Hippocampal synaptic connectivity in phenylketonuria

In humans, lack of phenylalanine hydroxylase (Pah) activity results in phenylketonuria (PKU), which is associated with the development of severe mental retardation after birth. The underlying mechanisms, however, are poorly understood. Mutations of the Pah gene in Pah/c57bl6 mice result in elevated levels of phenylalanine in serum similar to those in humans suffering from PKU. In our study, long-term potentiation (LTP) and paired-pulse facilitation, measured at CA3-CA1 Schaffer collateral synapses, were impaired in acute hippocampal slices of Pah/c57bl6 mice. In addition, we found reduced expression of presynaptic proteins, such as synaptophysin and the synaptosomal-associated protein 25 (SNAP-25), and enhanced expression of postsynaptic marker proteins, such as synaptopodin and spinophilin. Stereological counting of spine synapses at the ultrastructural level revealed higher synaptic density in the hippocampus, commencing at 3 weeks and persisting up to 12 weeks after birth. Consistent effects were seen in response to phenylalanine treatment in cultures of dissociated hippocampal neurones. Most importantly, in the hippocampus of Pah/c57bl6 mice, we found a significant reduction in microglia activity. Reorganization of hippocampal circuitry after birth, namely synaptic pruning, relies on elimination of weak synapses by activated microglia in response to neuronal activity.Hence,ourdatastronglysuggest that reducedmicroglial activity inresponse to impairedsynaptic transmission affects physiological postnatal remodelling of synapses in the hippocampus and may trigger the development of mental retardation in PKU patients after birth.


INTRODUCTION
Mutations in the genetic coding for phenylalanine hydroxylase (Pah), which converts phenylalanine (Phe) into tyrosine, may result in dysfunction of the enzyme.As a consequence, concentrations of phenylalanine in the plasma, in cerebrospinal fluid and in brain tissue are highly elevated (1-3).In humans, these mutations have long been known (4) and newborns suffering from this autosomal-recessive disorder known as phenylketonuria (PKU) develop severe mental retardation (5), unless they are treated with a phenylalanine-reduced diet.Previous studies on effects in the brain have shown that PKU is characterized by abnormalities in cell density, dendritic arborization, spine synapse density and hypomyelinization (6)(7)(8)(9).The underlying molecular mechanisms of mental retardation in PKU, however, remain elusive.
As a model for PKU, Shedlovsky and co-workers (10,11) generated the Pah enu2 mouse, in which the gene coding for the Pah is mutated.This mutation results in phenylalanine levels similar to those found in PKU patients.Even though the Pah enu2 mouse was developed in the early nineties, little is known about the morphological and functional alterations in the brain of this mutant.In addition, previous studies used Pah enu2 on a BTBR T+tf/J (BTBR) background, which questions the specificity of the effects (12)(13)(14)(15)(16). Basically, BTBR mice lack a corpus callosum, which could interfere with results on phenylalanine effects in the hippocampus (17).As the hippocampus is essential for learning and memory, and morphological and functional alterations in the hippocampus are typical of many prominent neurodevelopmental disorders associated with mental retardation, such as the fragile X-syndrome and the Down syndrome (18,19), we focussed on the hippocampus in this study.
Mental retardation in PKU develops after birth, because maternal Pah activity protects against elevation of phenylalanine concentration during embryonic development; hence, the consequences of the gene defect are compensated until birth.We therefore studied synaptic plasticity in the hippocampus of the Pah enu2 /c57bl6 mouse from birth until 12 weeks thereafter.We also analysed microglia activity in the hippocampus of Pah enu2 /c57bl6 mice (20), which recently turned out to be a key player in synaptic pruning, the physiological loss of synapses during early postnatal development.
In this study, we demonstrate dysfunctional synaptogenesis and synaptic transmission in response to elevated levels of phenylalanine, which resulted in reduced microglia activity and, as a consequence, in reduced synaptic pruning in the hippocampus of Pah enu2 /c57bl6 mice.We propose that the lack of synapse remodelling after birth may contribute to mental retardation in the PKU mouse.

