Introduction

The term angiogenesis describes the growth of new blood vessels from preexisting ones. After a pro-angiogenic stimulus, endothelial cells (ECs), which form the inner lining of a vessel, start to break cell–cell contacts and secret proteases to degrade the basal membrane [1–3]. This degradation causes a feedback loop and intensifies the angiogenic process. In parallel, the expression of VEGF-A increases the permeability of ECs. Plasma proteins are secreted and build up a temporary matrix for the migrating ECs. Vessel sprouting is a hierarchical organized and coordinated process. It starts with the activation of a few endothelial cells—named tip cells—and is regulated by the Delta-like 4 (DII4)-Notch pathway [4–6]. The sprouting terminates when a tip cell has been attached to a close by (neighboring) vessel. In the following, EC secret factors such as platelet-derived growth factor B (PDGF-B) recruit mural cells (pericytes and/or smooth muscle cells) to the vascular wall [2, 7–9].

Although several factors involved in the angiogenic cascade have been already identified, only few genome-wide expression analyses have been conducted using human samples, yet. In order to investigate angiogenesis in human adults, we induced a wound healing process in skin by removing the epidermal layer. Wound healing is a highly dynamic and complex process consisting of three partly overlapping steps: (i) inflammation phase, (ii) regeneration/tissue formation phase, and (iii) remodeling phase. All steps have to be aligned in a precise manner to restore tissue integrity [10–12]. The increased metabolic demands associated with wound healing require the formation and sprouting of new blood vessels to supply the surrounded tissue with oxygen and nutrients [11, 13–15].

In order to identify new angiogenic factors and to investigate how these factors are orchestrated during wound healing, skin samples from wounded and unwounded skin were obtained and genome-wide gene expression patterns were analyzed. A small set of genes that were differentially expressed during wound healing was identified. To characterize these genes regarding the angiogenic process in more detail, siRNA experiments using the scratch wound assay were performed. Thereby, we identified genes with a reduced proliferation and/or migration capacity after knockdown, suggesting that these genes—which have not yet been associated with angiogenesis—might be involved in dermal vasculature regeneration.

Materials and methods

Ethics statement

The recommendations of the current version of the Declaration of Helsinki as well as the international guidelines (e.g., FDA Regulations, GEP and AWB guidelines) were observed. Suction blistering was approved by the Freiburger Independent Ethics Committee (011/1973). Punch biopsies were obtained through a clinical study approved by the Hamburger Independent Ethics Committee (PV3424). Written informed patient consents were given from all volunteers.

Wounding of the skin by suction blistering and taking off the epidermal skin layer

In order to wound the skin, the epidermis was removed by suction blistering as described previously [16].

Tissue samples

Full-thickness skin samples (6 mm in diameter) were obtained from 14 volunteers by Sciderm GmbH (Hamburg, Germany). Punch biopsies of each volunteer were taken from wounded (suction blistering and removal of epidermis, sb) and unwounded skin (co) at four different time points (2, 4, 6, and 16 weeks after wounding). The wounded skin did not display any swelling or infection. All biopsies were isolated from the upper buttock to minimize depot specific variations.

In vivo capillary microscopy

In order to measure the density and morphologic patterns of capillary loops in the dermis, noninvasive capillary microscopy was performed using a CAM1V Video capillary microscope (Lawrenz Medizinelektronik). In total, eight female volunteers aged 66–72 years participated in the study. Before the investigations began, the patients were acclimated in a room with a constant temperature of 21 °C and 40 % humidity for 10 min. A drop of paraffin oil was applied to the skin to minimize reflection. To increase visibility of capillaries, venous stasis of 120 mm Hg on the forearm was done to maximize capillary blood load. Pictures were taken with the VRmagic CamLab Programm. Per volunteer and area, 7–9 pictures (0.65 mm2) with 240-fold magnification were captured. Analyses of the pictures were performed with manual counting plus algorithmic calculation of the red area in percentage via the software ImageJ.

