Evaluation of Extracellular Matrix Remodeling in Full-thickness Skin Grafts in Mice

Tissue Preparation

C57BL/6 mice had a FTSG implanted the IPOM or onlay position (n=20), as described previously. ¹⁸ Eight weeks after surgery, a complete section of the anterior abdominal wall, including the skin, muscles, peritoneum, and the FTSG, was excised with a scalpel and cut into smaller pieces for fixation in paraformaldehyde before paraffin embedding. Paraffin-embedded blocks with cross sections of the anterior abdominal wall were sectioned at 5 µm at room temperature using a sliding myotome. Sections were collected on a heated water bath and transferred to adhesion microscope slides (Superfrost Plus, Erie Scientific Co; Portsmouth, New Hampshire) and stored for later staining.

Staining for HA With HA-binding Probe (HABP)

The sections were deparaffinated and rehydrated and thereafter incubated in 3% H₂O₂ in methanol for 5 min. After washing in distilled water and phosphate-buffered saline (PBS), the sections were incubated with bovine serum albumin (BSA; 10 mg/ml) for 30 min to block nonspecific binding sites. The sections were incubated at 4C overnight with a biotin-marked HABP (1:40 dilution). The following day, the sections were incubated with Vectastain-Elite Avidin-Biotin complex reagent (Vector Laboratories; Burlingame, California) for 40 min, and after washing incubated for 5 min in a solution of 3,3′-diaminobenzidine (Vector Laboratories; Oxfordshire, UK). Nuclear counterstaining with Mayer’s hematoxylin was followed by dehydration and mounting on cover slips.

Masson Trichrome Staining

The sections were deparaffinated, rehydrated, and fixed in Bouin’s solution overnight. After washing in distilled water, the cell nuclei were stained with Weigert’s hematoxylin for 5 min. Biebrich Scarlet-Acid Fushin for 5 min stained the cytoplasm, muscles, and collagen. Collagen was decolorized for 5 min with phosphotungstenphosphomolbydenic acid solution from the Masson Trichrome stain kit (Sigma-Aldrich; Oakville, Canada). Collagen was stained for 5 min with aniline blue, followed by 2 min in 1% acetic acid. Finally, the sections were dehydrated and mounted on cover slips.