Phenylalanine hydroxylase knockout impairs long-term potentiation and paired-pulse facilitation
In view of the mental retardation found accompanying PKU, we studied the effects of PKU-like, elevated levels of phenylalanine by using electrophysiological methods.We firstly examined the induction and maintenance of long-term potentiation (LTP).As shown in Figure 1, the magnitude of potentiation in wild-type (WT) mice clearly differed from that in Pah enu2 mice.In WT animals, we found a significant increase of mean fEPSP slopes to 217 + 14% after 60 min compared with the baseline (Fig. 1A and B).Mean fEPSP slopes of the Pah enu2 mice significantly decreased to 173 + 13% after 60 min as compared with control animals (Fig. 1B).This reduction is already seen 10 min after theta-burst stimulation (TBS) (Fig. 1B).As a next step, we focused on the presynaptic mechanisms contributing to synaptic potentiation.We used different TBS protocols consisting of 6, 8 or 10 bursts, applied at 5 Hz in all experiments, according to Raymond (2007).In WT animals, TBS stimulation resulted in increased synaptic potentiation and therefore in increased activation of synapses.In contrast, in Pah enu2 mice, the potentiation was reduced as compared with the controls.As a result, increasing the strength of TBS correlated positively with increasing synaptic potentiation in the controls but not in the Pah enu2 mouse (Fig. 1C).Hence, stronger theta-burst stimuli did not result in stronger activation, which points to altered presynaptic mechanisms in Pah enu2 mice.A classical model for studying presynaptic effects is paired-pulse facilitation (PPF), indicating transmitter release (21).Both WT animals and knockout animals show PPF between 320 and 80 ms, as shown in Figure 2.While WT animals reach their maximal facilitation at an interstimulus interval of 40 ms, the maximum is reached in Pah enu2 mice at an interstimulus interval of 80 ms.Thus, knockout mice have a significantly reduced PPF at an interval of 40 ms compared with the control animals.

Reduced expression of the presynaptic proteins and increased expression of postsynaptic proteins in Pah enu2 mice
To substantiate our electrophysiological results in response to high concentrations of phenylalanine, we additionally studied presynaptic and postsynaptic proteins by quantitative immunohistochemistry and western blot analysis in the hippocampus of the Pah enu2 mice.For the presynaptic side, we chose synaptophysin, which is a constituent of transmitter vesicle membranes (22,23), and loss of synaptophysin function is associated with mental retardation (24).Synaptosomal-associated protein 25 (SNAP-25) is part of the SNARE complex, which is essential for transmitter release (25).
Quantitative immunohistochemical evaluation of presynaptic proteins in the hippocampus of adult (12 weeks) Pah enu2 mice and WT littermates revealed a significant reduction in the immunoreactivity of both proteins.The downregulation of the proteins was seen in both the CA1 (stratum radiatum) and CA3 (stratum lucidum) regions (Fig. 3A and B).The degree of downregulation of SNAP-25 immunoreactivity was region-specific; immunoreactivity was reduced by 15% in the CA1 region and by 40% in the CA3 region (Fig. 3C).With synaptophysin, quantification of immunoreactivity by image analysis revealed a downregulation by 25% in CA1 as well as in CA3 as compared with the situation in WT animals (Fig. 3D).Presumably due to region specificity of protein expression, densitometric evaluation of both proteins in lysates of the whole hippocampus by western blot analysis confirmed the downregulation of the protein with synaptophysin but not with SNAP-25 (Fig. 3E and F).
Synaptopodin and spinophilin were used as postsynaptic proteins.Synaptopodin is closely associated with the spine apparatus, a spine-specific organelle, which is typical for large mushroom spines, the so-called memory spines.Synaptopodin knockout mice show deficits in behavioural tests for learning and memory and impaired LTP (26).Spinophilin is an actinassociated protein, which is enriched in spines (27,28).Quantification of the immunofluorescence signal of synaptopodin and spinophilin revealed remarkably increased expression of both proteins in the hippocampus of adult Pah enu2 mice as compared with WT littermates (Fig. 4A and B).This increase was evident in the stratum radiatum of the CA1 region as well as in the stratum lucidum of the CA3 region.The staining intensity of synaptopodin showed a significant increase of up to 230% in the CA1 and 160% in the CA3 region of the hippocampus (Fig. 4C).Similarly, spinophilin was upregulated, up to 207% in CA1 and up to 224% in the CA3 region (Fig. 4D).Densitometric evaluation of whole hippocampal lysates by western blot analysis confirmed the immunohistochemical data.In whole hippocampal lysates, the expression of synaptopodin and of spinophilin was increased as compared with lysates of WT mice (Fig. 4E and F).

Increased spine synapse density in the hippocampus of the Pah enu2 mouse
We next stereologically determined spine synapse density in the stratum radiatum of the hippocampal CA1 and CA3 region in Pah enu2 mice and WT littermates during postnatal development.The synapses were defined by a bouton, a synaptic cleft, a postsynaptic density and a spine (Fig. 5A).Synaptic density increased up to 3 weeks after birth; no difference was found between both genotypes.From 3 weeks after birth onwards, the density of spine synapses decreases continuously in WT mice (Fig. 5B and C).This slow and continuous loss of synapses in mammals from birth to adulthood has been frequently described (29,30) and is known as synaptic pruning.As compared with WT littermates, synaptic pruning started later during postnatal development and was significantly reduced in the hippocampus of the Pah enu2 mice (Fig. 5B and C).From the age of 4 weeks, the number of spine synapses per volume was in general significantly higher in Pah enu2 mice compared with that of their WT littermates (Fig. 5B and C).

