Is TNF a link between aging-related reproductive endocrine dyscrasia and Alzheimer's disease?

Early demonstrations of inflammatory cells in AD brains did not threaten the decades-long primacy of research based on amyloid plaques being the direct primary cause of function loss in AD, since their association with these plaques [ 44 ; 45 ] was interpreted as facilitating plaque removal [ 46 ; 47 ]. This implied that this inflammatory response was to be encouraged, and immunotherapy designed to promote amyloid removal was investigated, albeit with disappointing results [ 48 ; 49 ]. With plaque removal becoming a increasingly questionable goal [ 50 ; 51 ], and a developing awareness that inflammatory mediators induce, and therefore precede, AβPP [ 52 – 57 ] expression and processing towards the amyloidogenic pathways [ 58 ], a fresh approach to understanding the sequence and roles of inflammation and amyloid in AD pathogenesis is warranted. An alternative approach, that of inflammation preceding and causing amyloid plaque deposition, begun with the work of Griffin and co-workers [ 59 ]. In 1989 this group demonstrated, in AD brains, that overexpression of interleukin 1 (IL-1), a cytokine that functionally overlaps with TNF, had a role in amyloid plaque formation. Four years later Tarkowski [ 60 ] reported that TNF levels in CSF from 56 individuals who had mild cognitive impairment, when tracked over a period, predicted which were much more likely to develop into frank AD. Newer studies continue to reaffirm this finding, with markers of inflammation showing in serum and CSF before any indications of increased Aβ or tau [ 61 ; 62 ]. Another group took advantage of the increased sensitivity of assaying for soluble TNF receptors rather than TNF itself. They found good evidence for levels of these receptors, which TNF induces, in serum and CSF predicting conversion to clinical AD over a 4–6 year period [ 63 ]. A more recent example utilized another acute phase protein, clusterin (apolipoprotein J), and found it to be intimately associated with onset, progression, and severity of this disease [ 64 ]. Clusterin, which is induced by TNF [ 65 ], was present 10 years earlier than fibrillar Aβ deposition. In addition a new experimental study notably reports that anti-TNF, not TNF, as earlier approaches would have suggested, reduce amyloid plaques in transgenic mice [ 66 ]. Taken together, these arguments for inflammation having a key role in the onset, rather than the dissipation, of AD pathology raise the question of what initiates these proinflammatory cytokines increases in the brain in AD.

Anxiety and aggression are often observed with AD, and after stroke and traumatic brain injury. The literature clearly associates these changes with increased TNF in a range of circumstances. For example, testosterone cells isolated from individuals with a generalized anxiety state generate more TNF on in vitro activation [ 107 ], and the ability of endotoxin-triggered human monocytes to generate TNF in vitro associates with aggressiveness [ 108 ]. Most strikingly, double TNF receptor knock-out mice essentially lack both anxiety and aggression [ 109 ]. These and similar data are consistent with the high anxiolytic capacity of anti-TNF biological agents in patients with rheumatoid arthritis [ 110 ] as well as in a mouse model of chronic gastrointestinal inflammation [ 111 ]. Inflammatory bowel disease is a well-known cause of anxiety disorder in humans [ 112 ]. The well-known anxiolytic actions of progesterone [ 113 ] may be mediated via TNF, with the decline in circulating progesterone with endocrine dyscrasia resulting in elevations in TNF. Progesterone has not been tested as a treatment for AD. In contrast to TNF, oxytocin attenuates experimental anxiety states [ 114 – 116 ], and its levels in human CSF have been inversely associated with life history of aggressive behavior [ 117 ]. This is plausibly explained by the ability of oxytocin to reduced production of TNF by monocytes from human volunteers [ 118 ]. Evidence for oxytocin being the driving force behind the effects of social interaction (i.c.v. oxytocin receptor inhibitor cancelled the effect), and its insufficiency explaining the effects of social isolation (i.c.v. oxytocin injection cancelled the effect) [ 119 ] is compelling. It may also explain, through the capacity of oxytocin to inhibit TNF production [ 118 ], why long-term social isolation exacerbates the impairment of spatial working memory in AβPP/PS1 transgenic mice [ 120 ]. In addition, social isolation makes experimental stroke worse, decreasing post-stroke survival rate and exacerbating infarct size and edema development [ 121 ]. The authors explained their data in terms of IL-6, a cytokine induced by TNF, being higher in the socially isolated animals. So far as we are aware oxytocin has not yet been tested to treat AD, stroke or traumatic brain injury.

