Applying Quantitative Immunogold Labeling Distributions of Cellular Compartments in Immunoelectron Microscopy Images to the Study of EGFR Intracellular Signaling

Immunoelectron microscopy has been shown to be a useful tool for quantifying
distributions of relevant proteins across cellular compartments. This methodology
has been recently applied to study the biological response of Heberprot-P in humans,
an epidermal growth factor (EGF)-based therapeutic product that enhances healing
of chronic wounds in patients with diabetic foot ulcers (DFU) safely and efficaciously.

can be mapped in the context of cellular ultrastructure, intracellular trafficking of these molecules may be efficiently monitored while showing details of cell fine structure. Therefore, we hypothesized that this approach may be particularly relevant to study key cellular signaling pathways with therapeutics effects.
Lower extremities ulcerations and the potential for amputation are major complications of diabetes [3]. Consequently, diabetes is considered as the leading cause of lower limb amputations globally [4]. In addition, impairment of the healing process is a hallmark of diabetes [5]. Thus, persistent hyperglycemia and related mitochondrial reactive oxygen species overproduction are involved in chronic tissue damage, aberrant wound healing response and diabetic complications [6]. Fibroblasts play a key role in the wound healing response as they proliferate, express a range of cytokines, growth factors, chemokines and secrete proteases, extracellular matrix proteins and other factors. Importantly, the functions of fibroblasts have been shown to be affected in diabetic foot ulcers (DFU) [7,8]. Moreover, key signaling pathways mediated by the epidermal growth factor (EGF) receptor (EGFR) have been shown to play a central role in wound healing and they are affected in DFUs [9][10][11].
We have recently applied this methodology to study the biological response to Heberprot-P in humans, an EGF-based therapeutic product that enhances healing of chronic wounds in patients with DFU safely and efficaciously. While EGFR was scarcely detected in fibroblasts derived from diabetic granulation tissue, EGF treatment of DFU induced a strong activation of EGFR in fibroblasts [8]. A key feature of this activation was the early immunolabeling increase of EGFR in the fibroblasts´s nucleus as compared to cells analyzed at later times following EGF treatment. However, at later times after EGF therapy, EGFR was augmented in mitochondria of fibroblasts. These findings indicated for the first time the relationship of EGFR localization in nucleus and mitochondria and the therapeutic effect of a world leading treatment for DFU directly in humans. It also points to the value of quantitative IEM for the study of relevant intracellular signaling pathways.

2-Quantitative Immunogold Labeling Distributions of Cellular Compartments
Several excellent reviews have described the various methods of quantitative IEM and its applications [12,13]. Here we will focus on those stereology-based methods that we have used to analyze EGFR signaling in DFU samples.

Randomized Sampling of Specimens
Several important considerations have to be taken into account when designing quantitative IEM experiments including the intrinsic variability of biological specimens. Importantly, biological structures tend to vary in appearance depending on their position and their orientation. TEM and IEM involve cutting ultrathin sections (40-80 nm thick) to reveal the internal structure of a specimen at a high level of resolution. Major consequences of producing ultrathin physical sections from 3-dimensional structures are: Figure 1: Workflow for quantitative IEM to study EGFR intracellular traffic. Note: Steps in a multistage random sampling scheme followed by stereology-based quantification. First, a set of specimens are selected from DFU, fixed and cut into serial slices. These are embedded (blocks) and sectioned for TEM followed by on-section immunolabeling. Then, these are sampled in turn to provide fields of view (FOVs) followed by quantification using stereology to count gold particles and test points in volume occupying compartments (VOC) and data handling involving descriptive and inferential statistics. Each step involves randomization of section location (SUR sampling).

