Methods
Angio PRP isolation and characterization.
We designed a sterile and closed class IIa device (NovySep), characterized by a collecting tube with an inert porous membrane of high-grade polyethylene, a rubber stopper to insert peripheral blood with a 2.5 ml syringe needle (21G) and a ring nut to adjust the plasma phase volume above the membrane after centrifugation (Fig. 1A). The device was designed for single use only and to collect blood-derived mononucleated cells and the plasma phase after centrifugation without opening the system. Peripheral blood was collected from healthy volunteers (n = 101) of the blood bank of Department of Transfusion Medicine and Haematology at Policlinico Hospital of Milan, after informed consent and according to the guidelines approved by the Ethics Committee on the Use of Human Subjects in Research of the Policlinico Hospital of Milan (Milan, Italy, Ethics Committee permission number 793/13). 2.5 ml of peripheral blood, collected in sodium citrate tube were filled into NovySep device and centrifuged at 1500 rpm for 10 minutes to induce the phase separation (EP20161201.7). The platelet-rich-plasma phase (PRP) and the cells at the interface between red cells and plasma were collected. We analysed pre-separation blood and AngioPRP by blood Coulter counter instrument (DxH 500, Beckman Coulter). Pre-separation blood and cell phase collected were directly labelled with monoclonal antibodies shown below. Cells were incubated with Syto 16, anti-CD45 V500, anti-CD3 V450, anti-CD3 APC, anti-CD56 PE-CY7, anti-CD14 APC-H7, anti-CD16 PE, anti-CD15 V450, anti-CD19 APC-R700 or anti-CD31 PE Cy7, anti-CD184 APC, anti-CD90 PerCP, anti- CD90 FITC, anti-CD146 PE, anti CD34 APC (BD Biosciences-Pharmingen, San Diego, California, USA). The controls were isotype-matched mouse immunoglobulins. After each incubation performed at 4 °C for 20 minutes, cells were washed in PBS 1X containing 1% heat-inactivated FCS and 0,1% sodium azide. The cytometric analyses were performed on a LYRIC flow cytometer using FACSuite software (BD Biosciences-Immunocytometry System). Each analysis included at least 1-2x104 events for each gate. A light-scatter gate was set up to eliminate cell debris from the analysis. The percentage of positive cells was assessed after correction for the percentage reactive to an isotype control conjugated to a specific fluorophore. Percentage of different cells subpopulations were calculated on the Syto 16 positive gate.
Ex vivo preclinical experimentation: EPC colony forming assay, HUVEC coculture and organotypic skin culture.
The angiogenic potential of AngioPRP was tested in 35-mm dishes using the Endothelial Progenitor Cell Colony-Forming Assay (EPC-CFA) (MethoCult SFBIT; STEMCELL Technologies Inc.) added with proangiogenic growth factors/ cytokines, as previously reported [39] (rh SCF 100 ng/ml, rh VEGF 50 ng/ml, rh b-FGF 50 ng/ml, rh EGF 50 ng/ml and rh IGF-1 ng/ml, all from Miltenyi Biotec; eparin 2 U/ml, STEMCELL Technologies Inc). Aliquots of AngioPRP were seeded at a cell density of 5x104 cells/dish (3 dishes per volunteer). 16 to 18 days after the beginning of the culture, the number of adherent EPC colonies per dish was counted under phase contrast light microscopy LEICA DMi8 (Leica, Germany). Primitive EPC colony-forming units (pEPC-CFUs) and definitive EPC-CFUs (dEPC-CFUs) were separately counted and expressed as a percentage of the total number. Pro-angiogenic potential of AngioPRP was evaluated in coculture system constructed using human umbilical vein endothelial cells (HUVECs) as previously described.[40] Briefly, 8x104 HUVEC (ATCC-LGC, VA, USA) were plated on 3D Matrigel (BD Biosciences-Pharmingen, San Diego, California, USA); 1.35x105 cells and 9.55x106 platelets for AngioPRP or 9.55x106 platelets for PRP were added to HUVEC culture; after 24 hours cells ramification was quantified by ImageJ software (NIH) [41]. To investigate the skin regeneration potential of AngioPRP, we used a multilayered model of human dermis and epidermis as previously described (MatTek’s EpiDermFT Full Thickness EFT-400) [42]. Epidermal-only wounds were induced using a sterile 5 mm dermal biopsy punch (Miltex Inc.,York, PA) and the epidermis was mechanically removed using forceps. After wounding, EpiDermFT tissues were cultured into 6-well plate with four different culture conditions: 1) an organotypic skin culture with 3.5x104 cells and 9.85x106 platelets for complete AngioPRP, 2) 3.5x104 cells of Angiocells suspended in PBS, 3) 9.85x106 platelets for PRP, 4) 50 µl of PBS 1X (as negative control) and analyzed after 24 hours, 2, 4, 5, 6 and 7 days of culture. Blood from 10 healthy volunteers was collected in NovySep device to obtain 10 individual AngioPRP as described above. The isolated 10 AngioPRP were pooled and further centrifuged at 1500 rpm for 10 minutes to obtain a pellet of cells which was suspended in PBS (Angiocells). The supernatant containing platelets rich plasma (PRP) was used as such. Wound closure was calculated via equation:
Wound healing (%) = (1-Open wound area / Initial wound area) x 100
In vivo wound healing experiments.
