Deep Immunophenotyping of Human Whole Blood by Standardized Multi-parametric Flow Cytometry Analyses

Immunophenotyping is proving crucial to understanding the role of the immune system in health and disease. High-throughput flow cytometry has been used extensively to reveal changes in immune cell composition and function at the single-cell level. Here, we describe six optimized 11-color flow cytometry panels for deep immunophenotyping of human whole blood. A total of 51 surface antibodies, which are readily available and validated, were selected to identify the key immune cell populations and evaluate their functional state in a single assay. The gating strategies for effective flow cytometry data analysis are included in the protocol. To ensure data reproducibility, we provide detailed procedures in three parts, including (1) instrument characterization and detector gain optimization, (2) antibody titration and sample staining, and (3) data acquisition and quality checks. This standardized approach has been applied to a variety of donors for a better understanding of the complexity of the human immune system. Supplementary Information The online version contains supplementary material available at 10.1007/s43657-022-00092-9.


DCs
Dendritic

Introduction
The human immune system is a complex network of molecules, cells, and tissues that provide effective host defense. Due to its plasticity, the immune system is highly variable between individuals (Brodin and Davis 2017;Liston et al. 2016Liston et al. , 2021. A major aim of phenomics is to quantitatively measure these multiscale networks. Immunophenotyping is essential for analyzing the components and functions of the immune system. The standardized deep immunophenotyping approach will provide an opportunity for longitudinal monitoring of human immune status (Hartmann et al. 2019).
Multi-parametric flow cytometry is a rapidly developing technology that stands as one of the most important analytical tools in the field of immunology (Delmonte and Fleisher 2019). It allows the simultaneous identification and quantification of distinct immune cell subsets at a single-cell level. New antibodies, new fluorochromes, and high-performing flow cytometers are expanding the possibilities for the identification and phenotypic characterization of specific cell populations. However, the increased complexity of immunophenotypic approaches requires optimized antibody panels and fully standardized procedures (Maecker et al. 2010;Maecker et al. 2012).

Selection of Markers and Clones
The design of reproducible antibody panels to produce optimal resolution data for multi-parametric flow cytometry is laborious and time-consuming. As the first step in panel design, it is necessary to clarify the expression level (low to high), expression pattern (bimodal, continuum), and co-expression patterns of the target antigen. Typically, target antigens can be divided into two groups: lineage markers, and function markers. The lineage markers often have known expression patterns, which are used to delineate the target immune cell populations. The function markers are related to the process of cell biology such as cell proliferation, differentiation, activation and exhaustion with unknown expression patterns. An important issue for consideration is the selection of antibody clones. Different antibody clones against the same target antigen have distinct staining patterns from one another . There are several resources available to help choose commonly used, validated antibody clones such as the series of optimized multicolor immunofluorescence panels (OMIPs) publications (Wang and Creusot 2021) and the OMIP database (https:// www. omipc ollec tion. com).

Choice of Fluorochromes
Fluorochrome selection is an essential step in designing multicolor immunofluorescence panels. Different fluorochromes display a wide range of brightness scales. In general, bright fluorochromes (e.g., PE, BV421) should be reserved for function markers with low expression level, or unknown expression patterns. They can also be chosen for the continuum markers that require clear discrimination between dimly stained and negative cell populations. For example, BV421conjugated CD14 antibody stained brightly and provided Fig. 1 Considerations of antibody clones and fluorochromes during panel development. a-d Comparison of different antibody clones for chemokine receptors detection on T cells. Plots showed CCR4 expression (a, b) and CCR7 expression (c, d) on CD4 + cells. The CCR4 dim population and CCR7 dim population were clearly separated from the negative population when the CCR4-clone L291H4 and CCR7-clone G043H7 were used. e Brightness of fluorochromes is essential in the discrimination of immune cell subsets. Non-classical monocytes (CD14 low CD16 hi ) could be identified when the bright (BV421) fluorochrome was used. f A representative spillover spreading matrix from an 11-color configuration of the CytoFLEX LX. The color coding was from no spillover spread (white) to high spillover spread (red). The three fluorophores contributing the least spillover spreading were BV421 (V450), FITC (B525), and PerCP/ Cy5.5 (B690). In turn, the three detectors receiving the least spillover spreading are V525, V450, and R660 ◂ 1 3 good separation between intermediate monocytes and nonclassical monocytes (Fig. 1e). While dim fluorochromes (i.e., FITC, PerCP/Cy5.5) can be assigned to lineage markers with high expression level such as CD3 and CD45.

