Background
Recent studies have unequivocally linked activation of the innate immune system with development of metabolic, subjective and behavioral components of sickness. Peripheral or central administration of the cytokine inducer LPS induces transient anorexia, social isolation, general malaise, an increase in non-rapid eye movement sleep and fever [
1]. All of these symptoms are dependent on neuroinflammation and the production of pro-inflammatory cytokines within the brain. Sustained activation of the innate immune system can lead to development of depressive disorders [
2]. Several conditions, such as aging and obesity, and diseases, such as rheumatoid arthritis, atherosclerosis and congestive heart failure, are associated with an increased prevalence of depressive disorders. These conditions have a common underlying chronic inflammatory component [
3]. Indeed, elevated levels of circulating pro-inflammatory cytokines, including TNFα, IL-6 and IL-1β, are frequently observed in patients with depression [
4]. Although an associative link between neuroinflammation and sickness behavior is now widely accepted, the above studies do not provide a cause-effect relationship between neuroinflammation and development of depression disorders.
There is increasing evidence that development of depression can be associated with activation of the innate immune system [
5,
6]. In particular, cytokine therapy for certain types of cancer and viral infections induces development of depressive symptoms in a significant percentage of the population under consideration [
7,
8]. Humans exposed to low-dose endotoxin elicited a depressed mood that correlated with cytokine secretion. Interestingly this low dose of endotoxin did not elicit symptoms of sickness [
9]. A similar reduction in mood occurs in humans injected with a typhoid vaccine, and this decline significantly correlates with an increase in IL-6 secretion and enhanced activity within subgenual anterior cingulate cortex [
10]. These findings provide a direct cause-effect relationship between neuroinflammation and depression and a distinction between overt sickness and depression. At the preclinical level, acute and chronic activation of the immune system reliably induces depressive-like behavior of mice. LPS induces transient sickness followed by depressive-like behavior, increased immobility in the forced swim test (FST) and the TST. These behaviors are reversed by anti-depressants and by minocycline which attenuates neuroinflammation [
11,
12]. These and other studies clearly suggest that development of anti-inflammatory regimes would be a viable strategy as a potential therapeutic for inflammation-associated depressive disorders.
IGF-I, a neurotrophic hormone, elicits a broad spectrum of biological activities [
13]. However, few studies have been reported that describe an anti-inflammatory action of IGF-I. IGF-I decreased expression of pro-inflammatory cytokines following treatment with galactosamine plus LPS, which results in IGF-I mediated liver protection [
14] and reduced atherosclerosis progression in ApoE mice [
15]. Also, IGF-I gene transfer attenuated glial activation and tau hyper-phosphorylation following spinal cord injury [
16]. These studies and others illustrate that IGF-I may be anti-inflammatory. However, the positive effects in these studies may reflect the ability of IGF-I to modulate macrophage or neutrophil infiltration into tissues, rather than a direct anti-inflammatory action on the immune system, which would be evident as decreased cytokine expression per cell. Only a couple of
in vitro studies have shown a direct anti-inflammatory potential for the IGF system. Microglia-derived IGF-II, which acts similar to IGF-I, inhibited TNFα-activation of JNK in oligodendrocytes [
17] and IGF-I increased dephosphorylation of IкB in astrocytes; thereby diminishing NF-кB activity following TNFα exposure [
18,
19]. IGF-I also depressed TNFα-induced NF-кB activation in colonic adenocarcinoma cells [
6,
20]. Importantly, IGF-I mediates the anti-inflammatory actions of the neural cell adhesion molecule (NCAM) mimetic, FGL, thus tempering glia activation associated with aging and glial activation following treatment with interferon-(IFN)γ [
21]. Similarly, IGF-I decreased IFNγ-induced and age-associated microglial activation, IL-1β induction and deficit in long-term potentiation [
22]. Exogenous IGF-I, given i.c.v., can even temper the development of sickness behavior induced by either TNFα or LPS [
23‐
25]. Taken together, these data indicate that IGF-I can attenuate an inflammatory response both at the cellular and subcellular levels and antagonize neuroinflammatory-induced behavioral changes.
