BASIC SCIENCE Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198 – 209 Original Article nanomedjournal.com Poly(propylene imine) dendrimers with histidine-maltose shell as novel type of nanoparticles for synapse and memory protection Ester Aso, PhD a, e , Isak Martinsson, MSc b , Dietmar Appelhans, PhD c , Christiane Effenberg, MSc c , Nuria Benseny-Cases, PhD d, e , Josep Cladera, PhD d , Gunnar Gouras, MD, PhD b , Isidre Ferrer, MD, PhD a, f , Oxana Klementieva, PhD a, b,⁎ a Institute of Neuropathology, IDIBELL-University Hospital Bellvitge, L'Hospitalet de Llobregat, Spain b Wallenberg Neuroscience Center, Lund University, Lund, Sweden c Leibniz Institute of Polymer Research Dresden, Dresden, Germany d Universitat Autònoma de Barcelona, Bellaterra, Spain e ALBA Synchrotron Light Source, Cerdanyola del Vallès, Spain f CIBERNED, Instituto Carlos III, Spain Revised 17 December 2018 Abstract Poly(propylene imine) dendrimers have been shown to be promising 3-dimensional polymers for the use in the pharmaceutical and biomedical applications. Our aims of this study were first, to synthesize a novel type of dendrimer with poly(propylene imine) core and maltose-histidine shell (G4HisMal) assessing if maltose-histidine shell can improve the biocompatibility and the ability to cross the blood- brain barrier, and second, to investigate the potential of G4HisMal to protect Alzheimer disease transgenic mice from memory impairment. Our data demonstrate that G4HisMal has significantly improved biocompatibility and ability to cross the blood-brain barrier in vivo. Therefore, we suggest that a maltose-histidine shell can be used to improve biocompatibility and ability to cross the blood-brain barrier of dendrimers. Moreover, G4HisMal demonstrated properties for synapse and memory protection when administered to Alzheimer disease transgenic mice. Therefore, G4HisMal can be considered as a promising drug candidate to prevent Alzheimer disease via synapse protection. © 2019 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons. org/licenses/by-nc-nd/4.0/). Key words: Dendrimers; Histidine; Maltose; Biocompatibility; Amyloid; Synapses Alzheimer’s disease (AD) is an age-dependent neurodegen- could stop or delay AD progression. It is already accepted that erative process characterized by the presence of senile plaques loss of synapses is considered the best pathological correlate of mainly composed of different species of aggregated β-amyloid cognitive decline 2–4 and the link between soluble Aβ aggregates (Aβ), and by the presence of neurofibrillary tangles, mostly (oligomers) and synapse degradation is already established. 5,6 composed of various isoforms of hyperphosphorylated tau Thus, one direction for AD therapies could be protection of protein. 1 At present, there is no effective treatment, which synapses, which includes finding molecules that can protect Abbreviations: AD, Alzheimer disease; BBB, Blood-brain barrier; G3Mal, Maltose modified PPI dendrimers of 3 rd generation; G4HisMal, Maltose modified PPI dendrimers of 3 rd generation; FITC, Fluorescein-5/6-isothiocyanate; LDH, Cytotoxicity assay based on colorimetric measurements of lactate dehydrogenase activity; MTT, Cell proliferation assay based on colorimetric measurements of tetrazolium dye 3- (4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; PAMAM, Polyamidoamine; PEG, Polyethyleneglycol; PPI, Poly(propylene imine); ThT, Thioflavin T; ThS, Thioflavin S; SAXS, Small angle x-ray scattering; MALDI-TOF, Matrix Assisted Laser Desorption/Ionization Time of Flight Mass Spectrometry Conflict of Interests: There is no financial or personal interest or belief that could affect authors’ objectivity or could be the source and nature of a potential conflict. ⁎Corresponding author at: Wallenberg Neuroscience Center, Lund University, BMC B10, Sölvegatan 19, 22184 Lund, Sweden. E-mail address: oxana.klementieva@med.lu.se (O. Klementieva). https://doi.org/10.1016/j.nano.2019.01.010 1549-9634/© 2019 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/). Please cite this article as: Aso E, et al, Poly(propylene imine) dendrimers with histidine-maltose shell as novel type of nanoparticles for synapse and memory protection. Nanomedicine: NBM 2019;17:198-209, https://doi.org/10.1016/j.nano.2019.01.010 E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 199 synapses against the damage associated with soluble Aβ Maltose monohydrate, L-histidine, sodium borate, boran oligomers. pyridine complex, and fluorescein-5/6-isothiocyanate (FITC) In this study, we used 3-dimensionally branched macromol- were purchased from Sigma-Aldrich (Germany). ecules called dendrimers. Dendrimers are built by a series of iterative synthetic steps from a small core molecule, propylene Aβ(1−40), with the amino acid sequence: [DAEFRHDS- imine in our case. Dendrimers have important features such as GYEVHHQKLVFFAEDVGSNKGAIIG LMVGGVV], was controlled structure, nanoscopic size and high tunable availabil- obtained from JPT Peptide Technologies GmbH (Germany). ity of multiple functional groups at their surface. 7 In medical applications, dendrimers have high potential as drug nanocar- Aβ(1−42), with the amino acid sequence: [DAEFRHDS- riers, imaging agents or as drugs per se. 8 However, the ability to GYEVHHQKLVFFAEDVGSNKGAIIG LMVGGVVIA], cross the blood−brain-barrier (BBB) has been shown only for was obtained from Tocris (USA). Amyloid peptides were few types of dendrimers, such as polyamidoamine (PAMAM) dissolved in 10 mmol HEPES buffer (Sigma-Aldrich Spain, dendrimer–drug conjugates, 9–11 and maltose poly(propylene Sweden) with 0.02 % NH3 at pH 12; stock concentration was imine) dendrimers. 