Immunohistochemistry

Immunohistochemistry labeling was carried out with the following antibodies: α-smooth muscle actin (ab32575, Abcam; Cambridge, UK), S100A4 (ab197896, Abcam), vimentin (ab92547, Abcam), and Ki-67 (14-5698-82, Invitrogen; Waltham, Massachusetts). Deparaffinization was carried out as follows: Slides were pre-heated for at least 20 min in an oven set at 60C and inspected to ensure that all remaining paraffin had liquefied. Heated slides were then immersed in Xylene three times for 5 min with new Xylene changes every 5 min. Sections were then rehydrated through immersion in pure ethanol for 3 min and then 95% ethanol for another 3 min. Deparaffinized rehydrated slides underwent heat-mediated antigen retrieval with pH 6 citrate buffer (Merck; Darmstadt, Germany) for vimentin staining or pH 8 Tris-ethylenediaminetetraacetic acid buffer (Merck) for S100A4 staining. Buffer-immersed slides were heated in a 70C water bath for 30 min and then rinsed three times for 5 min with PBS with 0.05% added Tween20 (Merck; hereafter referred to as PBST) at room temperature. To block nonspecific binding, sections were covered in 400 µl of 5% Goat normal serum (ab7481, Abcam), diluted by volume with PBST and incubated for 15 min at room temperature. Blocking solution was subsequently drained and the primary antibody (Table 1), diluted in PBST with 1% bovine serum, was applied to each section for overnight incubation at 4C. Sections were kept in a leveled moisture chamber to hinder evaporation of solutions from the sections. The following day, sections were returned to room temperature and rinsed three times for 5 min in PBST, followed by 15 min blocking at room temperature with Goat normal serum as above. A secondary antibody (Table 1) was applied for 30 min at 37C. Finally, the slides were rinsed with PBS before mounting on cover slips and a VECTASHIELD antifade media containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Newark, California). In a subset of nine specimens, a Ki-67 primary antibody was added together with the rabbit polyclonal vimentin primary antibody to create a solution with two different primary antibodies. Later, at the step where the secondary antibody is added, a solution of Alexa 488 goat anti-mouse together with Alexa 594 goat anti-rabbit secondary antibodies was used to double-stain for vimentin and Ki-67 simultaneously. Negative primary and secondary antibody controls consisted of slides processed as above, including the application of normal serum and secondary antibodies, but modifying the protocol by (1) omitting the primary antibodies outlined above and just adding secondary antibodies or (2) replacing the primary antibody with a concentration-matched polyclonal rabbit IgG isotype control (ab37415, Abcam). In negative controls without rabbit IgG, labeled cells with fibroblast-like appearance could not be seen in neither the host skin nor the graft. In primary antibody controls with nonspecific IgG rabbit antibodies, the sections showed some nonspecific reactivity with both acidic and alkaline pre-treatment. This nonspecific IgG binding was primarily seen in parts of the FTSG with disrupted keratin and degenerating hair follicles. A nonspecific reactivity could be seen in a minor subset of cells in the connective tissue of both normal skin and the collagenous part of the FTSG. This nonspecific reactivity was seen in <2% of all cells in both parts of the section and was deemed to be of minor impact for the quantification of cells identified with the anti-vimentin and anti-S100A4 rabbit IgG antibodies. Positive controls consisted of sections from abdominal skin from either animals that had not been operated (external positive control) or areas of skin further removed from where the incision was made (internal positive control). In these positive controls, smooth muscle actin, vimentin, and S100A4 antibodies labeled the cytoplasm of cells with stellate fibroblast-like appearance in the dermis of the skin. No sections were considered for further analysis unless these labeled cells were present and clearly stained in the dermis of the section.

OPEN IN VIEWER

Table 1.

Antibodies used in the study.

Antibody	Supplier	Type	Concentration	Pretreatment/Miscellaneous
Anti-vimentin	Abcam	Rabbit monoclonal	1:300	Citrate pH 6 for 30 min at 70C
Anti-S100A4	Abcam	Rabbit monoclonal	1:500	Tris-EDTA pH 8 for 30 min at 70C
Anti-Ki 67	Invitrogen	Rat monoclonal	1:500	Citrate pH 6 for 30 min at 70C
B-CHP	3Helix	—	1:20	PBS 1% BSA
Rabbit IgG isotype control	Abcam	Rabbit polyclonal	1:300–1:500	Citrate pH 6 for 30 min at 70C or Tris-EDTA pH 8 for 30 min at 70C
Goat anti-rabbit Alexa Flour 594	Abcam	Goat monoclonal conjugated	1:500	Secondary antibody
Goat anti-rat Alexa Flour 488	Abcam	Goat monoclonal conjugated	1:500	Secondary antibody
Alexa Flour 647 Streptavidin	Thermo Scientific	—	1:200	Secondary antibody

Abbreviations: CHP, collagen hybridizing peptide; EDTA, ethylenediaminetetraacetic acid.

Collagen Hybridizing Peptide (CHP)

CHP staining was performed according to manufacturer’s protocol (3Helix; Salt Lake City, Utah) after the embedding material was removed as described above. Briefly, a working amount of the biotin-conjugated CHP stock solution (0.3 mg/ml at 4C) was diluted in PBS containing 1% BSA to a final concentration of 15 μg/ml and thermally dissociated to single strands by heating for 6 min at 80C. To prevent thermally induced tissue damage, the heated solution was briefly placed in ice. Within 2 min after cooling, the CHP solution was applied to the tissue samples for 2 hr at 4C. The slides were then washed three times for 5 min in PBS, and an Alexa Flour 647 streptavidin conjugate antibody (Thermo Scientific; Waltham, Massachusetts) was added at a concentration of 0.05 mg/ml for 1 hr at room temperature. The slides were then washed three times for 5 min in PBS, and a mounting media with DAPI was applied before mounting on cover slips.