Figure 2:
Illustrations of multistage SUR sampling cascade for studying EGFR intracellular traffic. Note: Red circles represent selected samples for further sampling. A) The DFU specimen is cut into serial slices and every 6th slice is selected (at least two widely separated slices), and subsequently SUR sampled (not shown) to obtain tissue pieces to prepare embedded blocks for further sampling; B) Each block is overlaid with a lattice of lines (gross lines) and a group of ultrathin sections (thin lines, 1 and 2) from every 2nd slice (red circles) are selected (at least two widely separated groups of ultrathin sections); C) In this example a ribbon of three TEM sections (1-3) is placed on nickel grid and a SUR sample of fields of view from one section is obtained (red circles, section 2).  EGFR intracellular traffic, specimens were fixed and dehydrated as described elsewhere [8]. Then, samples were embedded in Lowicryl and polymerized by exposure to ultraviolet light at room temperature for 72 h to produce blocks. Ultrathin sections obtained from these blocks were scanned to generate fields of view (FOVs) for TEM. It has been proposed that in this scheme, each specimen must provide a minimum of two tissue blocks, at least two widely separated ultrathin sections per block and a minimum of two microscopical FOVs per section [12].
Then, the next step is to count particles and allocate them to appropriate compartments. At this step, the use of stereological test probes (see below) allows statistical comparisons between compartments within a cell or a group of similar cells. Random sampling of specimens is a critical requirement at each stage of this process ( Figure 3) (For a general detailed description, see [13]. Importantly, systematic uniform random (SUR) sampling is considered to be more efficient than independent random sampling, particularly for heterogeneous specimens, because it confer more even coverage across the entire specimen [14,15]. In SUR sampling, the position and the orientation of the first piece are randomized and a selected pattern (the sampling interval) determines the positions and orientations of other items. However, to avoid sampling inefficiency the sampling interval should not coincide with some natural pattern or periodicity within the specimen itself [15].

Defining Compartments
Generally, intracellular ultrastructural compartments include volume-occupying organelles (VOC) (e.g. mitochondria, nuclei, golgi complex, etc) and surface-occupying membranes (e.g. mitochondrial membranes, cisternal membrane of the rough endoplasmic reticulum, etc). It is worthy of note that randomization of positions within the specimen is sufficient when studying only VOC. However, when investigating surface-occupying compartments or a mixture of VOC and surface-occupying compartments both position and orientation must be randomized. The aim of our study involved the investigation of EGFR localization through a set of VOC and thus, the randomization of positions within the specimen was sufficient.

Methods for Quantifying Gold Particles
Two major methods, that had been thoroughly described before [2], were used to study the intracellular trafficking of EGFR [8].

Approach 1:
To compare the distributions of gold particles between different compartments within a cell or cell system. The aim of this method is to test whether the observed distribution of gold particles between compartments within a given cell type is random or non-random [2]. Firstly, an appropriate set of compartments (VOC for this study) should be chosen to suit the investigation aim.
Second, gold particles lying on compartments must be counted on all sampled fields. Thus, the numerical frequency distribution   and secondly, the corresponding partial 2 value must account for a substantial proportion of total2 (at about 10%).
Preferential labeling for EGFR in VOC within FLCs: Table   1 show an example showing the application of this approach to quantify EGFR immunolabeling within different compartments of fibroblast-like cells (FLCs). In this study ultrathin sections of granulation tissue were incubated with anti-EGFR mouse IgG monoclonal antibody followed by gold-labeled (15 nm) anti-mouse IgG. All sections were stained and analyzed by TEM as described elsewhere [8] (Figure 4). The example in Table 1 Figure 4. Ne (i.e. random distribution) and RLI for each compartment were calculated as described above. Table 1 illustrates a two-sample 2 analysis with two columns (No and Ne) and ¨r¨ compartments (arranged in rows) used to compare No and Ne from a FLC analyzed at T6. This is representative of the analysis performed at later time points (T6-T24) after EGF treatment to detect preferential labeling of compartments in a single group of cells (FLCs in this work) ( Figure 5A). For the dataset shown in Table 1, the total 2 value is 21.17 and, for 3 degrees of freedom, the probability level is P<0.001. This means that the null hypothesis (of no difference between distributions or that the observed distribution of gold particles is random) must be rejected.
Both ER+GC and mitochondria fulfilled the criteria for deciding on preferential labeling of a compartment described above, as they showed RLI values >1 and their corresponding partial 2 value accounted for more than 10% of total 2 (Table 1).  On the other hand, Figure 5A also illustrates the preferential labeling for EGFR labeled by gold particles in different compartments from the groups of FLCs analyzed at T15-T60 and at T6-T24 after EGF treatment. Results are shown only for those compartments satisfying the criteria for preferential labeling. Note that while ER+GC was predominantly labeled in the FLCs studied from T15-T60, ER+GC and mitochondria showed preferential labeling at later times of EGF treatment (T6-T24). The numerical frequency of these particles for each compartment represents the No. Then, the observed gold counts can be analyzed by means of a contingency table analysis with ¨g¨ groups (arranged in columns) and ¨r¨ compartments (arranged in rows) that will generate predicted gold particles (Ne) and, hence, partial 2 values, for each group and each compartment. As stated in method 1, compartments associated with 1-5 of predicted gold particles should comprise no more than 20% of the studied compartments and no expected value should be <1. Examination of the total 2 value, for (g-1) x (r-1) degrees of freedom, will indicate whether the gold labeling distributions are different. If they are different, DOI: 10.26717/BJSTR.2020.26.004352 examining the partial 2 values (using the 10% cut-off) will identify the compartments which contribute with a substantial proportion of total 2 (at about 10%).