Five-month-old severe combined immunodeficient (NOD.Cg-PrkdcScid/J) [43] mice were obtained from Charles River Laboratories International, Inc. (Calco, Italy); the use of animals in this study was authorized by the National Ministry of Health (authorization number 51/2018-PR). All experimental protocols were reviewed and approved by the University of Torino’s animal ethics research committee. The methods described below were carried out in accordance with those approved protocols, as well as the Italians ethical guidelines regarding the use of experimental animals. Wound healing model was obtained as described in Dunn and colleagues [44]. Briefly, animals were anesthetized with avertin and two full-thickness excisions of 5-mm that include the panniculus carnosus were created on the dorsum, one on each side of the midline of the mouse. A silicone splint was placed around the wound with the assistance of adhesive and the splint was then secured with interrupted sutures. Each mouse acts as its own control, with one wound receiving treatment (AngioPRP or Hyalomatrix, Anika Therapeutics Inc., Bedford, MA 01730, USA) and the other phosphate buffer saline (PBS 1X). A transparent occlusive dressing was applied to prevent contamination. Wounds were checked by taking photos every 2–3 days, and the area was quantified relative to a millimeter reference using ImageJ software (NIH) and expressed as the percentage of wound area measured at day 0, 4, 7, 10, 15 and 21 days after injury, corresponding to wound closure; mice were sacrificed by cervical dislocation under full deep anesthesia and the back skin lesions were removed; the biopsies have been divided into two group respectively for histological or proteomic analysis. One group was placed in isopentane and freezed at -80 °C for proteomic analysis. The other group was incubated in 4% paraformaldehyde in PBS at 4 °C overnight and after transferred to 30% sucrose in PBS 1X solution for a further 24 hours at 4 °C, embedded in O.C.T matrix and freezed at -80 °C. Serial sections of 12 µm thickness were cut and examined by immunofluorescence and histological analysis.
Histological and immunofluorescence stainings.
Serial section of 12 µm of skin tissue and organotypic skin were cut and stained with hematoxylin and eosin (H&E, Bio Optica Spa, Italy) Orcein (Sigma-Aldrich Inc., St. Louis, MO, USA) and Masson’s trichrome staining (Bio Optica Spa, Italy), according to the manufacturer’s instructions for morphological assessment. Images were captured with LMD6000B (Leica, Germany) at 12 regular intervals, representing the entire section and the epidermal thickness was quantified as area per interval using ImageJ software (http://rsbweb.nih.gov/ij/). For immunofluorescence analysis, transversal tissue sections were incubated with mouse monoclonal antibody anti-Citokeratin 10 (1:100, ab9025, Abcam, UK), rabbit monoclonal antibody anti-Vimentin (1:100, ab16700, Abcam, UK), rabbit polyclonal antibody anti-Involucrin (1:100, ab53112,Abcam, UK), mouse monoclonal antibody anti-Cytokeratin 14 (1:100, ab7800 Abcam, UK) rabbit polyclonal antibody anti-Cytokeratin 5 (1:100, ab53121, Abcam, UK), rabbit polyclonal antibody anti-β-Catenin (1:200, ab16051, Abcam, UK), rabbit polyclonal antibody anti-CD206 (1:200, ab64693, Abcam, UK), Alexa Fluor 594 rat monoclonal antibody anti-Ly-6G/Ly-6C (Gr-1)(1:50, 108448 BioLegend), rat monoclonal antibody anti-CD31 (1:50, 550274 BD Biosciences-Pharmingen, San Diego, California, USA), rabbit polyclonal antibody anti-Collagen VI (1:250, ab6588, Abcam, UK), mouse monoclonal antibody anti-alpha SMA (1:50, A2547, Sigma-Aldrich Inc., St. Louis, MO, USA), rabbit polyclonal antibody anti-VE-Cadherin (1:50, ab33168, Abcam, UK), mouse monoclonal antibody anti-eNOS (1:100, ab76198, Abcam, UK), rat monoclonal anti-E-Cadherin (1:100, ab11512, Abcam, UK), mouse monoclonal anti-Cytokeratin 10 (1:100, ab9025, Abcam, UK), rabbit polyclonal anti-Loricrin