12 Previously we have reported that maltose 250 μmol; aliquots were kept at −80°C until use. shell significantly reduces poly(propylene imine) (PPI) dendri- mers of 3 rd generation toxicity (G3Mal) in vivo and allows G3Mal to cross blood−brain-barrier. However, in spite of Synthesis of G4HisMal dendrimer confirmed anti-amyloidogenic properties, G3Mal has not been able to improve memory deficits in APP/PS1 transgenic mice. 12 Synthesis of G4HisMal was performed in two steps: First, For the present study, we synthesized a novel type of Poly(propylene imine) dendrimers of the 4th generation were dendrimers: poly(propylene imine) dendrimers with a histidine- modified with His (G4His), then G4His was modified with maltose shell (G4HisMal). Histidine was selected due to several maltose (G4HisMal). reasons: is selectively transported through the BBB 13 it has The whole conversion process was carried out under argon chelating properties for Cu 2+ ions 14 which is considered to be protection atmosphere. G4 dendrimer (0.135 g (1 equivalent); important since metal ion dyshomeostasis plays a detrimental 1.88×10 -5 mol) was freeze-dried overnight, while anhydrous role in oxidative stress related to AD progression 15; it has some DMSO (5 mL) was additionally pretreated under stirring in high neuroprotective capacity. 16 For dendrimers modification, histi- vacuum condition for several hours to dissolve freeze-dried G4 dine was combined with maltose, since G3Mal has been proved dendrimer and triethylamine (0.142 mL). A second reaction as non-toxic antiamyloidogenic agents capable to cross BBB. 12 solution was prepared to unify N-Boc-L-histidine (125 mg; 26 Here, we hypothesized that due to maltose PPI dendrimers may equivalents; 4.9×10 -4 mol) and BOP (283 mg; 34 equivalents keep anti-amyloidogenic properties meanwhile the added related to N-Boc-L-histidine; 6.4×10 -4 mol) in 10 mL anhydrous histidine may help PPI dendrimer to cross BBB and will add DMSO and, then, stirred for 1 hour at room temperature. The neuroprotective properties. ester-activated N-Boc-L-histidine solution was slowly added to Using in vitro and in vivo models of AD, we characterized the the dendrimer solution. Then, the corresponding reaction mixture anti-amyloidogenic and neuroprotective properties G4HisMal. was stirred for 24 hours at room temperature. The reaction Here we report that G4HisMal had significantly improved solution was intensively dialyzed in bidistilled water for 3 days biocompatibility and the ability of cross BBB. We proved that using dialysis tube with 1000 MWCO (ZelluTrans, Roth G4HisMal crossed BBB, and did not accumulate in the brain (Germany), Flat with 45 mm) by exchanging bidistilled water tissue being well-tolerated since treated no visible signs of at least 3 times per day (5l beaker glass). To obtain solid material weaknesses or apathy in mouse behavior were recorded during for G4His with 99% yield 2-step freeze-drying was done by all period of chronic treatment. Strikingly, G4HisMal treatment reconstitution of the dialyzed product in 5 mL bidistilled water prevented memory decline during AD-like pathology. Our data for the second freeze-drying process. G4His was characterized demonstrated that the positive cognitive effects induced by by 1H and 13C NMR and mass spectrometry (Supplementary G4HisMal in aged AD transgenic mice were not associated with Figures 1-3). insoluble Aβ load reduction but with synapse protection. G4His (0.1 g (1 equivalent); 7.5 × 10-6 M), D(+)-maltose monohydrate (5.513 g (nearly 32 equivalents for each amino group of G4His), 15.3 mmol), and borane-pyridine complex (3.4 Methods mL; 15.3 mmol; 8 mol solution) were taken up in a sodium Reagents borate buffer (15 mL; 0.1 mol). The reaction solution was stirred at 50°C for 7 days. Then, the crude product was purified by PPI dendrimer of the 5 th generation with 64 terminal amino dialysis towards bidistilled water for 3 days, exchanging water at groups and 4 th generation with 32 terminal amino groups was least three times per day. G4HisMal was obtained from freeze- obtained from SyMO-Chem (Eindhoven, The Netherlands) and drying process. The yield was (0.95 g, 72 %). Conversion of the specified as 4th generation PPI dendrimers (PPI G4) and 3 rd PPI G4HisMal with FITC was used for the detection and mapping of dendrimer (PPI G3) following the uniform nomenclature G4HisMal in cells and brain tissue. Synthesis of FITC-labeled description of polyamine dendrimers, 17 this nomenclature was dendrimers was done as described. 18 Synthetic pathways for also applied for other previously published PPI dendrimers and G4HisMal is presented in Figure 1. Description of characteri- cited in the present text (ref. 12, 35 and 36). zation of dendrimers is described in Supplementary Information. 200 E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 Figure 1. Structure and synthesis of G4HisMal. (A) Simplified structure of G4HisMal. Layers 1-4 indicate dendrimer’s branching points and generation of the dendrimer. Layer 4 shows N terminal groups with maltose as R1 and histidine as R2 substituents. (B) Distance distribution functions P(R) of G4HisMal in PBS calculated SAXS pattern of G4HisMal in PBS at 37°C using GNOM. 40 Insertion: De novo three-dimensional reconstruction of the scattering entity of G4HisMal using DAMMIF after 10 independent reconstructions. A line indicates the maximum dimension (Dmax), UCSF Chimera 41 was used for visualization. The SAXS raw data model fitting is shown in Supplementary Figure 7. (C) Reaction pathway for synthesizing glycodendrimer G4HisMal. Design of experiments in vivo BBB cross Short-term intranasal administration was conducted in non- The experiments carried out on male APP/PS1 and wild-type transgenic littermates aged 6 months