Microglia activity is impaired in the hippocampus of Pah enu2 mice
A recent study showed that microglia play an important role in synaptic pruning by phagocytosis of cellular remnants in this process (20), which prompted us to study microglia activity in the Pah enu2 mouse (Fig. 6).We used immunoreactivity of Iba-1, a specific protein in microglial cells.We took advantage of Iba-1 because the degree of immunoreactivity of Iba-1 allows for the determination of microglial activity in single cells, in addition to simple detection of microglial cells (31,32).Laser scanning microscopy and subsequent image analysis revealed no difference in microglial activity in Pah enu2 mice as compared with WT littermates at the age of 2 weeks, neither in the CA1 nor in the CA3 region (Fig. 6B).At an age of 12 weeks, however, the activity of microglia was significantly reduced in the stratum lacunosum-moleculare and stratum radiatum of the CA1 region, as well as in the stratum radiatum of the CA3 region (Fig. 6B).In the CA3 region, the activity was even reduced by as much as 50% (Fig. 6B).

Phenylalanine increases spine density and regulates synaptic protein expression in hippocampal dissociated cultures
To test whether the effects we found in the Pah enu2 mouse are direct effects of elevated levels of phenylalanine in this mutant or whether they are due to potentially altered concentrations of other metabolites, we finally studied the effects of various doses of phenylalanine on dissociated hippocampal neurones.We used hippocampal cultures from embryonic day 18, as these cultures are pure neuronal cultures and the contamination with glial cells is ,1% (33).After 2 weeks in culture, the neurones were either transfected with EGFP to visualize spines and to determine their density along dendrites after phenylalanine treatment (Fig. 7A  and B), or the neurones were processed for immunohistochemistry and examined for the expression of synaptic proteins.Phenylalanine was applied at doses to the medium that result in final concentrations of 1, 2 and 5 mM.In control experiments, we found no effects of these doses on the viability of the neurones, as tested by TUNEL assay to assess apoptosis (data not shown).Spine counting along dendrites revealed a dosedependent increase in spine density after phenylalanine treatment (Fig. 7C).When spines were counted according to their shapes (large mushroom spines, head diameter .0.8 mm, thin spines, head diameter ,0.8 mm and filopodia-like spines), we found that neither the density of large mushroom spines (Fig. 7D) nor the density of filopodia-like spines (Fig. 7E) were affected, but it was exclusively the density of thin immature spines that was increased in a clear dose-dependent manner (Fig. 7F).At a dose of 2 mM phenylalanine, we counted more than twice as many thin spines as compared with the control.In addition, the number of branched spines significantly increased dose-dependently (Fig. 7G).
As a functional read out, we evaluated immunohistochemical stainings of the synaptic proteins, which we had tested in the Pah enu2 mouse (Fig. 8).By quantitative immunohistochemistry, using confocal microscopy and subsequent image analysis, we found a downregulation of both SNAP-25 and synaptophysin after treatment with all doses of phenylalanine (Fig. 8A and B).In dissociated hippocampal cultures treated with phenylalanine to final concentrations of 1, 2 and 5 mM, no effect was seen with synaptopodin, but a clear-cut increase in the optical density of spinophilin immunoreactivity was observed (Fig. 8C and D).

DISCUSSION
In this study, we demonstrate, for the first time, that elevated levels of phenylalanine affect hippocampal synaptic remodelling after birth.Impairment of synaptic transmission and impairment of neuronal activity-dependent microglia recruitment and microglia activity underlie reduced synaptic pruning up to adulthood.Our data indicate a direct effect of elevated levels of phenylalanine on the development and maintenance of synaptic connectivity, which may add to our understanding of mental retardation in PKU.