E-Cadherin (1:100, ab85679, Abcam, UK). Cell nuclei were stained for 5 min at room temperature with DAPI (Sigma-Aldrich Inc., St. Louis, MO, USA). Slides were analyzed using a fluorescent microscope LEICA DMi8 (Leica, Germany), images were captured at regular intervals along the entire section and fluorescence intensity per single interval was quantified with Image J software (http://rsbweb.nih.gov/ij/); integrated density was measured using a ROI corresponding to epidermal region in each slice interval and plotted in the graphic after subtracting the corresponding background signal measured within the tissue-free area [45].
Strength Measurements.
Following sacrifice, the skins for mechanical testing were placed in metal screw clamps with rubber pieces covering the clamped ends. Clamps were placed into a Bose Electroforce 3100 instrument. Applying an initial traction of 0.15 N, the traction measured in MPa was increased by 0.2% per second up to the breaking point. Force (N) and displacement (mm) were measured on a xy plotter and these points were subsequently recorded as stress (σ = force per cross-sectional area) and strain (ε = change in length/initial length) and re-plotted in Excel [46].
Western Blot analysis.
WB analysis were performed as previously described [47]. Briefly, total proteins from skin tissue of five-month-old severe combined immunodeficient (NOD.Cg-PrkdcScid/J) mice were extracted and quantified with Bradford Assay. Samples were resolved on polyacrylamide gels (ranging from 10–14%) and transferred to nitrocellulose membranes (Bio-Rad Laboratories, California, USA). Filters were incubated overnight with following antibodies: Vinculin (1:600, MA5-11690 Thermo Fisher Scientific, MA, USA ); TGFβ (Rb pAb 1:600, ab92486, Abcam, UK); β-Catenin (Rb pAb 1:600, ab16051, Abcam, UK); Membranes were incubated with primary antibodies ON at 4 °C, then followed by washing, detection with horseradish peroxidase (HRP)-conjugated secondary antibodies (Dako Agilent, CA, USA) and developed by enhanced chemiluminescence (ECL) (Amersham Biosciences, USA). Bands were visualized using an Odyssey Infrared Imaging System (Li-COR Biosciences, USA). Densitometric analysis was performed using ImageJ software (http://rsbweb.nih.gov/ij/).
Proteomics analysis.
In-solution digestion
For proteomic analysis the epidermal and dermal layers of the treated skin enclosed by the silicone splint were removed 21 days after injury and frozen in isopentane. Samples were then suspended in 200 µL 0.1 M NH4HCO3 pH 7.9 buffer and homogenized in ice. The protein concentration was assayed using SPN-Protein assay kit (G-Biosciences, St. Louis, MO, USA) and the membrane proteins were solubilized by adding Rapigest SF reagent (Waters Co, Milford, MA, USA) at the final concentration of 0.2% (w/v). The resulting suspensions were incubated under stirring at 100 °C for 20 minutes and at 80 °C for 2 hours. The digestion was carried out on 50 ± 0.5 µg proteins of each sample by adding Sequencing Grade Modified Trypsin (Promega Inc., Madison, WI, USA) at an enzyme/substrate ratio of 1:50 (w/w) overnight at 37 °C in 0.1M NH4HCO3 pH 7.9 buffer with 10% CH3CN. An additional aliquot of 0,5 µg of trypsin (1:100 w/w) was added in the morning, and the digestion continued for 4 hours. Moreover, the addition of 0.5% Trifluoroacetic acid (TFA) (Sigma-Aldrich Inc., St Louis, MO, USA) stopped the enzymatic reaction, and a subsequent incubation at 37 °C for 45 min completed the RapiGest acid hydrolysis [48]. The water immiscible degradation products were removed by centrifugation at 13.000 rpm for 10 minutes. Finally, the tryptic digest mixtures were desalted using Pierce C-18 spin columns (Thermo Fisher Scientific - Pierce Biotechnology, Rockford, Il, USA), according to manufacturer protocol and were resuspended in 0.1% formic acid (Sigma-Aldrich Inc., St. Louis, MO, USA) in water (LC-MS Ultra CHROMASOLV, Honeywell Riedel-de Haen, Muskegon, MI, USA).
