Hyaluronan-carnosine conjugates inhibit Aβ aggregation and toxicity

Alzheimer’s disease is the most common neurodegenerative disorder. Finding a pharmacological approach that cures and/or prevents the onset of this devastating disease represents an important challenge for researchers. According to the amyloid cascade hypothesis, increases in extracellular amyloid-β (Aβ) levels give rise to different aggregated species, such as protofibrils, fibrils and oligomers, with oligomers being the more toxic species for cells. Many efforts have recently been focused on multi-target ligands to address the multiple events that occur concurrently with toxic aggregation at the onset of the disease. Moreover, investigating the effect of endogenous compounds or a combination thereof is a promising approach to prevent the side effects of entirely synthetic drugs. In this work, we report the synthesis, structural characterization and Aβ antiaggregant ability of new derivatives of hyaluronic acid (Hy, 200 and 700 kDa) functionalized with carnosine (Car), a multi-functional natural dipeptide. The bioactive substances (HyCar) inhibit the formation of amyloid-type aggregates of Aβ42 more than the parent compounds; this effect is proportional to Car loading. Furthermore, the HyCar derivatives are able to dissolve the amyloid fibrils and to reduce Aβ-induced toxicity in vitro. The enzymatic degradation of Aβ is also affected by the interaction with HyCar.


Results
Synthesis of Hy conjugates with car. Several methods of conjugating bioactive substances to hyaluronic acid have been reported in the literature 58 . We did not use any spacer linker to increase the resistance of the conjugate to carnosinase activity 60 ; therefore, we chose to link hyaluronic acid to Car through an amide bond between the carboxylic group of the d-glucuronic unit of the polysaccharide and the N-terminal amino group of Car. Accordingly, hyaluronic acid-Car conjugates were prepared following a synthetic route that involves the activation of the Hy-linked carboxylate groups and their subsequent conversion into amide bridges by reaction with the amino group of carboxyl-protected Car. The final products differ in terms of MW of the starting Hy (200 or 700 kDa) and Car loading (7, 10, 14, 35%), i.e., the percentage of D-glucuronic units drafted by the Car residues.
Subsequently, NMR analysis of the HyCar conjugates was performed to ascertain the chemical conjugation and to calculate the Car loading percentage of each new HyCar derivative. Finally, the medium molecular weight (MMW) and the intrinsic viscosity (I.V.) (Supplementary Table S1) of each conjugate were determined, and the results were in accordance with those obtained from the NMR analysis 59 . The greater the loading percentage of Car was, the higher the MMW, as expected. Moreover, Car conjugation clearly affects the final viscosity: all the HyCar derivatives showed an I.V. value lower than that of the parent polymer compounds. This difference clearly increased as the MMW increased after Car derivatization.
The structural characterization also confirmed that no fragmentation of the polysaccharide skeleton occurred as a consequence of the synthesis process. carnosinase resistance. Carnosinase (CN1) is a dipeptidase that is present in blood serum and degrades histidine-containing dipeptides such as Car 60 . Therefore, to assess the stability of the HyCar conjugates towards the enzyme carnosinase, the following experiments were carried out by using HyCar(200)7 as a model substrate. The time-dependent stability of this conjugate towards CN1, compared with that of Car and a mixture of Car plus Hy, was determined by incubating each of these substances (or mixtures) with the enzyme. The histidine released during each incubation period was determined by a fluorometric assay 60 . The stability of HyCar towards the action of carnosinase was approximately nine times greater than that of Car used either by itself or in a mixture with hyaluronic acid. The results of this test are shown in Fig. 3.
HyCar (200)7 is the smaller HyCar derivative tested in this work. To verify whether the MW of the hyaluronan moiety of the HyCar conjugates (or their Car loading) could have any effect on carnosinase resistance, we also tested the CN1-mediated degradation of the larger HyCar derivative (HyCar(700)35). The results ( Figure S1) clearly show that HyCar(700)35 is also resistant to enzymatic hydrolysis of the Car moiety.
