Effects of uremic toxins on hippocampal synaptic transmission: implication for neurodegeneration in chronic kidney disease

Patients affected by chronic kidney disease (CKD) have an increased risk of developing cognitive impairment. The cause of mental health disorders in CKD and in chronic hemodialysis patients is multifactorial, due to the interaction of classical cardiovascular disease risk factors, kidney- and dialysis-related risk factors with depression, and multiple drugs overuse. A large number of compounds, defined as uremic toxins that normally are excreted by healthy kidneys, accumulate in the circulations, in the tissues, and in the organs of CKD patients. Among the candidate uremic toxins are several guanidino compounds, such as Guanidine. Uremic toxins may also accumulate in the brain and may have detrimental effects on cerebral resident cells (neurons, astrocytes, microglia) and microcirculation. The present study aims to analyze the effect of Guanidine on hippocampal excitatory postsynaptic field potentials (fEPSPs) and in CA1 pyramidal neurons recorded intracellularly. Moreover, we compared these effects with the alterations induced in vitro by CKD patients derived serum samples. Our results show an increased, dose-dependent, synaptic activity in the CA1 area in response to both synthetic Guanidine and patient’s serum, through a mechanism involving glutamatergic transmission. In particular, the concomitant increase of both NMDA and AMPA component of the excitatory postsynaptic currents (EPSCs) suggests a presynaptic mechanism. Interestingly, in presence of the lower dose of guanidine, we measure a significant reduction of EPSCs, in fact the compound does not inhibit GABA receptors allowing their inhibitory effect of glutamate release. These findings suggest that cognitive symptoms induced by the increase of uremic compounds in the serum of CKD patients are caused, at least in part, by an increased glutamatergic transmission in the hippocampus.


INTRODUCTION
Patients affected by chronic kidney disease (CKD) have an increased risk for cognitive impairment when compared with the general population [1]. The prevalence of cognitive impairment ranges between 10% and 40% in patients with CKD and is significantly higher in patients with end-stage renal disease (ESRD) receiving chronic hemodialysis or peritoneal dialysis [1].
Multifactorial is the cause of cognitive impairment in CKD patients and in patients on chronic hemodialysis. These factors include several risk factors related to cardiovascular disease (older age, hypertension, diabetes), kidney (anemia, uremic toxins), dialysis (hypotension, inflammation) as well as depression, insomnia, and multiple drugs use [1][2][3].
A systematic review on cognition in ESRD patients on chronic hemodialysis showed that memory and executive function are impaired and that the domain of orientation and attention is particularly compromised [1]. This finding, together with the observation that, after kidney transplantation, there is an improvement in several cognitive performances [4], suggests that the cognitive deficits in patients on hemodialysis may be at least partially reversible [1].
A large number of compounds defined as uremic toxins, which are generally excreted by healthy kidney, accumulates in the circulations, in the tissues and in the organs of CKD and ESRD patients [5]. Uremic toxins are several guanidino compounds (GCs), such as creatinine, guanidine, guanidino succinic acid (GSA), and methyl guanidine (MG) [5] (see Fig. 1). These toxins may also accumulate in the brain [6], exerting detrimental effects on brain microcirculation and on neurons and glial cells [7]. Interestingly, pre-clinical studies using the surgical method to induce CKD, demonstrated alterations in the short-term memory and a deficit in working memory [8,9].
Although abnormal excitatory transmission might be implicated in cognitive dysfunctions induced by GCs, experimental findings on this issue are controversial [10][11][12]. It has been reported that, in animal models, uremic GCs induce seizures mimicking the epileptic activity observed in the uremic brain [13]. In particular, GSA and MG, were markedly more potent convulsants than guanidine and creatinine [14]. De Deyn and Macdonald have published in the CA1 region, the increased uremic GCs levels evoked the activation of N-methyl-D-aspartate receptors (NMDARs) in conjunction with the blockade of GABA A and glycine receptor-associated chloride channels [15]. In this condition, the pyramidal cells were depolarized enough to reduce the blocking action of Mg 2+ on NMDARs, causing the influx of Ca 2+ and the consequent increase of GSA-induced currents [15,16]. In the present study, among the several uremic toxins accumulating in the brain of patients affected by CKD, we have analyzed the dosedependent effect of Guanidine in bath application on field excitatory postsynaptic potentials (fEPSP) in the CA1 hippocampal area. Moreover, using patch-clamp whole-cell recordings from CA1 hippocampal pyramidal neurons, we found that Guanidine (100 μM) increases both NMDA and α-Amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) currents suggesting a presynaptic mechanism of action for this toxin. Interestingly, the lower dose of Guanidine (1 μM) did not induce this excitatory increase of glutamatergic transmission. Finally, we found that serum of CKD patients mimics the excitatory effects induced by Guanidine, further supporting the hypothesis that increased excitatory transmission is a key factor in the development of cognitive dysfunction in CKD patients.
These results suggest that a presynaptic mechanism is involved in the enhanced glutamatergic basal synaptic transmission after Guanidine bath application.