treated with FITC mice. Two experimental treatment settings were designed, −G4HisMal, or the PBS (n = 3 per group) by applying two covering short-term intranasal administration (BBB cross), and equal drops with a micropipet to each nasal cavity (5 μL), long-term intranasal administration (treatment). resulting in a total dose of 10 mg/kg/body weight. After 1 hour, E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 201 the animals were perfused through the left cardiac ventricle with cortex in which Aβ is deposited in APP/PS1 mice. (n = 5, 6 per cold PBS pH 7.4 under deep whole-body anesthesia (ketamine/ group).The Aβ burden was quantified using Adobe Photoshop xylazine, 5:1; 0.10 mL/10g/body weight, intraperitoneally). CS4 software (Adobe Systems, San Jose, CA, USA) Specific When blood was cleaned up from the brain blood vessels, the immunolabeling densities (glial response, oxidative stress) were brains were removed, one brain hemisphere was homogenized, calculated in reference to the Aβ plaque area (6F/3D-positive) in while the second one was snap frozen and stored at −80°C until 5 representative pictures taken from the cortex of each animal. use. Fibrillar amyloid burden was calculated as the percentage of the fibrillar (OC-positive) amyloid deposition area with respect to G4HIsMal treatment the total 6E10-positive area. At the age of 3 months animals were randomly divided as follows: 7 transgenic and 6 WT mice received 5 μg per day Cell culture G4HisMal; 5 transgenic and 6 WT mice received the same volume of PBS (5 μL). Long-term intranasal administration Human neuroblastoma cell line SH−SY5Y was purchased lasted 3 months until animals reached of age of 6 months. from the European Collection of Cell Cultures (ECACC) and Memory evaluation tests were performed at the end of long- were grown as described. 12 Primary neuronal cultures were term intranasal treatment using two object recognition test in a V- generated from wild-type (wt) mouse embryos as described. 20 maze (Panlab, Barcelona, Spain) as described. 19 Statistical analysis Fluorescent measurements of FITC-G4HisMal in mouse brain Results of memory tests were analyzed with two-way tissue extract ANOVA followed by Tukey’s post hoc test. The data from the rest experiments tests were analyzed with one-way ANOVA Frozen brain tissues were homogenized in 5 volumes (wt/vol) followed by Tukey’s post hoc test. In all the experiments, the of TBS extraction buffer [140 mmol NaCl, 3 mmol KCl, 25 significance level was set at P b 0.05. mmol Tris, pH 7.4, 5 mmol EDTA, and protease inhibitor cocktail (Roche, Madrid, Spain)]. Homogenates were spun at Ethical issues 100000 g for 1 hour, and the supernatants were saved as the brain soluble fractions. Quantitative determination of Aβ(1−40) and All in vivo experiments were approved by ethics committees Aβ(1−42) in brain soluble fractions was carried out using Aβ(1−40) of the Barcelona University (Principal Investigator: E. Aso) and and Aβ(1−42) human ELISA kits (Invitrogen, U.S.A.) according Lund University (Principal Investigator: G. Gouras). In all to the instructions of the manufacturer. experiments, we followed the guidelines of the Directive 2010/ Fluorescent measurements were performed immediately after 63/EU. Animals were housed in specific pathogen-free the preparation of tissue extracts, the fluorescence of FITC labeled Animal House of University of Barcelona and Lund University. dendrimers was recorded at 485 nm excitation and 500−700 nm The animals were maintained in 12 h light/dark cycles in a emission wavelength as described. 18 The background was set as temperature regulated animal facility with free access to water an intrinsic fluorescence of brain homogenate treated with PBS. and food. The content of FITC-labeled dendrimers was quantified with standard curves of FITC-labeled dendrimers and expressed per gram of tissue. Results Maltose-histidine shell significantly improved the ability of poly Double-labeling immunofluorescence and imaging (propylene imine) dendrimers to cross BBB Brain tissue sections were immnolabeled as described 19 with PPI dendrimers of 4th generation with primary surface amino combinations of primary antibodies (Supplementary Table 1) groups were sequentially partially modified with histidine and according to manufacturer protocols. For Thioflavin S (Th S) maltose as described 21,22 (Figure 1, A-C). G4HisMal dendrimer staining, sections were incubated with 1 % ThS (Sigma) in 70 % structure was characterized by NMR spectroscopy; mass ethanol for 10 minutes. For imaging, we used confocal spectrometry; dynamic light scattering; synchrotron-based microscopes Leica TCS SP8 or Leica TCS (Leica Microsystems) small angle x-ray scattering (SAXS) and MALDI-TOF mass equipped with Diode 405/405 nm and Argon (405, 488, 552, 638 spectroscopy: We determined the degree of substitution of nm) lasers with an HP PL APO 63x/NA1.2 water immersion histidine and maltose on the outer shell as 26 histidine and 61 objectives. maltose units, with resulting molecular mass of 34000 g/mol Immunofluorescence quantification (Supplementary Figures 1–7). We used SAXS to study the overall shape and structure of G4HisMal in solution. SAXS data Amyloid burden was calculated as the percentage of the area provided information on the shape and maximal diameter (6 nm) of amyloid deposition in plaques with respect to the total cortical of G4HisMal. Interestingly that three-dimensional reconstruction area. 9 pictures were taken from 3 different sections (-0.1 mm, of the scattering entity of G4HisMal revealed shell-like structure -1.5 mm and -2.0 mm from bregma) of each animal brain. The (Figure 1, B) pictures were taken from cingular/retrosplenial/motor