Human Serum Albumin Binds Native Insulin and Aggregable Insulin Fragments and Inhibits Their Aggregation

The purpose of this study was to investigate whether Human Serum Albumin (HSA) can bind native human insulin and its A13–A19 and B12–B17 fragments, which are responsible for the aggregation of the whole hormone. To label the hormone and both hot spots, so that their binding positions within the HSA could be identified, 4-(1-pyrenyl)butyric acid was used as a fluorophore. Triazine coupling reagent was used to attach the 4-(1-pyrenyl)butyric acid to the N-terminus of the peptides. When attached to the peptides, the fluorophore showed extended fluorescence lifetimes in the excited state in the presence of HSA, compared to the samples in buffer solution. We also analyzed the interactions of unlabeled native insulin and its hot spots with HSA, using circular dichroism (CD), the microscale thermophoresis technique (MST), and three independent methods recommended for aggregating peptides. The CD spectra indicated increased amounts of the α-helical secondary structure in all analyzed samples after incubation. Moreover, for each of the two unlabeled hot spots, it was possible to determine the dissociation constant in the presence of HSA, as 14.4 µM (A13–A19) and 246 nM (B12–B17). Congo Red, Thioflavin T, and microscopy assays revealed significant differences between typical amyloids formed by the native hormone or its hot-spots and the secondary structures formed by the complexes of HSA with insulin and A13–A19 and B12–B17 fragments. All results show that the tested peptide-probe conjugates and their unlabeled analogues interact with HSA, which inhibits their aggregation.


Introduction
Recently, there has been a focus on finding effective transporter vehicles for drugs. These systems, known as drug delivery systems (DDS), are characterized by properties that improve the biocompatibility, biodistribution, and physicochemical properties of a medicine once it has been administered [1]. One of the most difficult challenges associated with the design of new DDS is that they should not interact with the immune system [2,3]. This requirement is fulfilled by Human Serum Albumin (HSA), the most abundant plasma protein. This protein can reversibly bind various ligands such as lipids, hydrophobic compounds, and metal ions (copper, nickel, and calcium) with high binding affinity, acting as a carrier of various ligands in the circulatory system. It is composed of three structurally similar globular within biomolecules and biological membranes to be followed, while facilitating early diagnosis of diseases and enabling the biodistribution of drugs to be controlled [47,48].
Pyrene and its derivatives are very popular fluorescent markers used to study interactions with HSA [49][50][51][52]. Despite the fact that there is differentiated localization of pyrene derivatives in HSA, these systems are commonly applied as fluorescent probes, due to the high sensitivity of their fine vibronic structure to changes in the environment (the Ham effect) and long fluorescence lifetimes. The chromophore itself binds with albumins in a nonspecific way, with relatively low affinity (binding constant in the order of 10 4 dm 3 mol −1 ), and its polarity index (I 1 /I 3 ratio of vibronic peaks in the fluorescence spectrum) is close to that in ethanol. Negatively-charged pyrene derivatives also bind strongly within albumins [52]. In a study of the intermolecular proton-transfer reaction of tri-sodium 8-hydroxypyrene-1,3,6-trisulfonate (pyranine, HPTS) with HSA, Cohen et al. suggested that the probe is bound in Sudlow's site II [53]. There are also data in the literature concerning the design of pyrenyl probes attached to short peptides chains, which are capable of recognizing specific sites in proteins (e.g., in lysozyme and bovine serum albumin) [54][55][56] and the specific structural elements of cell membranes. These systems, also known as protein scissors activated by light, are responsible both for recognition of the binding site in the target molecule and for producing the photochemical energy necessary to selectively destroy peptide bonds inside the protein. Furthermore, derivatives of 4-(1-pyrenyl)butyric acid can be used to determine oxygen concentrations in biological systems [57,58].
The data on the ability of HSA to bind various compounds prompted us to undertake research on the use of HSA as a system capable of transporting peptidic inhibitors of insulin and amylin aggregation (both of these hormones are present in the amyloid deposits identified in people with diabetes). Before starting research with the use of peptidic inhibitors of the aggregation process, it was necessary to ascertain whether HSA interacts (binds) with native hormones as well as their fragments known as hot spots. Hot spots are structural elements that determine the aggregation of native polypeptides (proteins). The first goal of the project was therefore to determine whether HSA binds native human insulin and the A13-A19 and B12-B17 fragments of human insulin, which are responsible for aggregation of the whole hormone. We assumed that attaching 4-(1-pyrenyl)butyric acid as a fluorophore to insulin and both hot spots of insulin would enable localization of their binding sites within HSA because of its spectral properties. The use of pyrene derivative of butyric acid should enable the synthesis of fluorophore labeled peptides directly on solid support or giving the opportunity of attachment it to insulin under conditions assuring a preservation of the unchanged spatial structure of the hormone. Additionally, we expected that the binding of insulin and its hot spots by HSA would influence their susceptibility to aggregation. The impact of HSA could thereby be tested in this respect. We next investigated whether HSA interacted with insulin and its fragments, and whether HSA reduced the aggregation capacity of insulin and its hot spots. Confirmation of the ability of HSA to bind insulin and its fragments and reduce their susceptibility to aggregation might then provide the basis for further research on the use of HSA as DDS for peptidic inhibitors of the insulin aggregation process, which could lead to innovative approaches to diabetes treatment.