Elevated levels of phenylalanine impair synaptic transmission of glutamatergic synapses
In acute hippocampal slices of the Pah enu2 mouse, with elevated levels of phenylalanine, both LTP and PPF were impaired as compared with WT animals, thus demonstrating the clear-cut effects of phenylalanine on synaptic transmission.Impaired synaptic transmission in the Pah enu2 mouse, however, appears to be mediated by effects on the presynapse rather than effects on the postsynapse.When we studied PPF, which displays presynaptic neurotransmitter release probability, we determined a significant reduction of the facilitation after a time interval of 40 ms in Pah enu2 mice.This finding, together with the downregulation of synaptophysin, which is a constituent of the transmitter vesicle membrane, points to a reduced pool of readily releasable neurotransmitter vesicles, as shown by Dobrunz et al. (34).Reduced synaptophysin immunoreactivity is furthermore highly consistent with the downregulation of another presynaptic protein, namely SNAP-25, which we found by quantitative immunohistochemistry. SNAP-25, as a member of the SNARE complex, is essential for the fusion of transmitter vesicles with the presynaptic membrane, and its dysfunction should result in reduced transmitter release into the synaptic cleft (25).The combination of findings on the postsynaptic side, impaired LTP and upregulated postsynaptic proteins, appeared less consistent.In addition, there is an ongoing discussion as to whether changes in LTP also represent presynaptic effects (35,36).
Reduced synaptophysin expression was also found in dissociated cortical neurones after treatment with phenylalanine (37).Reduced synaptic synaptophysin expression, however, does not allow for the authors' conclusion on reduced synaptic density.A positive correlation of synaptic protein expression and density of synapses is not obligatory, as previously shown (38), and as is also shown in this study.With respect to deficient transmitter release, decreased levels of neurotransmitters, such as dopamine, serotonin and noradrenaline, have been consistently shown, and the possibility of their role in cognitive disabilities in children suffering from PKU has been discussed (12,(39)(40)(41)(42).In many of these studies, the authors focussed on transmitters 1012 Human Molecular Genetics, 2015, Vol. 24, No. 4 whose synthesis requires phenylalanine, such as dopamine.We studied the effects of elevated concentrations of phenylalanine in the CA1 and CA3 regions of the hippocampus, which primarily consist of glutamatergic synapses.Glutamate is an amino acid; unlike phenylalanine, it is not involved in the synthesis of neurotransmitters, such as dopamine and adrenaline, and it is the main transmitter of excitatory synapses in the CNS.Thus, our data suggest that the presynaptic effects of elevated concentrations of phenylalanine shown in our study are independent of the type of transmitter that is released and may thus be relevant for all types of synapses in the CNS.
Similarly consistent data, such as impaired PPF and downregulation of presynaptic proteins, were not found postsynaptically.Impaired LTP in the Pah enu2 mouse was associated with upregulation of synaptopodin and spinophilin.Disorders of mental disabilities like fragile-X syndrome, Down syndrome and Alzheimer's disease are commonly associated with reduced LTP (43)(44)(45).Upregulation of spinophilin, however, a protein enriched in spines, is consistent with increased spine synapse density in the Pah enu2 mouse, rather than with impaired LTP.This also holds true for synaptopodin, a marker protein for mature mushroom spines.This protein has even been shown to be essential to the induction of LTP ( 26), and synaptopodinknockout mice show hippocampus-related behavioural deficits.It appears from these findings that the upregulation of postsynaptic proteins is a secondary phenomenon upon deficient transmitter release.
Impaired LTP, together with increased synapse density, is a common feature of diseases associated with mental retardation, such as fragile-X syndrome (46) and Down syndrome (47), and it was recently also shown after knockout of the aPix/Arhgef6 gene, which causes X-linked intellectual disability in humans (48).Hence, the formation of new spines is not likely to indicate improved synaptic functioning.In fact, classification of spines according to their morphology in vitro revealed that the spinepromoting effect of phenylalanine was only seen with thin spines, but was not found with large, mushroom-like, mature spines.Mushroom spines are functionally stronger than thin spines; they persist for months and contain more NMDA and AMPA receptors (49)(50)(51).In addition, the overall increase in spine density also included branched spines.Branched spines could give rise to perforated synapses, which are thought to augment synaptic efficacy during neural transmission (52,53).Thus branched spines could represent a compensatory mechanism in response to reduced neuronal activity after exposure to high concentrations of phenylalanine.Increased spine and spine synapse densities in the presence of elevated levels of phenylalanine also correlate positively with the upregulation of NMDA and AMPA receptors in PKU, as previously shown by Glushakov and co-workers (54,55).Moreover, the upregulation of AMPA and NMDA receptors could also result from disturbed transmitter release, similar to our findings of upregulated synaptopodin and spinophilin.
In sum, enhanced expression of postsynaptic proteins, namely spinophilin and synaptopodin, together with increased spine and spine synapse density, appears to be a compensatory mechanism upon disturbed transmitter release, as shown by using electrophysiological parameters and reduced SNAP-25 expression (56).