Antiaggregation activity. The amyloid-type aggregation of Aβ starts with the formation of soluble and toxic oligomers; their dimensions increase during the process, and fibrillary and insoluble structures are formed at the end of this pathological pathway. Reducing or preventing the aggregation process of Aβ is one of the main challenges associated with attenuating or preventing the onset of AD. Therefore, all HyCar derivatives were tested for the in vitro self-induced aggregation of Aβ 42 by using thioflavin T (ThT), a fibril-sensitive dye.
As a first step, the extent of dose-dependent aggregation of the HyCar(200) derivatives and their parent compounds was monitored at the end of the incubation period (40 h). As reported in Fig. 4, the maximum ThT fluorescence variation due to amyloid aggregation was significantly reduced in the presence of Hy(200). In contrast, Car alone did not show any significant effect on the formation of ThT-sensitive species at the lower concentrations used (0.5 and 1 µM). However, both the underivatized Hy(200) polymer and the dipeptide (Car) were clearly able to partially inhibit the total amyloid species formation in a dose-dependent manner. The concentrations of Car were equivalent to the level of Car units in the samples containing HyCar(200)14. www.nature.com/scientificreports/ HyCar(200) derivatives inhibit the formation of amyloid species, and the decrease in fluorescence with respect to that of Aβ alone (CTRL sample) is directly proportional to their concentration. Interestingly, the higher the Car content covalently linked to the Hy(200) scaffold was, the better the inhibition activity. As the main result, HyCar(200)14 showed the best performance among the HyCar(200) derivatives in terms of reducing the amyloidtype aggregation of Aβ 42, and the higher concentrations of this compound yielded a final fluorescence increment as low as 10% compared to that shown by Aβ 42 alone.
Then, kinetic measurements of amyloid aggregation were carried out in order further study the antiaggregant capabilities of the HyCar(200) derivatives. The maximum fluorescence gain (F max -F 0 ) is proportional to the aggregation extent of Aβ, whereas the lag time (t lag ) value represents the starting point of the amyloid-type aggregation of Aβ. As a consequence, the lower F max -F 0 is and the higher t lag is, the better the antiaggregant activity.
When ThT is incubated with Aβ 42 , the amyloid-type aggregation mechanism results in a sigmoid-shaped fluorescence response (Fig. 5a). The lag phase lasts 5.9 h (Supplementary Table S2), and the maximum fluorescence gain (F max − F 0 ) reaches a value of 51.2.
For the treatment with Car, the kinetic parameters of the self-induced amyloid aggregation do not significantly change. When Hy(200) or a mixture of Hy(200) and Car was co-incubated with Aβ, the corresponding  The amyloid aggregation assay was also used to test the Car derivative with Hy(700) (HyCar (700)35). The molar concentration range used for the Hy(700) derivatives (0.4 µM) was lower than that tested for activity evaluation of the Hy(200) derivatives (10 µM) because the former compounds are less soluble than the latter ones. Compared to the kinetic trend for amyloid aggregation shown by Aβ 42 (Fig. 5b, CTRL sample), neither Hy(700) and Car (tested individually) nor the Hy(700) + Car mixture affected the lag phase of the process, although the presence of the latter slightly decreased the F max − F 0 value, while Hy(700) did not alter Aβ 42 fibril formation (Supplementary Table S3).
On the other hand, the F max − F 0 value decreased to 4.4 and the lag phase was lengthened to 24.5 h when HyCar(700)35 was tested.
The excellent antiaggregation effect of HyCar(700)35 on amyloid formation was evaluated for dose dependence (Supplementary Figure S2, Table 1). Both the F max − F 0 and t lag values progressively changed as a function of the HyCar(700) derivative amount, thus confirming the effect that Car derivatization of Hy(700) has on the antiaggregant activity.
Overall, HyCar(200)14 and HyCar(700) 35 show a prominent effect in delaying the formation and reducing the amount of amyloid-type aggregated forms of Aβ 42 .
We also evaluated the activity of Hy(200), Hy(700) and different Car conjugates on the dissolution of preformed Aβ 42 fibrils (Supplementary Figure S3). After treatment with HyCar(200)14, ThT fluorescence was reduced by only 20%, whereas twice that value (40%) was observed when the amyloid fibrils were treated with HyCar(700) 35.