Loading of the 2-Chlorotrityl Chloride Resin (GP1)
The amino acid (3 equiv. relative to the resin) and 6 equiv. of EtNiPr 2 were dissolved in CH 2 Cl 2 (10 mL/g resin). The 2-chlorotrityl chloride resin was pre-swollen in CH 2 Cl 2 for 1 h. The solution of amino acid and the resin were added and the suspension was shaken for 120 min. The resin was then filtered and washed with CH 2 Cl 2 /MeOH/EtNiPr 2 in the ratio of 17:2:1 (3×), DMF (2×) and CH 2 Cl 2 (3×). The functionalized resin was dried in a vacuum desiccator to constant mass.

Peptides Functionalization by 4-(1-Pyrene)butyric Acid (GP3)
To the obtained peptides on resin was attached 4-(1-pyrene)butyric acid as a fluorescent probe. Before coupling, the resin was pre-swollen in CH 2 Cl 2 for 1 h. Activation of the carboxyl group of 4-(1-pyrene) butyric acid was performed in DMF. First, 2 equiv. of DMT/NMM/TosO − was dissolved and stirred for 5 min at 0 • C. Then, 2 equiv. of fluorescent probe and 1 equiv. of NMM were added. The progress of the reaction was controlled using TLC (eluent CH 2 Cl 2 , visualization: UV-Vis and 0.1% NBP solution in EtOH). The lack of a red spot and R f = 0 indicated the end of the activation process. The solution was transferred to swollen resin without a Fmoc group on the N-terminal amino acid. Coupling was performed for 24 h. The resin was filtered and washed with DMF (3×) and CH 2 Cl 2 (3×). Completion of the reaction was monitored by the Kaiser test.

Cleavage from the Resin (GP4)
The conjugates functionalized with fluorescent probe were cleaved from the resin using a mixture of 95% TFA (2,2,2-trifluoroacetic acid)/2.5% H 2 O/2.5% TIS (triisopropylsilane), (2 mL/0.1 g resin). Cleavage was performed over 4 h, and then the resin was filtered off and the filtrate was evaporated under nitrogen. To the residue was added Et 2 O to precipitate the product. The obtained solid was filtered off and washed with Et 2 O. The crude product was lyophilized and identified by the LC-MS method.  (2), and Py-VEALYL (3), we were not able to determine the molar absorbance coefficients of these compounds in the buffer solution. The examined samples were obtained by dissolving small amounts of Py-Ins (1), Py-LYQLENY (2) and Py-VEALYL (3) in HSA solution and vortexing for less than 1 min. The prepared samples were used within 24 h. Py-Ins (1), Py-LYQLENY (2), and Py-VEALYL (3) were also dissolved in Triton TX ® solution (TX-100), methanol, and ethanol. The concentration of HSA was verified spectrophotometrically using the molar absorption coefficient 35500 M −1 cm −1 at 280 nm for HSA [61]. In order to ensure complete protein hydration, the solutions were stored at room temperature for at least 1 h before measurements. The samples for fluorescence measurements were saturated with argon or deaerated with a vacuum pump. The results were independent (within experimental uncertainty) of the deaeration procedure.

MST Measurements
HSA was used as a fluorescently labeled target molecule. Peptides 5 and 6 (ligands) without a fluorescent marker were used in the studies. The HSA (0.6 mg) was stained using RED-NHS second-generation dye, diluted in dimethyl sulfoxide. The remaining solutions and subsequent ligand dilutions were prepared in 1 × PBS buffer with the addition of 0.1% Pluronic ® F-127 solution and sodium chloride (150 mM). The final concentration of HSA used in experiments was 20 nM, whereas the highest concentration for both ligands was 0.25 mM. A serial 1:1 dilution was performed by transferring one volume of ligand solution to an equal volume of buffer, mixing, and repeating. The ligand concentration was reduced by 50% in each dilution step. After serial dilution of the ligand, 10 µL of labeled HSA was added to each tube from 16 to 1 and mixed by pipetting. After 30 min of incubation at ambient temperature, the capillaries were dipped into the appropriate tube with solution and then put in the device tray.

CD Measurements
Samples were prepared by dissolving 1 mg of the analyzed peptide in phosphate buffer pH 7.2 (10 mL), to a final concentration of 0.1 mg/mL All measurements were carried out at room temperature using a quartz cuvette (1 mm path length, Hellma). The experimental settings were as follows: measurements range, 190-270 nm; data pitch, 5 nm; scanning mode, continuous; scanning speed, 100 nm/min; bandwidth, 3 nm; integration time, 1 s.  in phosphate buffer, pH 7.2) was added to the all samples, which were incubated for another 3 days at room temperature.

Measurement Methods
Steady-state absorption and fluorescence spectra were recorded at a resolution of 0.5 nm using an Aminco Bowman AB2 Series spectrofluorometer (Thermo Fisher Scientific Inc., Waltham, MA, USA). Unless otherwise indicated, in emission measurements, the excitation wavelength was 337 nm, and excitation and emission slits were set to 2.0 and 0.5 nm, respectively.
Time-resolved measurements of HSA fluorescence were performed using a flash photolysis system based on a GL-3300 nitrogen laser (Photon Technology International, Birmingham, NJ, USA) providing single light pulses with a wavelength of 337.1 nm, duration~800 ps, and average energy 100 µJ.
The detection system consisted of a Baush&Lomb monochromator (Rochester, NY, USA), Hamamatsu 1P28 photomultiplier, and a PS325 power supply system (Stanford Research, Sunnyvale, CA, USA). Output signals from the photomultiplier were digitized and recorded using a Tektronix DPO 7254 oscilloscope (Beaverton, OR, USA) (40 GS/s) and transferred to a computer. All experiments were carried out at room temperature (25 ± 1 • C) using quartz cells with an optical length of 1 cm.
