Unpacking the aggregation-oligomerization-fibrillization process of naturally-occurring hIAPP amyloid oligomers isolated directly from sera of children with obesity or diabetes mellitus

The formation of amyloid oligomers and fibrils of the human islet amyloid polypeptide (hIAPP) has been linked with β- cell failure and death which causes the onset, progression, and comorbidities of diabetes. We begin to unpack the aggregation-oligomerization-fibrillization process of these oligomers taken from sera of pediatric patients. The naturally occurring or real hIAPP (not synthetic) amyloid oligomers (RIAO) were successfully isolated, we demonstrated the presence of homo (dodecamers, hexamers, and trimers) and hetero-RIAO, as well as several biophysical characterizations which allow us to learn from the real phenomenon taking place. We found that the aggregation/oligomerization process is active in the sera and showed that it happens very fast. The RIAO can form fibers and react with anti-hIAPP and anti-amyloid oligomers antibodies. Our results opens the epistemic horizon and reveal real differences between the four groups (Controls vs obesity, T1DM or T2DM) accelerating the process of understanding and discovering novel and more efficient prevention, diagnostic, transmission and therapeutic pathways.


Results
Hexamer oligomers as potential biomarkers of early β-cell failure. Biomarker identification of early β-cell failure is multidimensional and encompasses the need for a deep understanding of the aggregation/ oligomerization process in vivo, their dynamics and their molecular networks 9 . We avoid the polymorphism of oligomeric species by concentrating on those oligomers with low molecular weight that are related to cellular apoptosis and may have a role to play in the developing stages of metabolic syndrome and diabetes 26,31,32,56,57 . We used a potential "universal pre-treatment"; which is cost-effective and easy to implement in diagnostic laboratories in hospitals 34 .
A total of 15 patients from the pilot hIAPP oligomers study were chosen randomly, in the group of Diabetes Mellitus: 5 patients with type 1; 5 with type 2. 5 patients with obesity. All were compared with 5 children from the healthy group (Table 1).
The distribution of clinical characteristics (sex, age, evolution), biochemical characteristics (glucose, HbA1c, total cholesterol, HDL-C, uric acid) and treatment were presented by their diagnostics. Quantitative variables were summarized by median and range, and categorical variables were reported with their absolute and relative frequency. Numerical variables were compared by Welch's test due to the lack of homoscedasticity and incompatibility with the normal distribution assumption 58 . To compare categorical variables Pearson's chi-squared test was performed 59 .
We demonstrated for the first time the presence of soluble hIAPP oligomers in pre-treated samples (PTS) of sera of children patients with type 1 diabetes (Group A), type 2 diabetes (Group B) and obesity (Group C) (Fig. 1); we compared the PTS of patients (groups A-C) with the ones of healthy children (Group D) by Western Blotting (WB) using the anti-hIAPP antibody and anti-amyloid oligomer antibody (A11) (Fig. 1). The hIAPP formation of soluble amyloid oligomers is well documented in the literature in vitro and in vivo in animal models 14,26,41,42,56,[60][61][62][63] . Recently, Bram and coworkers showed in a small sample of patients that the natural auto-antibodies of patients with diabetes recognized hIAPP synthetic oligomers in vitro 34 .
Densitometric quantification of each of the oligomerization states labeled with anti-hIAPP and anti-amyloid oligomer showed different oligomerization-profiles. There is significative variation between obesity, type 1 diabetes and type 2 diabetes and the healthy children (Fig. 1B,D). The low molecular weights are predominant and Control (n = 5) Obesity (n = 5) T2DM (n = 5) T1DM (n = 5) P  Wide range of sizes of naturally-occurring hIAPP oligomers and fibers observed in sera of children with obesity or DM. The morphology of the RIAO from PTS of sera of pediatrics patients from the study population was analyzed by Transmission Electron Microscopy (TEM) (Fig. 2). In all the samples the presence of small (~1.05 nm), medium (~16.56 nm) and large (~400 nm) oligomers is evident. In the group A we found a large number of small oligomers, while in the group B the large oligomers predominate; group C had many medium oligomers and group D a large number of small and medium oligomers (Fig. 3).