RNA isolation of skin biopsies

RNA from punch biopsy samples was processed with the NucleoSpin RNA II (Macherey–Nagel, Dueren, Germany) at the company ImaGene GmbH, Berlin. The biopsies, stored in liquid N2, were put in −80 °C cooled tubes with glass beads and 4 µL 1 M dithiothreitol (DTT). After addition of 180 µL of lysis buffer (RA1), the tubes were restored in liquid nitrogen. To each sample, 170 µL RA1 buffer was pipetted and vortexed at 4 °C for 15 min to gain homogenization. Following, RNA was isolated as recommended by the supplier.

Quantitative RT-PCR

The TaqMan Gene Expression Assay (Applied Biosystems, Foster City, USA) was used to analyze the knockdown efficiency of the siRNA-transfected endothelial cells. RNA was reverse-transcribed into cDNA, and PCRs were carried out in triplicate as recommended by the manufacturer. Quantitative RT-PCR of the genes ABLIM2 (Assay ID Hs00287164_m1), FGFR-1 (Assay ID Hs00915142_m1), GGCT (Assay ID Hs01031800_g1), SULF2 (Assay ID Hs01016476_m1), TMEM (Assay ID Hs00846437_s1), and ZSCAN18 (Assay ID Hs00225073_m1) (all Applied Biosystems, Foster City, USA) were performed using the Applied Biosystems 7900HT Fast Real Time PCR System. C T values were calculated by the RQ Manager software 1.2. Target gene expression was normalized on expression of the endogenous control GAPDH (Assay ID Hs99999905_m1), and calculations were performed using the comparative C T method [17].

Cultivation of human dermal endothelial cells

Primary microvascular endothelial cells isolated from female skin were obtained from Lonza (Walkersville, USA). Cell populations of three donors aged between 32 and 50 were pooled (Lot 0000122821, 0000125583, and 0000125028) and used in passage 4. Endothelial cells were cultured as previously described [18].

SiRNA transfections of endothelial cells using Lipofectamine

SiRNAs (Qiagen, Hilden, Germany) and lipofectamine (Lipofectamine 2000, Invitrogen) were mixed according to the manufacturer’s instructions but only with half of the denoted lipofectamine volume. Multiwell pipetting was performed using Biomek FXp Laboratory Automation Workstation (Beckman Coulter, USA). The mix was loaded into 96-well plates with a final concentration of 50 nM siRNA. Manually, a total of 7500 cells were plated directly into the transfection mix. The next day, cells were washed and medium was changed to Endothelial Cell Basal Medium-2 (EBM-2) (Lonza, Walkersville, USA) containing 5 % fetal bovine serum (FBS) and gentamycine (GA) (both Lonza, Walkersville, USA). After 24 h of incubation, scratch wound assay was performed.

Three different siRNAs per target were investigated, while siFGFR-1 (SI03066098) served as a positive control and the scrambled siRNA (SI03650325) as a negative control (both obtained from Qiagen, Hilden, Germany) on every 96-well plate. The entire library of 144 siRNAs (for sequences, see supplemental table S2) was screened twice.

Scratch wound assay

After 48 h post-transfection, a scratch was created in the confluent cell monolayer using pipette tips (Biomek AP96 P250, Beckman Coulter, USA) driven by Biomek FXp Laboratory Automation Workstation (Beckman Coulter, USA). Afterwards, plates were washed once with PBS and refilled with EBM-2 medium supplemented with 10 ng/mL of recombinant fibroblast growth factor 2 (FGF-2) (Gibco, Karsruhe, Germany). Migration and proliferation of transfected cells were measured at a fixed time point 30 h after scratching. The monolayer of dermal microvascular endothelial cells was washed with phosphate-buffered saline (PBS, Lonza, Walkersville, USA) and fixed with 2 % formaldehyde solution (36.5–38 % formaldehyde solution from Sigma–Aldrich, Taufkirchen, Germany) for 15 min. The cells were stained with propidium iodide (5.5 µg/mL, Invitrogen, Karlsruhe, Germany) and analyzed using a fluorescence microscope Axiovert S100 (Carl Zeiss Mikroskopie, Goettingen, Germany). Migration/proliferation rates were monitored by measuring the reduction of cell-free area by thresholding for dark objects at Wimasis Image Analysis GmbH (Munich, Germany). The rate of migration and proliferation was measured by quantifying the total area that the cells moved from the edge of the scratch toward the center of the scratch. Directly after wounding, the scratch area was determined in 36 repeats and regarded as a constant with 5.72 mm2.