Spillover Spreading Error
Spillover spreading has a critical impact on the quality of high-dimensional fluorescent antibody panels for flow cytometry (Nguyen et al. 2013). The signal from one fluorophore spilling into non-target detectors reduces the sensitivity of detectors. An important strategy to minimize spillover spreading is assigning the weak marker to the channel that receives less spread and assigning the backbone marker to the channel that contributes less spread. Figure 1f shows the spillover spreading matrix (SSM) of all possible combinations in the 11-color space for the CytoFLEX LX flow cytometer (Beckman Coulter).

Standardized Deep Immunophenotyping Workflow
For deep immunophenotyping of human peripheral whole blood, we developed and validated six different 11-color flow cytometry panels (Table 1). This assay characterizes immune cell subsets that circulate in the peripheral blood including all major immune cells such as neutrophils, eosinophils, basophils, monocytes, dendritic cells (DCs), natural killer (NK) cells, T cells, and B cells. It is suited for capturing thousands of immune cell traits including immune cell subset events, frequency, ratio, morphologic properties, and immune cell-surface protein expression levels.
The multi-step procedure must be standardized for flow cytometric immunophenotyping. These steps involve the instrument characterization (Nguyen et al. 2013;Perfetto et al. 2006Perfetto et al. , 2012, detector gain determination (Kalina et al. 2012;Maciorowski et al. 2017), antibody titration, sample staining (Berhanu et al. 2003), and data analysis (Monaco et al. 2016). Here, we describe in detail aspects of the procedures that are crucial for the deep immunophenotyping of human whole blood by multiparametric flow cytometry.

Calibration of Detector Linearity and Gain Range
1. Vortex the 8-peaks RCPs vigorously. Add 2-3 drops of particles to 1 mL of ddH 2 O in a 5 mL 12 × 75 mm polypropylene tube. 2. Check CytoFLEX LX flow cytometer's lasers, mirrors, and filters. Complete daily instrument-specific start-up and quality control (QC) procedures. 3. Adjust the Forward scatter (FSC)/Side scatter (SSC) gain to place the beads on a scale in the light scatter plot and set a gate around the singlet cell population on the FSC-area (FSC-A) vs FSC-height (FSC-H) dot plot to exclude aggregates. 4. Set gains for all detectors to 25 V (Except FSC/SSC channel). 5. Run 8-peaks RCPs and negative control compensation particles (CCPs), respectively. Acquire 5000 events for each kind of particle. 6. Increase gain by 50 V for each detector and repeat Step 5. 7. Export data in FCS3.1 file format and load into FlowJo. 8. Calculate the median fluorescence intensity (MFI) of both the second brightest bead peak (referred to as P7_MFI) and the brightest bead peak (referred to as P8_MFI). Calculate the detector linearity as follows: linearity = P8_MFI−P7_MFI P7_MFI (Fig. S1a).
9. Calculate the robust coefficient of variance (rCV) for the second brightest bead peak (referred to as P7_ rCV). 10. Calculate the MFI of the negative CCPs and defined them as background (referred to as B_MFI). Calculate the signal-to-background ratio (SBR) as follows: SBR = P7_MFI B_MFI (Fig. S1a). 11. For each detector, plot the detector linearity, P7_rCV, and SBR on one graphic. Here the detector gain range can be defined as the gain point showing the highest SBR co-occurring with the lowest rCV and a slope of zero on the linearity curve (Fig. S1b). 12. Repeat this procedure when a new laser, detector, or filter is installed.

Calculation of Minimal Detector Gain
13. Gate the second dimmest peak of the 8-peaks RCPs and calculate the rCV of that peak (referred to as P2_ rCV) in each fluorescence detector for gain ranges as described in the previous section. 14. Generate a plot showing the P2_rCV on the y-axis relative to the gain on the x-axis. When the slope of the curve is zero, the gain is minimal and brings the negative population out of the detector's electronic noise range (Fig. S1c).