Reproductive hormones are well known for their cell growth and differentiation properties. Age-related declines in the production of sex steroids and inhibins by the gonads leads to a decrease in negative feedback inhibition on the hypothalamus and pituitary, resulting in an elevation in the production and circulating concentrations of gonadotropins. Changes in the blood-brain barrier and cell cycle described earlier induced by endocrine dyscrasia may be mediated by alterations in expression of TNF (and Aβ), a molecule with known neurogenic and inflammatory properties. Links between the gonadotropins and TNF in brain function are supported by the findings that TNF, like gonadotropins, is a very pleiotropic cytokine, important in reproductive physiology [ 67 ], as well as being a physiological gliotransmitter [ 68 ] and central to neurogenesis [ 69 ]. Evidence for a role of gonadotropins in regulating TNF expression is demonstrated by the ability of FSH to induce TNF expression in investigations into the illness caused by chronic kidney dialysis [ 70 ]. Others, studying the reasons for the exacerbation of rheumatoid arthritis at the onset of menopause, have correlated the high circulating FSH and LH seen at this time with increases in TNF, interleukin-1β (IL-1β) and monocyte chemoattractant protein (MCP)-1 [ 71 ]. These data are consistent with the anti-gonadotropic actions of leuprolide rendering it an anti-mitotic and anti-inflammatory agent when it is used to treat endometriosis. In this context leuprolide has been reported to reduce a number of inflammatory cytokines, namely IL-1β [ 72 ], IL-6 [ 73 ; 74 ], and MCP-1[ 75 ], all of which are induced by TNF [ 76 – 78 ]. Moreover, anti-TNF treatment lowers levels of these cytokines [ 79 ; 80 ; 77 ]. Estradiol has been shown to inhibit the release of TNF from monocytes [ 81 ] while both estradiol and progesterone reduce TNF expression in mid-brain astrocytes [ 82 ]. In addition, estrogen receptor and estradiol agonists inhibit microglial activation, part of the evidence being reduced production of TNF and similar pro-inflammatory cytokines [ 83 ]. The signalling pathways have begun to be studied [ 84 ; 85 ]. In vivo , outcomes are consistent with these steroids inhibiting TNF production through their capacity to negatively feed back on the hypothalamus and lower gonadotropin-releasing hormone, LH and FSH production. Whether sex steroids directly mediate their effects via nuclear steroid receptors, membrane receptors, or via the regulation of gonadotropins or other hormones is unclear. Functionally, estradiol and progesterone can be regarded as anti-TNF agents that act before TNF is generated. These agents have been reported to protect against AD [ 35 – 37 ], stroke [ 86 ; 87 ], and traumatic brain injury [ 88 – 92 ]. A considerable basic literature from a number of autonomous groups argues for increased brain TNF [ 93 ] having a primary role in the pathogenesis of AD [ 94 – 101 ]. The area has recently been reviewed [ 102 ]. It therefore seems plausible from the reasoning in the above paragraphs that these two areas of AD research, hitherto considered unrelated, are in fact the upstream and downstream ends of a single disease mechanism (see Fig. 1 ). Indeed, both the loss of sex steroids/inhibins and the elevation of GnRH/gonadotropins with age-related endocrine dyscrasia would serve to elevate TNF expression in the brain. This possibility is testable experimentally, and might explain variations in brain TNF levels. Although, as recently reviewed [ 102 ], systemic inflammation and trauma may exacerbate pathogenic brain TNF levels in AD, the explanation for the presence of the underlying TNF in AD brains is little discussed, and still relies on arguments concerning viral infections [ 103 ], which does not sit well against the apparent non-infectious nature of AD. Induction of TNF by gonadotropins could explain why changes achieved in Aβ deposition and cognition with leuprolide treatment [ 12 ; 13 ], are consistent with those reported for anti-TNF approaches in similar circumstances [ 97 ; 104 ; 101 ; 66 ]. At this stage it is not possible to determine whether endocrine dyscrasia is regulating cell cycle signalling independent of TNF. It should be noted, however, that some 15 genes integral to TNF signalling have been found to regulate the G 2 /M stage of the human cell cycle [ 105 ]. Interestingly, a single nucleotide polymorphism in TNF has been associated with the risk of developing AD; that risk was further increased in those individuals positive for APOE E 4 [ 106 ], a gene involved in cholesterol metabolism and steroidogenesis.

These widespread functional links between the gonadotropins, sex steroids and TNF imply that it is worth investigating all three together when developing an understanding of, and therapy for, neurodegenerative diseases. Anti-TNF agents have been tested in an experimental AD model [ 66 ] and open-labeled human trials [ 122 ; 123 ], in an experimental traumatic brain injury model [ 124 ], and small, but impressive, open-labeled human stroke trial [ 125 ]. In a significant case report of TNF suppression, a patient being treated with the anti-TNF agent etanercept for ankylosing spondylosis prior to T7 complete paraplegia demonstrated remarkable sensory-motor recovery, improving from A to D on the American Spinal Injury Impairment Scale (AIS) within the first year [ 126 ]. A point of immediate interest arises from a report that TNF generation in the brain, once initiated, continues there for much longer (months) than it does systemically (hours) [ 127 ], exposing a specific target for anti-TNF agents for a much longer period in brain than in the rest of the body. This could rationalize the long reported intervals between stroke onset and apparently successful treatment with an anti-TNF agent [ 125 ]. For these reasons synergistic and other therapeutic studies using leuprolide, estrogens, progesterone and oxytocin in conjunction with anti-TNF agents are warranted.