The CD spectra were recorded using a Jasco J-1500 instrument (Jasco, Cracow, Poland) and quartz cuvettes with 1 mm optical path lengths. The measurements were acquired in the range of 190-270 nm at ambient temperature by taking points every 5 nm, with a scan rate of 100 nm per min, an integration time of 1 s, and a bandwidth of 4 nm. The samples were prepared in phosphate buffer (0.1 M, pH 7.2) with a final concentration of 0.1 mg/mL.
The MST measurements were carried out using a Monolith NT.115 (NanoTemper Technologies GmbH, München, Germany) instrument in special Monolith NT.115 capillaries.
The microscopic assays were performed for samples stained with Congo Red dye. The samples were transferred on a microscopic slide and visualized using a Delta Optical Genetic Pro (Warsaw, Poland) light microscope.
The UV-Vis measurements were carried out using a Hitachi spectrophotometer (Hitachi, Tokyo, Japan).
The fluorescence intensity measurements were carried out using a FLUOROMAX-4 system from Horiba Scientific (Edison, NJ, USA), measurement range of 470-600 nm, excitation wavelength 440 nm.

Results and Discussion
A derivative of insulin labeled with 4-(1-pyrene)butyric acid Py-Ins (1) was used for the tests, as well as two labeled insulin amyloidogenic cores: Py-LYQLENY-OH (2) analog of A13-A19 and Py-VEALYL-OH (3) derived from the B12-B17 fragment ( Figure 1). Py-Ins conjugate (1) was obtained by coupling human insulin with 4-(1-pyrene)butyric acid, using 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methyl-morpholinium p-toluenosulphonate (DMT/NMM/TosO − ) [59] as a coupling reagent. The synthesis of derivatives (2) and (3) was performed under microwave-assisted standard solid phase peptide synthesis conditions, also in the presence of DMT/NMM/TosO − as a coupling reagent. This strategy enabled pure products to be obtained with high efficiency and avoided undesirable side reactions, such as racemization and folding of peptide chains. Conjugates 1-3 with at least 98% purity were used in the studies. Biomolecules 2020, 10, x 8 of 25 The first stage of the research concerned the determination of the spectral characteristics of the HSA-4-(1-pyrene)butyric acid (PBA) complex. 4-(1-pyrene)butyric acid was used as a negative control in studies of the interaction of HSA with derivatives 1-3. We have analyzed the interaction of PBA with HSA in some detail ( Figure S5). The observed changes of fluorescence intensity are complex. Addition of HSA to aqueous PBA solution the intensity (I) decreased, reaching a minimum at the PBA/HSA molar ratio 2, and then increased to some extent. A similar trend was also observed for the complex of 1-pyrene sulfonic acid (PSA) with HSA [52]. The intensity (I) decreasing can be due to the interaction with quencher Trp214 residue located in the domain IIA. However, the quenching efficiency of PBA decreased at HSA concentrations higher than 8 µM. This finding can suggest that PBA may also be located in another site (sites) distant from Trp214 (non-specific binding). The decay of the excited singlet state of PBA (15 µM) in a wide range of HSA concentrations for deaerated samples are shown in Figure S6. Additionally, it has been found that, in the case of a sample of PBA in buffer (without of HSA), fluorescence decay can be described as a classical firstorder kinetics, and its lifetime is equal to 126 ns (Table S1). The biexponential kinetic patterns and the lifetime data as a function of the HSA concentration (Table S1) might be interpreted in terms of different microenvironments of the probe. The fast component, τ1 (in the range of 16-20 ns), most likely corresponds to PBA molecules embedded in the subdomain IIA (near Trp214) [52]. The higher HSA concentrations led to successively docking of the probe to the protein; τ2 increases monotonically reaching the limit value 163 ns. The important observation from time-resolved measurements is the change in the primary fluorescence signal intensity ( Figure S6) in the function of HSA concentration. The literature data show that PBA itself has a weak affinity for albumin [56,62].
We began our research by determining the spectral characteristics of the Py-Ins conjugate in the phosphate buffer (PBS), with HSA or TX-100 as additives (conditioning the formation of micellar structures) (Figure2, panel I). In the buffer solution, the maximum of the band corresponding to the S0 (ν = 0) → S2 (ν = 2) transition of the pyrene moiety occurred at 342 nm. The addition of 30 µM HSA or 20 mM TX-100 resulted in characteristic red-shifts of 2 and 3 nm, respectively, and an increase in absorbance with characteristic better separation of the vibrational structure. The observed spectral changes are characteristic for the transfer of the fluorescent marker to the more hydrophobic environment. The structure of labeled insulin (1) and insulin amyloidogenic cores (2, 3) with 4-(1-pyrene)butyric acid.
The first stage of the research concerned the determination of the spectral characteristics of the HSA-4-(1-pyrene)butyric acid (PBA) complex. 4-(1-pyrene)butyric acid was used as a negative control in studies of the interaction of HSA with derivatives 1-3. We have analyzed the interaction of PBA with HSA in some detail ( Figure S5). The observed changes of fluorescence intensity are complex. Addition of HSA to aqueous PBA solution the intensity (I) decreased, reaching a minimum at the PBA/HSA molar ratio 2, and then increased to some extent. A similar trend was also observed for the complex of 1-pyrene sulfonic acid (PSA) with HSA [52]. The intensity (I) decreasing can be due to the interaction with quencher Trp214 residue located in the domain IIA. However, the quenching efficiency of PBA decreased at HSA concentrations higher than 8 µM. This finding can suggest that PBA may also be located in another site (sites) distant from Trp214 (non-specific binding). The decay of the excited singlet state of PBA (15 µM) in a wide range of HSA concentrations for deaerated samples are shown in Figure S6. Additionally, it has been found that, in the case of a sample of PBA in buffer (without of HSA), fluorescence decay can be described as a classical first-order kinetics, and its lifetime is equal to 126 ns (Table S1). The biexponential kinetic patterns and the lifetime data as a function of the HSA concentration (Table S1) might be interpreted in terms of different microenvironments of the probe. The fast component, τ 1 (in the range of 16-20 ns), most likely corresponds to PBA molecules embedded in the subdomain IIA (near Trp214) [52]. The higher HSA concentrations led to successively docking of the probe to the protein; τ 2 increases monotonically reaching the limit value 163 ns. The important observation from time-resolved measurements is the change in the primary fluorescence signal intensity ( Figure S6) in the function of HSA concentration. The literature data show that PBA itself has a weak affinity for albumin [56,62].