Cell viability assay

After siRNA transfection, cell viability was measured using Cell-Titer-Blue-Cell Viability Assay (Promega, Fitchburg, USA). The protocol was performed according to the manufacturer’s instructions. The reagent was added 48 h after transfection to the cells, and viability was monitored. Cells treated for 24 h with cytotoxic cis-diammineplatinum(II)dichloride (10 µg/mL) (Sigma-Aldrich, Taufkirchen, Germany) served as a negative control.

Caspase-Glo 3/7 assay

After siRNA transfection, apoptotic activity was measured using the Caspase-Glo 3/7 Assay (Promega, Fitchburg, USA). The protocol was performed as recommended by the supplier. Transfected cells were analyzed 48 h after knockdown. Cells treated for 24 h with staurosporine served as negative control.

Array-based gene expression analysis

In order to identify differentially expressed genes in wounded and unwounded skin, samples were analyzed using one-color Whole Human Genome Microarray 4 × 44 K chips from Agilent Technologies which were carried out by ImaGene GmbH (Berlin, Germany).

Immunohistochemistry staining (IHC) of human skin biopsies

To analyze CD31 protein expression in skin tissues, samples were paraffin-embedded, sectioned at 2–5 μm, and mounted on glass slides. Sections were deparaffinized by heating slides to 100 °C and incubating in DAKO Target Retrieval Solution (Glostrup, Dänemark, pH 6.1, diluted 1:10) for 20 min. Afterwards, slides were carefully washed with demineralized water. All immunohistochemical steps were performed using EnVision G|2 doublestain system as recommended from the supplier (DAKO, Glostrup, Denmark). Monoclonal CD31 antibodies (DAKO, Glostrup, Denmark) were diluted 1:30 in DAKO antibody-diluting agent and incubated on slides for 30 min. The hematoxylin staining was performed according to DAKO´s instruction. After staining, sections were air-dried at RT. In the following, sections were sealed with Leica CV Ultra mounting medium (Nussloch, Germany) and glass coverslips.

Immunofluorescence microscopy of human skin biopsies

To verify expression of identified genes on protein level in wounded skin, appropriate paraffin-embedded tissue Sections. (2–5 μm, mounted on glass slides) of punch biopsies obtained from wounded skin (suction blister) were stained by immunofluorescence technique. TMEM121 and ABLIM2 primary antibodies were purchased from ABCAM (ab151077, ab100926, Cambridge, Great Britain). Endothelial cells were stained using CD31 antibody from DAKO (M0823 Glostrup, Denmark). AF488 and AF594 from Life Technologies GmbH (Darmstadt, Germany) were used as secondary antibodies. Sections were deparaffinized by a 100 °C incubation in DAKO Target Retrieval Solution (Glostrup, Denmark, pH 6.1) for 20 min. Afterwards, slides were carefully washed with demineralized water and stained as follows: blocking with antibody dilution buffer (DCS Innovative Diagnosik-Systeme, Hamburg, Germany) for 90 min, treating with primary antibody for 60 min, washing two times with antibody dilution buffer for 5 min each, incubating with secondary antibody for 60 min, washing once as described above, and finally staining the nuclei with Hoechst 33342 (Life Technologies GmbH, Darmstadt, Germany). Between all steps, slides were rinsed thoroughly with DPBS (Sigma-Aldrich, Steinheim, Germany).

Data deposition

The microarray data set is available in the Gene Expression Omnibus database http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=ctipmieixtsrhcd&acc=GSE56803, under the accession number GSE56803.

Statistical analyses

The preprocessing and quality control of microarray data were performed as previously described [18]. To identify genes differentially expressed in wounded versus unwounded skin, the limma package for the R statistical software was used. Few samples were excluded because they showed “outlier” characteristics. Supplement Table 1 summarizes the information about the cleaned set of samples.