The RIAO from PTS from sera of children form amyloid fibers. In many PTS there are observed small (~55.71 nm), medium (~69.21 nm), and large (~252.31 nm) amyloid fibers building a network (Fig. 2). It is a striking result because it revealed the capacity of hIAPP amyloid oligomers to form fibers, and it is an evidence that the aggregation/oligomerization process continues to be active even in the test tube (Fig. 2). The studied groups show: A: small fibres; B: small fibres and medium fibres C: very long fibres located around the oligomers, D small fibres (Fig. 3).
The RIAO and fibres are recognized by anti-hIAPP and anti-oligomers antibodies in the ultrastructural immunolocalization studies. Amylin and oligomers antigens recognized by monoclonal www.nature.com/scientificreports www.nature.com/scientificreports/ antibodies (mAbs), anti-amylin, and A11 (anti-oligomers) were immunolocalized by TEM. Both mAbs were localized on fibres or small clusters of fibres (Figs. 4 and 5) The loss of oligomers is due to the washing process; the fibers adhere better to the formvar membrane. In the immunolocalization of anti A-11 ( Fig. 5) we can see a greater amount of gold particle deposits than in Fig. 4. In group A we can see dispersed granules. In group B there are gold clusters on the fibres. Group C and D is very similar to group A, the gold particles are found on small fibre clusters.
The aggregation of RIAO from samples is fast and delays the process of fiber formation of synthetic hiApp. We explored the kinetics of aggregation of the RIAO samples, and we found that they aggregated quickly. In Fig. 6A we observed an immediate augmentation in the intensity of the Thioflavin T (ThT) fluorescence that implies that the sample is already aggregated 64,65 . This behaviour is concentration dependent. Then we tested the seeding effects of those oligomers from several samples in the synthetic hIAPP aggregation kinetics (Fig. 6B). We demonstrated that the oligomers from groups A and B delayed the process of fibrillization of the hIAPP (Fig. 6B). They showed a significant increase in the lag phase compared to synthetic hIAPP alone (t-lag 22.8 min to 77.8 min for group A, and t-lag 22.8 min to 97.8 min for group B), thus delaying the toxic effect of the oligomers. Whereas the oligomers from group D showed a smaller delay of the aggregation process (Fig. 6B).
Exploring the biomolecular assembly of RIAO. We decided to study the hIAPP oligomers from PTS of sera of children with obesity and diabetes from the protein folding perspective. The aggregation/oligomerization/fibril formation process usually implies conformational changes from monomer native structure to the www.nature.com/scientificreports www.nature.com/scientificreports/ β-sheet-rich structures of amyloid polymorphism. Furthermore, we explored the biomolecular assembly of hIAPP by far-UV Circular Dichroism spectroscopy (CD).
The CD of freshly and old PTS from each group showed diversity ( Fig. 7(A-D)). We followed the development of a β-structure signature and the decrease of α-structure signature. In Fig. 7(A-D) we compare representative samples of each group. We used synthetic hIAPP oligomers as reference molecules. The secondary structure percentages obtained for synthetic hIAPP were 35% turn/sheet, >50% random coil, and 10% helix, which are consistent with the reported in the literature 63,66,67 . The synthetic hIAPP oligomers have >88% turn/sheet, 10% random coil and <8% helical structure 63 .
In the features of CD of secondary structure elements from the samples of all groups A-D (Fig. 7) we found a positive peak at ~190-195 nm and two negative peaks at 208 and 222 nm corresponding a signal of α-helix; also a positive peak ~195-200 nm, and a negative peak ~215-220 nm corresponding with a signal of β-sheet. In general, the CD signal at 216-218 nm indicates the sheet content.