We began our research by determining the spectral characteristics of the Py-Ins conjugate in the phosphate buffer (PBS), with HSA or TX-100 as additives (conditioning the formation of micellar structures) (Figure 2, panel I). In the buffer solution, the maximum of the band corresponding to the S 0 (ν = 0) → S 2 (ν = 2) transition of the pyrene moiety occurred at 342 nm. The addition of 30 µM HSA or 20 mM TX-100 resulted in characteristic red-shifts of 2 and 3 nm, respectively, and an increase in absorbance with characteristic better separation of the vibrational structure. The observed spectral changes are characteristic for the transfer of the fluorescent marker to the more hydrophobic environment. Broad absorption bands can indicate the presence of Py-Ins aggregates. Therefore, a solution of Py-Ins (1) complex with HSA was filtered through a 0.2 µm syringe filter in order to remove any Py-Ins aggregates, and the stability of the complex was tested for a further 6 days. On the absorption spectra recorded after 2, 4, and 6 days ( Figure 2, panel IIa), no changes were observed, illustrating the stability of the complex and a lack of Py-Ins aggregation. This may indicate that the Py-Ins (1) conjugate was completely bound to the HSA, and that the protein has a protective role in the aggregation process. Additional confirmation of this was provided by the observed linear increase in Py-Ins absorbance for the samples diluted with 30 µM HSA solution (Figure 2, panel IIb).
The binding capacity of the Py-Ins (1) conjugate by HSA was also demonstrated by the results of fluorescence studies (Figure 3, panel I). Measurements were made in PBS and with the addition of TX-100 or HSA. Typically, the transfer of a pyrene marker from an aqueous environment to a less polar environment leads to higher emission intensity and longer lifetime in the excited state, due to the lower impact of nonradiative mechanisms for energy dissipation in the hydrophobic environment. This was confirmed in the case of Py-Ins (1) in TX-100 solution. The different behavior of Py-Ins in the presence of HSA suggests more effective fluorescence quenching of the pyrene Broad absorption bands can indicate the presence of Py-Ins aggregates. Therefore, a solution of Py-Ins (1) complex with HSA was filtered through a 0.2 µm syringe filter in order to remove any Py-Ins aggregates, and the stability of the complex was tested for a further 6 days. On the absorption spectra recorded after 2, 4, and 6 days (Figure 2, panel IIa), no changes were observed, illustrating the stability of the complex and a lack of Py-Ins aggregation. This may indicate that the Py-Ins (1) conjugate was completely bound to the HSA, and that the protein has a protective role in the aggregation process. Additional confirmation of this was provided by the observed linear increase in Py-Ins absorbance for the samples diluted with 30 µM HSA solution (Figure 2, panel IIb).
The binding capacity of the Py-Ins (1) conjugate by HSA was also demonstrated by the results of fluorescence studies (Figure 3, panel I). Measurements were made in PBS and with the addition of TX-100 or HSA. Typically, the transfer of a pyrene marker from an aqueous environment to a less polar environment leads to higher emission intensity and longer lifetime in the excited state, due to the lower impact of nonradiative mechanisms for energy dissipation in the hydrophobic environment. This was confirmed in the case of Py-Ins (1) in TX-100 solution. The different behavior of Py-Ins in the presence of HSA suggests more effective fluorescence quenching of the pyrene molecule docked to the HSA binding site. In the presence of TX-100 and HSA, the pyrene fluorescence spectra show a prominent fine structure in the form of five predominant peaks, while, on the spectra for the pyrene derivatives in buffer solution, this vibrational structure was not detected in full. The lack of vibronic peaks I 2 and I 3 on the fluorescence spectrum of Py-Ins (1) may point to the presence of aggregates of the Py-Ins conjugate or a different microenvironment. The spectra presented in Figure 3, panel I show a characteristic broad band in the spectral range of 450-500 nm, which may be associated with the formation of excimers of pyrene fragment bounded to insulin or exciplexes between pyrene residue and an aromatic amino acid of the HSA. In homogeneous buffer solutions, the Py-Ins (1) fluorescence decay can be described by the bi-exponential function:G where A 1 , A 2 are amplitudes of the lifetime components τ 1 , τ 2 . We applied a 4-parameter fitting procedure and the best fit results are summarized in Table 1, together with the average lifetimes calculated as: lack of vibronic peaks I2 and I3 on the fluorescence spectrum of Py-Ins (1) may point to the presence of aggregates of the Py-Ins conjugate or a different microenvironment. The spectra presented in Figure 3, panel I show a characteristic broad band in the spectral range of 450-500 nm, which may be associated with the formation of excimers of pyrene fragment bounded to insulin or exciplexes between pyrene residue and an aromatic amino acid of the HSA.  Table 1 for parameters of the best-fit lines (smooth).
In homogeneous buffer solutions, the Py-Ins (1) fluorescence decay can be described by the biexponential function:G  Table 1 for parameters of the best-fit lines (smooth).