Synaptic pruning and microglia
Physiologically, the initial increase in the number of spine synapses shortly after birth is followed by a continuous prolonged synapse loss up to adulthood.This synapse loss during early postnatal development, namely synaptic pruning, has been described for different brain areas and species (57).In this study, we show for the first time that synaptic pruning is reduced in the hippocampus of Pah enu2 mice.After an increase in synapse density shortly after birth, we found higher hippocampal synaptic density in the Pah enu2 mice as compared with WT animals.Higher synapse density in the hippocampal CA1 and the CA3 regions was seen up to 12 weeks of age.Most importantly, enhanced synapse density in CA1 and CA3 is accompanied by decreased activity of microglia in both regions.Prior to the start of synaptic pruning, the expression of microglial proteins was unaffected and did not differ from WT animals.
A prominent role of microglia in synaptic pruning has been elegantly demonstrated in very recent studies.It was shown that microglia modify postnatal circuits in an activity-and complement-dependent manner (58)(59)(60)(61)(62)(63).Abolishment of microglia activation by knockout of the fractalkine receptor (Cx3cr1) results in a significantly higher synapse density in the hippocampal CA1 region shortly after birth and thereby shows deficits in synaptic pruning (20).Similarly, in the C3 complement knockout mouse, synapse density was also increased and paralleled by lower neurotransmitter release of glutamatergic synapses.Our data are consistent with these findings, because synaptic pruning is initiated at a later time point during postnatal development in the Pah enu2 mouse as compared with the WT and is associated with reduced microglia activity and increased synapse density in our mutant.In contrast to the findings by Paolicelli et al. (20), showing that increased synapse density is a transient phenomenon, reduced synaptic pruning, together with reduced microglia activity, persists in the hippocampus of Pah enu2 mice up to 12 weeks of age.As synaptic pruning is aimed at elimination of weak synapses (59), one would assume persistent synaptic dysfunction in the Pah enu2 mouse.In fact, deficient synaptic pruning is associated with weak synaptic transmission and decreased functional brain connectivity, as recently shown by Zhan et al. (60).Microglia-dependent elimination of synapses during early postnatal development is obviously a prerequisite for the formation of strong synaptic contacts in adulthood (60).
In the visual system, it was shown for the first time that developmental microglia-associated synaptic pruning depends on neuronal activity.Surprisingly, under conditions of reduced neuronal activity, the engulfment of synaptic elements by microglia was increased (59) and weaker synapses were eliminated.Prior to phagocytosis, however, the recruitment of microglia and their contact to synapses is obligatory, and both depend on neuronal activity.In the healthy developing brain, microglia make brief (5 min) and periodical (1/h) contacts with synapses (64).Reduction of neuronal activity by light deprivation or TTX treatment reduced contact frequency as well as microglia motility, and it was possible to reverse this effect by light  re-exposure or under TTX-free conditions (64,65).As phenylalanine affects synaptic transmission in dissociated microgliafree hippocampal cultures, our findings strongly suggest that reduced neuronal activity in the hippocampus of the Pah enu2 mouse accounts for insufficient microglia recruitment and motility and, as a consequence, for deficient synaptic pruning.
Together, we propose that elevated levels of phenylalanine affect neurotransmitter release and subsequently impair neuronal activity, which results in deficient neurone-microglia signalling and, as a consequence, in insufficient synaptic pruning and thus in impaired brain connectivity.

Animals
Homozygous Pah enu2 mice and WT C57BL/6 mice were conceived from heterozygous mating.The mice were housed on a 12-h light -dark cycle with food and water ad libitum.Genetic characterization was performed on DNA prepared from tail tissue using the Extract-N-Amp TM Tissue PCR Kit (Sigma-Aldrich).Subsequent PCR amplification of exon 7 of the Pah gene (fw primer cttgtactggtttccgcctcc, rev primer ccagcctgcaatgagcctgatc) was carried out, succeeded by digestion of the PCR product with the FastDigest Enzyme Alw26I (Fermentas) in order to detect the Pah enu2 mutation.All experiments were carried out in accordance with the institutional guidelines for animal welfare.

Transfection of hippocampal dispersion culture
The neurones were transfected with EGFP after 10 days in vitro (DIV10) for dendrite and spine evaluation.The Effectene Transfection Reagent Kit (Qiagen), with some alterations to the recommended protocol, was used for transfection.Briefly, pEGFP DNA was diluted with Buffer EC to a concentration of 0.1 mg/ml.Enhancer was added (1 : 8) and the mixture was vortexed.Finally, Effectene Transfection Reagent was added to the solution and incubated for 10 min.The transfection complex was then added dropwise to the cells.After 1.5 h, the neurones were washed with pre-chilled PBS and cultivated for another 24 h in culture medium.