CRITICAL
Gain titration to deliver optimal resolution (see supplementary optional procedure subheading Gain Titration (Voltration) Experiment).

Antibody Titration and Sample Staining
CRITICAL 1. Human peripheral whole blood should be processed as soon as possible after collection. 2. Human peripheral whole blood should be transported and stored at RT. 3. For biosafety and to prevent sample contamination, whole blood samples should be processed in a biosafety cabinet.

Antibody Titration
15. Plan on using 120 μL/tube to stain, and a five-point, twofold dilution series. . 38. Generate a plot showing the SI on the y-axis relative to the dilution on the x-axis. The dilution that represents the best SI with the lowest concentration of antibody is the dilution to use (Fig. S2).

Compensation Setup Experiment
39. Vortex the BD CompBeads vigorously. In the meantime, label 5 mL 12 × 75 mm polypropylene tubes for each fluorochrome-conjugated antibody. 40. Pipette 80 μL of stain buffer into each tube and then add 20 μL of CompBeads (containing 10 µL positive control beads, along with 10 µL negative control beads). 41. Add the correct concentration of the antibodies to each tube, which is determined by the titration experiments (see subheading Antibody Titration). 42. Prepare one additional tube as a negative control (without any antibody

CRITICAL
If necessary, prepare three times 100 µL of whole blood stained in parallel and pooled directly before measurement to obtain sufficient cell numbers.

T Cell Subsets
T cells are important mediators in cell-mediated immunity, and numerous different T cell subsets were identified and characterized. To identify T cells the pan marker CD3 was included in panels 2, 3, 4, and 5. CD4 + T cells, CD8 + T cells, the double-negative T cells (CD4 − CD8 − T cells), and double-positive T cells (CD4 + CD8 + T cells) were identified by their differential CD4 and CD8 expression (Fig. 4a). The CD8 + T cells (Fig. 4b) and CD4 + T cells (Fig. 4c) were To determine the activation status of T cells and their subsets (CD4 + T cells, CD8 + T cells, CD4 − CD8 − T cells, and CD4 + CD8 + T cells), the expression of activation markers CD69, CD28, HLA-DR, and CD38 was analyzed in Panel 4 (Fig. 5a). The inhibitory receptor (CD85j), exhaustion marker programmed cell death protein 1 (PD-1), and senescence marker CD57 expression was also examined in these T cell subsets (Fig. 5b) (Healy and Murdoch 2016; 1 3 Swanson and Seder 2020). CD85j + T cells have been shown to be associated with aging (Alpert et al. 2019). The PD-1 is upregulated after T cell activation to prevent an excessive immune response. The CD57 was a marker of terminal differentiation and associated with autoimmune diseases, infectious diseases, and cancer (Characiejus et al. 2008;Palmer et al. 2005;Pedroza-Seres et al. 2007).

Non-Classical
Normalized cell count

B Cell Subsets
To identify total B cells, two B cell pan markers CD19 and CD20 were included in panel 6 (Fig. 7a). CD21 −/low B cells were identified in Fig. 7b, which represent an innate-like B cell population (Rakhmanov et al. 2009). CD11c + memory B cells were gated on total B cells (Fig. 7c), which are precursors of antibody-secreting cells (Golinski et al. 2020). The CD27 + memory B cells, naïve B cells, transitional B cells, and founder B cells were characterized in a two-step process by their expression of IgD, CD27, CD38, and CD24 (Fig. 7d). In addition, we identified unswitched B cells, marginal zone B cells, IgD only memory B cells, and IgM only memory B cells based on their differential expression of IgD and IgM (Fig. 7e). Plasmablasts, plasma cells, and classswitched B cells were identified from IgD − IgM − B cells (Fig. 7f). The gating strategies for B cell subsets presented here refer to the studies of OMIP-003 (Wei et al. 2011), OMIP-047 (Liechti et al. 2018), OMIP-051 (Liechti and Roederer 2019a), and OMIP-068 (Cascino et al. 2020).