To date, IGF-I has been tested for anti-depressant activity using naïve rodents and following chronic mild unpredictable stress (CUS). Chronic IGF-I administration, s.c. at 50 μg/kg/d, decreased immobility in the FST of naïve mice, decreased consumption latencies in the novelty-induced hypophagia test of naïve mice and increased sucrose consumption by mice following CUS [
26‐
29]; all suggesting anti-depressant activity. In the same study, exercise-induced improvement in the FST was blocked by an anti-IGF-I antibody [
29], suggesting that exercise has an anti-depressant activity that is dependent on IGF-I. Again with naïve mice, IGF-I and NBI-31772 (a drug which frees endogenous IGF-I from a natural inhibitor) decreased immobility in the TST. The action of IGF-I and NBI-31772 was blocked by the IGF type 1 receptor antagonist JB1 [
28]. These actions were independent of changes in general locomotor activity, which is used an index of sickness [
30,
31], suggesting behavioral specificity. Using naïve rats, a single i.c.v. injection of 1 μg IGF-I decreased immobility in the FST [
26]. JB1 blocked the decreased immobility in the FST that was present after a single i.c.v. 1 μg dose of IGF-I [
27]. These data clearly indicate that IGF-I has anti-depressant activity using a variety of measures (ex., FST, TST, sucrose consumption; which parallel each other as reliable indices of depressive-like behaviors of rodents treated with IGF-I).
However, IGF-I has not yet been evaluated for anti-depressive actions on the important background of acute neuroinflammation. Here we have filled this void by defining the ability of exogenous IGF-I to modulate depressive-like behavior of LPS-challenged mice. LPS and IGF-I were both administered i.c.v. to directly test if IGF-I was able to modulate a central inflammatory response, independent of actions at the periphery. Based on the association of neuroinflammation with depression, we then examined whether the anti-depressive effect of IGF-I was associated with its ability to temper the neuroinflammatory processes within the brain. Our data show that central IGF-I significantly impairs development of depressive-like behavior and this action was related to an anti-inflammatory response in the brain measured as measured by a reduction in expression of inflammatory markers. Moreover, IGF-I induced expression of BDNF, which has well-characterized anti-depressant activity. These results provide strong evidence that IGF-I within the brain tempers depressive-like behavior in both naïve and LPS-challenged mice.
Discussion
The most important finding of this study was that central administration of 1,000 ng IGF-I significantly reduces the duration of immobility after the injection of either vehicle or LPS. This is the first evidence to establish that IGF-I tempers the innate immune response within the brain following central administration of LPS. IGF-I acts to reduce expression of central inflammatory markers while increasing expression of the neurotrophin BDNF. There was no statistical interaction between IGF-I and LPS on depressive-like behavior (immobility during the TST), thus IGF-I did not act to specifically block LPS-induced depressive-like behavior. However, these data do establish that the anti-depressant activity of IGF-I is present in both naïve and LPS-challenged mice. To assure that immobility in the TST could be used to model depressive behaviors, we confirmed that immobility was responsive to classic anti-depressants, fluoxetine and desipramine. Both of these drugs had anti-depressant activity 9 h after LPS, fluoxetine eliminating the LPS response and desipramine acting more like IGF-I; i.e. lowering immobility in the absence or presence of LPS. These findings are in line with other work with fluoxetine and desipramine [
33,
34]. The anti-depressant effect of IGF-I was also observed in the absence of LPS, which is consistent with other studies with rodents where increasing IGF-I activity decreases depressive-like behaviors, including diminished immobility during the FST or TST [
26‐
28].
Activation of the innate immune system of the brain contributes to pathophysiologic changes that occur in depression due to the generation of inflammatory mediators including nitric oxide, via iNOS, and pro-inflammatory cytokines such as IL-1β and TNFα. Although tested at much higher doses (500 to 25,000 ng) than used for the current experiment, numerous experiments have shown that LPS given centrally induces steady-state mRNA expression of cytokines that are directly related to an increase in the content of the transcribed proteins n the rodent brain [
35‐
42]. Our results showed that as little as 10 ng LPS induces expression of cytokines and iNOS. In contrast, central administration of IGF-I reduced the expression of IL-1β, TNFα and iNOS (Figure
4). The effects of LPS and IGF-I on pro-inflammatory cytokine and iNOS expression directly paralleled depressive-like behavior. Although having a similar inhibitory trend, central IGF-I did not significantly affect IL-4, IL-6 or IL-10. IGF-I also decreased the expression of GFAP. GFAP expression is recognized as a marker of astrocyte activation [
43] and LPS has been shown to increase GFAP expression [
44] and cytokine expression [
45] by astrocytes. However, our low i.c.v. dose of LPS was not capable inducing GFAP expression. These data do, however, provide strong evidence that the general degree of activation of the innate immune system within the brain was depressed by IGF-I.