Immunohistochemistry and image analysis
For the experimental approach, Pah enu2 mice and WT littermates of different ages (2, 3, 4, 8 and 12 weeks) were deeply anesthetized with an intraperitoneal injection of 2 ml/kg of a ketamine-xylazine mixture (ketamine 12 mg/ml; xylazine 0.16% in saline).The anesthetized animals were perfused with physiologic salt solution, followed by a 4% paraformaldehyde (PFA) solution.The brains were removed and postfixed in 4% PFA overnight at 48C.Immunohistochemistry was performed on cryostat sections (20 mm) of these brains.Dispersed hippocampal neurones were fixed with 4% PFA for 15 min, then washed with PBS and stored at 48C until further use.The following antibodies were used: mouse anti-synaptophysin (1 : 300, Millipore), rabbit anti-SNAP-25 (1 : 300, Abcam), rabbit anti-synaptopodin (1 : 300, Synaptic Systems), rabbit anti-spinophilin (1 : 300, Sigma), mouse anti-MAP2 (1 : 500, Sigma), rabbit anti-MAP2 (1 : 500, Millipore) or rabbit anti-Iba-1 (1 : 800, Wako).The sections, as well as the fixed cultures, were incubated over night at 48C with the primary antibody, followed by incubation with the secondary antibody solution for 2 h at room temperature (Alexa 488conjugated goat anti-rabbit or goat anti-mouse IgG, 1 : 500, Cy3-conjugated goat anti-rabbit or goat anti-mouse IgG, 1 : 500, Molecular Probes).The nuclei were stained with DAPI (Sigma-Aldrich).The stained tissue sections were scanned with a Zeiss LSM 510 Meta confocal microscope (Zeiss, Jena, Germany) using a ×63 oil immersion objective.The specificity of the immunostaining was demonstrated by the absence of signals in sections processed after omission of the primary antibody.An image field of a defined size was placed on the stratum radiatum of CA1 and the stratum lucidum of the CA3 region for protein analysis.To evaluate microglia activation, the staining intensity of Iba-1 was measured in the area of the stratum lucidum of CA3 and stratum radiatum, as well as in the stratum lacunosum moleculare of the CA1 region.The intensity of the staining was calculated according to the activity of microglia in each area.Quantification of the staining intensity was carried out using the software ImageJ (NCBI).The relative staining index was calculated by the software.The software Biorevo BZ-9000 (Keyence) was used for spine evaluation.After image acquisition, spine lengths and widths, as well as dendrite lengths, were manually measured.To avoid bias, all analyses were performed with the investigator blind to the protocol of the sample under study.

Electron microscopy
Pah enu2 mice and WT littermates of different ages (2, 3, 4, 8 and 12 weeks) were deeply anesthetized with an intraperitoneal injection of 2 ml/kg of a ketamine -xylazine mixture (ketamine 12 mg/ml; xylazine 0.16% in saline) and transcardially perfused with 2.5% glutaraldehyde in PBS.The brains were removed and postfixed overnight at 48C.The hippocampi were dissected out of the brains and divided into posterior (CA3) and ventral (CA1) parts.Following fixation in 2.5% glutaraldehyde overnight, the samples were postfixed in 1% OsO 4 for 30 min.Subsequently, the samples were dehydrated in graded ethanol Human Molecular Genetics, 2015, Vol. 24, No. 4 starting with 10% and progressing to 100%.The dehydrated samples were then embedded in Epon 820 (Serva).The blocks were trimmed and cut in thin sections on a Reichert-Jung OmU3 ultramicrotome.The ultrathin sections were stained with uranyl acetate, followed by lead citrate.
Spine synapse density was calculated using unbiased stereological methods, as previously described (67).In brief, consecutive serial ultrathin sections were cut and placed on formvar-coated single grids.The sections contained sections either from the CA1 stratum radiatum or the CA3 stratum lucidum.Digitized images were taken at a magnification of ×5800 (CM100 electron microscope, Philips, Aachen, Germany).The dissector technique was used to obtain a comparable measure of synaptic numbers, unbiased for possible changes in synaptic size (68).To calculate the spine synapse density on pyramidal cell dendrites, a reference grid was superimposed on the electron microscopic prints.Only those spine synapses were counted that were on the reference section but not on the consecutive section.For each animal and each area, at least 20 neuropil fields were analysed.The observer was blinded to the experimental group the whole time.