The ability to dissolve the ThT-sensitive amyloid forms was also shown by Hy(700) and by a mixture of Hy(700) and Car, though the effect was significantly weaker than that exhibited by HyCar(700) 35. This finding indicates that the covalent conjugation between Hy(700) and Car clearly results in a gain-of-function effect of both the Aβ antiaggregation and disaggregation activities.
Enzymatic hydrolysis of Aβ. The non-covalent interactions between the HyCar derivatives and the amyloid peptide could also have an effect on the clearance of Aβ. Insulin-degrading enzyme (IDE) 61 is a zinc metalloprotease that catalyses the hydrolysis of several polypeptides, such as insulin, IAPP, bradykinin, glucagon and Aβ.
For this reason, the IDE-mediated hydrolysis of Aβ 28 was also assayed in the presence HyCar(700)35, the HyCar derivative with the highest antiaggregant activity among all the newly synthesized HyCar conjugates. Aβ 28 was chosen as a model peptide for Aβ 42 because it not only encompasses the N-terminus and CHD regions important for the oligomerization process but also shows an aggregation rate lower than that of Aβ 42 ; these features make Aβ 28 a suitable substrate for the time scale of the enzymatic assay.
The main purpose of this assay was to both further prove the HyCar-Aβ interaction and obtain indirect information about the portions of the Aβ sequence involved in the non-covalent interaction with the HyCar derivatives.
As reported in Supplementary Figure S4, Aβ alone (CTRL) was hydrolysed by IDE, mainly forming 1-13, 1-19, 1-15 and 5-17 hydrolytic peptides. Their normalized intensity changed during the reaction, and after 40 min, the hydrolytic pattern was very different from that detected a few min after the reaction began. In particular, 1-13 was the main fragment at the beginning of the reaction, whereas the intensity of 1-19 exceeded that of all the other hydrolytic peptides as early as 10 min after the reaction started. IDE-mediated digestion was also monitored in the presence of HyCar(700)35 (0.1 and 0.4 µM) or the parent compounds (Hy and Car), alone or in a mixture (Hy + Car). Hy and Car do not change the hydrolytic pattern of the amyloid substrate. Both the HyCar derivative and the mixture significantly affect the kinetic trend of the hydrolytic pattern. To better describe this effect, the ratio between the intensity of the main Aβ fragments (1-19 and 1-13) is reported in Fig. 6, showing the time-dependent effect of IDE-mediated hydrolysis on the different Aβ domains.
The 1-19/1-13 ratio rapidly increased when Aβ was incubated with IDE (CTRL). The same trend was observed when Hy or Car was co-incubated with the substrate. The presence of HyCar(700)35 at the lowest dose tested (0.1 µM) significantly slowed the formation and hydrolysis of 1-19 and 1-13, respectively, thus reducing the ratio 1-19/1-13. This effect was more prominent when the concentration of HyCar was higher (0.4 µM). This dose-dependent effect confirms once more the non-covalent interaction between HyCar(700)35 and the amyloid  Figure S5) were analysed by using Atomic Force Microscopy (AFM). Figure 7 shows the AFM height micrographs of Aβ assemblies for freshly dissolved and 24 h-incubated solutions in MOPS buffer at 37 °C, either in the absence or presence of the different Hy derivatives. A mixed population of small oligomers was found for freshly dissolved Aβ 42 samples (Fig. 7a). The formation of oligomer species could reasonably be attributed to the interaction of the monomer forms with the mica surface, as previously reported 62 . After 24 h of incubation, bundled assemblies of fibrils were visible for Aβ 42 alone (Fig. 7b) and Aβ incubated in the presence of Hy(200) (Fig. 7c) or Hy(700) (Fig. 7d). In contrast, an evident fibril antiaggregant effect was detected for Aβ 42 co-incubated with HyCar conjugates, since Aβ 42 + HyCar(200)14 samples exhibited only oligomers (Fig. 7e), while few coiled fibrils and oligomers were found for Aβ 42 + HyCar(700)35 (Fig. 7f).