For the representative spectrum of the group A samples in Fig. 7A, we showed that the CD signals at 195 nm, 208 nm, and 222 nm were about 30% higher in the fresh sample. This means that the secondary structure variation on group A samples (expressed in Δ CD signals of old sample -fresh sample in %) is negative for α-helix at 195 nm. The negative signs stand for a decrease in the corresponding structure as described by Juárez. On the other hand, the β-sheet signal at 216 nm increased, which implies that the corresponding structure is increasing. Group B samples display CD signals four times higher at 195 nm, 208 nm, 218 nm and 222 nm in the fresh samples (Fig. 7B). The ΔCD signal is negative at 195 nm and positive at 218 and 222 nm. The positive signs stand for an increase in the β-sheet structure and the unordered structure. Group C samples do not display differences in the CD signals (Fig. 7C), whereas group D showed CD signals around three times higher at 195 nm, 208 nm, 218 nm and 222 nm in the fresh sample (Fig. 7D). The ΔCD signal is positive at 208 nm, 218 nm and 222 nm.
Within the samples from groups A-D analyzed here, the content of α-helical structure is negative and larger for group B, meaning a loss of the structure. The average R1 parameter (ratio 195/208) that characterizes the individual helices is bigger in the old sample of group D, and is ordered: D > A > C > B. The ratio 222/208 nm stands for the presence of single helices, and in all the samples it was about 0.9, indicating a lower inter-helix interaction. Meanwhile, the increase of β-sheet structure and the unordered structure was ordered: B > D > A > C. There  www.nature.com/scientificreports www.nature.com/scientificreports/ was a loss in α-helix structure and formation of both β-sheet and unordered conformation. Regarding the hIAPP homo-oligomers, these findings are well appreciated when we obtained the fraction of secondary structure using the CDSSTR method 68 (Supplementary Fig. 1).
the RiAo from sera are homo and hetero-oligomers. Eleven PTS of sera from patients and controls were analyzed by Size Exclusion Chromatography (SEC) 69 . The samples were in absence and presence of 3 M guanidine hydrochloride (Gn-HCl); insoluble aggregates were removed by centrifugation (Fig. 8(A-D)).
The samples in absence of GuHCl eluted at several peaks, and they were very similar to each other from one sample to another (Elution volume of ~5.8 ml-peak 1; ~7.8 ml-peak 2; and ~10.86 ml-peak 3) in the chromatogram of SEC column (Fig. 8A). The size of the peak is what mostly varies between each sample. This indicates that in some samples there is significantly more of the aggregate species corresponding to that peak than in others. In the first peak (~232 kDa) the maximal area is for the samples from group A, and the minimal area for the samples www.nature.com/scientificreports www.nature.com/scientificreports/ of group D (Fig. 8). It is important to point out that this peak displays strong tyrosine (Tyr) and tryptophane (Trp) signals in the samples of group A, and fair in the samples of group D and the synthetic oligomers. A striking result is that the third peak (~35 kDa) did not display Trp signal. www.nature.com/scientificreports www.nature.com/scientificreports/ We compared each group peaks individually by area and by elution time. The results were as shown in Fig. 8. We can observe on the graphs comparing the different peaks' area percentage at 210 nm that there are greater differences in area between the groups.
Fractionating the PTS with 3 M Gn-HCl yields few elution peaks. The first peak corresponding to the ~208 kDa aggregates disappears, whereas the other elution peaks (elution volume of 6.5 ml, 8 ml and 11.6 ml) correspond to medium and low molecular weight aggregates. The peak corresponding to ~35 kDa is higher taking into account that the concentration of the sample was half of the concentration in absence of Gn-HCl (Fig. 8). We demonstrated that the hIAPP oligomerization is a process of co-aggregations forming homo and hetero-oligomers, since there are peaks with Trp signal and we consider that the hIAPP's do not have Trp 63 .
Furthermore, we analyzed a sample of a patient from group A by mass spectroscopy (nano electro-spray translational impact. The translational process of this study starts with a changing paradigm of seeing DM as PCD 30,70 . The discovery of the amyloid oligomers in the STZ-induced diabetes in rats allows the identification of rIAPP oligomers as biomarkers 26 . This new knowledge leads to the search for markers in humans to be able to give short-time diagnostic results, which is the way to overcome the biomarker dilemma 9 . It also allows to study for the first time RIAO from sera of pediatric patients with obesity or DM and shift the paradigm that the conformational disease are linked to aging 2,71 . This opens the window to clinical confirmation that the hIAPP oligomers are in the sera and can be isolated, stabilized, and performed the initial biophysical characterization of them; the next step is the assay development to identify them in sera in a patient population. The accompanied paper is the continuation of the translational process.

