Histological and Transcriptional Expression differences between Diabetic Foot and Pressure Ulcers

Yssel Mendoza-Mari1, Calixto Valdés-Pérez2, Elena Rodríguez-Corrales3, José Suárez-Alba4, Ariana García-Ojalvo1, Diana García del Barco1, Gerardo Guillén-Nieto1, Luis Herrera-Martínez1 and Jorge Berlanga-Acosta1* 1Wound Healing, Tissue Repair and Cyto-protection Laboratory, Center for Genetic Engineering and Biotechnology, Cuba 2Diabetic Angiopathy Ward, National Institute of Angiology and Vascular Surgery, Cuba 3Diabetic Angiopathy Ward, National Institute of Dermatology “Dr. Ramón Hernández Vaquero”, Cuba 4Department of Pathology, Center for Genetic Engineering and Biotechnology, Cuba


Introduction
Pressure or decubitus ulcers (PU) and diabetic foot ulcers (DFU) are among the most common chronic wounds representing individual clinical entities but sharing a significant socioeconomic impact [1]. These chronic ulcers translate in systemic repercussion as they may act as pro-inflammatory and pro-oxidant organs super-imposed to a host that could evolve to a chronic low-grade inflammatory response [2].
Pressure ulcers continue to be a worldwide common health problem, particularly among the neurologically impaired or bedridden elderly. Its medical history dates back to the renaissance when the French surgeon Ambrose Paré enlisted; nutrition, pain relief and debridement as key factors to heal these ulcers. Surprisingly, it is not quite different than the present modality to some extent [3]. To date, a large percent of grades 3 and 4 pressure ulcers evolve to chronification and lead the patient to death due to ulcer complications such as sepsis or osteomyelitis [4].
From the ethiopathogenic perspective, cutaneous ischemia appears to be the proximal trigger of a downstream cascade of molecular events that converge to impose a chronic evolution. These include the overactivation of molecular regulators toward a pro-apoptotic program [5], imbalance in matrix metalloproteinases regulation (MMP2 and MMP9) [6], and adhesion molecules overexpression [7]. At the experimental level, a reduction of the cutaneous cells constitutive expression of endogenous cytoprotective molecules such as hypoxia-inducible factor-1 alpha (HIF-1α), VEGF, HSP 70 and 90 and hemeoxigenase-1 [8] have been demonstrated. Despite the medical and social problem that the pressure ulcer represents, still remains a paucity of data aimed to characterize its molecular pathophysiology in clinical samples.
Within the diabetic context, hyperglycemia stands as the proximal ethiopathogenic trigger for the onset and progression of biochemical disturbances that steers the systemic complications, in which ulcer healing failure and amputation are included. The vicious circle between wound chronicity and deficient local infection control, determines that 85% of all non-traumatic lower extremity amputations are rendered by diabetic subjects [9]. At the molecular level, this high glucose burden unleashes acute and chronic self-perpetuating loops, which include ROS-lipid peroxidation, hyperinflammation/disimmunity, AGE-RAGE toxicity, mitochondrial dysfunction, nytrosilation end-products accumulation, FOXO genes deregulation and a perspicuous reduction of growth factors physiology. The concerted action of these elements enforces granulation tissue-producing cells to senescence, arrest and apoptosis. Furthermore, the failure of the agonistic stimulation of the insulin axis and other tyrosine kinase receptors, negatively impacts on the biology of diabetic wound cells [10].
This study intends to be a modest contribution to the characterization of the chronic wound microenvironment, by characterizing and comparing granulation tissue samples histology, as the gene expression profiles in these two common and devastating types of human chronic wounds.