The fast component (τ 1 , A 1 ) and the slow component (τ 2 , A 2 ) can be assigned to Py-Ins (1) molecules in the aggregate and monomer forms, respectively. Transfer of the probe from the buffer solution to the TX-100 matrix or HSA was accompanied by an increase in the mean lifetime of the singlet excited state from 86.6 ns to 147.9 ns and 133.6 ns, respectively. Filtering of the solution also increased the fluorescence lifetime for all the examined systems. The data presented in Table 1 indicate that the A 1 amplitude reflecting the contribution of the aggregate fraction decreased significantly after filtering the solution and for Py-Ins/TX-100 the kinetic decay became nearly monoexponential. However The results of our research on the ability of HSA to bind the conjugate Py-Ins (1) prompted us to undertake further studies using labeled 4-(1-pyrene)butyric acid of amyloidogenic insulin cores: Py-LYQLENY-OH (2) and Py-VEALYL-OH (3). In order to investigate the binding of Py-LYQLENY (2) and Py-VEALYL (3) to HSA, absorption studies as well as steady-state and time-resolved fluorescence studies were performed, without and in the presence of the serum protein, using PBS solution s of 2 and 3. The spectral properties of peptides 2 and 3 are presented in the Supplementary Material ( Figures S7 and S9). The absorption spectra of both pyrene-labeled insulin fragments are presented in Figure 4, panel I), in which a 2-3 nm bathochromic shift in the presence of HSA compared to the buffer solution may be observed. This finding indicates the preferred location of the pyrene label within the less polar area within HSA. Moreover, peptidic fragments could interact with the HSA surface, enhancing docking of the probe inside the protein. Similar observations for pyrene-short synthetic peptide conjugates have been presented by Kumar et al. [54][55][56].    (2) and (3) in the presence and absence of HSA. In the presence of HSA, there was a significant change in the oscillation structure compared to the PBS solutions, involving the appearance of previously invisible vibrational bands II and III in the pyrene moiety. This confirms the interaction of the pyrene-labeled insulin   (2) and (3) in the presence and absence of HSA. In the presence of HSA, there was a significant change in the oscillation structure compared to the PBS solutions, involving the appearance of previously invisible vibrational bands II and III in the pyrene moiety. This confirms the interaction of the pyrene-labeled insulin fragments with HSA, and the change of the microenvironment of the fluorescence probe from strongly polar (buffer solution) to hydrophobic (less polar) inside of HSA. In the presence of HSA, the pyrene moiety ceases to be exposed to a polar medium and starts to be located in a hydrophobic microenvironment. Stiffening of its structure probably took place (the carbon atoms of the pyrene residue are situated in the same plane), which accounts for the sharpening of the oscillating structure in the spectrum.
Additionally, we examined the quenching of Py-LYQLENY and Py-VEALYL by acrylamide (ACR) in the buffer solution and HSA solution ( Figure 5 Stern-Volmer plots). In the presence of HSA, both conjugates 2 and 3 were protected against interaction with the quencher, since the slope of the relationship I 0 /I vs. ACR concentration is much lower with respect to the analogous slope in the case of the protein. These results indicate that pyrene fragment of both conjugates Py-LYQLENY (2) and Py-VEALYL (3) is located inside of HSA.  For Py-LYQLENY conjugate (2), the binding constants Kb were calculated according to Equation  Figure 6b) and I is the fluorescence intensity for subsequent acrylamide (ACR) concentrations used as a fluorescence quencher.
Based on measurements of fluorescence intensity as a function of HSA concentration (Figure 6a), it was found that Kb of Py-LYQLENY (2) equals 2 × 10 5 dm 3 mol −1 . The calculated value also confirmed existence of interactions between 2 and HSA with pyrene fragment presented on less polar environment inside protein. It has been proved that the trapping of PSA inside HSA led to two possible probe/protein binding modes with the binding constants 6.5 × 10 6 dm 3 mol −1 (a specific receptor site), and 3.8 × 10 5 dm 3 mol −1 (non-specific binding) [52]. The binding constants 1.5 ± 0.3 × 10 6 dm 3 mol −1 to 6.5 ± 0.4 × 10 7 dm 3 mol −1 were calculated for complexes of HSA with Py-hydrophobic peptide. Attachment of 4-(1-pyrene)butyric acid to peptide chain containing Phe and Gly residues can increase the hydrophobicity of docked molecules and improve the more specific binding [56]. For Py-LYQLENY conjugate (2), the binding constants K b were calculated according to Equation  The results of steady state fluorescence studies were also confirmed by time-resolved measurements of fluorescence decay for the fluorescence marker both in the buffer and in the presence of HSA (Figure 7). In buffer, the fluorescence of the probe decayed according to classical first-order kinetics, and its lifetime τ0 was determined by fitting the experimental decay with the exponential function. For HSA conjugates 2 and 3, a bi-exponential function was applied and the average lifetimes were calculated from Equation (2). The times of the fluorescence decay for conjugates 2 and 3 are summarized in Table 2. Based on measurements of fluorescence intensity as a function of HSA concentration (Figure 6a), it was found that K b of Py-LYQLENY (2) equals 2 × 10 5 dm 3 mol −1 . The calculated value also confirmed existence of interactions between 2 and HSA with pyrene fragment presented on less polar environment inside protein. It has been proved that the trapping of PSA inside HSA led to two possible probe/protein binding modes with the binding constants 6.5 × 10 6 dm 3 mol −1 (a specific receptor site), and 3.8 × 10 5 dm 3 mol −1 (non-specific binding) [52]. The binding constants 1.5 ± 0.3 × 10 6 dm 3 mol −1 to 6.5 ± 0.4 × 10 7 dm 3 mol −1 were calculated for complexes of HSA with Py-hydrophobic peptide. Attachment of 4-(1-pyrene)butyric acid to peptide chain containing Phe and Gly residues can increase the hydrophobicity of docked molecules and improve the more specific binding [56].