Fluoxetine and desipramine, which blocked the increase in duration of immobility caused by LPS, both have anti-inflammatory activity [
46‐
49]. This finding supports a possible direct role of the innate immune system in depressive-like behavior and the possible importance of the anti-inflammatory impact of IGF-I on behavior. Whether the decrease in cytokine expression induced by IGF-I was causative, remains speculative. However, administration of the anti-inflammatory agent, minocycline, similarly depressed cytokine expression in the brain of mice and this anti-depressant was more effective in diminishing depressive-like behavior than decreasing the sickness response [
12]. Similar to the results reported here, this finding indicated dissociation between sickness and depressive-like behavior. Such dissociation was also found with cancer patients wherein paroxetine, a SSRI class anti-depressant, had little activity against neurovegatative symptoms of sickness following IFNγ treatment, whereas symptoms of depression were more responsive to paroxetine [
8].
The neurotrophic hypothesis of depression considers that functional changes in existing neurons and their synapses (neuronal plasticity) and the organism's ability to adapt/respond to the environment via neurogenesis or reorganization of neuronal networks (brain plasticity) are dysfunctional because of a deficient degree of neurotrophic support and classic anti-depressants, such as fluoxetine, reactivate plasticity in the adult brain [
50,
51]. The IGF system is one of the most characterized neurotrophic networks yet described with IGF-I inducing the proliferation, survival, differentiation and maturation of all cells of the central nervous system [
52]. In addition, IGF-I directly supports neuronal and brain plasticity [
53], key components of the neurotrophic hypothesis. In the brains of naïve mice, IGF-I mRNA is expressed primarily by neurons [
54‐
56]. IGF-IEa is the primary transcript in the rodent brain throughout development although IGF-IEb (also sometimes referred to as IGF-IEc) expression can be induced [
57]. We confirmed the prominence of IGF-IEa above IGF-IEb in the mouse brain (Table
1). The effect of central administration of LPS on IGF-I expression has not been reported. Although not a major source of IGF-I in the naïve rodent brain, LPS depressed the secretion of IGF-I from cultured microglia [
58]. This inhibitory effect of LPS agrees with the reduction in IGF-IEb mRNA expression in the brain reported here.
Pro-inflammatory cytokines induce IGF resistance of multiple cell types (reviewed in [
13]), but this possibility has never been tested with cells of the central nervous system. Whether activation of the immune system by central LPS and the subsequent expression of pro-inflammatory cytokines decrease IGF-I activity in the brain is unknown. However, it is important to note that addition of exogenous IGF-I strikingly diminished depressive-like activity. Thus, our IGF-I treatment paradigm may have been successful in LPS-challenged mice because it either restored the effect of lowered IGF-I mRNA expression, caused by LPS, or overcame the effect of cytokines-induced IGF resistance. Further work is necessary to clarify the mode by which LPS/cytokines-induced and IGF-I-reduced depressive-like behavior.
In a previous study, 1,000 ng IGF-I attenuated the decrease in exploration of a novel object induced by i.c.v. administration of 100 ng LPS to CD-1 mice [
23]. This was accompanied by an IGF-I induced reduction in the body weight loss that occurred following LPS treatment. IGF-I also attenuated the decreased social exploration induced by centrally administered TNFα or IL-1β although IGF-I was more potent against TNFα than IL-1β and only blocked the reduction in body weight loss that was induced TNFα [
24]. These results were replicated in a different set of experiments using only TNFα [
25]. Since the decrease in social exploration caused by inflammation is essentially mediated by IL-1β, with TNFα having only an accessory role [
59], the present results, wherein IGF-I did not block sickness, are not necessarily in contradiction with our previous results on the effects of IGF-I on sickness behavior. A sub-optimal very low dose of LPS was chosen for the current study in order to elicit a minimal and transient bout of sickness and thus a lower metabolic challenge to the mice. In addition, this very low dose of LPS was carefully chosen not to induce IGF-I sensitive sickness behavior, but to cause reliable depressive-like behavior at a time at which the mice are recovering from sickness. The time interval of 9 h post-LPS was shorter than the 24 h time interval that was used in our previous experiments with LPS administered i.p. [
12] for the simple reason that a low i.c.v. dose of LPS caused more transient behavioral alterations than systemically administered LPS. The important finding of this work clearly still remains that IGF-I decreased depressive-like behavior independent of an effect on the sickness response. This indicates that although linked, sickness behaviors and depressive-like behaviors of mice have subtle yet distinct underlying etiologies. This dissociation between sickness behavior and depression has already been noted in both preclinical [
12] and clinical studies [
8].