Western blot
The hippocampi of 12-week-old Pah enu2 mice and WT littermates were used for protein analysis by western blot.After cervical dislocation, the hippocampi were immediately removed from the brain and frozen on liquid nitrogen.After thawing at 48C, the hippocampi were manually homogenized in a buffer containing 50 mM Tris-base, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mM EDTA and a mixture of proteinase and phosphatase inhibitors (Promega).The samples were centrifuged at 13 000 rpm for 20 min at 48C, and the supernatants were frozen until further use.Ten micrograms of each sample were run on a 10% SDS-PAGE and blotted on nitrocellulose membranes.Unspecific binding of antibodies was prevented by treatment of the membrane with 5% non-fat milk powder solution (in TBS plus 0.3% Tween).The incubation with the primary antibodies was carried out in blocking solution for 12 h at 48C.The same antibodies that were used for the immunohistochemistry were used in the following dilution: mouse anti-synaptophysin (1 : 1000), rabbit anti-SNAP-25 (1 : 4000), rabbit antisynaptopodin (1 : 1000) and rabbit anti-spinophilin (1 : 1000).As an internal standard, a-tubulin expression was determined (mouse anti-a-tubulin, 1 : 40 000, Sigma).The secondary antibodies goat anti-mouse IgG (1 : 2500, Thermo scientific) or donkey anti-rabbit-IgG (1 : 5000, Thermo scientific) conjugated to horseradish peroxidase were applied for 2 h at room temperature.Immunodetection was accomplished using the Immobilon Western Chemiluminescent HRP Substrate (Merck Millipore), and the chemiluminescent signal was detected by the FUSION-SL4 advanced imaging system (Vilber) and quantified by densitometry using the software ImageJ (NCBI).

Electrophysiology
For electrophysiological examination, adult (12 weeks) Pah enu2 mice and their WT littermates were used.The animals were anaesthetized with isoflurane and decapitated.The brains were removed and separated in left and right hemisphere and put in ACSF (48C).With the aid of a microvibratome (VT1000, Leica), the hemispheres were cut in 300-mm slices.The slices were kept in oxygenized ACSF at 378C for equilibration.The ACSF consisted of (in mM): 125 NaCl, 25 NaHCO 3 , 2.5 KCl, 1.25 NaH 2 PO 4 , 2 CaCl 2 , 1.5 MgCl 2 and 22.7 D-glucose.A Zeiss Axioskop 2FS and micromanipulators (M0-103, Narashige; SM1+Unit Mini3, Luigs and Neumann) were used for the recordings.Evoked field excitatory postsynaptic potentials (fEPSPs) were recorded extracellularly from the stratum radiatum of the CA1 region.Synaptic responses of these pyramidal neurones were evoked via the Schaffer collaterals from the CA3 region.For both regions, ACSF-filled glass electrodes with a resistance between 1 and 4 MV were used.Stimulation was carried out with an EPC-10 patch clamp amplifier (HEKA).Synaptic response was preamplified by a Philips PM 5170 amplifier.In this connection, the Schaffer collaterals were activated by TBS.To induce LTP, a pattern consisting of eight bursts (5 Hz), each burst consisting of four pulses (100 Hz), was used.Each pulse lasted for 0.2 ms [modified according to (69,70)].At the beginning of each recording, stimulation intensity was adjusted to elicit an amplitude of ≤50% of the maximal fEPSP slope.Test pulses were therefore presented (0.2 Hz, every 2.5 s) to establish stable baseline fEPSPs.After 20 min of stable baseline recording, TBS was administered to evoke LTP in CA1.After this high-frequency stimulation, the fEPSPs were continuously monitored for another 90 min.Changes in fEPSP slopes (60-80 min post-theta-burst) compared with the baseline were used to measure LTP.Presynapse function was assessed with the paired-pulse paradigm, delivering two closely spaced stimuli at different interstimulus intervals (ISI of 320, 160, 80, 40, 20 and 10 ms), determining the relative facilitation of the second response.The ratio of PPF was calculated as EPSP2 slope/EPSP1 slope, averaging over three responses per pulse pair.Data analysis was performed using Pulsefit v8.80 (HEKA), SPSS 18.0 (SigmaPlot) and Excel (Microsoft).All values are presented as means + SEM.N indicates the number of experiments (acute slices), comprising at least three animals.

Statistics
The data of dendritic spine synapse quantification, immunoreactivity of the proteins and microglia, spine evaluation and quantification by western blot are presented as means + SEM.The paired t-test was used for statistical analysis.In all experiments, Pah enu2 mice were always compared with WT littermates.In experiments with hippocampal neuronal dispersion cultures, neurones treated with phenylalanine were always compared with control conditions without elevated phenylalanine levels.The level of significance was set at P , 0.05.

Figure 1 .
Figure 1.LTP in acute hippocampal slices of WT and Pah enu2 mice.(A) In slices of WT mice, the average course of fEPSP slopes increased to 189 + 6% after 10 min and to 217 + 14% after 60 min.In Pah enu2 mice, the slopes increased only up to 150 + 9% after 10 min and up to 173 + 13% after 60 min in response to HFS. (B) Quantitative evaluation of the slopes in WT and Pah enu2 mice, showing a significant reduction of the fEPSP slopes after 10 and 60 min.(C) Quantitative evaluation of the slopes in WT and Pah enu2 mice, showing that the fEPSP slope increase correlate with the number of bursts in slices of WT animals but not in slices of the mutant (mean + SEM, N ¼ 3 animals, * P , 0.05, * * P 0.01, Students t-test).

