphology throughout the brain tissues of the IfitmDel animal. Most notable was the reduced elaboration of microglia morphology accompanied by extremely long and usually mono-polar processes that extended from the cell body with few bifurcations. Also, the cell bodies appeared smaller and less distinct than their control counterparts. How altered microglia cell morphology could specifically affect leptin signaling and the production of neuropeptides such as Pomc or Npy is not known. However, it is possible that IfitmDel microglia are unable to produce normal interactions with other cells types including neurons and astrocytes such as trophic interactions or clearance of debris thereby leading to altered and potentially toxic microenvironments.
The function(s) of the Ifitm proteins in development and maintenance of the central nervous system is a newly developing field. Elevated expression of IFITM family members has been noted in the brains of schizophrenic patients, patients with autism, bipolar disorders and Alzheimer's disease [43][44][45][46]. Besides being reported in neurodegenerative diseases, Ifitm3 also has been shown to respond to Poly I:C as an inducer of type I interferons, by increasing gene expression in astrocytes [47]. The Ifitm3 protein is found in the endosomes of astrocytes and knockdown of Ifitm3 expression inhibits clathrin dependent uptake in such cells [47], similar to our description for cells obtained from the IfitmDel animal [14]. It appears that the expression of Ifitm proteins is crucial to function in astrocytes. However, our observations failed to reveal any gross morphological abnormalities in astrocytes of IfitmDel animals (stained by GFAP, data not shown). This will require further evaluation to assure astrocytes are not functionally compromised despite the overall appearance of normal morphology.
The dysregulation of the signaling pathways in regulating energy homeostasis in the hypothalamus leading to metabolic disorders are well-known. This includes neuronal dysfunctions or inflammation in certain brain regions may also be linked to metabolic disorders [48][49][50]. Activated microglia are capable of secreting pro-inflammatory cytokines that serve to recruit even more microglia to the site of inflammation that can further influence the metabolic response in the brain [51][52][53][54]. In neuronal dysfunctions such as schizophrenia, autism, bipolar disorders and Alzheimer's disease [43][44][45][46], the activation of microglia in addition to astrocytes can worsen the disease progression by secreting excessive inflammatory cytokines. One byproduct of these activation-dependent cytokines is the development of metabolic syndrome phenotypes characterized by weight gain and/or leptin resistance [52,55,56]. While there is a growing body of evidence suggesting the Ifitm proteins may influence neuronal function through poorly defined mechanism, our study suggests they actually play a role in normal brain cellular architecture and interaction.
A final point is that the IfitmDel animal is lacking all five of the Ifitm genes. As shown in Fig  4C, the Ifitm1, 2, 3 and 6 genes are expressed in the hypothalamus and the expression of Ifitm2 and Ifitm3 normally increases with age. As the IfitmDel animal does not produce this agerelated alteration in expression, there is the intriguing possibility that these increases produce normal compensatory functions towards control of the age-related increases in obesity and altered leptin modulation of metabolic homeostasis. Since an obesity phenotype has not been described for single Ifitm gene deletions (Ifitm3 deficient or Ifitm1 deficient) [16,53] nor have any brain anomalies such as the deficiency in microglia as shown in this report been described for any of these single Ifitm gene deletion strains, this will require further investigation. For example, Ifitm6 is primarily expressed in osteoclasts and macrophage lineages (of which microglia are related [6,57]). But whether or not the microglia deficiency seen in the IfitmDel animal is due to the lack of Ifitm6 during microglial cell development remains to be evaluated. Also other possibilities such as the indirect regulation of other modulators of adipose cell signaling by Ifitm (e.g., regulation of leptin signaling through carbonic anhydrase activation [58]) will need to be investigated. What does appear important is that Ifitm genes are implicated in modulating important endocrine functions. In the framework of the microglia and hypothalamic interactions, our data also suggest that the phenotype could vary depending upon exposure to Ifitm-specific pathogens.