The role of BDNF in depression was recently reviewed by several investigators. A genetic link has been identified between BDNF variants and response of humans to anti-depressants [
60] and lower BDNF expression was associated with the development of major depressive disorder [
61]. Using animal models, anti-depressant administration, therapeutic regimes such as exercise, and IGF-I all increased BDNF expression within the brain. BDNF and BDNF receptor (TrkB) deficient mice or those carrying a human variant, associated with resistance to anti-depressants, did not respond to anti-depressants. BDNF deficiency in the forebrain also caused a deficit in the TST [
62,
63] and anti-depressants induced BDNF expression [
51]. These studies clearly identify a role for BDNF in depressive-like behavior; lower BDNF increased depressive-like behavior whereas higher BDNF lowered depressive-like behavior. Relative to the current work, LPS either decreased or did not change BDNF expression within the hippocampus [
64‐
67] or decreased expression in the whole brain 3 days after i.p. LPS [
68]. One of the hottest aspects of BDNF biology is the differential expression and regulation of specific transcript; all of which produce the same mature protein. The studies mentioned above quantified total BDNF expression. However, transcripts initiating from exons I, II and III are expressed predominantly in neurons whereas transcripts initiating from exons IV, V and VI are expressed by both neurons and astrocytes within the naïve brain [
69]. Normal expression of all transcripts may be necessary for wellness, as a knockout of even one of the 9 BDNF transcripts caused depressive-like behavior of mice [
70]. Duloxetine, an SNRI class anti-depressant, increased the expression of only 4 of the 9 transcripts [
71]. Clearly, considerable effort is needed to clarify the regulation of BDNF expression, especially relative to immune activation where little is known except that TNFα induced transcript IV expression in cultured astrocytes, whereas LPS and IL-1β did not change expression [
72]. There were no reports of IGF-I regulating the expression of specific BDNF transcripts. Here we show that LPS decreased expression of the exon I, not exon VI, driven BDNF transcript, whereas IGF-I increased expression of both transcripts. These specific transcripts were chosen since exon I was known to be neuron-specific, but exon VI transcripts were also present in astrocytes. It was tempting to speculate that LPS, by some uncharacterized mechanism, decreased neuronal, but not astrocyte, expression of BDNF as part of its ability to induce depressive-like behavior. Proof of this relationship awaits further evidence. IGF-I increased expression of BDNF I and VI; transcripts found in neurons or both neurons and astrocytes, respectively; and IGF-I decreased depressive-like behavior. Elevated BDNF expression may be a major part of the mechanism by which IGF-I acted as an anti-depressant.
Since BDNF has anti-depressant activity, the increase in BDNF expression may be associated with the anti-depressant actions following central administration of IGF-I. However, IGF-I also enhances BDNF signaling within neurons. We had previously shown that IGF-I sensitizes cortical neurons to BDNF: increasing BDNF induction of Erk phosphorylation and synergizing with BDNF to decrease calcium uptake following treatment with glutamate [
73]. Those data support earlier work showing that IGF-I and BDNF synergize to support neuron survival [
74]. Together they clearly indicate that IGF-I and BDNF work together within the brain. Also, IGF-I directly supports neuronal plasticity [
53] as does BDNF (reviewed in [
51]). It will be interesting to determine if IGF-I and BDNF synergize to increase neuronal plasticity thus offsetting the well-characterized loss of synaptic plasticity following treatment with LPS [
75‐
78]. Such a finding may further define a precise mechanism by which either growth factor regulates depressive-like activities.
Competing interests
RD and KWK have received honoraria from Astra-Zeneca, RD has received honoraria from Bristol-Myers Squibb and Lundbeck Laboratories. RD works as a consultant for Lundbeck Laboratories
Authors' contributions
RHM designed the experiments, performed surgery, treated animals, performed behavior tests, analyzed data and wrote much of the manuscript. SP performed surgery, treated animals, performed behavior tests, analyzed data, did all the PCR analyses and wrote some of the manuscript. KWK and RD helped in the design and interpretation of the experiments and editing the manuscript. All authors have read and approved the final version of this manuscript.