Imaging Acquisition and Analyses

Slides were scanned with PANNORAMIC 250 Flash III (3DHISTECH, Budapest, Hungary) using the associated software Pannoramic Scanner Software v3.03. The scanned slides were evaluated in SlideViewer v2.5 (3DHISTECH). Evaluation of HA staining was performed by three independent assessors (authors A.E.T., A.G., and U.H.). The dense connective tissue (DCT) of the graft (see the Results section) was localized according to HTX and eosin staining and Masson Trichrome staining, and evaluated using the SlideViewer program according to the following grading protocol: HABP DCT (−) indicating no HABP in the DCT of the graft; HABP DCT (+/−) indicating a speckled appearance where less than half of the fascicles/strands in the DCT were positive for HABP; HABP DCT (+) where 50–75% of the DCT were positive for HABP; and HABP DCT (++) where HABP staining in the DCT encompassed >75% of the fascicles/strands.

Graft cross-sectional area (CSA) and areas of different tissue types were estimated with the inbuilt measurement tools provided with the software. Figures were assembled and annotated in Adobe Photoshop v23.5 (Adobe; San Jose, California).

The CHP intensity units were measured by evaluating three areas per section of normal skin or graft using ImageJ (64-bit; v.2.9.0/1.53t; Java 1.8.0_322). The intensity units were calculated by normalizing the mean intensity with the background intensity.

Quantification of Vimentin-positive and S100A4-positive Cells

Cross sections of the host skin were identified on each section stained with the abovementioned markers. Regions of 1091 × 682 µm² were preselected along the length of the skin section, omitting areas with section folding, excessive tearing, and where the skin overlay the graft itself. One or two well-preserved areas were randomly selected for quantification. Our goal was to select two well-preserved areas from each specimen but in some cases this was not possible due to excessive tearing of the section. The dermis of the skin was delineated with an area tool, leaving out epidermis, subcutis, and hair follicles in each preselected area. This was done to achieve a quantitative estimate that mirrored the properties of the dermis. The number of DAPI-positive cells was quantified in the area, followed by a second quantification of those DAPI-positive cells that were positive for vimentin or S100A4. Between 34 and 335 cells were counted in each specimen. The total skin area quantified in each specimen was 19,912–258,263 µm². In the areas of the section encompassing the graft, only areas containing DCT were quantified. DCT was identified by simultaneous reviewing of HTX and Masson Trichrome staining on corresponding serial sections. Areas of DCT were identified by their high collagen content and fibrillar structure (see the Results section for details). For all grafts studied, approximately four areas 1091 × 682 µm (as with normal skin) were preselected that contained at least 25% DCT. One or two areas were randomly selected for quantification. Quantification was carried out similarly to that for the dermis in the host animal skin, except that in the DCT there were no hair follicles and visible subcutis to be removed from the region of interest. The total area of DCT was delineated in each region and the cells counted using a counting frame for unbiased quantification. In the graft, 106–804 cells were counted, encompassing 25,820–1,050,932 µm², and the proportion of vimentin- and S100A4-positive cells was calculated, using the total number of DAPI cells counted in the skin and the graft, respectively. The density of vimentin- and S100A4-positive cells was calculated with the number of cells counted per total area of areas analyzed, and reported as positive cells/1000 µm².