AFM analyses were also performed to test the capability of the HyCar compounds to dissolve the preformed Aβ fibrils (Supplementary Figure S6). Indeed, for Aβ 42

Effect of hyaluronic acids and their Car conjugates on Aβ 42 cytotoxicity. It has been previously
shown that oligomeric aggregates of Aβ exhibit profound neurotoxic effects and the ability to form insoluble amyloid deposits in the brain, particularly in the cerebrum and hippocampus 63 . The neurotoxic effects of Aβ 42 oligomers with both in vitro and in vivo models result in the cognitive deficiencies prevalent in the pathogenesis of AD 64 . Thus, to assess the cytotoxicity of the oligomeric species, the effect of Aβ 42 at different levels of oligomerization on the viability of undifferentiated SH-SY5Y cells was investigated via the MTT assay. Cells were treated for 24 h with the different compounds in the presence of Aβ 42 that was pre-incubated for 6, 24 or 48 h (see "Material and methods"). Consistent with previously reported results 65 , we found a decrease in SH-SY5Y cell viability after treatment with 5 μM pre-incubated Aβ for 6 h and 24 h. According to the data obtained from the above ThT assays, showing amyloid aggregate formation after 6 and 24 h of pre-treatment (Supplementary Figure S5), the MTT assay revealed that the presence of exogenous Aβ 42 oligomers decreased cell viability by 15 ± 1% and 34.3 ± 0.7%, respectively. Notably, freshly dissolved or pre-incubated Aβ 42 (for 48 h) did not have a significant effect on neuroblastoma cell viability. This suggests the existence of a different Aβ 42 status during the 48 h of pre-incubation. How HyCar(200)14 and HyCar(700)35 affect Aβ 42 oligomer-induced neuronal toxicity was examined. The results (Fig. 8) show that Hy(200) and Hy(700) as well as their Car conjugates significantly increase cell viability. This effect could be ascribed to a decrease in the toxic oligomer species that directly reduces the Aβ-induced cell toxicity.

Discussion
The potential role of Car in brain-related disorders has recently been reviewed 66 , suggesting that reduced plasma Car levels 67 characterize patients with probable AD, while the carnosinase level has been proposed as a CSF biomarker for early stages of AD 68 . To avoid or delay the carnosinase degradation effect on Car, different derivatives of the dipeptide have been synthesized by conjugating it with derivatives of vitamin E or small carbohydrates [69][70][71] .
Our findings also show that two Hy chains with different molecular weights (200 and 700 kDa) are able to protect Car from the degradative action of carnosinase when the dipeptide is covalently drafted on the polymer www.nature.com/scientificreports/ to form HyCar conjugates. In contrast, the mixture of the parent compounds does not prevent or delay the enzymatic degradation of Car. Polymeric carbohydrates are among the most abundant components of the extracellular matrix, such as the GAG chains in proteoglycans (PGs) 72 . They are associated with different amyloid deposits [73][74][75][76] and are involved in the formation of fibrils. Sulphated GAGs stabilize mature fibrils against dissociation 77 and prevent their proteolytic degradation 78 ; at the same time, sulphated GAGs attenuate the neurotoxic effect of Aβ in cellular assays 79 . AD therapeutic approaches are usually designed to prevent Aβ fibrillogenesis through reduction of Aβ production, inhibition of fibril formation, or removal of fibril aggregates 80 .
It has been suggested that GAG-mediated neuroprotection occurs due to the sequestration of Aβ and increased fibril formation 81,82 , consistent with the hypothesis that the generation of senile plaques in AD characterized by fibrils is a protective response aimed at reducing soluble Aβ neurotoxicity 79 . The promotion of Aβ fibril formation has been attributed to the negative charges of sulphated GAGs 76 and to other GAGs, including Hy, bearing carboxyl groups instead of sulphate groups 82 . Conversely, some findings indicate that no fibril formation occurs in the presence of Hy 21 .