Discussion
Understanding the phenotypes of human immune system is important to define metrics of immunological health. Here, we present a standardized workflow for high-dimensional single-cell immunophenotyping of human whole blood using multi-parametric flow cytometry. Our six optimized panels can also be freely combined with other panels for specific hypothesis-driven studies. The researchers can replicate the procedures easily and avoid costly, time-consuming mistakes during panel development. The robust platform approach of using high-throughput flow cytometry can be applied to biological and medical research.
Our panels enable deep phenotyping of most of the immune cell populations in human peripheral blood. Several groups have designed multicolor flow cytometry panels for the monitoring of immune cells from peripheral blood mononuclear cells (PBMCs) (Moncunill et al. 2014;Nogimori et al. 2021;Park et al. 2020;Payne et al. 2020). However, polymorphonuclear leukocytes such as granulocytes are depleted in PBMCs. Granulocytes are important components of the human innate immune system. The identification of granulocytes including neutrophils, eosinophils and basophils was designed in our assay. Our panels also included the assessment of unconventional T cells (γδ T cells, NKT cells, MAIT cells). Unconventional T cells share functional profiles of both innate and adaptive immunity that play critical roles in many diseases (Bae et al. 2018;Feng et al. 2015;Godfrey et al. 2019Godfrey et al. , 2018Godfrey et al. , 2016Mayassi et al. 2021;Petley et al. 2021;Silva-Santos et al. 2019;Wilkinson and Cerrone 2020). It is important to understand the immune functional status of the unconventional T cells. Furthermore, we identified different Th and Th-like subsets by their expression of CXCR3, CCR4, and CCR6 on classical Th cells, Tfh cells, Tregs, and Tc cells. These designed panels enable us to capture thousands of immune cell traits.
Flow cytometry is a powerful tool that provides rapid multi-parametric analysis of cells at the single-cell level. Additional standardization efforts are also needed in the selection of reagents, instrument setup, sample handling, and data analysis to decrease variability in a longitudinal study or multi-site study (Gratama et al. 2002;Gratama et al. 1997;Kalina 2020;Maecker et al. 2010;Maecker et al. 2012). Several international consortia are developing standardization of flow cytometry protocols and applications such as Fig. 2 Gating strategies for the PMNs, Monocytes and DCs panel (Panel 1). a CD45 + cells were identified by their CD45 expression after exclusion of doublets by cross-checking the forward scatter (FSC) signal for its area (A) versus height (H) and width (W) characteristics. b CD45 + leukocytes were distinguished into CD15 + and CD15 − populations, then neutrophils and eosinophils were identified by their CD16 expression on CD15 + cells. c Identification of monocytes. Total monocytes were separated on an CD14 versus CD16 dot plot from CD15 − SSC low cells. Monocytes were identified and further distinguished into classical monocytes (CD14 hi CD16 − ), intermediate monocytes (CD14 hi CD16 + ) and non-classical monocytes (CD14 + CD16 hi ) by their CD14 and CD16 expression. d Identification of basophils and dendritic cells. Lineage negative cells were identified by their expression of CD3, CD14, CD19 and CD56. Then the basophils (HLA-DR − CD123 + ) were gated on an anti-HLA-DR versus anti-CD123 dot plot. mDCs (CD11c + CD123 − ) and pDC (CD11c − CD123 + ) were identified by their differential CD11c and CD123 expression. The mDCs were further subdivided into CD16 + mDC and CD16 − mDC by their CD16 expression. e Expression of functional markers (CD69, CD28, HLA-DR, and CD38) by different PMNs, monocytes, and DCs subsets. A reference population of CD3 − CD56 − cells from the same sample was served as a negative expression control. Basophils can be used as a negative expression control for CD64, CD86, and HLA-DR