LC-MS/MS
Analysis were performed as previously described [49]. Briefly, trypsin digested mixtures were analyzed using Eksigent nanoLC-Ultra 2D System (Eksigent, part of AB SCIEX Dublin, CA, USA) combined with cHiPLC-nanoflex system (Eksigent) in trap-elute mode. Briefly, samples (0.8 µg injected) were first loaded on the cHiPLC trap (200 µm x 500 µm ChromXP C18-CL, 3 µm, 120 Å) and washed with the loading pump running in isocratic mode with 0.1% formic acid in water for 10 minutes at a flow of 3 µL/min. The automatic switching of cHiPLC ten-port valve then eluted the trapped mixture on a nano cHiPLC column (75 µm x 15 cm ChromXP C18-CL, 3 µm, 120 Å) through a 87 minutes gradient of eluent B (eluent A, 0.1% formic acid in water; eluent B, 0.1% formic acid in acetonitrile) at a flow rate of 300 nL/min. In depth, gradient was: from 5–10% B in 3 min, 10–40% B in 80 min, 40–95% B in 17 min and holding at 95% B for 7 min. Trap and column were maintained at 35 °C for retention time stability. Mass spectra were acquired using a QExactive mass spectrometer (Thermo Fisher Scientific, San Josè, CA, USA), equipped with an EASY-Spray ion source (Thermo Fisher Scientific, San Josè, CA, USA). Easy spray was achieved using an EASY-Spray Emitter (Dionex Benelux BV, Amsterdam, The Netherlands) (nanoflow 7 µm ID Transfer Line 20 µm x 50 cm) held to 1.9 kV, while the ion transfer capillary was held at 220 °C. Full mass spectra were recorded in positive ion mode over a 400–1600 m/z range and with a resolution setting of 70000 FWHM (@ m/z 200) with 1 microscan per second. Each full scan was followed by 10 MS/MS events, acquired at a resolution of 17,500 FWHM, sequentially generated in a data dependent manner on the top ten most abundant isotope patterns with charge ≥ 2, selected with an isolation window of 2 m/z from the survey scan, fragmented by higher energy collisional dissociation (HCD) with normalized collision energies of 30 and dynamically excluded for 10 sec. The maximum ion injection times for the survey scan and the MS/MS scans were 100 and 200 ms and the ion target value were set to 106 and 105, respectively.
Data Analysis
All data generated were searched using the Sequest HT search engine contained in the Thermo Scientific Proteome Discoverer software, version 2.1. The experimental MS/MS spectra were correlated to tryptic peptide sequences by comparison with the theoretical mass spectra obtained by in silico digestion of the Uniprot Mus musculus proteome database (54109 entries), downloaded in February 2019 (www.uniprot.org). The following criteria were used for the identification of peptide sequences and related proteins: trypsin as enzyme, three missed cleavages per peptide, mass tolerances of ± 10 ppm for precursor ions and ± 0.6 Da for fragment ions. Percolator node was used with a target-decoy strategy to give a final false discovery rates (FDR) at Peptide Spectrum Match (PSM) level of 0.01 (strict) based on q-values, considering maximum deltaCN of 0.05 [50]. Only peptides with high confidence, minimum peptide length of six amino acids, and rank 1 were considered. Protein grouping and strict parsimony principle were applied.