PGs and Aβs colocalize in senile plaques in the AD brain. MD simulation has shown that the GAGs interact by means of their negative charges with the Aβ domain, characterized by the presence of a basic residue cluster (HHQK) and a contiguous tyrosine, while chitosan, with its positive charges, interacts with the Aβ regions rich in Asp and Glu residues 82 . www.nature.com/scientificreports/ It has been suggested that low-molecular-weight heparins can perturb the interactions between PGs and Aβs, blocking or preventing the process of amyloidogenesis 21 . In on-pathway fibrillation, Aβ monomers give rise to fibrils through many metastable steps of oligomeric species formation characterized by β-rich structures 83,84 .
Among natural inhibitors of Aβ amyloid formation, polyphenols such as epigallocatechin-3-gallate (EGCG) and curcumin have been extensively studied 22,85,86 . EGCG could interact with different residues of Aβ 42 (Phe4, Arg5, Phe19, Phe20, Glu22, Lys28, Gly29, Leu34-Gly37, and Ile41) and form hydrogen bonds with Aβ 85 , while curcumin was able to bind to the 12 th and 17 th to 21 st residues of Aβ 42 86 . Mono-curcumine, mono-EGCG and the dual-EGCG-curcumine conjugated with Hy (MW = 1 × 10 6 kDa) have been synthesized 87,88 . The mono-conjugated derivatives 87 inhibited Aβ fibril formation more strongly than free polyphenols, while Hy modified with the two polyphenols was more efficient than the single conjugates with EGCG and curcumine 88 . All the modified polyphenols decreased the amyloid cytotoxicity more efficiently when conjugated with Hy. The formation of Hy conjugate nanogels induced an isolation effect, hindering the interactions between Aβ molecules, while hydrophobic interactions of Aβ with EGCG or curcumin and electrostatic repulsion between the bound Aβ and like-charged Hy altered the Aβ monomeric species structures, giving rise to off-pathway aggregation. These two effects were correlated with the conjugated nanostructures that changed with the loading percentage of Hy.
In this context, our results provide evidence that Hy(200) inhibits Aβ 42 aggregation in a dose-dependent manner (Fig. 5) and protects cells from the toxicity induced by amyloid oligomeric species. Hy(700) induces cell viability in the presence of amyloid species, while it is a modest antiaggregation agent, probably due to the lower concentrations employed (nanomolar range) in the ThT assay than those used for Hy(200) (micromolar range). Furthermore, both the low-and high-molecular-weight hyaluronan chains are able to dissolve fibril formation (Supplementary Figure S3). These protective effects of Hy(200) and Hy(700) found in our experiments are in contrast with those reported in recent years, where the hyaluronic acids did not affect Aβ aggregation and toxicity. The high MW of Hy (1 × 10 6 Da) employed to examine the Aβ antiaggregation effect of polyphenol conjugates with hyaluronic acids and the related difficulties in handling the nanogels could explain the different findings 87,88 . Moreover, the use of two different sizes of Hy should have an important impact on their potential applications because the biological function of Hy and its derivatives also depends on the MW 89 .
All the HyCar derivatives that we synthesized have a loading percentage of Car higher than the highest substitution degree obtained for the synthesis of other antiaggregant Hy conjugates 88 .