CD3-CD56-cells negative expression control
Normalized cell count the EuroFlow Consortium (Kalina et al. 2012;van Dongen et al. 2012), the ONE study consortium (Streitz et al. 2013), the Human Immunology Project Consortium (HIPC) (Brusic et al. 2014;Courtot et al. 2015;Finak et al. 2016;Maecker et al. 2012), the PRECISAIDS project (Jamin et al. 2016), and several other groups (Hasan et al. 2015;Ivison et al. 2018). There are a variety of aims for these projects, and the focuses of standardization differ. These studies provided a reference framework for standardized flow cytometry and inspired our research. Establishing robust flow cytometry panels requires careful selection of antibody clones and fluorochrome combinations. For researchers starting Panel design, ranking target antigens is a good place to start. Antigen ranking should mainly consider the following three points, including (1) the expression level of the target antigen on the cells of interest, antigens with low expression levels should be given priority; (2) the required resolution of the target antigen, specifically, for functional markers, which usually expression occur along a continuum and therefore need to be considered first, while for lineage markers (e.g. CD3, CD4, etc.), which usually have good separation pattern, can be considered later; and (3) gating strategies, the markers in the back of the gating order should be noted to prevent the influence of the preceding markers. Based on the above antigen ranking, further fluorochromes selection and testing are performed. We conducted extensive antibody testing (we presented a few examples of antibody selection in the Fig. 1) and found significantly different performances of antibody reagents from different suppliers. We suggest choosing suitable antibodies from different suppliers according to the testing results. For instrument setup, we describe the optimized methods based on existing data and experience. Usually, it is subjective to adjust the voltage gains based on the unstained sample. We describe a practical procedure on how to set the optimal voltage gains for each fluorescence detector (Maecker and Trotter 2006;Perfetto et al. 2006Perfetto et al. , 2012. Furthermore, we noticed the compensation by software automated algorithms needs to be carefully reviewed, and if necessary, adjusted compensation after sample acquisition. This protocol was optimized for direct staining of human peripheral blood samples which is time-saving and minimizes variations in sample preparation. The procedures for previous studies on immunophenotyping assays were including isolation of PBMCs, freezing and thawing cells. However, several studies have demonstrated that the isolation of PBMCs by Ficoll density gradient centrifugation will alter the expression of cell surface markers, cell subset distribution and function (Appay et al. 2006;Hoffmeister et al. 2003;Maecker et al. 2012Maecker et al. , 2005Renzi and Ginns 1987;Valle et al. 2012). It can also include additional variations owing to freezing/thawing steps. Our immunophenotyping assay was developed for an easy, fast procedure in only 2 mL of human peripheral blood.
Manual gating of flow cytometry data is a major source of variability in flow cytometry analyses. Our gating strategy was designed for batch analysis using FlowJo and minimized the gating adjustment. It is well suited for centralized data analysis (Maecker et al. 2005). Recently the advances in computational flow cytometry make it possible to further explore multiparametric flow cytometry data using highdimensional analysis methods, such as t-distributed stochastic neighbor embedding (t-SNE) (van der Maaten and Hinton 2008), viSNE (Amir el et al. 2013), Spanning-tree progression analysis of density-normalized events (SPADE) (Qiu et al. 2017), FlowSOM (Van Gassen et al. 2015, FLOW-MAP (Zunder et al. 2015), and PhenoGraph (Levine et al. 2015). These computational methods will be more efficient, objective and have better reproducibility (Brummelman et al. 2019;Mair et al. 2016;Saeys et al. 2016). However, there are several challenges that need to be solved such as automated population identification, mapping cell types across samples, etc. (Saeys et al. 2016).
In conclusion, we present six multi-parametric flow cytometry panels for the deep immunophenotyping of human whole blood. The standardized approach and protocols for instrument setup, antibody titration, sample staining, and data quality checks were described. However, this protocol is not designed to investigate the secreted proteins and intracellular proteins of the immune cells such as cytokines, and transcriptional factors. The deep immunophenotyping approach can generate a high informative value of datasets. These huge and high-dimensional data can be analyzed by computational flow cytometry methods. Computational flow cytometry is emerging as an important new field for profiling immunity in humans. It will promote a deeper understanding of the complex heterogeneity of cellular Fig. 3 Gating strategies for the NK, NKT, MAIT and γδ T cells panel (Panel 2). Exclusion of doublets and gating of CD45 + cells as shown for panel 1 (Fig. 2a). a Lymphocytes were gated on the FSC-A versus SSC-A dot plot, and T cells were identified by their CD3 expression. b Identification of αβ T cells and γδ T cells (anti-CD3 versus anti-TCR γδ plot). The γδ T cells were further subdivided into Vδ1 cells and Vδ2 cells (anti-TCR Vδ2 versus anti-TCR γδ plot). c Identification of MAIT cells (anti-CD161 versus anti-TCR Vα7.2 plot) and MAIT cells in αβ cells, CD8 + T cells and CD8 − T cells (d). e Identification of NKT cells (anti-CD3 versus anti-CD56 plot) and NKT cells in αβ cells, CD8 + T cells and CD8 − T cells (f). g, h Two gating strategies to identify NK cells. NK cells can be determined by CD3 − CD56 + cells (e) and subsequently distinguished into two subsets by their CD56 and CD16 co-expression into CD56 bright NK cells (CD56 hi CD16 −/low ) and CD56 dim NK cells (CD56 low CD16 + ). NK cells can also be determined by CD3 − NKp46 + cells (f), which were divided into early NK cells (CD56 hi CD16 −/low ), effector NK cells (CD56 low CD16 + ) and terminal NK cells (CD56 − CD16 + ). i Expression of activating receptors (NKG2D and NKp46) by different cell subsets  Exclusion of doublets and gating of CD45 + cells as shown for panel 1 (Fig. 2a). CD4 + T cells, CD8 + T cells, CD4 − CD8 − T cells, and CD4 + CD8 + T cells were identified by their expression of CD4 and CD8 (shown in Fig. 4a). a Expression of activation markers (CD69, CD28, HLA-DR, and CD38) by different T cell subsets. b Expression of the inhibitory receptor (CD85j), exhaustion marker PD-1, and terminal differentiation marker CD57 by different T cell subsets. Lower histograms represent the fluorescence minus one (FMO) control expression for the lineage-gated CD3 + T cells. FMO controls for the functional markers were performed on one whole blood sample