The results of steady state fluorescence studies were also confirmed by time-resolved measurements of fluorescence decay for the fluorescence marker both in the buffer and in the presence of HSA (Figure 7). In buffer, the fluorescence of the probe decayed according to classical first-order kinetics, and its lifetime τ 0 was determined by fitting the experimental decay with the exponential function. For HSA conjugates 2 and 3, a bi-exponential function was applied and the average lifetimes were calculated from Equation (2). The times of the fluorescence decay for conjugates 2 and 3 are summarized in Table 2.  Our   Our results show that the fluorescence lifetimes of both conjugates were about four times longer in the presence of HSA compared to the PBS solution. The remarkable prolongation of pyrene fluorescence lifetime in both cases confirmed preferred probe localization in a less polar environment inside protein.
To examine possible interaction with Trp214, we employed the time-resolved technique with excitation at 295 nm to track the fluorescence changes of Trp214 in subdomain IIA (Sudlow's I site). The results presented in Figure 8 reveal a slight decrease in the Trp214 fluorescence lifetime in the presence of Py-LYQLENY.
The decay curves were fitted using the bi-exponential function and the average lifetimes were 5.4 ns and 3.7 ns for HSA/PBS and HSA/Py-LYQLENY, respectively. HSA fluorescence (*Trp214) was quenched by Py-LYQLENY via Förster resonance energy transfer (Figure 8). The parameters of FRET analysis for HSA/Py-LYQLENY (including r DA = 22 Å) fall in the range of reported data [52]. This can be evidence of Py-LYQLENY specific binding in the Sudlow's site 1. Additionally, we noticed another confirmation of this thesis in the pulse radiolysis measurements of the HSA-Py-LYQLENY. complex. It has been claimed [64] that ligands located at the Sudlow's site 1 are not reduced by the hydrated electron. We found that Py-LYQLENY (2) is efficiently reduced by hydrated electron in a buffer solution even in concentration as low as 20 µM (Figure S10), while, in the presence of HSA, the reduction process of Py-LYQLENY has been completely eliminated. The decay curves were fitted using the bi-exponential function and the average lifetimes were 5.4 ns and 3.7 ns for HSA/PBS and HSA/Py-LYQLENY, respectively. HSA fluorescence (*Trp214) was quenched by Py-LYQLENY via Förster resonance energy transfer (Figure 8). The parameters of FRET analysis for HSA/Py-LYQLENY (including rDA = 22 Å) fall in the range of reported data [52]. This can be evidence of Py-LYQLENY specific binding in the Sudlow's site 1. Additionally, we noticed another confirmation of this thesis in the pulse radiolysis measurements of the HSA-Py-LYQLENY. complex. It has been claimed [64] that ligands located at the Sudlow's site 1 are not reduced by the hydrated electron. We found that Py-LYQLENY (2) is efficiently reduced by hydrated electron in a buffer solution even in concentration as low as 20 µM ( Figure S10), while, in the presence of HSA, the reduction process of Py-LYQLENY has been completely eliminated.
Based on the results presented so far, it can be concluded that HSA shows the ability to transport and inhibit undesirable aggregation of Py-Ins, Py-LYQLENY, and Py-VEALYL conjugates. In order to confirm the results of spectroscopic measurements, we also analyzed the interactions of native insulin (4) and hot spots of insulin (LYQLENY (5), VEALYL (6)), using three independent research techniques covering circular dichroism, microscale thermophoresis techniques, and a set of nonspecific assays recommended for the study of protein/peptide aggregation (the Congo Red (CR) assay, Thioflavin T (ThT) assay, and microscopic examination).
Circular dichroism (CD) spectroscopy can be a very valuable source of information about differences in the secondary structure of peptides and proteins [65]. It has been shown that all αhelical proteins show a strong minimum at 222 nm and 208-210 nm and a strong maximum at 191-193 nm. In the case of β-sheet protein, a single negative band may be observed in the 210-225 nm wavelength range and a strong positive band in the 190-200 nm wavelength range, although their intensities are noticeably lower than those for α-proteins [66]. The CD spectra recorded for buffered solutions of insulin (4) and hot-spot fragments (5), (6) in the presence of HSA as well as for native insulin are presented in Figure 9a,b. The measurements were performed after 30 min (sample 0), and then after two and four days of incubation at 37 °C. Based on the results presented so far, it can be concluded that HSA shows the ability to transport and inhibit undesirable aggregation of Py-Ins, Py-LYQLENY, and Py-VEALYL conjugates. In order to confirm the results of spectroscopic measurements, we also analyzed the interactions of native insulin (4) and hot spots of insulin (LYQLENY (5), VEALYL (6)), using three independent research techniques covering circular dichroism, microscale thermophoresis techniques, and a set of nonspecific assays recommended for the study of protein/peptide aggregation (the Congo Red (CR) assay, Thioflavin T (ThT) assay, and microscopic examination).
Circular dichroism (CD) spectroscopy can be a very valuable source of information about differences in the secondary structure of peptides and proteins [65]. It has been shown that all α-helical proteins show a strong minimum at 222 nm and 208-210 nm and a strong maximum at 191-193 nm. In the case of β-sheet protein, a single negative band may be observed in the 210-225 nm wavelength range and a strong positive band in the 190-200 nm wavelength range, although their intensities are noticeably lower than those for α-proteins [66]. The CD spectra recorded for buffered solutions of insulin (4) and hot-spot fragments (5), (6) in the presence of HSA as well as for native insulin are presented in Figure 9a,b. The measurements were performed after 30 min (sample 0), and then after two and four days of incubation at 37 • C.