For the quantitative RT-PCR analyses of the genes ABLIM2, FGFR-1, GGCT, SULF2, TMEM121, and ZSCAN18, as well as for the FGF-2-induced cell migration, and for the scratch wound analyses, Statistica 9.1 (StatSoft, Hamburg, Germany) and GraphPad Software Prism 5 (La Jolla, USA) were used. Normal distribution was checked by means of Shapiro–Wilk’s test, and unpaired t test was performed. A significance level of 0.05 (alpha) was chosen for statistical analyses, based on two-sided hypothesis testing.

For analyses of vascular changes after suction blistering and removal of the epidermis, GraphPad Software Prism 5 (La Jolla, USA) was used and repeated-measures ANOVA was performed. A significance level of 0.05 (alpha) was chosen for statistical analyses. Functional categories and pathway analyses of differentially expressed genes were generated by the use of IPA (Ingenuity® Systems, www.ingenuity.com).

Results

Capillary density is increased after human epidermal wounding

To establish an appropriate in vivo model to analyze angiogenic processes in the human skin, a pilot study was conducted. Therefore, we wounded human skin by removing suction blister roofs from the upper arm of eight human volunteers [19, 20]. Besides ethical reasons, the advantage of wounding only the upper skin layer is that even to an early time point (after 2 weeks) skin was completely healed and the epidermis was regenerated (Fig. 1a). As the epidermal compartment lacks vessels and is supplied with oxygen and nutrients by the papillary dermal vasculature [21–23], we first investigated whether the wounding of the epidermis is sufficient to stimulate dermal vessel growth and remodeling. Therefore, vessel density was examined before wounding, and 2, 4, 6, and 16 weeks after wounding using capillary microscopy (Fig. 1b). Unwounded skin of volunteers aged between 66 and 72 years (control areas) was characterized by unstructured morphological arrangements and an average capillary density of 13 capillaries per mm2 (Fig. 1c). In contrast, 2, 4, and 6 weeks after wounding, the microvessel density in the dermis increased significantly compared to control areas. In addition, a more structured and organized morphology and distribution was observed. The maximum increase in microvessel density (28 capillaries/mm2) was determined 2 weeks after wounding. After 16 weeks, the number of capillaries decreased comparable to control levels. To verify the results with an independent analysis method, we determined the percentage of total visible vessels using the same sample set. After 2, 4, and 6 weeks post-wounding, the “visible vessel area” increased significantly from 6.8 to 17.0 % per mm2 (Fig. 1d). Again, after 16 weeks post-wounding, the visible vessel area was insignificantly elevated, indicating that vessel formation is induced and stable up to 6 weeks after suction blistering.

Fig. 1
figure 1

Vascular changes after epidermal wounding. a Images of IHC staining (CD31) of biopsies obtained before wounding, 2 and 16 weeks after suction blistering. Bar: 200 µm. b Images captured by capillary microscopy are shown for one representative volunteer at five different time points. Per volunteer, 7–9 pictures were captured each, with an area of 0.65 mm2. c Numbers of capillaries and d area of visible vessels per mm2 were quantified to the according time points. The horizontal black lines denote medians and whiskers the 2.5th and 97.5th percentiles. Significant differences are indicated by asterisks. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (repeated-measures ANOVA). n = 8

Full size image

Taken together, removing the epidermis without wounding the dermis is sufficient to stimulate capillary reorganization and/or regeneration.