Experimental Procedures Population and study enrollment criteria
The enrollment of patients was based on the following criteria addressed to preserve as "clean" as possible the experimental substrate for a neat constitutive gene expression profile. Diabetic co-morbids and pressure ulcer-affected patients with concurrent clinical conditions, requiring medication that we understood as potentially gene expression modifiers were not enrolled in the study. Thus, for diabetics, only those with oral hypoglycemic treatment were sampled. Insulin-dependent and uncompensated patients were not included. Furthermore, patients with clinical depression, heart failure, renal insufficiency, unstable angina, asthma, angiotensin-converting-enzyme inhibitors-medicated, and subjects with auto-immune processes were excluded. As all the pressure ulcer samples population had a spinal injury (lumbar segment) as the primary process, the only subjects included were those free of lower limbs hyperspastic episodes, simply receiving physical therapy/postural rotation and no pharmacological intervention. Obviously, patients with malignant diseases and history of chemotherapy were discarded. No pregnant or breast feeding women were conceived to be included.
All the patients were consulted about the experimental procedure and approved to offer a biopsy fragment for the study. For the chronic wounds, the granulation tissue fragment was obtained from a productive area, with no macroscopic evidence of infection, during a routine wound debridement and cleansing. The investigation material was collected from: (1) sacrolumbar pressure ulcers (stage IV) from nonemaciated patients with history of spinal trauma [11], and (2) diabetic foot ulcers of both neuropathic and ischemic origin from reasonably compensated in-hospital patients [12]. As control of an acute wound we used a fragment derived from granulation tissue, obtained from a healthy female donor (20-yrs old) who experienced wound dehiscence following esthetic breast surgery. The sample was collected during the final surgical wound closure when the exuberant granulation was trimmed.

Tissue processing
Six-millimeter diameter punch biopsies were collected days after a series of surgical debridements and ulcers bed preparation, as part of the standard wound care procedures; using disposable sterile biotomes (Acuderm Inc., USA), washed in normal saline to remove blood and immediately hemisectioned along its longitudinal axis. A hemisection was fixed in 10% buffered formalin, horizontally paraffin-embedded, processed, and 5-µm slides stained with H/E and Mallory trichrome for a better definition of collagen accumulation. The other hemisection was preserved in RNA later solution (Ambion, Life Technologies, USA) and processed for gene expression analysis.

Gene expression analyses by semiquantitative RT-PCR
Total RNA was purified according to TRI Reagent standard procedure (Sigma, USA), followed digestion with RQ1 DNase I (Promega, USA) to remove contaminating genomic DNA. Afterward, 500 nanograms of DNA-free RNA were reverse transcribed using a commercial available kit (Omniscript RT, Qiagen, Germany) with oligo-dT primer. The RT reaction was performed at 42°C for 60 min. PCR mixtures contained 2 μL cDNA, 1 μL of each primer (10 μM), 12.5 μL 2X Taq MasterMix (Qiagen, Germany) in a final volume of 25 μL. Specific sense and antisense primers, annealing temperatures and number of repeating cycles are referred in Table 1. Amplifying conditions were performed as followed: a first step of 95°C for 5 minutes, thereafter repeating cycles comprised of 95°C for 30 seconds, specific annealing temperature for 30 seconds and 72°C for 30 seconds, and a final extension step of 5 minutes at 72°C. PCR bands (8 µL of PCR product with 2 µL of gel loading buffer) were resolved on a 1.5% (w/v) agarose gel electrophoresis and visualized under ultraviolet light subsequent to being stained with ethidium bromide. PCR products were quantified using the Kodak ID 3.6 software package (Kodak Inc, USA). Ribosomal subunit 18S was used as housekeeping gene for normalization. Despite the broad spectrum of biological functions of some of the genes studied, for practical reasons we have grouped them as follows:

Statistical analysis
Statistical analysis of normalized values was carried out using GraphPad Prism 6 for Windows, version 6.01. For gene expression data, normal distribution (Kolmogorov-Smirnov) and variance homogeneity (Brown-Forsythe) tests were performed. Once normality was demonstrated, differences between pressure and diabetic foot ulcers for each gene were analyzed using two-tailed unpaired Student's t-test. In all cases, p-values <0.05 were considered statistically significant. The values shown represent mean ± S.D. (error bars).