DISCUSSION
Chronic kidney disorder represents a severe risk factor for cognitive impairment development [17]. The cognitive impairment has a prevalence of 30-60% in the CKD patients with respect to agematched controls [18]. One of the possible factors underlying cognitive decline in CKD patients is the accumulation in the blood of uremic toxins. These toxins easily cross the blood-brain barrier (BBB) and the blood-cerebrospinal fluid (CSF) barrier through specific transporters, and their high levels in the various brain regions could cause detrimental neurological effects [19]. It has been shown that, in patients with renal failure [20] or in hemodialysis patients [21], serum guanidine levels are 10 to ≥14 times higher than in control and, above all, a significant positive correlation exists between creatinine and guanidine levels in serum and cerebrospinal fluid subjects [21]. Herein, we used an electrophysiological approach to model this increase in uremic toxin content in CA1 slices of control rats. Among the several uremic toxins accumulating in the brain of patients affected by CKD, we analyzed the effect of Guanidine.
In the present study, we obtained three major findings. First, we found that Guanidine induces a dose-dependent increase of glutamatergic transmission measured by fEPSP slope. This increase of glutamatergic transmission was confirmed by the analysis of the I/O correlation, which shows the increased response of the field potential, under the same range of electrical stimulation, in the presence of Guanidine. We then moved to the analysis of pre-and post-synaptic effects of Guanidine administration. Paired-pulse ratio (PPR) changes in fEPSP responses to two subsequent stimuli is attributed to a presynaptic alteration in release probability [22,23]. An increase in the ratio of the second pulse response (fEPSP2) to the first pulse response (fEPSP1) indicates a decrease in the release probability. The suggested reduction in the transmitter release is consistent with the observations that manipulations depressing transmitter release usually increase the magnitude of PPR [24,25]. When the interstimulus interval is increased, the PPR decreases. In fact, the neurotransmitter release is no more influenced by the first pulse [26]. Bath application of 100 μM Guanidine, which induces a significant increase of glutamatergic transmission (as shown in Fig.  2D), affects PPR inducing an increase of fEPSP2 pulse with respect to the first one. The massive increase of glutamatergic transmission induced by the last dosage used (1 mM), is so strong that it does not anymore influence the PPR, which, in this condition, remains around 1 (fEPSP2 = fEPSP1) in all the interstimulus amplitudes.
group [6,15,27], we hypothesized that Guanidine exerts a dosedependent inhibitory effect on GABAergic transmission. From lower dose of Guanidine (1 μM), we obtained a decrease of NMDARs and AMPARs currents as expected in the presence of a physiological GABAergic inhibitory action. Conversely, when the recordings were made in 100 μM of Guanidine, we observed an increase of both NMDA and AMPA components of the EPSCs, further supporting the idea that this toxin might increase excitatory glutamatergic transmission at a presynaptic level, generating excitotoxicity and harming GABA inhibitory activity. This hypothesis has been already suggested by authors postulating that uremic GCs could act as competitive antagonists at the GABA A receptor transmitter recognition site [10]. It should be kept in mind that the extracellular recordings, shown in Fig. 2, are much less affected by GABAergic transmission than the analysis of NMDA and AMPA currents shown in Fig. 3. Indeed, in Fig. 3A the inhibitory effect of GABAergic transmission on glutamate release can be very well appreciated.
Finally, we found that the serum derived from dialyzed patients induces a significant increase of glutamatergic transmission, thus mimicking the effect of Guanidine. This finding further supports the hypothesis of a pathological increased excitatory drive in the uremic brain. The fact that we observed this effect in the hippocampus, a brain area implicated in several memory-related processes, has an additional pathophysiological relevance.
Our results confirm the initial assumption that the serum derived from CKD patients contains guanidino compounds acting as uremic neurotoxins. These toxins might trigger excitotoxic mechanisms due to the increase of both NMDAR and AMPARmediated currents [27]. This combined excitotoxic mechanism, possibly associated with the inhibition of GABAergic transmission [10], could ultimately result in an abnormal increase of calcium and sodium intracellular influx, leading to neuronal death. Further studies are required to validate and clarify this hypothesis.