Identification of differentially expressed genes after epidermal wounding

In order to identify genes being involved in angiogenesis during human wound healing, we determined the mRNA transcription profile of unwounded and wounded skin using Agilent Whole Human Genome Microarrays, which allow the analysis of 45,015 gene probes. For the analyses, ten female and four male volunteers were wounded by suction blistering, and 2, 4, 6, and 16 weeks after wounding, punch biopsies were obtained from wounded (sb) and unwounded skin (co) of each volunteer (Fig. 2a). The time points have been chosen because the epidermis of wounded skin needed at least 2 weeks to recover completely. Therefore, an earlier time point—with an incomplete recovered epidermis—would have caused biased microarray results, simply because different tissue compositions would have been analyzed (Figs. 1a, 2b). Before starting the analyses of the transcription profiles, the microarray data were preprocessed and filtered to remove low-quality probes. Next, the processed data were subjected to a panel of sample quality metric functions (supplement Fig. S1). The processed microarray data (table S1) comprised on each array 30,317 significant gene probes, which corresponded to 15,222 genes. In total, 810 transcript variants were significantly differentially expressed between co and sb samples during the time course of 16 weeks, as visualized by the heat map (Fig. 2c). The 810 transcript variants corresponded to 577 up-regulated and 233 down-regulated transcripts in sb compared to co samples (Benjamini Hochberg adj. p val <0.01). The strongest expression change was detected 2 weeks after wounding. After 4 and 6 weeks following wounding, expression change intensities decreased, resulting in a comparable level between co and sb samples after 16 weeks. The course of expression changes is therefore in good agreement with the changes in microvessel density which also decreased at later time points (Fig. 1c, d). The 810 transcript variants could be assigned to 616 genes. Only three gene probes were differentially expressed between male and female samples, indicating that gender-specific differences play only a minor role.

Fig. 2
figure 2

Differentially expressed genes during human wound healing. a Schematic overview of the combined wounding (suction blistering) and punch biopsy study. Two, 4, 6, and 16 weeks after suction blistering, full-thickness skin samples were obtained and RNA was isolated for microarray analyses. b Images of wounded skin areas are shown for two representative volunteers (before wounding (d0), 7 and 14 days after suction blistering (d7 and d14). c The heat map shows the 616 differentially expressed genes between wounded (sb) and unwounded (co) skin samples at the four different time points. The rows are scaled according to increasing p values (BH adj. p ≤ 0.01). d Top five functional categories for differentially regulated genes in wounded (sb) versus unwounded (co) skin samples derived from Ingenuity Pathway Analysis (IPA). e Top network built by up-regulated and down-regulated genes in sb samples obtained by IPA. Genes with a blue asterisk are expressed in human dermal microvascular endothelial cells

Full size image

Supplement table S2 summarizes those genes out of the 616 differentially expressed genes between co and sb samples that are known to be involved in angiogenesis. To further analyze the biological relevance of the 616 differentially expressed genes, network and functional analyses were performed using IPA (Ingenuity Pathway Analysis, Ingenuity® Systems, www.ingenuity.com). Differentially expressed genes were enriched in functional categories such as “dermatological disease and conditions, genetic disorders, cancer, embryonic development,” and “hair and skin development and function” (Fig. 2d). These functions were further specified by subcategories such as “development of epidermis, synthesis of melanin, differentiation of epidermal cells and melanocytes, and innervation of skin.” The third top-score network named “cellular movement, tumor morphology, cell to cell signaling and interaction” (Fig. 2e). Interestingly, a variety of interacting proteins of the vascular endothelial growth factor (VEGF) family (f. e. VEGF-A, VEGF-B, VEGF-C, VEGF-D) were up-regulated or down-regulated during the 16-week progress of wound healing. VEGF proteins play important roles in angiogenesis [24–26], and although the members of this family were not directly affected in our model system, it can be speculated that the deregulation of interacting proteins could contribute to angiogenesis.

Overlap between differentially expressed genes during wound healing and gene expression in cultured endothelial cells

For our analysis, full-thickness skin samples have been used containing a variety of different cell types such as keratinocytes, fibroblasts, immune cells, and endothelial cells. To investigate whether the differentially expressed genes are also expressed in cultured human microvascular ECs, we compared the 616 differentially expressed genes during wound healing to 13,406 genes which are known to be expressed in cultured endothelial cells [18]. A total of 438 genes overlapped, as indicated by the Venn diagram (Fig. 3), suggesting that a great proportion (>70 %) of differentially expressed genes might be relevant for endothelial cell behavior. To further characterize this gene set, we selected those genes that showed in addition a twofold change in gene expression during wound healing. In total, 48 genes have been selected by these criteria (table S3 and S4).