cortex, Since nanoparticles can pass rapidly from the nose into the somatosensory cortex and piriform/entorhinal cortex per each brain along olfactory nerves, and the brain and brain stem along section. The areas selected were the main regions of the cerebral branches of the first and second trigeminal nerve structures, 23 202 E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 Figure 2. G4HisMal dendrimers cross the BBB and inhibit Aβ(1−40) induced cell toxicity. (A) FITC fluorescence of brain tissue homogenates after i.n. administration of FITC−G4HisMal and FITC−G4Mal. Statistics: Student’s t-test, data are expressed as a mean ± SD; n = 3 per group. (B) Fluorescence microscopy images show the presence of FITC−G4HisMal in the cortex after i.n. administration of FITC−G4HisMal. The bar is 10 μm. (C) Aggregation of 20 μm Aβ(1−40) in the presence of G4HisMal. (D) Transmission electron microscopy (TEM): samples were collected at the end of ThT kinetics as shown in (C). Bar =100 nm. (E) MTT assay of Aβ(1−40) toxicity in the absence (dashed bar) and the presence of G4HisMal (gray bars). White columns correspond to control SH −SY5Y cells; these values are taken as 1 of cell viability. The experiment was repeated twice in triplicate. Statistics: one-way ANOVA followed by Tukey’s post hoc test, data are expressed as a mean ± SD. (F) Colocalization of FITC-G4HisMal dendrimers with Aβ(1−40) in SH−SY5Y cells. SH−SY5Y cells were treated with 10 μmol Aβ(1−40) and 1 μmol G4HisMal after 24 hours of incubation, then cells were fixed and labeled with specific antibodies against Aβ(1−40) (red) 1 μmol and FITC against FITC-labeled dendrimers (green). Merge shows co-localization of Aβ(1−40) and FITC-G4HisMal. Aβ(1−40) fibrils and monomers were used as controls as shown in Figure S9. Scale bar is 20 μm. intranasal (i.n.) administration was chosen as a noninvasive way G3Mal in brain tissue does not exceed 6 % of a single dose of administration. To compare the ability to cross the BBB of administered nasally to the mouse. 12 To evaluate BBB G4HisMal with precursor G3Mal, dendrimers were conjugated penetration rate of FITC-G4HisMal, we measured FITC with fluorescein isothiocyanate (FITC) as described. 18 FITC- fluorescence of brain homogenates prepared from animals G4HisMal and FITC-G3Mal dendrimers were administered treated with FITC conjugated G4HisMal, G4Mal and G3Mal. intranasally to six-month-old wild-type mice at a dose of 10 We documented that FITC fluorescence of G4HisMal was 40 % mg/kg/(body weight). Animals treated with PBS were used as higher compared to maltose dendrimers, indicating that BBB controls. One hour after treatment, animals were perfused with penetration rate was 8.4 % from the total dose of 10 mg/kg/body PBS, when blood was cleaned up from the brain blood vessels, weight used for intranasal administration. (Figure 2, A, the brains were removed FITC fluorescence in brain homogenate Supplementary Figure 8). Importantly, we measured BBB was measured. Our result demonstrate that FITC-G4HisMal has cross of FITC-conjugated dendrimers, due to observed thera- significantly improved ability to cross BBB compared to the peutic effect of G4HisMal we suggest that non-conjugated precursor dendrimer, FITC-G3Mal. The efficiency of FITC- G4HisMal may have higher BBB penetration rate. Of note, if G3Mal to cross BBB is relatively low, as BBB penetration rate of brain perfusion were not sufficient to clean brain vessels from the E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 203 blood completely, then detected fluorescence might come from mice did not demonstrate visible signs of weakness and apathy, FITC-G4HisMal, which partially bypassed BBB. However, we indicating that the compound was well-tolerated. At the end of believe that both type of dendrimers did cross BBB since the treatment, after one week of drug wash-out period, all transport of large molecular weight biologics into the brain along animals were exposed to the two-object recognition test. trigeminal and olfactory nerves as occurs within a minute. 24 Strikingly, mice from APP/PS1 G4HisMal group demonstrated The presence of FITC-G4HisMal in brain tissue was significant memory improvement when compared with the confirmed by immunofluorescent labeling with an antibody control mice, from APP/PS1-PBS group (Figure 3, A). specific to FITC (Figure 2, B). Thus, immunolabeling and direct Surprisingly, Aβ plaque quantification did not show a reduction measuring of FITC fluorescence demonstrated the presence of in plaque load in the neocortex of treated APP/PS1 mice (Figure FITC-G4HisMal in the brain tissue after i.n. administration, 3, B). However, we noticed distinct Aβ plaque morphology in indicating that FITC-G4HisMal dendrimers had crossed the the brain tissue of treated animals when compared to the Aβ BBB. Importantly, modification of G4Mal with histidine plaques seen in the cortex of APP/PS1-PBS mice. To evaluate significantly improved the BBB cross 20 efficiency (Supple- amyloid plaque morphology, total Aβ deposits were labeled with mentary Figure 8). G4HisMal did not accumulate in the brain antibody 6E10; fibrillar Aβ was labeled with antibody OC. 28 tissue after daily repeated i.n. administrations during one week Analysis of the plaque area as a ratio of the fibrillar Aβ (OC and measured by FITC fluorescence in brain homogenate (data positive) against the total Aβ (6E10 positive) demonstrated a not shown). significant elevation of fibrillar content in amyloid plaques after Next, we tested G4HisMal capacity to interact with G4HisMal treatment (Figure 3, C). Thus, G4HisMal modulated Alzheimer's amyloid peptides. As demonstrated by Thioflavin plaque morphology in vivo. Of note, we could not detect direct T (ThT) binding assay and electron transmission micrographs, colocalization of amyloid plaque and G4HisMal in vivo, similar G4HisMal dendrimers did not prevent fibril formation but to our in vitro experiment, since for the treatment we used clumped Aβ(1−40) fibrils together (Figure 2, C, D). Interestingly, untagged G4HisMal. the intensity of ThT was significantly decreased following the Since inflammation and oxidative are strongly linked to aggregation of Aβ(1−40). This observation can be explained that AD, 29,30 we compared inflammation and oxidative stress G4HisMal may compete with ThT for binding to Aβ(1−40) or markers in the brain tissue around the amyloid plaques between G4HisMal may change Aβ(1−40) fibrillar resulting in lower ThT APP/PS1-G4HisMal and APP/PS1-PBS mice. Using GFAP and fluorescence quantum yield. 