HyCar clearly inhibits the Aβ 42 aggregation process; this effect increases proportionally with Car loading, influencing both the lag phase and the final growth. Moreover, the HyCar conjugates are more efficient than the parent compounds (individually or as a mixture) in inhibiting Aβ 42 aggregation, suggesting the existence of a synergistic effect between the polymer and the dipeptide. A prominent role is played by Car, whose antiaggregant properties observed in vitro 37,41 have been attributed to the interaction with the CHD domain 14,15 and turn region 16,17 . Furthermore, molecular docking 41 , MD simulation, and NMR results 37 indicated perturbation of the non-covalent interactions between the Aβ side chains of amino acids involved in the intermolecular salt bridge between two Aβ monomers 90 . Moreover, the electrostatic effect has already been proposed to justify the antiaggregant activity of chitosan-hyaluronan nanoparticles towards self-induced amyloid aggregation 91 . Additionally, in the case of the HyCar derivatives, the increase in the number of positively charged sites (imidazole groups) and the decrease in the number of negatively charged carboxylic groups with respect to the unconjugated www.nature.com/scientificreports/ polymer could account for the antiaggregant activity of the HyCar derivatives being higher than that reported for Hy. The capacity of the Hy scaffold to segregate the Aβ molecules should also be an important aspect that makes the polymer, conjugated or not, more prone than Car to interfere with the on-pathway aggregation of Aβ 88 . Increasing the loading percentage of Car on the Hy scaffold could represent a promising strategy to further improve the antiaggregant properties of the HyCar derivatives, as well the protection against the cytotoxic effects of the amyloid oligomer species. The His13-Lys16 region, encompassing two histidine residues of the Aβ domain, has been suggested to be involved in the molecular interaction responsible for the inhibitory effect of negatively charged GAG 82 . The same Aβ region is involved in the physiological degradation of Aβ catalysed by IDE 61 . Both IDE and Aβ show conformational transitions during the proteolytic process. An Aβ region switches from an α-helix to a β-strand structure to fit the IDE catalytic site. The same conformational shift accounts for the formation of oligomers and aggregates, both in vitro and in vivo, starting from amyloidogenic substrates. This clearly highlights the potential role of IDE in preventing the aggregation process that originates from the abnormal misfolding of amyloidogenic proteins.
The IDE-mediated hydrolysis of Aβ results in a hydrolytic pattern characterized by a sequential mechanism. In the early stage of hydrolysis, a conserved exosite domain of IDE binds the N-terminal region of Aβ. Therefore, the primary C-truncated Aβ fragments contain no less than 9-10 amino acid residues, and the main targeted regions are those encompassing the hydrophobic Phe residues (Phe19-Phe20), as well as the vicinal His residues (His13-His14). Longer incubation times promote proteolysis on secondary cleavage sites of Aβ, mainly moving close to the primary cleavage sites 46 . For this reason, monitoring the formation of the hydrolytic peptides Aβ 1-13 and Aβ 1-19 represents a new and important approach for testing the effects of Hy, Car and HyCar derivatives in the early stage of IDE-mediated Aβ hydrolysis.
The outcome of the IDE-dependent proteolysis assay is further proof of the interaction between the HyCar derivatives and the amyloid peptide. The hydrolytic pattern of amyloid degradation is not markedly modified by HyCar or by the mixture of the parent compounds, thus supporting the non-covalent type of HyCar-Aβ interaction. Moreover, when Car is covalently linked to Hy, the resulting effect on the kinetic trend of the Aβ hydrolytic pattern is significantly different from that exhibited by the Car + Hy mixture. This clearly shows that the chemical conjugation of Car to Hy has a gain-of-function effect on the interaction with Aβ. The synergistic effect between the polymer and the dipeptide, which is associated with the antiaggregant properties of HyCar, also has a clear impact on the Aβ hydrolysis catalysed by IDE. The observed dose-dependent effects of the HyCar conjugate could also be due, at least in part, to the direct effect of viscosity changes that could in turn influence the enzymatic hydrolysis and on-pathway aggregation process of the amyloid peptide.
From AFM analyses, both height and amplitude micrographs of Aβ grown in the presence of the Hy conjugate at low molecular weight demonstrate the capability of the hyaluronic polymer chains to interfere with amyloid fibril formation, thus hindering their aggregation and/or disentangling some previously associated fibrils.

concluding remarks
Finding pharmacological approaches that cure or, better, prevent the onset of AD, the most common neurodegenerative disorder, is an important challenge for researchers. Many efforts have recently been devoted to developing multi-target ligands that could counteract the multiple facets of this devastating disorder, such as toxic aggregation, oxidative stress and metal dyshomeostasis. Moreover, investigating the effect of endogenous compounds or a combination thereof is a promising approach to prevent the side effects of entirely synthetic drugs.
The new HyCar derivatives combine the beneficial properties of both hyaluronic acid and Car. The synergistic effect of the parent compounds confers antiaggregant activity towards the amyloid-type aggregation of Aβ. IDE-dependent Aβ degradation, as well as the cytotoxic effect of the oligomeric species, is also affected by the non-covalent interaction with HyCar. This encouraging outcome is also favoured by the resistance of the Car moiety to the action of carnosinase.