Statistics

Analyses were carried out using the software SPSS 28.0 (IBM; Armonk, New York). Normality of data variables was graphically evaluated with Q–Q plots. Differences between groups (only IPOM and onlay in this study) were assessed with independent t-tests for variables with normal distributions, and with Mann–Whitney U test for variables where normality could not be assumed. For paired tests, paired t-tests were used in normally distributed data and Wilcoxon signed-rank test in data where normality could not be assumed. In analysis of bivariate correlations, Spearman’s rank correlation coefficient was used in all correlations except those bivariate correlations that concerned HABP staining intensity evaluations. Here, Kendall’s tau coefficient (which treats tied ranks more correctly) was used instead. Missing data due to loss of sections from tearing were treated in a pairwise exclusion manner, for example, if one single measurement was missing from one mouse, that did not preclude it from inclusion in correlational testing for other measurements. Therefore, the number of mice included could be less than 20 for some statistical tests. For HABP evaluations, inter-rater congruence was evaluated with Fleiss’ kappa. Generation of bar graphs was carried out in Graphpad prism (Graphpad software; Boston, Massachusetts).

For correlations and differences, a p value <0.05 was considered significant.

General Morphology

Despite the mice being sacrificed at the same time, there were marked differences in appearance of the FTSGs (Fig. 1). All skin grafts showed some signs of remodeling. In the most mature cases, hair, hair follicles, and sebaceous glands had all disappeared entirely and become replaced with extensive, relatively cell-sparse, DCT. In other cases, hair follicles and glands were in different states of remodeling (Fig. 1). Despite various states of remodeling, two different types of connective tissue were present and recognizable in 8 of 10 FTSGs—one characterized by DCT with fewer cells, seemingly derived from the deeper parts of dermis (row 4, Fig. 1), and one characterized by loose connective tissue (LCT) with high cell density (row 5, Fig. 1). In the latter, intact or degenerating hair follicles and sebaceous glands (Fig. 2) were present, but much lesser organized connective tissue. In Masson Trichrome staining, the DCT was intensively stained blue (collagen) whereas LCT had weaker staining. In general, the more voluminous the DCT, the more organization (with clear orientation of collagen and cells) was present. Integration into the host abdominal wall was most visible in the corners of the graft (Fig. 2) and generally had a more integrated appearance in mice operated with the onlay method. In one mouse, the FTSG differed dramatically in appearance (Fig. 3). It was somewhat smaller in CSA than the other grafts, contained very little DCT, and in eosin and HTX staining appeared to contain many lymphocytes and phagocytic macrophages. Circular cell arrangements, suggestive of remnants of former hair follicles and cystic cavities, were much more frequent than in other grafts. Furthermore, this graft showed a clean separation from the enclosed cavity within the host animal, with no sign of tissue integration. This specimen was interpreted as a “failed” graft of the 20 studied, that for some reason had degenerated after implantation. Staining indicated low presence of both collagen and HA, and it was excluded from further grouped analysis in the study, for example, when calculating mean CSA.

OPEN IN VIEWER

Figure 1.

Sections of different specimens showing cross sections of host mouse skin (A1–C1), full-thickness skin graft (FTSG) with limited remodeling (A2–C2), FTSG with extensive remodeling (A3–C3), an example of a region with dense connective tissue (A4–C4), and a region with loose connective tissue (LCT; A5–C5). The first column shows panels with eosin and hematoxylin (HTX) staining (Panel A), the second column shows HABP staining, where hyaluronic acid is stained brown (Panel B), and the third column shows Masson Trichrome staining (TCM), collagen is blue, elastin and keratin are red, and hair is yellow (Panel C). The epidermis of the host is indicated with ‘Sk’ where applicable. Black arrowheads indicate sebaceous glands (rows 1, 2, and 3). Long black arrows indicate hair follicles (rows 1, 2, and 3). A region of LCT with high cellularity is indicated by a short black arrow (bottom row). In the second (A2–C2) and third (A3–C3) rows, host skin and FTSG are in continuum and delineated by the dotted line. Bars A–C = 100 µm. Abbreviation: HABP, HA-binding probe.

OPEN IN VIEWER

Figure 2.