25,26 Similar results were obtained Iba1, as the markers of glial response, we did not observed for aggregation kinetic of Aβ(1−42) (Supplementary Figure 9). significant differences around the amyloid plaques between Importantly, using electron microscopy, we did not detect any G4HisMal treated and PBS groups (Figure 4, A). Using HNE Aβ oligomers in the presence of G4HisMal. Our data and neuroketal as oxidative stress markers which have been demonstrated that G4HisMal significantly inhibit Aβ(1−40) cell shown to be increased in AD, 31,32 we did not detect significant toxicity for human neuroblastoma cells (Figure 2, E): The differences in oxidative stress markers around amyloid plaques presence of 1 μmol G4HisMal, eliminated the toxicity of 10 (Figure 4, B) nor in brain homogenate (Supplementary Figure μmol Aβ(1−40). Immunofluorescent labeling with a specific 11). Taken together, results, shown in the Figures 3 and 4, antibody against Aβ(1-40), showed co-localization of Aβ(1−40) indicated that more fibrillar amyloid plaques observed in and FITC-G4HisMal (Figure 2, F, Supplementary Figure 10). G4HisMal group induced similar glial response and oxidative Interestingly, as detected by immunofluorescence, monomeric stress effects compared to less fibrillar plaques observed in the Aβ(1−40) was taken up by endocytic uptake, 27 while in the control PBS group. Therefore, we concluded that G4HisMal presence of G4HisMal, Aβ(1−40) was not taken up by the cells. modulated morphology of amyloid plaques was not a direct Therefore, our data confirmed that when added to SHSY-5Y, cause of memory improvement after treatment. G4HisMal did not affect cell viability itself and inhibited Aβ(1 Next, we quantified the levels and the ratio of soluble Aβ42 to −40) induced cell toxicity. Aβ40, using specific enzyme-linked immunosorbent assays (ELISA). Surprisingly, the ratio and levels of soluble Aβ42 to Aβ40 also were not changed after the G4HisMal treatment Long-term G4HisMal treatment protects memory performance (Figure 5, A) indicating that as measured by ELISA, soluble Aβ in APP-PS1 mice without reducing amyloid plaque load were not related to the memory improvement of APP/PS1 mice. To test anti-amyloid properties of G4HisMal in vivo, APP/ One possible explanation could be that for the memory loss in PS1 mice were treated intranasally (i.n.) 5 μg/day of G4HisMal control APP/PS1-PBS animals were responsible soluble Aβ same volume of PBS was used as a control. G4HisMal treatment oligomers which could not be discriminated by ELISA. We was started at the early pre-symptomatic stage when APP/Ps1 hypothesized that G4HisMal could protect neurons form soluble mice were 3 months of age and no memory impairment, nor Aβ oligomers. To test this hypothesis, we treated primary amyloid plaques are normally detected. 19 Administration of neurons derived from wild-type mouse embryos with 1μmol of G4HisMal was repeated three times per week and continued freshly prepared of soluble Aβ(1-42) oligomers in the presence of during three months until the APP/PS1 mice reached the age of 6 G4HisMal. Using a release of lactate dehydrogenase (LDH) months when significant amyloid plaque load, synaptic loss, and assay, we evaluated cell viability of cultured primary neurons in memory dysfunction were manifested. 19 Age-matched wild-type the presence of Aβ(1-42) and G4HisMal. As it was reviled by mice were used as a control. Importantly, during the complete LDH assay, soluble Aβ(1-42) oligomers were the most toxic period of G4HisMal treatment, neither APP/PS1 nor wild-type species for cultured neurons, whereas monomers and fibrils did 204 E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 Figure 3. Long-term G4HisMal treatment protects memory performance in APP-PS1 mice and attenuates the morphology of amyloid plaques. (A) Memory performance in the V-maze shows significant improvement after treatment with G4HisMal. Statistics: two-way ANOVA with genotype and treatment as between factors followed by Tukey’s post hoc test, data are expressed as a mean ± SEM. (B) Representative immunohistochemical images of 6F/3D-positive amyloid depositions. Scale bar is 100 μm. Statistics: one-way ANOVA followed by Tukey’s post hoc test; data are expressed as a mean ± SEM. (C) Representative immunohistochemical images of OC and 6E10-positive amyloid depositions in the cortex. Scale bar is 50 μm. Statistics: Student’s t-test; data are expressed as a mean ± SD. E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 205 Figure 4. G4HisMal treatment does not modulate glial response and oxidative stress around Aβ plaques. (A) Glial response around amyloid plaques was not changed after G4HisMal treatment. Representative images of double immunofluorescence labeling of GFAP (green) and Aβ (red) (left panel) and Iba1 staining (red) and Aβ (green) (right panel) did not reveal a reduction of microglia and astrocytes