Overall, new compounds were obtained by combining the well-known protective features of Car with the ability of Hy to recognize its CD44 receptor, present at high levels in AD brains 92,93 . Therefore, we can speculate that this new class of Aβ aggregation antagonists can pave the way to a possible translational scenario of new HyCar compounds.
Anhydrous methanol was kept for 24 h on 4 Å molecular sieves before use. Carnosine methyl-ester (Car-OMe) was synthesized in anhydrous methanol, as previously reported 94 .
All the fluorescence and UV-Vis spectra were reordered by using a multimode microplate reader (Varioskan Flash, Thermo Scientific). www.nature.com/scientificreports/ N-(3-dimethylaminopropyl)-N'-ethyl-carbodiimide hydrochloride (EDC*HCl; 1.3 mmol) in 10 ml of water was added to the mixture. After 20 h at 5 °C under stirring, the reaction mixture was treated with 0.1 N NaOH (100 ml) for 3.5 h at 5 °C. The pH was then adjusted to 7.0 with 2 N HCl, and the conjugate was precipitated by the addition of acetone (800 ml). After centrifugation, the product was dissolved in water and dialyzed for 60 h. The conjugate was then recovered and lyophilized (yield 75%). 1  The amount of Car linked to hyaluronic acid was determined from the ratio between the integration value of the signal at 2.06 ppm (related to the acetyl groups of Hy) and that of the H-2 or H-5 signal of the imidazole ring of histidine (at 8.66 and 7.34 ppm, respectively).

Synthesis of the
For the synthesis of HyCar(200)7,10,14, the procedure was not different from that adopted to synthesize HyCar(200)35, except for the amounts of Car-OMe and EDC: Finally, HyCar(700)35 was synthesized as described for the production of HyCar(200)35, except for a few aspects: the volume of the solvents was doubled, and 20 ml of water was added to the final reaction mixture.

Molecular weight distribution and intrinsic viscosity of the Hycar conjugates.
The medium molecular weight of each conjugate was determined by means of size exclusion chromatography on a Visctek GPCmax VE 2001 chromatographic system (Malvern) equipped with two TSK-GEL GMPWXL columns (7.8 mm ID × 300 mm; TOSOH BIOSCIENCE) installed in series. The system was coupled with three detectors placed in series: a refraction index detector, a light scattering detector and a four-capillary differential viscometer. Each sample was eluted using 0.1 M sodium nitrate in water containing 0.5 g/L sodium azide with a flow rate of 0.6 ml/min at 40 °C. Omnisec 4.1 software (Marvern Panalytical, https ://www.malve rnpan alyti cal.com/ en/produ cts/produ ct-range /visco tek-range /visco tek-syste ms/visco tek-tdama x/acces sorie s/omnis ec-softw are) was used for data acquisition and analysis. Triple-detector detection was performed on a Viscotek TDA 302 (Malvern). enzymatic hydrolysis by carnosinase. The time-dependent stability of the conjugate HyCar(200)7 towards CN1, compared to that of the parent compounds, was determined by incubating these substances (900 μM) in Tris/HCl buffer (50 mM, pH 8.0) at 37 °C with the enzyme, purified from the culture medium of stably transfected HeLa cells, as previously reported 57 . Several aliquots were sampled until 5 h, and the reaction was stopped by adding trichloroacetic acid (TCA). The amount of histidine was determined after a reaction with ortho-phthalic aldehyde (OPA; Fluka) by means of a fluorometric assay 60 , adapted to multiwell plate detection 70 . Aβ 42 sample preparation. Monomeric Aβ 42 was obtained using a variant of Zagorsky's protocol [95][96][97] .