Demographic characterization of the study population
As shown in Table 2, diabetic foot granulation tissue donors included both genders; in contrast, pressure ulcers samples were solely from male patients who had suffered spine compressive traumas. Although the study population is small, the racial profile is representative of the Cuban ethnicity and for the case of diabetics; the basic disease duration almost tripled the one detected for pressure ulcers. Thus, we succeeded in confronting a substrate of chronicity for the basic disease as for the wounds evolution age. As for the later, all the wounds studied met clinical criteria of non-infection, chronicity given the ulcer bed heterogeneity, and of complexity given the strata depth involvement as illustrated by the grades and stage of each classification scale. Importantly, diabetics' derived samples included the most important ethiopathogenic classes: neuropathics and ischemics (above 0.5 ankle-brachial index). Furthermore, the samples were taken from a selected population of relatively well-compensated subjects in which at least the acute effects of a hyperglycemic burden may not be an interference factor for gene expression studies. We also ensured to include only pressure ulcer patients with normal glycemia. All these measures, although somewhat tended to reduce the sample size, were necessary to render confident experimental data.

Histological imprinting in pressure and diabetic foot ulcers
Pressure ulcers, ischemic and neuropathic diabetic foot ulcers show particular histological aspects that seem to distinguish each of these lesions. The granulation tissue sections from the group of pressure ulcers appeared characterized by the presence of thick, densely packed, wavy, and Mallory positive collagen fibers. Scattered fat vacuoles embedded within the granulation tissue were seen along with an outsized hyperplasic epidermal layer at the wound edge indicating epithelial edge migration abortion. Moreover two main qualitative findings appeared notorious: (1) the limited amount of immunoinflammatory cells infiltrating the granulation tissue as compared to other forms of chronic wounds, and (2) the relative scarcity of fibroblasts in relation to the amount and density of the collagen deposited. The fibroblasts population appeared polarized within the field of evolving collagen bundles. In other microscopic fields, closer to the wound surface, irreversible nuclear changes in fibroblasts were distinguished ( Figure 1A and 1B). Ischemic diabetic foot ulcers-derived granulation tissue exhibited the onset of a matrix of hyaline aspect and general scarceness of functional neovessels. An impressive constellation of neo-angiogenesis defects appears since early stages which can be described as: endothelial cells co-opting failure, precocious thickening of vascular wall with hypertrophic tunica media, endothelial cells nuclei hypertrophy, perivascular collagenization, venular lumen collapse, and thick fibrin-like perivascular cuffs. Furthermore a diffuse infiltration of mononuclear cells was observed. Some of these features are shown in Figure 1C. In sharp contrast to the ischemic ulcers-derived material, neuropathic lesions appeared to granulate earlier; exhibiting a poor collagen deposition as a relatively reduced density of extracellular matrix-producing cells ( Figure 1D). These neuropathic ulcers matrix often provided the image of a spider web, made up by thin collagen fibers that weakly reacted to Mallory staining. Definitively a substantial difference between both types of wound extracellular matrix was observed. As opposed to ischemic ulcers, a large number of small capillaries were observed, often with endothelial hypertrophy and/ or peripheral fibrin cuffs suggesting hyperpermeability. As noted for ischemic diabetic ulcers, these wounds also exhibited an abnormal, mixed inflammatory infiltration.

Gene expression differences between pressure and diabetic ulcers
Differences in gene transcriptional expression were detected between pressure ulcers and diabetic foot ulcers samples. Table 3 shows only those genes with significant different expression.
EGFR and downstream signaling targets: A marginally lower although not significant difference (p=0.07) of the EGFR gene expression was detected in the diabetic patients' samples as referred to pressure ulcers counterparts. A similar observation is applicable to AKT1 gene transcription (p=0.06). Surprisingly, the expression levels of PIK3CA and its antagonistic regulator PTEN were found very similar between diabetic and pressure ulcers (p=0.12 and 0.13 respectively). The metabolic master switch MTOR appeared far more underexpressed in diabetic ulcers than in pressure ulcers (p=0.001) which is expected given the theoretical anabolic depression of the cells in diabetes ( Figure  2A).
Cell proliferation control: CCND1 expression appeared statistically similar between the two groups of wounds (p=0.17). However the expression levels of other cell proliferation-promoting genes as CDK4, PCNA, and E2F1 appeared significantly depressed in diabetic ulcers as compared to pressure ulcers (all p<0.05) ( Figures  2B-D). Furthermore, c-myc which is a well-renown cell proliferationcommitted transcription factor also exhibited a marginal reduction (p=0.06) in diabetic foot ulcers. Canonic cell cycle inhibitors as RB1 (p=0.55) and TP53 (p=0.177) showed no transcriptional differences between pressure and diabetic ulcers.