around Aβ plaques after treatment. Scale bar is 50 μm. (B) Oxidative stress around amyloid plaques was not altered after G4HisMal treatment. Representative images of double immunofluorescence labeling of HNE (red) and Aβ (green) (left panel) and SOD2 staining (red) and Aβ (green) (right panel) did not detect a reduction of oxidative stress around Aβ plaques after treatment. Scale bar is 50 μm. Statistics: one-way ANOVA followed by Tukey’s post hoc test; data are expressed as mean ± SEM. not reduce cell viability during the incubation period. G4HisMal regulated in AD. 33,34 Western blotting with specific antibodies significantly reduced the neurotoxicity of soluble Aβ oligomers against synaptic markers demonstrated that synaptophysin, (Figure 5, B). Therefore, we conclude that G4HisMal was able to Psd95, and drebrin were up-regulated in brain homogenate of interfere with soluble neurotoxic Aβ(1-42) fraction reducing the APP/PS1 mice treated with G4HisMal, compared to the controls, neurotoxicity of soluble Aβ(1-42) oligomers. mice treated with PBS (Figure 6). Importantly, wild-type animals treated with G4HisMal did not show a difference in the level of Preventive G4HisMal treatment protects synapses synaptic markers as compared to wild-type PBS group (Supplementary Figure 11). Thus, G4HisMal treatment protected To understand a possible mechanism behind the memory synapses in AD transgenic mice and was safe for wild-type mice. rescue documented after G4HisMal treatment, we evaluated the To study one of the possible mechanisms of synapse protection, level of pre- and postsynaptic markers using specific antibodies we assayed the soluble fraction of brain homogenate using dot against drebrin, Pds95, and synaptophysin. All those synaptic blot with specific oxidative stress markers. Taken together, our markers play a role in the synaptic plasticity and are down- data demonstrated that first, G4HisMal protected synapses in the 206 E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 Figure 5. G4HisMal reduces the toxicity of Aβ oligomers for cultured primary neurons. (A) Soluble Aβ quantification: Enzyme-linked immunosorbent assay (ELISA), demonstrated that soluble Aβ was not changed in the brain extract of APP/PS1 G4HisMal mice. Statistics: one-way ANOVA followed by Tukey’s post hoc test; data are expressed as mean ± SEM; n = 5 - 7 per group. (B) Left panel: Recombinant 10 μmol Aβ(1−42) was incubated at in PBS at 37°C. ThT fluorescence variation was used to monitor aggregation of Aβ(1−42) (black line), red line corresponds to ThT alone. The arrows indicate the time when aliquots of Aβ(1−42) monomers (Aβ M), oligomers (Aβ O), and fibrils (Aβ F) suspensions were taken for neuronal viability assay. Right panel: Wild-type primary neurons were treated for one hour with 1 μmol Aβ(1−42) monomers (Aβ M), oligomers (Aβ O), and fibrils (Aβ F), G4HisMal significantly reduces cytotoxicity of Aβ O. Statistics: one-way ANOVA followed by Tukey’s post hoc test, data are expressed as a mean ± SD, n = 3 embryos per group. brains of the treated APP-PS1mice; second, synapse protection biocompatibility and the ability to cross BBB of the precursor in presymptomatic phase of AD could be an avenue for AD dendrimer G3Mal. The second objective was to investigate the treatment and third, G4HisMal is a potential drug candidate for, potential of G4HisMal to protect AD transgenic mice from synapse protection in presymptomatic phase of AD. memory impairment. In vivo evaluations demonstrated that histidine maltose shell significantly improved biocompatibility and ability to cross BBB of PPI dendrimers; G4HisMal were able Discussion to interfere with Aβ fibril formation in vitro and in vivo; finally, chronic treatment with G4HisMal protected APP/PS1 mice from Alzheimer’s disease is a fatal neurodegenerative disorder and memory impairment. Since synaptic markers such as Psd95, the most common cause of dementia in the elderly. Genetics synaptophysin, and drebrin were preserved after the G4HisMal strongly supports the hypothesis that aggregated Aβ is one of the treatment in AD transgenic mice, we suggest that the mechanism players triggering degradation of synapses. 5,6 AD presymptom- behind memory protection could be synapse shielding from atic phase may last a decade or even more, and it is during this soluble Aβ neurotoxicity. early phase, before the irreversible changes occur, therapies are The loss of synaptophysin immunoreactivity is the best- most likely to be effective. In this study, we tested if preventive known correlate of cognitive decline in human AD 2 and, in AD treatment with G4HisMal can protect APP/PS1 mice from transgenic models, synaptic markers such as synaptophysin and memory impairment. Previously, maltose modified PPI dendri- Psd-95 are also shown to be reduced. 34 Moreover, Psd95 mers (G3Mal and G4Mal) have been suggested as promising knockout animals have learning defects and impaired basal modulators for the formation of the amyloid fibrillar structures synaptic transmission 37; transgenic animals lacking synaptophy- related to certain proteinopathies diseases such as AD and prion sin have increased exploratory behavior and reduced novel diseases. 12,35 The interaction of G3Mal and G4Mal dendrimers object recognition. 38 with amyloid peptides is mainly tailored by hydrogen bonds Interestingly, it has been shown that loss of synaptophysin depending on the density of hydrogen bond-forming sugar immunoreactivity clearly preceded plaque formation, raising the groups on the dendrimer surface. 