Figure 2 .
Figure 2. Mean PPF curves of WT and Pah enu2 mice.Mean slopes ratios of WT (solid line and open circles) and Pah enu2 mice (dotted line and grey squares) are plotted against the interstimulus interval.While WT animals reach their maximal facilitation at an interstimulus interval of 40 ms, Pah enu2 mice reach their maximum at an interstimulus interval of 80 ms (mean + SEM, N ¼ 3 animals, four slices per animal, Students t-test).

Figure 3 .
Figure 3. Immunoreactivity of presynaptic proteins in hippocampal sections of WT and Pah enu2 mice.Example figures of SNAP-25 (A) and synaptophysin (B) in the CA1 hippocampal region of 12-week-old animals.(C) Quantitative evaluation of immunohistochemical stainings of SNAP-25 in CA1 and CA3.In both regions, the protein is significantly downregulated.(D) Quantitative evaluation of immunohistochemical stainings of synaptophysin in CA1 and CA3.In both regions, the protein is significantly downregulated.(E and F) Quantification of protein expression by western blot analysis revealed no difference between SNAP-25 in WT and in the mutant, but revealed a significant downregulation of synaptophysin in Pah enu2 mice (mean + SEM, N ¼ 3 animals, 10 sections per animal, * * P , 0.01, Students t-test).

Figure 4 .
Figure 4. Immunoreactivity of postsynaptic proteins in hippocampal sections of WT and Pah enu2 mice.Example figures of synaptopodin (A) and spinophilin (B) in the CA1 hippocampal region of 12-week-old animals.(C) Quantitative evaluation of immunohistochemical stainings of synaptopodin in CA1 and CA3.In both regions, the protein is significantly upregulated.(D) Quantitative evaluation of immunohistochemical stainings of spinophilin in CA1 and CA3.In both regions, the protein is significantly upregulated.(E and F) Quantification of protein expression by western blot analysis revealed a significant upregulation of both proteins in Pah enu2 mice (mean + SEM, N ¼ 3 animals, 10 sections per animal, * P , 0.05, * * P , 0.01, Students t-test).

Figure 5 .
Figure 5. Synapse density in the hippocampus of WT and Pah enu2 mice.(A) Example figure of a spine synapse; these were stereologically counted in electron micrographs.(arrow: bouton, arrow head: synaptic density, asterisk: spine) (B and C) In CA1 (B) and in CA3 (C), there is no difference in spine synapse density up to 3 weeks after birth.Thereafter, up to 12 weeks of age, spine synapse density is increased in Pah enu2 mice as compared with WT (mean + SEM, N ¼ 3 animals, 20 neuropil fields, * P , 0.05, * * P , 0.01, Students t-test).

Figure 6 .
Figure 6.Immunoreactivity of Iba-1 in the hippocampus of the Pah enu2 mouse as compared with WT. (A) Example figures of Iba1 immunoreactivity in the CA3 hippocampal region of 12-week-old Pah enu2 and WT mice.(B) Quantitative evaluation of Iba1 immunoreactivity per cell by image analysis of immunohistochemical stainings.(mean + SEM, N ¼ 3 animals, five sections per animals, * * P , 0.01, Students t-test).

Figure 7 .
Figure 7. Spine density in dissociated neuronal hippocampal cultures.(A and B) Example figures of spine density after transfection of the neurones with EGFP under control conditions (A) and after treatment of the cultures with 5 mM phenylalanine (B).Spine density is increased.(C) Quantification of all spines revealed a significant increase in the number of spines per 100 mm length of dendrite.(D-G) Counts of spines according to their shapes revealed no differences in the density of mushroom spines (D) and filopodia (E), but a dose-dependent increase in the number of thin spines (F) and branched spines (F) (mean + SEM, N ¼ 3 cultures, 40 cells per culture, * P , 0.05 to control, * * P , 0.01 to control, Students t-test).

Figure 8 .
Figure 8. Immunoreactivity of pre-and postsynaptic proteins in dissociated neuronal hippocampal cultures.(A and B) Presynaptic proteins (SNAP-25 and synaptophysin) are significantly downregulated at all doses of phenylalanine.Postsynaptic protein expression is unchanged with synaptopodin (C) and upregulated with spinophilin (D) (mean + SEM, N ¼ 3 cultures, 40 cells per culture, * P , 0.05 to control, * * P , 0.01 to control, Students t-test).