MATERIALS AND METHODS Ethical approval
All procedures on animals (male Wistar rats, 2-months old) were performed in strict accordance with a protocol approved by the Animal Care and Use Committee at the Italian Ministry of Health and European Communities Council Directive of September 2010 (2010/63/E), and every effort was made to minimize animal suffering and reduce the number of animals used for the experiments (n = 35 rats).

Extracellular recordings.
Recording electrodes were made of borosilicate glass capillaries (Harvard Apparatus, Holliston, Massachusetts) and filled with 2M NaCl (resistance, 10-15 MΩ). Under visual control, a bipolar tungstenstimulating electrode (World Precision Instruments, Friedberg, Germany) was positioned into the Schaffer collateral fibers and extracellular field excitatory postsynaptic potentials (fEPSPs) were recorded with a recording electrode into the CA1 region. Testing stimuli of 0.1 Hz, 10 μs duration, and 20-30 V amplitude evoked fEPSPs that were 50-70% of maximum slope. An Axoclamp 2B amplifier (Molecular Devices, USA) was used for extracellular recordings. Input/Output (I/O) relationships were measured at the start and after drug application of each experiment by applying a series of stimuli of increasing intensity to the Schaffer collaterals. Paired-pulse ratio (PPR) indexes were calculated as mean (±SEM) as the ratio of the slope of fEPSP2nd/fEPSP1st) at various interstimulus intervals (60, 100, 200, and 300 ms).
Whole-cell patch-clamp recording. For patch-clamp recordings, neurons were visualized using infrared differential interference contrast microscopy in the CA1 region (Eclipse FN1, Nikon). Whole-cell recordings were performed with borosilicate glass pipettes (resistance, 6-9 MΩ) (Harvard Under visual control, a stimulating electrode was inserted into the Schaffer collateral fibers, and a recording electrode was inserted into the pyramidal CA1 region of the hippocampal slice [28]. Signals were amplified with a Multiclamp 700B amplifier, recorded, and stored on PC using pClamp 10.4 (Molecular Devices, USA). Whole-cell access resistance was 15-30 MΩ. Input resistances and injected currents were monitored throughout the experiments. Variations of these parameters >20% lead to the rejection of the experiment. For the NMDA and AMPA current experiments, neurons of the CA1 region were voltageclamped at −70 and +40 mV to record, respectively, AMPA-mediated and NMDA-mediated EPSCs [29]. All the experiments have been done in the absence of picrotoxin, a GABA receptor inhibitor. The NMDA component of the EPSC was individuated by using the kinetic method, considering the peak amplitude at 50 ms after the beginning of the event.
The serum of the healthy subjects (n = 5) and dialysis patients (n = 10) was provided by the Hemodialysis Unit, Division of Nephrology, Università Cattolica del Sacro Cuore, Rome (Dr. Maurizio Bossola). Table 1 reported the concentration of routinely analyzed clinical parameters in healthy and dialysis serum.
Guanidine and serum were applied by dissolving them to the desired final concentration in oxygenated Krebs' solution and were bath applied by switching the standard solution.

Statistical analysis
Animal sample size has been calculated with G*Power software (5% type I error, 80% power). The study was not classified as randomized and blinded because it does not expect random assignment of animals to treatment groups and the investigator knows the experimental procedure of each group and assesses the electrophysiological outcome. Data analysis of electrophysiological experiments was performed offline using Clampfit10.4 (Molecular Devices, USA) and GraphPad Prism 6.0 (GraphPad Software). We applied the assumption of a normal distribution with a similar variation within each group of data statistically compared.
Values in the text and figures are mean ± standard error of the mean (SEM), n representing the number of recorded neurons. Paired Student's t-test was used for the electrophysiological analysis of the effect of synthetic uremic Guanidine in vitro application two-way ANOVA with multiple comparisons was utilized for statistical analysis between different experimental groups over time or at a specific time point, respectively. When time × group interaction was significant, group means for each time point were compared using Bonferroni's post hoc test. The significance level was established at p < 0.05.

DATA AVAILABILITY
The data presented in this study are available on request from the corresponding author.  Differences between groups are not statistically significant except for: *p < 0.5; ***p < 0.001. Data we reported as mean ± SD. Student's t test for unpaired data was used to compare the demographic and clinical features of the experimental groups. A p value of <0.5 was considered statistically significant.