Extracellular matrix biology:
Pressure ulcers exhibited far more gene transcriptional activity for COL1A1, COL3A1 and TGFB1 than diabetic ulcers (all p ≤ 0.04) (Figure 3). No statistical differences were observed between the two ulcer groups for ILK gene expression (p=0.12).
Glucose metabolic pathways: A significant reduction in the expression of genes involved in glucose metabolic regulation in diabetic foot ulcers was detected in granulation tissue cells. Accordingly, INSR, SLC2A1, and enzymes as HK1 and HK2 isoforms, PFKP, PKM isoforms 1 and 2, and PDHA1 appeared significantly underexpressed in diabetic ulcers (all p ≤ 0.03) in relation to pressure ulcers. In biological line with the latest finding, PDK4 gene which downregulates PDH, showed higher expression in diabetic ulcers than in pressure ulcers tissue (p=0.017) (Figure 4).

Survival and anabolism:
No statistical differences were detected in FOXO1 gene transcription between diabetic and pressure ulcers (p=0.6). However it is notorious that FOXO3 gene appeared significantly underexpressed in diabetic granulation tissue as compared to pressure ulcers tissue (p=0.031) ( Figure 5A). Again, in close biological correspondence, other metabolic regulator gene as SIRT1 exhibited significantly less expression in diabetic ulcers than in nondiabetic counterparts (p=0.042) ( Figure 5B). Finally, no significant differences were noted for PPARG (p=0.88) nor PPARGC1 (p=0.088) gene expression between the two types of ulcers. Note the precocious onset of vascular anomalies including wall thickening (thick arrows) and endothelial cells nuclear hypertrophy (thin arrows) within an ischemic granulation tissue of about 10 days. An immuno-inflammatory infiltrate by round cells is typical of these types of wounds (frame). Magnification x 40, H&E staining Figure 1D: Histological aspect of neuropathic diabetic foot ulcer.

Anti-oxidant and anti-hypoxia defenses: SOD2 enzyme isoform
The extracellular matrix here is ordinarily less dense than in ischemic diabetic foot ulcers. Thin collagen fibers appeared interconnected with fibrin material. The frame includes incipient capillaries with wall distortion, luminal occlusion and endothelial nuclear hypertrophy. As indicated by the arrow precocious wall thickening takes along the angiogenic process. As mentioned in the text, at this stage, relative fibroblast scarceness is detected. An apparently active secreting fibroblast is shown in the circle. Magnification x 40, H&E staining. exhibited significantly higher transcriptional activity in pressure ulcers than in diabetic ulcers (p=0.026) ( Figure 5C). No statistical differences were observed in HIF1A gene between the two ulcers populations (p=0.9).
Pro-inflammatory response: TNF gene expression levels were similar in both forms of chronic wounds (p=0.9).

