In order to investigate how the EPO system influences cognitive performance and synaptic plasticity, and to experimentally prove that previously reported effects of rhEPO on cognition in patients are independent of EPO effects on hematopoiesis/brain oxygen supply, we created a novel mouse model with constitutive EPOR expression in cortical and hippocampal neurons that are defined by α-CaMKII promoter activity. This way, we were able to specifically mimic EPO system function independent of any ligand in pyramidal neurons of cortex and hippocampus, that is regions pivotal for learning and memory processes. In other words, we 'over-accentuated' the endogenous action of the EPO system in specific cortical and hippocampal layers, particularly the CA1 subregion , to delineate the contribution of these neuronal subpopulations to the EPO effects on cognition. We found that selective constitutive expression of EPOR in forebrain neurons leads to a phenotype with superior performance in higher cognitive tasks. Behaviorally, this phenotype is accompanied by slightly increased activity and impulsivity. Electrophysiologically, both short- and long-term plasticity at the Schaffer collateral CA1 synapses are significantly increased in cEPOR expressing TG mice.
Field potential recordings revealed augmented paired-pulse facilitation, short-term potentiation and LTP in cEPOR TG mice. One possible cause for the differences observed in the twin-pulse stimulation results might be the slight right shift of the input-output curves which suggests somewhat reduced baseline excitability and thus leaves more room for potentiated response. Even though the detailed underlying molecular mechanisms cannot be determined on the basis of our extracellular recordings, the enhanced paired-pulse facilitation confirms that an increased number of transmitter quanta can be released in cEPOR TG mice (for review see ). Accordingly, the dynamic range of synaptic plasticity (and hence efficacy) is clearly extended as compared to WT mice.
The gain in paired-pulse facilitation is comparable to changes observed earlier in mice receiving EPO injections . However, in cEPOR TG mice, LTP stabilized at higher levels. Since both paired-pulse facilitation and LTP are augmented, both pre- and postsynaptic mechanisms seem to contribute to the improved synaptic plasticity. Upon high frequency stimulation (LTP induction), the immediate phase of post tetanic potentiation is known to be independent of kinase activity . The following 60 minutes of LTP reflect 'early LTP', which at the Schaffer collateral CA1 synapse is NMDA receptor dependent, involves activation of various kinases leading to AMPA receptor phosphorylation, but is independent of protein synthesis and/or gene transcription [30, 31]. Interestingly, some of the kinases involved in the phosphorylation events during early LTP (MAPK, PI3K and ERK) are indeed part of the EPO signaling cascades  and thus might constitute putative sites of signal convergence. Nevertheless, the enhanced spatial learning and cognitive flexibility observed in cEPOR TG mice suggest that, even though not evaluated electrophysiologically, the late phase of LTP, that is the phase requiring gene transcription and protein synthesis, is augmented as well [30, 31].
A detailed and comprehensive behavioral-cognitive analysis of cEPOR TG mice was performed to demonstrate that increased EPO signaling in cortex and hippocampus enhances a whole array of learning and memory processes, as well as cognitive flexibility and attentional capacities, reflected by shorter reaction times and reduced distractibility through competing irrelevant auditory stimuli. Very similar higher cognitive tasks were found improved in human patients upon several months of weekly high-dose intravenous EPO treatment [13–15], pointing to specific targets of EPO action on cognition that are common to both mice and humans. Augmented EPOR signaling in cEPOR TG mice also improved social memory, which is partly dependent on hippocampal functions . We note that a recent study reported on better facial recognition performance in patients with major depression following high-dose EPO application , supporting social cognition as another selective target of EPO effects across species.
It is important to point out that there are clear differences between EPO effects on higher cognition upon systemic administration to healthy mice  as compared to the selective and specific stimulation of the EPO system in forebrain pyramidal neurons reported here. In contrast to mice receiving intraperitoneal EPO injections , the cEPOR TG mice did not show improved performance in the (still relatively basic) initial 5-CSRTT training phases. Their superiority, however, was pronounced in the highest cognitive challenge tasks, demanding tremendous attentional capacities.
Surprisingly, under cognitively most challenging conditions, cEPOR TG mice demonstrated more premature responses as readout of impaired behavioral impulse control . The slightly hyperactive and impulsive phenotype of cEPOR TG mice was further confirmed by a simple additional assay - the marble burying test. Behavioral consequences of this kind were not noted upon high-dose EPO treatment where the cellular target is defined by the almost ubiquitous presence of EPOR throughout the brain. A potential explanation for the cEPOR TG phenotype of impulsivity and hyperactivity might be the continuous stimulation of the EPO system exclusively in cortical projection neurons. Since the frontal cortex has reciprocal projections to subcortical and basal brain regions [for example [35–37]], responsible for locomotion, motivation and impulsivity (for review see for example [38–40]), the excitation of frontal pyramidal neurons might lead to a relative deficit in the simultaneous inhibitory regulation of these areas, consistent with a disturbance of the homeostatic balance within neuronal networks, resulting in impulsivity and hyperactivity of TG mice. This hypothesis is presently under systematic investigation in our laboratory through selected cEPOR expression in subpopulations of inhibitory interneurons.
Interestingly, there is an ongoing debate regarding the nature of brain EPOR. The fact that non-hematopoietic but neuroprotective EPO variants have been identified makes the additional existence of a different EPOR in the brain very likely (for review see [7, 8]). Indeed, the work of Xiong and others [41, 42] may well be interpreted along these lines. These authors demonstrated in a traumatic brain injury model worse outcome of neural EPOR-deficient mice, but, surprisingly, beneficial effects of EPO treatment even in the absence of the 'classical' neural EPOR. Moreover, they found that EPOR null mice per se are not impaired in spatial learning, indicating that the 'classical' brain EPOR may not be crucial for this task. This again emphasizes that the superior performance of cEPOR TG mice in spatial learning and memory, induced by over-activation of one selected population of neurons only, is partly explained by the provoked dysbalance in neuronal networks. In fact, some of the cognitive phenotypes observed here may also be derived from over-activation of downstream signaling cascades in neurons which are not endogenously triggered by EPO, thereby creating a somewhat 'artificial' scenario. Nevertheless, the cEPOR approach taken here may ultimately help to delineate the role of discrete neuronal subpopulations in cognitive processes.
EPO and EPOR are expressed at very high levels in the developing central nervous system [43–47]. In contrast, their expression is markedly reduced postnatally and remains low in the normal adult brain . Both genes are upregulated under disease conditions in various different cell types in the brain, possibly to exert neuroprotective effects [49, 50]. Our present data indicate that activated EPOR serves a role in neuroplasticity, independent of and in addition to its anti-apoptotic neuroprotective tasks. Since the steady-state level of both receptor and ligand is nevertheless low in the uninjured brain, we suggest that this function may also be disease-relevant. We propose a model in which EPO-EPOR induction under disease conditions not only prevents neuronal cell death, but also triggers the enhanced neuronal plasticity that is required to functionally compensate for lost neuronal functions. It is intriguing that this could indicate a strategy of the neocortex, which is known to provide striking functional compensations after injury.