The aging of the brain is a multifaceted phenomena, encompassing structural, neuronal, glial and microenvironmental changes. The causative drivers of the cellular changes are difficult to untangle, as alterations such as loss of neuronal circuits and plasticity could be primarily driven by cell intrinsic neuronal-dependent aging or could be derived from the impact of aging glial support. The role of inflammation in this process is likely to be a primary driver of at least some aspects of the aging process, based on correction experiments. For example, defects in learning and memory capacity in the contextual fear conditioning and radial arm water maze assays can be partially corrected in 18 month-old mice through the repeated injection of plasma from young mice [47]. Likewise, heterochronic parabiosis, joining the circulatory system of a 21 month-old mouse to a 2 month-old mouse, improved the performance of old mice in olfactory sensitivity assays [48]. Thus, while a neuronal cell intrinsic aging effect is likely, the emergent feature of cognitive decline must at least partly depend on environmental factors, such as a build up of toxic or inflammatory products. T cells are a key potential player in the development of this neuroaging environment. T cells can directly produce, or stimulate the production of, key inflammatory mediators, including cytokines, ROS and autoantibodies, a process that can directly contribute to emergent features such as the constraint of the neuronal stem cell niche [26]. The identification of T cells as a potential culprit opens the possibility of Tregs as a cure. Tregs have direct anti-inflammatory properties, preventing excessively exuberant responses from conventional T cells. In addition, Tregs can produce pro-repair orientated cytokines, such as amphiregulin and osteopontin [39, 49], which are neuroprotective following injuries such as stroke. As Treg numbers in the brain are low [36], it is expected that their influence will be mediated, at least in part, through the reprogramming of glial cells. In the peripheral context, Tregs are capable of reprogramming monocytes into the more pro-repair anti-inflammatory profile [50]. In the brain, an analogous program may be imparted onto microglia [51]. Microglial polarization can, in turn, impact astrocyte polarization [52]. Tregs can also stimulate OPCs and promote myelination, in a pro-repair state [53]. When considering the efficacy of PHP.GFAP-IL2 in mitigating aging effects, the parsimonious explanation would be a primary effect via the local expansion of Tregs. When used in the context of traumatic brain injury, the beneficial effects of PHP.GFAP-IL2 were only observed in the presence of an adaptive immune system, suggesting that the levels of IL2 achieved (~ 2pg/ml) are too low to trigger the activation of the lower affinity receptor expressed in non-Treg lineages [43]. While it would be attractive to speculate on the role of individual downstream mediators, the complexity and interdependencies of aging make a multifactorial function more likely, integrating pro-repair, anti-inflammatory and glial reprogramming functions.
Single cell transcriptomics of aging glia identified a shared molecular age-induced signature across all major glial cell types. The pathways identified concord with the current molecular understanding of aging, led at the cellular level by mitochondrial dysfunction, loss of proteostasis and cellular senescence. Mitochrondrial dysfunction (here identified as changes in the oxidative phosphorylation pathway and mitophagy pathway) has been identified as a key cell-intrinsic marker of aging, with dysfunction in microglia a potential contributor to neurodegeneration in dementia [54]. Indeed, an increased burden of mitochrondrial dysfunctional can hasten the progression of brain atrophy in mouse models of AD [55], while mitophagy of defective mitochondria reverses cognitive decline [56]. Loss of proteostasis is another reoccurring theme in aging across tissues, and is consistent here with changes to protein production (RNA transport, spliceosome and ribosome), processing (protein processing in endoplasmic reticulum) and degradation (proteasome) pathways. The mechanistic link between proteostasis and aging is unclear, with a lead contender being the toxic or inhibitory build-up of misfolded and/or damaged proteins. The genetic link between proteostasis network components and neurological diseases, such as ALS, AD and PD, strongly suggest that a failure of normal proteostasis is a distinct risk factor for neuropathology [57]. Related to proteostasis is autophagy, a process intimately linked to aging [58]. AD patients exhibit dysregulated autophagy [59] and loss of autophagy causes severe neurodegeneration in mice [60]. For the specific signaling pathways identified, the Neuregulin-ErbB pathway, Ras pathway and PI3K/AKT/mTOR [61–63] are all integral to neuroaging, and potential drivers for the cellular phenotypes developing in aging glial cells.
Surprisingly, the local provision of IL2 in the brain was sufficient to substantially revert almost all of these age-induced pathway changes across microglia, oligodendrocytes and astrocytes. This was despite relatively few immune pathways being identified as altered, apart from the upregulation of MHCII, previously linked to PHP.GFAP-IL2 treatment [43], and alterations in the TLR and IL17 signaling pathways in microglia. Whether this effect is mediated via direct reprogramming, such as Treg cytokine-mediated effects, or indirect effects of microenvironmental cleansing, remains to be seen. It is, however, highly promising that such cell-intrinsic molecular signatures of aging are responsive to late-stage intervention.
While this study suggests that an analog of the PHP.GFAP-IL2 treatment could be of use to avert cognitive decline in ageing in humans, there are several key barriers to translation. First, the degree to which the aging process is conserved across the species barrier is unclear. Many of the biological processes occurring in the aged brain, such as accumulation of DNA modifications, mitochondrial dysfunction and loss of proteostasis, are shared across species [64]. Likewise, cognitive decline is common between mice and humans [65], including reduction in spatial navigation performance (such as the Morris water maze in mice, or the equivalent in humans) [66]. Despite this, mouse cognitive decline seems to appear relatively sooner and faster than for humans [67], and decline in elaborate cognitive abilities specific to human, such as language, cannot be tested in animal models. Further, the neurogenic niche inhibited by T cells in aged mice [26] may not exist in humans [68]. Thus, even if the restoration of spatial navigation decline was achieved through brain-specific IL2 delivery in humans, it is not clear that other aspects of cognitive decline, such as language deficits, share a common cellular or molecular cause, or would respond to the same treatment approach. A second potential limitation to translation is the kinetics involved in cognitive decline. Here we treated aged mice for two months prior to assessment, with the treated aged mice performing similar in spatial navigation capacity to young control mouse. It remains unknown, however, whether the treatment actively reversed the molecular, cellular and behavioral processes of aging, or whether it merely prevented the decline. The kinetics of cognitive decline are compressed into a mouse lifespan, raising the question of whether treatment in humans could impact a decline occurring over the course of decades. If cognitive decline is primarily driven by active changes in the brain microenvironment, such as increased basal inflammation induced by a lifelong stimulation of the immune system [18, 19], then even transient pulses of purification could reverse or prevent the decline. On the other hand, if cognitive decline is driven primarily by programmed senescence, then such a treatment could be expected to, at most, delay cognitive decline. Fortunately, experiments in mice largely support the former model over the latter, as murine cognitive decline occurs with age even though individual mouse neurons can long out-survive a mouse, when transplanted into a longer-lived host [69]. Finally, the molecular pathways of the treatment itself need to be considered. The delivery system used here, the PHP.B capsid, performs poorer in blood-brain-barrier crossing in non-human primates than it does in mice [70]. Fortunately, alternative AAV capsids are available that perform well in humans [71], and sustained cargo expression in the brain has been observed to span years [72], an advantage for a potential longevity treatment. Tregs are found in both the mouse and human brain [36], and the IL2 pathway is highly conserved across the species, so it is likely that the expansion of brain Tregs could be induced in humans if cargo delivery was achieved. Nonetheless, substantial technical and regulatory barriers exit when considering the development of any longevity treatments that are designed for the use of healthy individuals, with long-term safety data being essential.