Briefly, 1 mg of peptide was dissolved in TFA (1 ml). The solution was sonicated for 10 min, and TFA was then evaporated under a gentle stream of N 2 . One millilitre of HFIP was added to the peptide. After 1 h of incubation at 37 °C, the peptide solution was dried under a stream of N 2. This procedure was repeated 2 times. The dried peptide was then dissolved in HFIP (1 ml) and fast-frozen at -80 °C before lyophilization (10 h). The lyophilized sample was re-suspended in 5 mM NaOH solution in ultrapure water (Millipore, 18.2 MΩ·cm resistivity at 25 °C) to a final concentration of 1.5 mM and stored at − 80 °C.
Aβ aggregation assay. The effect of the compounds of interest on the formation of amyloid-type fibrils of Aβ has been assayed as previously reported 98 . Briefly, all the HyCar derivatives (or their parent compounds), Aβ 42 (15 µM)and ThT (45 µM) were incubated in MOPS buffer (50 mM, pH 7.4) at 37 °C in the multiplate reader. The collected fluorimetric data (excitation and emission wavelengths were 450 nm and 480 nm, respectively) of all the measurements, carried out in triplicate, were fitted to Eq. (1).
F max -F 0 is the maximum fluorescence gain due to the amyloid aggregation value, whereas the lag phase (t lag ), calculated by using Eq. (2), represents the time interval before the formation of ThT-sensitive amyloid species.
The parameters of each set of measurements were expressed as the mean ± SD. Aβ disaggregating assay. Solutions 99 . A saturated solution of α-cyano-4-hydroxycinnamic acid (CHCA) in water:acetonitrile 2.3:1 with 0.3% TFA was used as the matrix. Each sample was spotted three times, and four spectra were acquired for each spot, thus obtaining a set of twelve spectra for each sample. To register AFM images, we used a previously described procedure 96 . Briefly, 10 μL of each sample was placed on freshly cleaved muscovite mica (Ted Pella, Inc.) and incubated for 5 min at room temperature. Then, samples were washed with 1 ml of ultrapure water, dried under a gentle nitrogen stream and immediately imaged. A Cypher AFM instrument (Asylum Research, Oxford Instruments, Santa Barbara, CA) operating in tapping AC-mode was used. Tetrahedral tips made of silicon and mounted on rectangular 30-μm-long cantilevers were purchased from Olympus (AT240TS, Oxford Instruments). The probes had nominal spring constants of 2 N/m and driving frequencies of 70 kHz. Images were scanned, and the sizes of the aggregates were measured using a free tool in the Asylum Research offline section analysis software program. cell culture maintenance and treatments. Human neuroblastoma (SH-SY5Y line) cells were treated as previously reported 100 . Briefly, cells were cultivated for no more than 20 passages in full medium, i.e., Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 medium (F12) supplemented with 10% foetal bovine serum (FBS), 2 mM l-glutamine and 100 μg ml −1 streptomycin. The cell culture was grown in tissue culture-treated Corning flasks (Sigma-Aldrich, St. Louis, MO) in a humidified atmosphere (5% CO 2 ) at 37 °C (Heraeus Hera Cell 150C incubator). cytotoxicity assays. The MTT method with the undifferentiated SH-SY5Y cell line was employed for the cell viability assays. The day before the experiment, cells were seeded at a density of 2.5 × 10 4 cells per well (50% confluence) in full medium on Corning tissue culture treated 48-multiwell plates (Sigma-Aldrich, St. Louis, MO) until cellular adhesion was attained. Samples of Aβ 42 (15 µM) were pre-incubated for 6, 24 and 48 h at 37 °C alone or in the presence of HyCar(200)14, HyCar(700) 35,Hy(200) and Hy(700) with or without Car. Cells were treated for 24 h with pre-incubated Aβ 42 (5 µM) in the presence of the studied compounds in full medium supplemented with 1% FBS. Then, cell viability was determined at 37 °C by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (Sigma-Aldrich, St. Louis, MO) method (MTT assay) as previously described 101 . After 90 min, the formazan salts produced by succinate dehydrogenase activity in live cells were solubilized by the addition of DMSO, and absorbance was measured at 570 nm by a microplate plate reader. The results were expressed as % of viable cells. At least three replicate experiments were conducted, and the averaged data with standard deviations were reported. Student's t-test was performed for statistical comparisons to analyse the variance, and a p-value less than 0.05 was considered to be statistically significant.