Overview (Panel A) and high-magnification views (Panels B–F3) of a section from a mouse with the full-thickness skin graft in the onlay position (outlined in blue). Areas of high-magnification views are marked with a yellow frame with corresponding letters as in the lower panels. In the top panel, the ventral orientation is facing up and dorsal orientation is facing down. The striated muscles of the RA are indicated in Panel A. The section is delimited by the host skin ventrally (Panel B) and the peritoneal wall dorsally (Panel E), where the border between the graft and linea alba (dotted line in Panel E) can be seen. Normal hair follicles of the host skin are indicated in Panel B (black arrows). A view of the graft with mostly dense connective tissue is demonstrated in Panel C. A view of the graft in a cellular region of loose connective tissue is demonstrated in Panel D, also displaying remnants of hair follicles and hair (white arrows). A corner of the graft is indicated in Panels F1–F3, showing an eosin and hematoxylin (HTX) staining (F1), HA-binding probe (F2), and Masson Trichrome (F3), respectively. Panels F1–F3 display different types of connective tissue in proximity to various levels of cellularity, presence of collagen, and presence of hyaluronan. Degenerating muscle fibers in the graft are indicated with white arrows. Bars A–F = 200 µm. Note that due to different magnification, the scale bar in the top panel’s lower right corner is relatively small. All remaining panels (B–F) have identical magnification. Abbreviation: RA, rectus abdominis.

OPEN IN VIEWER

Figure 3.

Sections of a full-thickness skin graft from an animal where the graft had failed to integrate. Panels A1–C1 are low-magnification views of the same area with eosin and hematoxylin (HTX) (A1) staining, HA-binding probe (B1) staining, and Masson Trichrome (C1) staining. The remaining panels (D1–F1) are high-magnification views of areas of the same graft. Only a miniscule proportion of the graft had developed dense connective tissue (long black arrow in A1–C1). Large parts of the graft were made up of dense groups of inflammatory monocytes (white arrow in D1), vacuoles surrounded by phagocytized melanin (black arrows in E1), or macrophage-like cells reminiscent of foam cells (white arrows in F1). Bars A1–F1 = 100 µm but note the different lengths of the scale bars due to the use of two magnifications.

On cross sections of the anterior abdominal wall, the area of the grafts varied between 2.3 and 11.7 mm² (mean 4.25 ± 2.20 mm²). The two smallest grafts (2.4 and 2.3 mm²) were either grafts with little tissue remodeling or where the graft itself had failed (see above). The largest graft (11.7 mm²) stood out from the rest due to a high degree of both inflammation, remodeling, and angiogenesis together with extensive signs of integration along the edges of the graft. As mentioned above, prominent areas of DCT were present in a majority of specimens (15 out of 20). In the failed specimen, it accounted for approximately 5% of the CSA whereas in the remaining 19 specimens, DCT accounted for 52.2 ± 22.6% of the CSA. In general, those specimens where the proportion of DCT was highest were those where the loss of hair and hair follicles was most complete. Organization of the LCT was generally more dubious, clear borders between actual LCT and degenerating remnants of hair follicles or groups of inflammatory cells meant that a precise estimate of the proportion of LCT was not possible in all specimens. For this reason, further analyses focused on the DCT.

Histochemical Staining for HA and Collagen

In the host’s skin, regardless of whether that skin was directly overlying the FTSG or situated on the anterolateral abdominal wall beside the area of implantation, the dermis was uniformly stained brown with HABP (indicating the rich presence of HA) and moderately blue with Masson trichrome (TCM) (indicating the presence of collagen fibrils; Fig. 1). In the FTSGs, the patterns of HABP and TCM staining were more variable. In FTSGs showing less remodeling (row 2, Fig. 1), HABP staining and TCM staining were both present with similar intensity. In FTSGs with more extensive remodeling, intense blue TCM staining was not necessarily associated with intense brown HABP staining. Instead, HA staining in DCT areas in the same animal varied considerably, usually in bundle-like structures. HA staining was generally absent in LCT tissue (Fig. 1). HABP staining in the DCT was rated from (–) indicating no HABP staining to (++) indicating HABP staining in >75% of the DCT. Three raters (coauthors A.E.T., A.G., and U.H.) analyzed 16 specimens and the inter-rater reliability was deemed to be good (Fleiss’ kappa 0.683; p<0.001). However, HABP ratings did not differ between specimens implanted with the IPOM or onlay method (p<0.114) and in the total study group, HABP grade did not correlate with the total graft area (p<0.511), proportion of DCT (p<0.171), or any other variable studied (data not shown).