36 However, being able to cross possibility that Aβ can induce structural and functional neuronal BBB and to interfere with Aβ, G3Mal does not improve clinical deficits independent of plaque formation. Therefore, a treatment symptoms in vivo. 12 Therefore, we synthesized a novel type of which may protect synapses is one of the avenues to fight against dendrimers: 4 th generation of PPI dendrimers with maltose- memory decline during AD. 3,39 histidine shell, G4HisMal. In this study, we explored in vitro and Here, three months of treatment with G4HisMal was able to in vivo the effects of G4HisMal. Our first objective was to assess prevent a decrease in synaptophysin and Psd95 synaptic markers whether histidine on the dendrimer shell can improve the compared to PBS-treated APP/PS1 mice. G4HisMal did not E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 207 Figure 6. G4HisMal treatment protects synapses in APP/PS1 mice. Representative dot and Western blots of Psd95, drebrin, and synaptophysin. β-actin was used for protein normalization Statistics: Student’s t-test. (n = 4-6 animals per group, dot and western blotting was done in triplicate). Data are expressed as a mean ± SD. Dot Blots and Western blots of Psd95, drebrin, synaptophysin for wild-type animals are shown in the Supplementary Figure 12. change the level of Psd95, Synaptophysin, Drebrin in wild-type that during chronic administration (during three months) of mice indicating that most likely the levels of mRNA expression G4HisMal was able to cross BBB and G4HisMal treatment of synaptic markers were not changed. Therefore, the increased rescued spatial memory deficits in APP/PS1 mice possibly via levels of Psd95, Synaptophysin, Drebrin in treated APP/PS1 shielding of synapses against soluble Aβ oligomers. Thus, mice compared to the control group are more likely an effect of G4HisMal is an effective and safe agent suitable for treatment of reduced degradation of synapses. Importantly, our results were the central nervous system. reproducible, in two independent experiments we got similar results demonstrating memory protection after G4HisMal treatment. Acknowledgments However, since the analysis of mRNA expression of synaptic markers was not possible in this study due to brain homogenate We thank the Saxon Ministry of Science and Art (SMWK), preparation, future experiments are required to determine the German Federal Ministry for Education and Science thoroughly what mechanism(s) is/are behind the protection of (BMBF), COST-Action program TD0802 (Biodendrimers in synapses by G4HisMal. Also, the interaction of G4HisMal with biomedical applications), Swedish Research Council 2017- soluble Aβ pool should be further assessed in vivo including the 01539 and MultiPark (Medical Faculty, Lund University). response of neurons, astrocytes, and microglia. To conclude, we demonstrated that histidine-maltose shell significantly improves biocompatibility of PPI dendrimers and Appendix A. Supplementary data their ability to cross BBB. Therefore, we suggest that maltose- histidine shell may be used to improve biocompatibility and Supplementary data to this article can be found online at ability to cross BBB of other types of dendrimers. We proved https://doi.org/10.1016/j.nano.2019.01.010. 208 E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 References 20. Willén K, Edgar JR, Hasegawa T, Tanaka N, Futter CE, Gouras GK. Aβ accumulation causes MVB enlargement and is modelled by dominant 1. Ferrer I. Defining Alzheimer as a common age-related neurodegener- negative VPS4A. Mol Neurodegener 2017;12, https://doi.org/10.1186/ ative process not inevitably leading to dementia. Prog Neurobiol s13024-017-0203-y. 2012;97:38-51. 21. Boas U, Heegaard PMH. Dendrimers in drug research. Chem Soc Rev 2. Terry RD, Masliah E, Salmon DP, Butters N, DeTeresa R, Hill R, et al. 2004;33:43. Physical basis of cognitive alterations in Alzheimer’s disease: Synapse 22. Fischer M, Appelhans D, Schwarz S, Klajnert B, Bryszewska M, Voit B, loss is the major correlate of cognitive impairment. Ann Neurol et al. Influence of surface functionality of poly(propylene imine) 1991;30:572-80. dendrimers on protease resistance and propagation of the scrapie prion 3. Mucke L, Masliah E, Yu GQ, Mallory M, Rockenstein EM, Tatsuno G, protein. Biomacromolecules 2010;11:1314-25. et al. High-level neuronal expression of abeta 1-42 in wild-type human 23. Liu Q, Shen Y, Chen J, Gao X, Feng C, Wang L, et al. Nose-to-brain amyloid protein precursor transgenic mice: synaptotoxicity without transport pathways of wheat germ agglutinin conjugated PEG-PLA plaque formation. J Neurosci 2000;20:4050-8. nanoparticles. Pharm Res 2012;29:546-58. 4. Masliah E. Recent advances in the understanding of the role of synaptic 24. Lochhead JJ, Thorne RG. Intranasal delivery of biologics to the central proteins in Alzheimer’s Disease and other neurodegenerative disorders. nervous system. Adv Drug Deliv Rev 2012;64:614-28. J Alzheimers Dis 2001;3:121-9. 25. Jimenez JL, Nettleton EJ, Bouchard M, Robinson CV, Dobson CM, 5. Tu S, Okamoto S, Lipton SA, Xu H. Oligomeric Aβ-induced synaptic Saibil HR. The protofilament structure of insulin amyloid fibrils. Proc dysfunction in Alzheimer’s disease. Mol Neurodegener 2014;9:48. Natl Acad Sci 2002;99:9196-201. 6. Klementieva O, Willén K, Martinsson I, Israelsson B, Engdahl A, 26. Sulatskaya AI, Kuznetsova IM, Turoverov KK. Interaction of Thioflavin Cladera J, et al. Pre-plaque conformational changes in Alzheimer’s T with amyloid fibrils: fluorescence quantum yield of bound dye. J Phys disease-linked Aβ and APP. Nat Commun 2017;8:14726. Chem B 2012;116:2538-44. 