Fig. 3
figure 3

Comparative gene expression analysis. The Venn diagram shows the overlap of genes with a differential expression pattern in wounded (sb) versus unwounded (co) samples as well as expression in human dermal endothelial cells

Full size image

Functional characterization of differentially expressed genes during wound healing using siRNA transfection in cultured endothelial cells

To further analyze the functional relevance of the 48 selected genes on angiogenesis, siRNA-induced knockdown experiments have been conducted in primary endothelial cells. To characterize the endothelial phenotypes after siRNA treatment, we performed scratch wound assays. This method considers two aspects of wound healing as endothelial cells migrate and proliferate after activation by FGF-2 into a denuded area to close the wounded cell layer [27, 28].

As a positive control for impaired endothelial migration and proliferation, we targeted the receptor of FGF-2, FGFR-1 (FGF-2 receptor, fibroblast growth factor receptor 1) [28]. The siRNA-induced knockdown of FGFR-1 resulted in down-regulation of FGFR-1 mRNA up to 70 % (supplement Fig. S2 a). Compared to scrambled siRNA treatment (set as 100 %), knockdown of FGFR-1 significantly reduced endothelial cell proliferation and/or migration as the scratch area was only closed by 57 % (supplement Fig. S2 b). Viability was comparable between transfected and control-transfected cells, indicating that the reduced wound closure was not caused by an increased cell death (see supplement Fig. S2 c).

In the following, we used the scratch wound assay to analyze the knockdown phenotypes of the 48 genes. To reduce off target effects, 3 siRNAs per gene were investigated, and each of the 144 siRNA transfections was performed in sextuple. To be scored as a hit, at least 25 % decreased wound closure has to be observed after transfection with two different siRNAs. In total, five genes—ABLIM2 (actin-binding LIM protein family, member 2), ZSCAN18 (zinc finger and SCAN domain containing 18), GGCT (γ-glutamylcyclotransferase), SULF2 (Sulfatase 2), and TMEM121 (transmembrane protein 121)—have been identified, which showed a decreased migration and/or proliferation capacity in primary human endothelial cells (Fig. 4a–e). The significant reduction of mRNA levels after knockdown of these genes can be seen in supplement Fig. S3 a–e. In addition, we could exclude the possibility that the observed phenotypes have been caused by an increased cell death of the transfected cell populations. Neither cell viability nor apoptotic activity was significantly affected after knockdown of the corresponding gene (supplement Fig. S4 a and b).

Fig. 4
figure 4

Genes showing a pro-angiogenic phenotype in wound healing assay in vitro. Primary blood endothelial cells were transfected with three different siRNAs per gene, indicated by the number 1, 2, or 3 (only the two most potent ones are depicted). After 48 h post-transfection, a scratch was created in the confluent cell monolayer and the Relative Wound Density (RWD) was measured in % 30 h post-wounding. The RWD of scrambled siRNA-transfected cells was set as 100 %. The RWD is shown for the knockdown of genes a ABLIM2, b GGCT, c SULF2, d TMEM121, and e ZSCAN18 induced by the two most effective siRNAs. The average value of the scrambled control is shown as the dotted line. The horizontal black lines denote medians and whiskers the 2.5th and 97.5th percentiles. Significant differences are indicated by asterisks. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 (unpaired t test). n (per siRNA) = 6

Full size image

To investigate whether altered proliferation or migration resulted in the observed phenotype, we determined the total cell number by propidium iodide staining (supplement Fig. S5). The siFGFR-1-transfected cells showed an average decrease of 26 % in cell number compared to scrambled siRNA-treated cell populations. However, the average decrease in wound closure was calculated to 57 %. The knockdown of the other genes impaired cell proliferation as well. For all genes, we observed that the phenotype “Reduced wound closure” was slightly more prominent than the proliferation defects (reduced cell number). The results of all genes are summarized in supplement Fig. S5. Therefore, we speculate that the identified genes affect not only EC proliferation but also EC migration although to a lower extent and could therefore contribute to two important processes during angiogenesis.

Histological analysis of AMBIM2 and TMEM121 during wound healing

To determine whether the identified genes are expressed during wound healing, histological analyses were performed using skin samples taken 2 weeks after wounding. For GGCT, SULF2, and ZSCAN18, a variety of different antibodies have been tested. Unfortunately, we could not detect a specific staining for those proteins. For ABLIM2 and TMEM121, a colocalization with CD31-positive structures (Fig. S6 A-D) was observed, suggesting that these two proteins might be involved in the regeneration of the dermal vasculature. For GGCT, SULF2, and ZSCAN18, additional analyses are needed to clarify their exact role during wound healing.