Discussion
We begin to unpack the ultrastructural morphology, aggregation/oligomerization process and protein assembly of RIAO between the groups of patients studied. We demonstrated the presence of homo and hetero-RIAOs in the sera of children and adolescents by WB, TEM and biophysical studies (Figs. 1-8). We found that the aggregation/oligomerization process is active in the sera; the main aggregates are dodecamers, hexamers, and trimers in the case of homo-RIAO 26,72 (Fig. 1), and for the hetero-RIAO we identified by mass spectroscopy the main co-aggregation proteins: serum albumin, immunoglobulins and haptoglobin, which are similar to the findings of amyloid interactome in vitro using plasm 73 . This work offered crucial information regarding human homo and hetero-oligomers of hIAPP, they are recognized by the anti-amyloid oligomers antibody (A11) that not bind to native proteins, monomers or mature amyloid fibers 41,74 . Furthermore, they are small oligomers (trimers and hexamers as shown in Fig. 1) that several research groups demonstrated that as the sizes of the oligomeric assembly decreases, its toxicity and deleterious membrane effects increases 28,29,32,48,56,75 . The striking results that the densitometry of WB from the samples of healthy children has less amount of small oligomers confirm this reasoning (Fig. 1B,D). The amyloid oligomers can form fibers and these fibers react with anti-hIAPP and anti-amyloid oligomers antibodies which imply the existence of self-catalysis and the creation of cytotoxic oligomers as consequence of fiber formation (Figs. 1-5).
We tracked the fibrillization process and showed that it happens very fast for the patients' samples and it is a dynamic process that continues in the test tube (Fig. 6A), and contrary to what was expected, they acted almost like chaperones in the presence of synthetic hIAPP, slowing down the process of aggregation despite being unable to cross-seeding the process 72 (Fig. 6B). A further and more detailed analysis will be necessary to explain this behavior.
The CD studies demonstrated that the homo and hetero oligomerization-fibrillization is a dynamic process with large structural changes and there are differences in the spectra of the RIAOs from one group to another. And last but not least, from the results of the SEC presented in Fig. 8 we can also obtain some interesting conclusions. The sample from the type 1 diabetes (A) was one of the most uniform in the elution times of their peaks, and these peaks were in general bigger in area than the other groups (A > C > B > D). In the samples from obesity (C) and the diabetes type 2 (B) the elution times of peaks of the different samples are more scattered, which suggests that they have more varied compositions (Fig. 8B-C).
Integrating all results we are able to get a more complete picture of the amyloid formation. The control group D in general has the less oligomers and fibers in the TEM and WB experiments (Figs. [1][2][3][4][5], and the peaks and the areas from the SEC experiments are smaller (Fig. 8). Furthermore, the conformational changes with time are smaller than in the patients' groups (Fig. 7). Each different experiment is like a blind man feeling different a part of an elephant, telling us important information about different aspects of this complex phenomenon. A homework for all the protein scientists is to discover and identify with which other biomolecules the hIAPP co-aggregates, the co-aggregation network and the co-aggregation pathway.
By using real patients' samples, we obtain novel and relevant results that complement the research using synthetic proteins 2 www.nature.com/scientificreports www.nature.com/scientificreports/ process in vivo, so that we can advance more quickly towards the finding of potential novel diagnostic tools, prevention strategies, and more effective treatments of this fastly-growing and harrowing disease 27,40,64,72,76,79,80,85,86 .

Experimental Section
Study design. A cross-sectional, analytical, ambispective, blinded pilot study was carried out.

Study participants. Children and adolescents were recruited from two main Pediatric Hospitals in Mexico:
Instituto Nacional de Pediatría (INP) and UMAE-Hospital de Pediatría, Centro Médico Nacional Siglo XXI, Instituto Mexicano del Seguro Social (IMSS). Informed consent was obtained from all participants. The study was conducted with the approval of the Ethics and Research Committees of IMSS and the INP.