7. Madaan K, Poonia N, Kumar S, Lather V. Dendrimers in drug delivery 27. Wesén E, Jeffries GDM, Matson Dzebo M, Esbjörner EK. Endocytic and targeting: Drug-dendrimer interactions and toxicity issues. J Pharm uptake of monomeric amyloid-β peptides is clathrin- and dynamin- Bioallied Sci 2014;6:139. independent and results in selective accumulation of Aβ(1–42) 8. Svenson S, Tomalia DA. Dendrimers in biomedical applications— compared to Aβ(1–40). Sci Rep 2017;7, https://doi.org/10.1038/ reflections on the field. Adv Drug Deliv Rev 2012;64:102-15. s41598-017-02227-9. 9. Kannan RM, Nance E, Kannan S, Tomalia DA. Emerging concepts in 28. Kayed R, Head E, Sarsoza F, Saing T, Cotman CW, Necula M, et al. dendrimer-based nanomedicine: from design principles to clinical Fibril specific, conformation dependent antibodies recognize a applications. J Intern Med 2014;276:579-617. generic epitope common to amyloid fibrils and fibrillar oligomers 10. Huang R-Q, Qu Y-H, Ke W-L, Zhu J-H, Pei Y-Y, Jiang C. Efficient gene that is absent in prefibrillar oligomers. Mol Neurodegener delivery targeted to the brain using a transferrin-conjugated polyethy- 2007;2:18. leneglycol-modified polyamidoamine dendrimer. FASEB J 29. Birch AM, Katsouri L, Sastre M. Modulation of inflammation in 2007;21:1117-25. transgenic models of Alzheimer’s disease. J Neuroinflammation 11. Sharma R, Kim S-Y, Sharma A, Zhang Z, Kambhampati SP, Kannan S, 2014;11:25. et al. Activated microglia targeting dendrimer–minocycline conjugate as 30. Cheignon C, Tomas M, Bonnefont-Rousselot D, Faller P, Hureau C, therapeutics for neuroinflammation. Bioconjug Chem 2017;28:2874-86. Collin F. Oxidative stress and the amyloid beta peptide in 12. Klementieva O, Aso E, Filippini D, Benseny-Cases N, Carmona M, Alzheimer’s disease. Redox Biol 2018;14:450-64. Juvés S, et al. Effect of poly(propylene imine) glycodendrimers on β- 31. Dalleau S, Baradat M, Guéraud F, Huc L. Cell death and diseases amyloid aggregation in vitro and in APP/PS1 transgenic mice, as a model related to oxidative stress:4-hydroxynonenal (HNE) in the balance. of brain amyloid deposition and Alzheimer’s disease. Biomacromole- Cell Death Differ 2013;20:1615-30. cules 2013;14:3570-80. 32. Gonzalo H, Brieva L, Tatzber F, Jové M, Cacabelos D, Cassanyé A, 13. Yamakami J, Sakurai E, Sakurada T, Maeda K, Hikichi N. Stereo- et al. Lipidome analysis in multiple sclerosis reveals protein selective blood-brain barrier transport of histidine in rats. Brain Res lipoxidative damage as a potential pathogenic mechanism. J 1998;812:105-12. Neurochem 2012;123:622-34. 14. Oshima T, Oshima C, Baba Y. Selective extraction of histidine 33. Shim KS, Lubec G. Drebrin, a dendritic spine protein, is manifold derivatives by metal affinity with a copper(II)–chelating ligand complex decreased in brains of patients with Alzheimer’s disease and Down in an aqueous two-phase system. J Chromatogr B 2015;990:73-9. syndrome. Neurosci Lett 2002;324:209-12. 15. Wang H, Wang M, Wang B, Li M, Chen H, Yu X, et al. The distribution 34. Poirel O, Mella S, Videau C, Ramet L, Davoli MA, Herzog E, et al. profile and oxidation states of biometals in APP transgenic mouse brain: Moderate decline in select synaptic markers in the prefrontal cortex dyshomeostasis with age and as a function of the development of (BA9) of patients with Alzheimer’s disease at various cognitive Alzheimer’s disease. Metallomics 2012;4:289. stages. Sci Rep 2018;8, https://doi.org/10.1038/s41598-018-19154- 16. Tang S-C, Arumugam TV, Cutler RG, Jo D-G, Magnus T, Chan SL, et y. al. Neuroprotective actions of a histidine analogue in models of ischemic 35. McCarthy JM, Franke M, Resenberger UK, Waldron S, Simpson JC, stroke: Histidine analogue in models of ischemic stroke. J Neurochem Tatzelt J, et al. Anti-Prion drug mPPIg5 inhibits PrPC conversion to 2007;101:729-36. PrPSc. PLoS One 2013;8e55282. 17. Mark JE, editor. Polymer data handbook. 2nd ed. Oxford; New York: 36. Klementieva O, Benseny-Cases N, Gella A, Appelhans D, Voit B, Oxford University Press; 2009. Cladera J. Dense shell glycodendrimers as potential montoxic anti- 18. Janaszewska A, Ziemba B, Ciepluch K, Appelhans D, Voit B, Klajnert amyloidogenic agents in Alzheimer’s disease. Amyloid–dendrimer B, et al. The biodistribution of maltotriose modified poly(propylene aggregates morphology and cell toxicity. Biomacromolecules imine) (PPI) dendrimers conjugated with fluorescein—proofs of 2011;12:3903-9. crossing blood–brain–barrier. New J Chem 2012;36:350-3. 37. Nagura H, Ishikawa Y, Kobayashi K, Takao K, Tanaka T, 19. Aso E, Lomoio S, López-González I, Joda L, Carmona M, Fernández- Nishikawa K, et al. Impaired synaptic clustering of postsynaptic Yagüe N, et al. Amyloid generation and dysfunctional immunoprotea- density proteins and altered signal transmission in hippocampal some activation with disease progression in animal model of familial neurons, and disrupted learning behavior in PDZ1 and PDZ2 Alzheimer’s disease: amyloid generation and UPS in FAD mice. Brain ligand binding-deficient PSD-95 knockin mice. Mol Brain Pathol 2012;22:636-53. 2012;5:43. E. Aso et al / Nanomedicine: Nanotechnology, Biology, and Medicine 17 (2019) 198–209 209 38. Schmitt U, Tanimoto N, Seeliger M, Schaeffel F, Leube RE. 40. Petoukhov MV, Franke D, Shkumatov AV, Tria G, Kikhney AG, Detection of behavioral alterations and learning deficits in mice Gajda M, et al. New developments in the ATSAS program package lacking synaptophysin. Neuroscience 2009;162:234-43. for small-angle scattering data analysis. J Appl Crystallogr 39. Tampellini D, Capetillo-Zarate E, Dumont M, Huang Z, Yu F, Lin 2012;45:342-50. MT, et al. Effects of synaptic modulation on amyloid, synapto- 41. Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng physin, and memory performance in Alzheimer’s disease trans- EC, et al. UCSF Chimera? A visualization system for exploratory research genic mice. J Neurosci 2010;30:14299-304. and analysis. J Comput Chem 2004;25:1605-12.