Collagen Evaluation (CHP and Masson Trichrome Staining)

Masson Trichrome staining was used to locate areas with collagen within the section. Most CHP-positive staining was seen in the DCT areas of the grafts. In all mice, CHP intensity was low in the overlying skin of the host animal. There was a notable variation in CHP staining intensity of the graft between specimens (Fig. 4), with the lowest two specimens displaying intensity units of 2.3 and 5.8 above background staining intensity whereas the highest two specimens displayed staining intensity units of 82.9 and 84.0 above background staining intensity, respectively.

OPEN IN VIEWER

Figure 4.

Sections of grafts stained with Masson Trichrome (TCM; Panels A1 and A2) or collagen hybridizing peptide (Panels B1 and B2) in two animals with markedly different fluorescence collagen hybridizing peptide (CHP) staining intensities. In both instances, the rich presence of collagens (blue) can be seen in the dense connective tissue (DCT; shown in the panels). In one instance, CHP intensity is high (B1) suggesting the presence of unfolded collagen helices, whereas in the other (B2) CHP intensity is low, suggesting a DCT with less active remodeling. Bars A and B = 100 µm.

Quantification of CHP intensity in the grafts showed a non-significant (p<0.097) but notable difference in intensity between the IPOM (47.5 ± 30.0) grafts and the onlay (22.5 ± 21.1) grafts.

Immunofluorescence Cell Markers

Vimentin-positive cells were widespread in the dermis and deep subcutaneous layers of the host skin, and generally had a stellate or elongated shape (Fig. 5). These cells were identified as typical fibroblasts. Vimentin-positive cells were also sporadically encountered in association with striated muscle fibers close to the FTSG, but these were ignored because activated muscle satellite cells are also known to express vimentin. ¹⁷ In our samples, the morphology of vimentin-positive cells differed between graft locations and specimens. In FTSGs with little remodeling, vimentin labeling was more diffuse showing less densely packed fibroblasts. In LCT areas usually containing hair follicle remnants, there were dense clusters of overlapping vimentin-positive cells, making precise quantification or characterization difficult or impossible. On the other hand, vimentin-positive cells were easy to identify and quantify in host skin dermis and in the DCT areas of the grafts. Most cells in both skin dermis and DCT areas that were quantifiable were vimentin-positive (Fig. 5). Dermal endothelial cells and pericytes of capillaries have also been reported to express vimentin.^19,20 Capillaries were sparse but present to some degree in all FTSG samples. Given that some tangential sectioning of capillary walls is unavoidable, and that bona fide fibroblast is also likely to be found close to capillaries, we also quantified vimentin-positive cells in association with capillaries. This means that our quantitative data likely include some vimentin-positive pericytes and endocytes besides classic fibroblasts, but given that capillaries made up a minor portion of the sections the skewness in data is suggested to be of minor importance for the overall results. For all specimens, the mean percentage of vimentin-positive cells in the dermis of host skin was 81.9 ± 12.2% and 61.7 ± 13.8% in the DCT areas of the graft (Fig. 6). The corresponding densities of vimentin-positive cells were 1.61 ± 1.01 cells/1000 µm² in host skin and 1.02 ± 0.84 cells/1000 µm² in the DCT areas of the graft (Fig. 6). Neither percentage (p<0.441) nor density (p<0.141) of vimentin-positive cells in the DCT areas of the graft differed between the IPOM and onlay positions.