On the CD spectra of the HSA-insulin complex (Figure 9), a regular increase was observed in signal intensity at 194 nm and a stronger minimum at 210 and 224 nm. This means that, in the presence, of the native form of insulin does not tend to form amyloid fibrils, for which the β-sheet structure is the main growth factor [60,67]. We can suppose that possible interactions between HSA and the native hormone inhibit undesirable insulin aggregation and increase the content of the α-helix structure in the sample. Changes recorded for the HSA-LYQLENY complex (Figure 9c,d) were slightly different compared to the HSA-insulin complex. Most importantly, no systematic increase was noticed in the intensities of the maximum and minimum. After four days of incubation, the signal intensity reached the highest volume (195, 210, and 223 nm), which also confirmed the increase in the amount of the α-helix structure as a result of binding of LYQLENY by HSA. On the other hand, the signal intensity at 223 nm showed the same enhancement as the signal at 210 nm. The observed increase of the minimum at 223 nm indicates the existence of a small quantity of misfolded LYQLENY. Moreover, this fraction showed a tendency to form β-sheet and/or loop structures in buffer solution, resulting in a significantly signal increment at 223 nm. There was also a lack of regular changes in signal intensity at 193 nm observed for the HSA-VEALYL complex (Figure 9e,f), although for this hot spot the increase in the intensity of the minimum at 210 nm was more pronounced in comparison to that of the minimum at 224 nm. All changes found on the CD spectra for the HSA-VEALYL complex confirmed that this hot spot also interacts with HSA. Binding of VEALYL by HSA prevents the formation of amyloid fibrils. Additionally, we can suppose that the non-bound fraction of VEALYL in the presence of HSA has less tendency to misfold than LYQLENY, as evidenced by the smaller content of β-sheet structures in the buffer solution. On the CD spectra of the HSA-insulin complex (Figure 9), a regular increase was observed in signal intensity at 194 nm and a stronger minimum at 210 and 224 nm. This means that, in the presence, of the native form of insulin does not tend to form amyloid fibrils, for which the β-sheet structure is the main growth factor [60,67]. We can suppose that possible interactions between HSA and the native hormone inhibit undesirable insulin aggregation and increase the content of the αhelix structure in the sample. Changes recorded for the HSA-LYQLENY complex (Figure 9c,d) were slightly different compared to the HSA-insulin complex. Most importantly, no systematic increase was noticed in the intensities of the maximum and minimum. After four days of incubation, the signal intensity reached the highest volume (195, 210, and 223 nm), which also confirmed the increase in the VEALYL complex confirmed that this hot spot also interacts with HSA. Binding of VEALYL by HSA prevents the formation of amyloid fibrils. Additionally, we can suppose that the non-bound fraction of VEALYL in the presence of HSA has less tendency to misfold than LYQLENY, as evidenced by the smaller content of β-sheet structures in the buffer solution.
The second technique used to assess the inhibition of aggregation of insulin (4) and insulin hot spots 5 and 6 in the presence of HSA was the microscale thermophoresis (MST) assay. This assay [68,69] is a very powerful and practical technique for quantifying interactions between small molecules and protein fibrillar aggregates. For both complexes of HSA with 5 and 6, it was possible to estimate the dissociation constant (Kd), which confirmed the attachment of the external ligand to the labeled protein. For LYQLENY (Figure 10a), Kd equals 14.4 µM, whereas, for the VEALYL fragment (Figure 10b), Kd equals 246 nM. The results indicate stronger interactions for the HSA-VEALYL complex, which is consistent with the results obtained from CD measurements. Moreover, the regular fluorescence changes observed with increasing concentrations of a non-fluorescent ligand show the limited tendency of both hot spots to aggregate in the presence of HSA.  In the last stage of our study, we tested the effect of HSA on the aggregation capacity of insulin and its hot spots. Three independent research techniques recommended for non-specific testing of peptide/protein aggregates were used. One of the methods was based on the examination by microscopy of peptide aggregates stained with CR. In the sample containing only HSA (Figure 11a), we recorded a crystal structure characteristic for proteins. Hot spot fragments 5 and 6 after incubation at 37 • C formed typical amyloid fibers (Figure 11b,d), which were very clearly visible after staining with CR [60]. In the presence of HSA, we observed significantly less content of fibers characteristic for amyloid deposits (Figure 11c,e). Moreover, the observed structures have different sizes and shapes and were less efficiently stained by CR in comparison to fibers formed by H-LYQLENY-OH and H-VEALYL-OH in the absence of HSA. Thus, the results of microscopy assays also proved that HSA can bind hot spots fragments, simultaneously enhancing their solubility and limiting the formation of amyloid deposits in buffer solution. and its hot spots. Three independent research techniques recommended for non-specific testing of peptide/protein aggregates were used. One of the methods was based on the examination by microscopy of peptide aggregates stained with CR. In the sample containing only HSA (Figure 11a), we recorded a crystal structure characteristic for proteins. Hot spot fragments 5 and 6 after incubation at 37 °C formed typical amyloid fibers (Figure 11b,d), which were very clearly visible after staining with CR [60]. In the presence of HSA, we observed significantly less content of fibers characteristic for amyloid deposits (Figure 11c,e). Moreover, the observed structures have different sizes and shapes and were less efficiently stained by CR in comparison to fibers formed by H-LYQLENY-OH and H-VEALYL-OH in the absence of HSA. Thus, the results of microscopy assays also proved that HSA can bind hot spots fragments, simultaneously enhancing their solubility and limiting the formation of amyloid deposits in buffer solution.