All the subjects underwent a complete physical examination that included weight measurement, waist circumference, and blood pressure according to the established standards. In addition a blood sample (after 10 hours of fasting) was obtained and stored at −80 degrees.
The participants were divided in 4 groups: Control Group (D): 5 healthy adolescents (without apparent acute or chronic disease).
Diabetes Mellitus groups: 5 adolescents with type I DM (Group A) and 5 adolescents with type 2 DM (Group B).
isolation of amyloid oligomers from sera: "Universal pre-treatment of samples". 1000  Western blot. Total protein concentration of oligomers from pre-treated samples (PTS) of the sera of either control or patients was determined by BCA assay using as standard curve BSA and hIAPP 26 . The concentration of the samples was normalized to 10 μg. The membranes were probed overnight at 4 °C with the purified anti-amyloid oligomers antibody ab126892 (Abcam, 1:1,000) or anti-amylin antibody (Santa Cruz Biotechnology, 1:200) diluted in phosphate-buffered saline with Tween-20 (PBS-T). To control the charge, we performed immunoblotting with Anti-Transferrin (Santa Cruz Biotechnology sc-30159 1:500). The membrane activity was detected by substrate chemiluminescence (Immobilion Chemiluminescence HRP Substrate 1:1) and revealed by Licor C-digit. The intensity of proteins bands was quantified by Image Studio Lite by scanning densitometry. Data was managed with MS Excel, while statistics and graphs were obtained with SAS JMP 9 statistical package.
Transmission electron microscopy (TEM). Negative staining. The PTS containing the amyloid oligomers were sonicated in ice (5 min) and applied (5 μL) to 300-mesh formvar/copper grids and left for 5 minutes and then the excess solution was removed. Grids were stained with 3% filtered uranyl acetate for 1 min. After air drying, grids were examined with an electron microscope JEM-1010, (JEOL,Japan). The images were taken by a CCD Gatan Orius SC600 and Digital Micrograph software.
Ultrastructural immunolocalization. For immunogold labelling 6 μL pre-treated sample were sonicated in ice for 5 minutes and applied onto 200-mesh formvar/gold grids for 10 minutes and then the excess solution was removed with filter paper. Without letting the samples dry, they were floated over the primary corresponding antibody (anti-Amylin or A11) in a PBS 1X solution 1:100 and left all night at 4 °C in a humid chamber. Afterwards, the grids were washed with PBS 1X and floated over the secondary antibody connected to 20 nm gold particles for the antibody against Amylin and 12 nm for the antibody against A11 (Jackson ImmunoResearch) in solution 1:100 for an hour at room temperature in a humid chamber. Then the grids were washed with water by flotation and left to dry to be contrasted with uranyl acetate at 3% for 30 seconds. The grids were observed under a JEM-1010 (JEOL Japan) electronic microscope and the image capture was carried out with a CCD camera model Gatan Orius SC600 and the digital micrograph software. CD spectroscopy. CD spectra were obtained by using a Jasco J-815 spectrometer (Tokio, Japan). PTS from controls (4) and patients (12) were analyzed by using PBS buffer as a blank. Five iterations of each spectra were recorded in the range of 190 to 260 nm; data interval of 0.5 nm, with a scanning speed of 50 nm/min, and a bandwidth of 1 nm; with protein concentrations of 0.1 mg/mL in PBS buffer (pH 7.4) and 0.1 cm flow-cell at Room Temperature (25 °C), coupled to the ASU-autosampler accessory. Results were analyzed in DichroWeb to obtain the fraction of secondary structure using CDSSTR and K2D methods 68 . fluorescence. Kinetics of aggregation-oligomerization. The aggregation process was monitored by ThT Fluorescence, Tyr and Trp fluorescence, and turbidity using a M1000 Tecan (Austria). Briefly, the pre-treated sera of patients and healthy controls were added into prewarmed ThT-buffer at 37 °C (PBS, pH 7.4 + 20 μM ThT). For synthetic hIAPP, the 1 mg/ml of peptide was dissolved in DMSO 64 . All of the experiments were performed in triplicate.