Discussion
The factors that "wipe away" from the cells the intrinsic mechanisms for a physiological repair and consequently lead to wound chronification remain unclear. In an attempt to gain further insights into the biology of chronic wounds, we have characterized the histology and compared the transcriptional activity of various wound healing-relevant genes, between diabetic and pressure ulcers as clinical paradigms of pernicious ulcers. The major limitation of this work resides in the reduced number of samples of chronic ulcers studied. Yet, the donor patients were rigorously selected so that they were representative of elemental features as ethnic homogeneity, long data of basic disease, wound chronicity and metabolic control for the case of diabetics. An additional limitation we confronted is the lack of "control" granulation tissue samples from acute, full-thickness wounds derived from nondiabetic, age-matched healthy donors. This has therefore restricted our views on the transcriptional activity of the genes studied in acute wounds from healthy subjects and excluded the possibility of statistical comparisons as a single acute wound sample was obtained. Thus, this control wound sample included is used solely as a limited reference.
As previously mentioned, we found that although a stagnant granulation process is a common factor for diabetic and pressure ulcers, each type of ulcer shows a particular morphological imprinting, thus establishing differences between pressure ulcers, diabetic-ischemic and diabetic-neuropathic; so that the two later classes appear as histological unrelated entities. The early onset of dense fibrotic matrix made up by thick collagen fibers had been previously described in pressure ulcers Gene expression was assessed by semi-quantitative RT-PCR and normalized to ribosomal subunit 18S mRNA. SD: Standard Deviation; min: Minimum; max: Maximum; dif: Difference; CI: Confidence Interval. *p< 0.05, **p < 0.01, ***p< 0.001.  [13]. However the driving forces behind this morphological appearance does not seem to be clarified. Whether these microstructural changes are a consequence of constant/repeated cutaneous pressure [14]; a sort of "lasting mechanical memory" could then be hypothesized as pressurerelief does not acutely restores the physiological healing trajectory [15]. An intriguing qualitative finding is that this fibroplastic induration appeared inversely correlated with the local fibroblast population, suggesting an imbalance between a consolidated fibrotic matrix and the matrix-producing cells. The fact that pressure ulcers-cultured fibroblasts exhibit a short replicative life becoming prematurely senescent [16] and that these cells are "homed" into a pro-apoptotogenic environment [5] incite to speculate that these wounds become "stalled" within the fibrogenic phase with no subsequent turnover or remodeling by a deficit of healthy effector cells. Globally speaking, fibroblasts dysfunction and apoptosis appear as pivotal factors toward wound chronification [17]. Although still persists scarcity of literature focusing on the histopathology of pressure ulcers [18], previous descriptions include an intense immune-inflammatory infiltrate [13]. Contradictorily, we rarely detected inflammatory infiltration across the microscopic fields. This discrepancy may be related to the characteristics of our study population: solely male, relatively younger, and non-emaciated subjects with spinal cord trauma as a basic condition, versus other studies based on elderly/debilitated patients [19].
Although persistent hyperglycemia is the common and proximal factor that disrupts fibroblasts, pericytes, keratinocytes and endothelial cells physiology thus hindering wound healing in diabetics [20,21]; it became clear that foot ulcer ethiopathogenic component (ischemic or neuropathic) is associated to a privative granulation tissue histopathological image. This finding may be a contribution of this work. To the best of our knowledge only Loots et al. [22] and Piaggessi et al. [23] had conducted histopathological studies of diabetic ulcers describing only the neuropathic lesions as "frozen" in a chronic low- grade inflammatory state associated to a scarce provisional extracellular matrix. These studies did not include ischemics-derived samples. The mechanisms mediating the differences of each type of extracellular matrix, angiogenic response, its organization and density; and even in fibroblasts cytology appear far to be elucidated. Similarly, it remains to be explained how and why, long-term evolution events like diabetic microvascular disease (i.e., media thickening or neointimal hypertrophy) can be so rapidly recreated by a growing neovessel within an incipient granulation tissue. Since the most aberrant morphological changes we found emerged from ischemic granulation tissue samples; hypoxia and its downstream biochemical disturbances could be raised as an instrumental ethiopathogenic ingredient for these vascular aberrancies.