Label-free differential analysis
In order to improve the identification of differentially expressed proteins, two different and complementary label-free approaches were adopted: an in-house algorithm, Multidimensional Algorithm Protein Map (MAProMa)[51] and Linear Discriminant Analysis (LDA). The protein lists obtained from the SEQUEST algorithm were aligned and compared by means of the average spectral counts (aSpC), corresponding to the average of all the spectra identified for a protein and, consequently, to its relative abundance, in each analyzed condition (Healthy, PBS control, AngioPRP, Hyalomatrix). In depth, to select differentially expressed proteins, subgroups were pairwise compared, applying a threshold of 0.2 and 5 on the two MAProMa indexes DAve (Differential Average) and DCI (Differential Confidence Index), respectively. DAve, which evaluates changes in protein expression, was defined as (X-Y)/ (X + Y)/0.5, while DCI, that evaluates the confidence of differential expression, was defined as (X + Y) x (X-Y)/2. The X and Y terms represent the SpC of a given protein in two compared samples.
Linear Discriminant Analysis and Hierarchical Clustering
The average SpC (aSpC) values of the identified proteins were calculated by MAProMa software. The spectral count (SpC) values were normalized using a total signal normalization method and compared using a label-free quantification approach, as previously reported.[52] In detail, the considered protein lists (D, n = 6; AC, n = 6; BC, n = 6; A, n = 6; B, n = 6) were first processed by linear discriminant analysis (LDA) and proteins with the largest (≥ 4) F ratio and smallest P-value (≤ 0.01) were retained and considered differentially expressed with high confidence. Proteins selected by LDA were processed by hierarchical clustering applying Ward’s method and a Euclidean distance metric using JMP15 software.
Proteome remodeling index
Proteome Recovery and Proteome Activation Index For each protein it was evaluated its variations (modules) due to perturbation (Healthy, D, vs. Disease, C(B)) and/or treatment (Disease, C(B) vs. treatment, A); if the two modules present similar values but opposite sign, the protein recovers its level to the reference (in our case Healthy condition). The extraction of so-called “recovered proteins” was simplified by calculating the Proteome Remodeling index (PRi) for each protein using an unbiased procedure. The PRi is calculated by the following formula:
PRi = M(b-p)J/M(p-t)J
where J: each identified protein;
M(b-p)J: Perturbation Module = Difference Healthy vs. Disease
(b: baseline; p: perturbed);
differential abundance of specific protein J comparing healthy (D) vs. Disease (C(B)) conditions.
M(p-t)J: Treatment Module = Difference Disease vs. treatment
(t: treated, A)
differential abundance of specific protein J comparing perturbed (disease) vs. treated (A) conditions.
Theoretically, if protein level after treatment remodeled to reference (healthy) the two modules (perturbation and treatment) should have similar value and opposite sign; then the PRi will be negative and close to unit (-1). In our case, we considered proteins with a PRi in the range − 0.5 to -2. Differential abundance of each module may be expressed as DAVE value, from MAProMa algorithm or ln[fold change]). However, because in some cases the fold-change calculation returns non-sense values (such as n/0 or 0/n), we preferred to use DAVE value to calculate modules. Moreover, MAProMa platform permits the filtration by absolute variation, using DCI algorithm, excluding very low expressed protein (very low spectral count, multiple DCI < |5|) confusingly with noise [49, 53].
Network analysis
A protein-protein interaction (PPI) network was built by combining differentially expressed proteins and the Mus Musculus PPI network retrieved from STRING database; only experimentally and database defined PPI with a score > 0.15 were considered. The resulting sub-networks were visualized and analyzed by Cytoscape and its plugins, as previously reported [54]. Specifically, BINGO 2.44 Cytoscape plugin[55] was used for evaluating the most represented GO terms; Mus musculus organism, hypergeometric test, Benjamini–Hochberg FDR correction and a significance level ≤ 0.01 were applied.
Statistics
Sample size was determined considering a statistical test power of 0.80 and an alpha value of 0.05. Results indicated that a sample size of 15 animals (n = 5 per group) would enable to detect a minimum difference in protein expression of 0.35 with an expected standard deviation of 0.15. To detect outliers, Grubb’s test was applied for each parameter. A probability value < 0.05 was considered significant. All analyses were performed as previously described [49] using Sigma Stat 11.0 dedicated software (Systat Software Inc., San Jose, CA, USA). Identified proteins were evaluated by LDA (JMP15 software SAS; F ratio > 4 and a p-value < 0.01) and MAProMa platforms. Finally, proteins extracted by PRi algorithms were statistically evaluated by ANOVA and Tukey’s test.