Biomolecules 2020, 10, x 20 of 25 In the CR assay, no decrease in absorbance typical for amyloid structures was observed in the case of any HSA complexes with insulin and hot spots 5 and 6 ( Figure 11f). A shift in the absorption maximum was observed, which might suggest the formation of amyloid aggregates. However, we found that this is the typical course of the UV-Vis curve for HSA.
The results of the ThT assay also without cross-talk showed the absence of amyloid aggregates of HSA complexes with insulin and peptides 5 and 6 ( Figure 11g). In all cases, we found no increase in fluorescence intensity, which was comparable to the values obtained for ThT.

Conclusions
In this study, we labeled insulin and its hot spots 5 and 6 with 4-(1-pyrene)butyric acid using In the CR assay, no decrease in absorbance typical for amyloid structures was observed in the case of any HSA complexes with insulin and hot spots 5 and 6 ( Figure 11f). A shift in the absorption maximum was observed, which might suggest the formation of amyloid aggregates. However, we found that this is the typical course of the UV-Vis curve for HSA.
The results of the ThT assay also without cross-talk showed the absence of amyloid aggregates of HSA complexes with insulin and peptides 5 and 6 ( Figure 11g). In all cases, we found no increase in fluorescence intensity, which was comparable to the values obtained for ThT.

Conclusions
In this study, we labeled insulin and its hot spots 5 and 6 with 4-(1-pyrene)butyric acid using DMT/NMM/TosO − as a coupling reagent. Both in a solution and in solid phase, this strategy was found to very effective, allowing products to be obtained with the desired spectroscopic properties. For the derivatives Py-Ins (1), Py-LYQLENY (2), and Py-VEALYL (3), labeled with the pyrene marker, fluorescence spectra showed a decay of the pyrene probe oscillation structure. We also observed a shortening of the fluorescence lifetime for Py-Ins (1), Py-LYQLENY (2), and Py-VEALYL (3) in PBS to 86.6 ns, 38.3 ns, and 43.9 ns, respectively. These results indicate peptide aggregation. However, in the presence of HSA, the solubility of Py-Ins (1), Py-LYQLENY (2), and Py-VEALYL (3) increased significantly in aqua media. There was also a noticeable change in the oscillation structure of the fluorescence spectra compared to those for samples without the serum protein. Prolongation of the fluorescence lifetimes was also observed, to 133.6 ns, 186.2 ns, and 153.9 ns for Py-Ins (1), Py-LYQLENY (2), and Py-VEALYL (3), respectively. All of these data indicate the inhibition of aggregation by derivatives 1, 2, and 3 in the presence of HSA. Moreover, for all labeled conjugates, the pyrene chromophore was able to penetrate deeper within the HSA. In turn, the peptidic fragment could interact with HSA surface moieties, stabilizing probe docking. The slight reduction of fluorescence lifetime of tryptophan residue Trp-214 (located in Sudlow's I site) in the presence of Py-LYQLENY (2) proves possible localization of pyrene core in this less polar area. However, the possibility of docking pyrene labeled peptides in different hydrophobic HSA pockets cannot be ruled out. Pyrenyl modified peptides bind to HSA simultaneously in a specific and non-specific manner with similar binding constant of 10 5 dm 3 mol −1 .
The results for unlabeled native insulin (4) and its hot spot fragments H-LYQLENY-OH (5) and H-VEALYL-OH (6) were consistent with data from spectroscopic measurements. Circular dichroism (CD) studies indicated increased amounts of the α-helical structure in all analyzed samples after incubation. It can be supposed that, in the presence of HSA, native insulin (4) as well as hot spot fragments (5) and (6) have less tendency to misfold, as evidenced by the smaller content of β-sheets in comparison to the results for phosphate buffer solutions. Moreover, it was possible to determine the dissociation constant in the presence of HSA for both unlabeled hot spots (5) and (6), as 14.4 µM (5) and 246 nM (6), respectively, using the microscale thermophoresis (MST) assay. The ability to determine both dissociation constants proves the lack of aggregation by peptides 5 and 6 in the presence of HSA. The lack of aggregation by HSA complexes with insulin and its hot spot fragments H-LYQLENY-OH (5) and H-VEALYL-OH (6) was confirmed in studies with Congo Red (CR), Thioflavin T (ThT), and in microscopic tests.
All peptide-probe conjugates 1, 2, and 3 together with their unlabeled analogues 4, 5, 6 were able to interact with HSA. The inhibition of the aggregation process for peptides forming amyloid structures suggests that HSA may be an internal stabilizing factor of insulin and its fragments, and could therefore offer a natural, internal inhibitor of the process of their aggregation. Therefore, HSA has the potential to be used as an effective inhibitory system for amyloidogenic cores of peptides in the bloodstream. This study provides the first indication that HSA could be used as a drug delivery system for peptidic analogues of hot spots of insulin, to inhibit the aggregation of this hormone. Studies on the preparation of HSA complexes with peptide inhibitors of insulin aggregation and on the use of HSA as a transporter are ongoing.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2218-273X/10/10/1366/s1, Figure S1: UV-Vis spectra recorded on the first, second, and third days of incubation of HSA-LYQLENY complex in the presence of Congo Red, Figure S2: UV-Vis spectra recorded on the first, second, and third days of incubation of HSA-VEALYL complex in the presence of Congo Red, Figure S3: The intensity of fluorescence spectra recorded on the first, second, and third days of incubation of HSA-LYQLENY complex in the presence of Thioflavin T, Figure S4: The intensity of fluorescence spectra recorded on the first, second, and third days of incubation of HSA-VEALYL complex in the presence of Thioflavin T, Figure

Conflicts of Interest:
The authors declare no conflict of interest.