Testing for Shifts in Compartment Labeling for EGFR in
FLCs Taken at Different Times After EFG Treatment: To compare the distribution of EGFR immunolabeling in various compartments of FLCs at different times after treatment with EGF (early T15-T60 and late T6-T24 times), a contingency table analysis was used ( Table 2). The null hypothesis of 'no difference of gold distributions between FLCs at different times after treatment´ was tested. Thus, a contingency table (g columns x r rows) was created ( Table 2). In Corresponding partial 2 values and total 2 were calculated as shown above.

Figure 5A:
A. Preferential labeling for EGFR labeled by gold particles in different compartments of Fibroblast-like cells (FLCs) analyzed at T15-T60, and at T6-T24 after EGF treatment.
For the dataset shown in Table 2 groups are represented by FLCs analyzed at T15-T60 and at T6-T24. Note that for this approach statistical handling can be distorted by large discrepancies in the total numbers of observed gold particles between groups. So, to get similar column sums for total gold counts in each group of cells as shown in the example (Table 2), gold particle counts from the original work were normalized to yield totals of roughly 950 per group. Total 2 for these two groups is 65.35 and for 3 degrees of freedom, the probability level is P<0.001. Therefore, the null hypothesis of no difference in distributions between groups can be rejected.
Analysis of partial 2 values revealed that the nucleus and mitochondria compartments were responsible for the difference (  Figure 5B). Data indicate a shift in gold labeling of EGFR from the early to late times after EGF therapy. Thus, fewerthan-expected gold particles were detected in mitochondria while more-than-expected immunogold labeling was shown in nucleus at T15-T60 (Table 2 & Figure 6B). Interestingly, at T6-T24, a shift in labeling showed more-than-expected gold particles in mitochondria and less-than-expected EGFR immunolabeling in the nucleus. This analysis supports the intracellular trafficking of the EGFR from nucleus at T15-T60 to mitochondria at T6-T24.
Nevertheless, a disadvantage of this approach is its limited ability to permit mechanistic interpretations of shifts in labeling patterns [12,16,17]. Thus, a compartment might contain many particles due to either its large size or because it is a small compartment with a high density of particle localization. In such situations, it is recommended to complement these results with the RLI analysis [12] In the EGFR study reviewed here, preferential labeling of EGFR in mitochondria at T6-T24 supports findings of approach 2 showing accumulation of EGFR in mitochondria at later times after EGF infiltration. Therefore, these evidences strongly point to the shift in compartment labeling for EGFR in FLCs analyzed at different times after EFG treatment. Note that FLCs at T15-T60 have fewer-than-expected gold particles at mitochondria (No/Ne ratio 2 0.52; 19.18% of total 2) and more-than-expected gold particles in nucleus (No/Ne ratio 2 1.25; 24.46% of total 2) while FLCs analyzed at T6-T24 have more-than-expected gold particles at mitochondria (No/Ne ratio 2 1.48; 19.20% of total 2) and fewer-thanexpected gold particles in nucleus (No/Ne ratio 2 0.75; 24.48% of total 2). Endoplasmic reticulum and Golgi complex (E); Mitochondria (M); Rest of Cytoplasm (C); Nucleus (N).

Conclusion
Development of rigorous sampling, stereological estimation and statistical evaluation procedures have strengthened quantitative analysis of IEM images [12]. EGFR is a multifunctional regulator in various subcellular organelles including plasma membrane, cytoplasm, mitochondria and nucleus [18]. Different roles of EGFR may depend on its subcellular localization and specific cellular microenviroment. By applying the methodology described in this review to the study of intracellular dynamics localization of EGFR, we have recently described that intralesional EGF therapy induces the activation and nuclear translocation of EGFR in FLCs early after EGF administration (8). Further characterization of distributions of immunogold labelling for EGFR at later times after EGF treatment showed increased labeling of EGFR in mitochondria and decreased immunolabeling in nucleus. These findings suggest that EGFR´s functions in nucleus prevailed at early times while they predominated in mitochondria at later times following EGF infiltration. The use of the two quantitative methods described here demonstrated that they complement each other, thus supporting their value for the study of intracellular signalling pathways of therapeutic value in vivo.