CD3-cells negative expression control
Normalized cell count behaviors, individual disease states, and perturbations of human immune system.
Author Contributions JG and FQ designed and coordinated the study. JG, YL, HL, YZ, JZ, XH, JH and HL performed experiments. JG and FQ contributed to scientific discussion. JG and FQ wrote the first draft of the manuscript. All authors reviewed, revised, and approved the final manuscript. Fig. 6 Gating strategies for the Th, Tc, Tfh and Treg cells panel (Panel 5). Exclusion of doublets and gating of CD45 + cells as shown for panel 1 (Fig. 2a). CD4 + T cells, CD8 + T cells, CD4 − CD8 − T cells, and CD4 + CD8 + T cells were identified by their expression of CD4 and CD8 (shown in Fig. 4a). a Identification of the Th subsets. Th1 (CXCR3 + CCR4 − CCR6 − ), Th2 (CXCR3 − CCR4 + CCR6 − ), Th9 (CCR4 − CCR6 + ), Th17 (CXCR3 − CCR4 + CCR6 + ), and Th17/Th1 (CXCR3 + CCR4 − CCR6 + ) subsets are identified on CD4 + T cells as indicated. b Tfh cells were identified on the anti-CXCR5 versus SSC-A plot. c CD4 + T cells were divided into Tconv cells and Treg cells, which were subsequently distinguished into CXCR5 + Tfr cells (anti-CXCR5 versus SSC-A plot). d CXCR5 + CD8 + T cells were gated on the anti-CXCR5 versus SSC-A plot. e-g Different Th-like subsets were identified based on the expression of CXCR3, CCR4, and CCR6 for Tfh cells (e), Treg cells (f), and Tc cells (g) with similar gating strategies for Th cells. h The expression levels of ICOS of different T cell subsets were evaluated. A reference population of CD3 − cells from the same sample was served as a negative expression control  (Fig. 2a). a Lymphocytes were gated on the FSC-A versus SSC-A dot plot, and total B cells (CD19 + and CD20 + ) were identified on an anti-CD19 versus anti-CD20 dot plot for subsequent analyses. b CD21 −/low B cells (anti-CD19 versus anti-CD21 plot) were identified on total B cells. c CD11c + memory B cells (anti-CD19 versus anti-CD11c plot) were identified on total B cells. Data Availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflict of interest
The authors declare that they have no conflict of interest.
Consent to participate All the participants provided informed consent.

Consent for publication Not applicable.
Ethics approval The study was approved by the Institutional Review Board of School of Life Science, Fudan University.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.