OPEN IN VIEWER

Figure 5.

Immunofluorescence staining with primary antibodies against vimentin and Ki-67 (Panels A1–C2) or S100A4 (Panels D1–F2). Panels A1–F1 show overlapping 4′,6-diamidino-2-phenylindole (DAPI) counterstain to visualize nuclei, whereas Panels A2–F2 show the images without DAPI counterstain. Panels on the left show labeling patterns in the skin of the host, middle panels show labeling patterns of DCT areas, and the panels on the right show the labeling patterns of LCT. Vimentin-positive cells (white arrows), S100A4-positive cells (white short arrows), and vimentin-negative/Ki-67-positive cells (long white arrows) are shown in each panel. Please note that DAPI labeling is dimmer in the S100A4 panels, this is due to the alkaline pre-treatment of the slides necessary for the S100A4 primary antibody. Bars A–F = 50 µm. Abbreviations: DCT, dense connective tissue; LCT, loose connective tissue.

OPEN IN VIEWER

Figure 6.

Bar graph displaying the number of labeled cells per 1000 µm² (Panel A) and the proportion of cells (Panel B) positive for vimentin and S100A4 in host skin and DCT of the graft, respectively. Asterisks indicate significant pairwise differences between estimates, tested with Wilcoxon signed-rank test. Error bars indicate 95% CI. Abbreviations: DCT, dense connective tissue; CI, confidence interval.

In both host skin and FTSGs, S100A4-positive cells were significantly fewer than vimentin-positive cells (Fig. 6) in both highly differentiated grafts and in those less differentiated. For all specimens, the mean percentage of S100A4-positive cells was 50.0 ± 26.7% in the dermis of host skin and 18.8 ± 14.3% in the DCT areas of the grafts. The corresponding densities of S100A4-positive cells were 0.61 ± 0.45 cells/1000 µm² in host skin and 0.22 ± 0.14 cells/1000 µm² in the DCT areas. Neither percentage (p<0.187) nor density (p<0.696) of S100A4-positive cells in the DCT areas differed between IPOM and onlay positions.

A subset of nine specimens was double-stained for vimentin and Ki-67 (Fig. 5). Ki-67-positive cells were found in all specimens but not always in the DCT areas. Between 0 and 10 positive cells (median 6) were found in the DCT areas of each full cross section and the majority of these were also vimentin positive. However, most Ki-67-positive cells were found in areas of the graft other than the DCT. Ki-67-positive cells were frequently seen in the LCT areas but were not always double-stained for vimentin. Ki-67-positive cells were also seen in the skin of the host but seldom as double-stained vimentin-/Ki-67-positive cells, usually in the epithelia or the hair follicles (Fig. 5).

All specimens were also stained for α-smooth muscle actin. Generally, this antibody labeled cells throughout the dermis and graft in approximate correspondence with the vimentin staining described above. In addition, it also labeled the cytoplasm in glandular structures of the dermis and epithelia. This, in addition to a quite variable labeling intensity, made a reliable and informative quantification of smooth muscle actin cells difficult. Therefore, no further quantification was carried out with this marker.

Correlations

No significant correlations were seen between the total graft CSA and any other variable. However, the proportion of graft that had developed into DCT was positively correlated to the proportion (p<0.033, ρ = 0.673) but not density (p<0.651, ρ = 0.164) of vimentin-positive cells within the DCT. Interestingly, there was a negative correlation between the proportion of S100A4-positive cells in the DCT and the density of vimentin-positive cells (p<0.023, ρ = −0.705) indicating that grafts with fewer vimentin-positive cells had a higher proportion of S100A4-positive cells. The proportions of S100A4-positive cells in DCT areas and host skin were positively correlated to the intensity of CHP staining (p<0.029, ρ = 0.720 and p<0.050, ρ = 0.667, respectively). A review of all correlations tested is found in Appendix Table A1.