The data derived from statistical comparisons described above indicate that the expression of different functional classes of genes, appear far more affected in diabetic patients ulcer cells rather than in decubitus ulcers counterparts. EGFR and some of its downstream phosphorylation targets were selected as paradigms of tyrosine kinase signaling pathways which may be disrupted by hyperglycemia [24]. Globally speaking our evidences support the notion that these pathways are dysfunctional in diabetic wound cells as compared to decubitus ulcers. This alteration has been attributed to glucose toxicity, receptors' non-enzymatic glycation or by the local inflammation [10,25,26] all of which appear to reduce kinase receptors phosphorylation activation. Furthermore, although we found no significant differences in the expression of two major cell proliferation cycle inhibitors (TP53 and RB1) between diabetic and pressure ulcers; critical mitosis promoters as CDK4, PCNA and E2F1 appeared underexpressed by diabetic wound cells. CDK4 and PCNA expression failure have been previously invoked as molecular markers of chronic wounds fibroblasts proliferation reluctance [27,28]. In line with these findings is that TGF-β1 appears significantly down-regulated in diabetic ulcers as compared to pressure ulcers. This growth factor is a well-reputed actor for fibroblasts proliferation, migration and secretion, which offers theoretical explanation for the remarkable transcriptional reduction in collagens I and III as for the fibroblasts proliferation promoters, observed in the diabetic wound samples [29]. As curious note with no major biological translation, the expression of all the above mentioned genes was arithmetically similar between pressure ulcers and the single acutewound granulation tissue sample.
Perhaps, the most relevant contribution of this work with no previous anticipation in the field; is that we found that diabetic granulation tissue cells, although "de novo" cells exhibit a sort of "genetic or epigenetic imprinting" for the deranged expression of glucose-metabolism related genes which have been largely implicated in type-2 diabetes pathophysiology [30]. As compared to decubitus ulcers, diabetic wound cells express far less insulin receptor, hexokinase isoforms 1 and 2, phosphofructokinase, pyruvate kinase isoforms 1 and 2, pyruvate dehydrogenase, and significantly more of its inhibitor enzyme pyruvate dehydrogenase kinase isoform 4. Interestingly, all of them have been considered as molecular markers of insulin resistance, glucose intolerance and predictors for the onset of type-2 diabetes [30]. Most remarkable is the fact that all of these and other genes deregulations had solely been described in the liver, skeletal muscle and adipose tissue as the most important glucose consumers and major insulin targets [30][31][32]. Thus, for the first time it is shown that a transient, de novo multicellular material, as the diabetic granulation tissue, somewhat "inherits" and reflects the same molecular derangements detected in the most important glucose clearance tissues/organs. In other words, granulation tissue, despite its early age, is also a diabetic tissue. Again, and as a curious comment, all these genes exhibit similar expression between decubitus granulation tissue and the single-acute granulation tissue control sample; suggesting that the above described derangements are diabetics' privative traits.
Although we found no differences between pressure and diabetic ulcers in critical genes involved in energy homeostasis and cytoprotection as PPARG and PPARGC1; a complementary finding around the metabolic disorders that prevail in diabetic granulation tissue is the significant underexpression of other important genes such as FOXO3 and SIRT1. The observation that SOD2 expression is significantly lower in diabetic granulation tissue, may contribute to explain the welldescribed pathogenic role of the pro-oxidant environment in diabetic complications [33]. The fact that FOXO transcription factors reduce the level of oxidative stress by the transcriptional induction of SOD [34] may assist us to explain why SOD2 appeared underexpressed in diabetic wound samples. A meaningful finding to suggest that diabetic granulation tissue cells may be under a negative pro-anabolic and energetic balance is supported by the complementary underexpression of SIRT1. SIRT1 is known to regulate metabolism, to stimulate mitochondrial protein synthesis and activity, as to selectively regulate FOXO to certain targets genes involved in cells self-defense [35]. Taken together these data incite to suggest that diabetic foot ulcer cells may be endowed with a myriad of molecular disorders that may encompass from insulin receptor underexpression to mitochondrial functional failure.
This study, has allowed for a more comprehensive idea of the intrinsic histological aspects of different types of chronic ulcers, has offered a broader notion of the transcriptional profile of wound healing and metabolically-relevant genes, and has further strengthened our hypothesis that wound chronification is a distal destiny reachable from different or alternative roads. It has the merit to show for the first time that not only the major glucose-clearance organs, but the young-aged granulation tissue cells also exhibit a molecular "imprinting" toward glucose homeostasis failure. Moreover, we still miss however to gain a major enlightening on the core of pressure ulcers pathobiology.