Discussion

To analyze angiogenic processes during cutaneous wound healing, we removed the epidermis without wounding the dermis to stimulate capillary reorganization. While we observed a recovery of the initial state already after 16 weeks, wounding of the epidermis and dermis led to an increase in capillary density even after 52 weeks post-surgery as Brown et al. described [29]. This difference can be explained by the depth of wounding and might correlate with stronger wound healing reactions. However, the advantage of wounding only the upper skin layer is that even to an early time point (after 2 weeks) skin was completely healed and skin biopsies of wounded and unwounded skin were homogenous in structure and thus comparable for genome-wide mRNA expression profiles.

So far, there are only a few dermatological studies available, analyzing genome-wide expression patterns of wounded and unwounded skin in humans. One study summarized results regarding angiogenesis in chronic human wounding. The authors used laser microdissection to isolate blood vessels from unwounded and wounded human skin [30]. The overlay between this analysis and our data set is minor which is most likely caused by the different study designs.

Several microarray experiments have been performed in mice to identify genes that are differentially expressed during wound healing [31, 32]. In one of these studies, Roy and coworkers investigated the acute temporal changes in excisional murine cutaneous wound inflammation by analyzing the wound-edge transcriptome [33]. Although rodents have dense hair, relatively thin epidermal and dermal layer, and wound healing happens mainly by contraction and not by re-epithelialization [34], an overlap of differentially expressed genes has been observed between both studies. Interestingly, five genes named CCL3, MX1, ARG1, MARCKSL1, and PIK3CD are overexpressed in wounded skin of mice and humans. All genes are highly expressed in macrophages which are known to be important players in wound healing [35]. However, it should be noted that these similarities have only be observed on the mRNA expression levels. Therefore, the biological relevance for mice and humans has to be determined in more detail in future studies.

To investigate the molecular processes of human angiogenesis more precisely, we selected genes (1) showing at least a twofold change in gene expression during wound healing and (2) being expressed in primary human endothelial cells. To test the functional relevance, siRNA experiments have been performed. The knockdown phenotypes were analyzed using the well-established scratch wound assay [28]. We could identify five genes that—to the best of our knowledge—have not yet been described to be involved in endothelial cell proliferation and/or migration. The expression patterns of these genes during the wound healing process are shown in detail in supplement table S5. In addition, histological analysis revealed that ABLIM2 and TMEM121 colocalize with CD31-positive structures (supplement Fig. S6). Although all genes are expressed in a variety of different cell types (supplement Fig. S7), most of them are rarely described in the scientific literature. For ZSCAN18 (zinc finger and SCAN domain containing 18) and GGCT (γ-glutamyl cyclotransferase), it is known that they are involved in proliferation and/or migration. However, these analyses have been performed in cancer or primary kidney cells [36]. TMEM121 (transmembrane protein 121) has been shown to be a repressor of MAP kinase signaling [37]. For ABLIM2 (actin-binding LIM protein family, member 2), it is speculated that this gene is important in the guidance of axons [38]. As endothelial sprouting and axonal growth share some similarities [39], ABLIM2 might also play a role in the guidance of tip and neighboring cells. SULF2 (sulfatase 2) was another interesting gene that has been identified in our study. Factors of the FGF family act with heparin and heparan sulfate proteoglycans to activate FGF receptors [40]. This interaction depends among others on sulfatases, which was f. e. studied in chondrocytes [41]. A contrary view was published by Peterson et al. [42], showing that SULF2 impairs tumor growth without any effect on vessel density. This contrary observation might be explained as cancerous blood vessels show divergent cellular abnormalities [43, 44]. Heterozygous SULF1−/− and SULF2−/− mice suffer from reduced neurite sprouting [45], and again a parallel between vessel and nerve guidance might be drawn.

Although future studies will be needed to clarify the exact role of these genes in stimulating angiogenesis during wound healing, the set of differentially regulated genes which additionally has impact on endothelial cell behavior in vitro may provide important insights into how human wound angiogenesis is orchestrated.