Variability of MMP/TIMP and TGF-β1 Receptors throughout the Clinical Progression of Chronic Venous Disease

Chronic venous disease (CVeD) is a prevalent condition with a significant socioeconomic burden, yet the pathophysiology is only just beginning to be understood. Previous studies concerning the dysregulation of matrix metalloproteinases (MMPs) and their inhibitors (tissue inhibitors of metalloproteinases (TIMPs)) within the varicose vein wall are inconsistent and disregard clinical progression. Moreover, it is highly plausible that MMP and TIMP expression/activity is affected by transforming growth factor (TGF)-β1 and its signaling receptors (TGFβRs) expression/activity in the vein wall. A case–control study was undertaken to analyze genetic and immunohistochemical differences between healthy (n = 13) and CVeD (early stages: n = 19; advanced stages: n = 12) great saphenous vein samples. Samples were grouped based on anatomic harvest site and subjected to quantitative polymerase chain reaction for MMP1, MMP2, MMP8, MMP9, MMP12, MMP13, TIMP1, TIMP2, TIMP3, TIMP4, TGFβR1, TGFβR2, and TGFβR3 gene expression analysis, and then to immunohistochemistry for immunolocalization of MMP2, TIMP2, and TGFβR2. Decreased gene expression of MMP12, TIMP2, TIMP3, TIMP4, and TGFβR2 was found in varicose veins when compared to controls. Regarding CVeD clinical progression, two facts arose: results across anatomical regions were uneven; decreased gene expression of MMP9 and TGFβR3 and increased gene expression of MMP2 and TIMP3 were found in advanced clinical stages. Most immunohistochemistry results for tunica intima were coherent with qPCR results. In conclusion, decreased expression of TGFβRs might suggest a reduction in TGF-β1 participation in the MMP/TIMP imbalance throughout CVeD progression. Further studies about molecular events in the varicose vein wall are required and should take into consideration the venous anatomical region and CVeD clinical progression.


Introduction
Matrix metalloproteinases (MMPs) are a large family of endopeptidases that are secreted in their latent form by different cells in the venous wall (including fibroblasts, vascular smooth muscle cells, and leukocytes) and have proteolytic activities that participate in cellular homeostasis, adaptation, Table 1 summarizes the main demographic and clinical features of the participants. Ethnicity is not presented as all participants were Caucasian. In order to control demographic and clinical variability between groups, two subsamples of the 44 participants were used to study differences in gene expression among the control, CEAP2-3, and CEAP4-6 groups (n = 29) and between the CEAP2-3 and CEAP4-6 groups (n = 31). Therefore, differences regarding sex, age, BMI, and pregnancies presented p > 0.05.

MMP, TIMP, and TGFβR Gene Expression in Healthy and Varicose Vein Walls
PCR analysis confirmed the gene expression of all MMPs, TIMPs, and TGFβRs in the vein samples, except for MMP8 and MMP13 (Figure 1). The absence of MMP8 and MMP13 gene expression was reconfirmed after using umbilical arteries and the PC3 prostate cell line cDNA as positive controls. Also, MMP1 and TGFβR1 were excluded from further qPCR analyses due to very low gene expression in the cDNA pools.
Int. J. Mol. Sci. 2018, 19, 6 3 of 14 Table 1 summarizes the main demographic and clinical features of the participants. Ethnicity is not presented as all participants were Caucasian. In order to control demographic and clinical variability between groups, two subsamples of the 44 participants were used to study differences in gene expression among the control, CEAP2-3, and CEAP4-6 groups (n = 29) and between the CEAP2-3 and CEAP4-6 groups (n = 31). Therefore, differences regarding sex, age, BMI, and pregnancies presented p > 0.05.  Figure 1). The absence of MMP8 and MMP13 gene expression was reconfirmed after using umbilical arteries and the PC3 prostate cell line cDNA as positive controls. Also, MMP1 and TGFβR1 were excluded from further qPCR analyses due to very low gene expression in the cDNA pools.    Similarly, the gene expression of TIMP4 (p < 0.001) and TGFβR2 (p = 0.019) was significantly decreased in the CEAP4-6 veins when compared to controls. Comparisons between the CEAP2-3 and CEAP4-6 vein groups ( Figure 3) from different anatomic harvest sites (tibiotarsal, saphenofemoral, and tributaries) were also performed. No significant differences in MMP, TIMP, and TGFβR gene expression were found between the two CVeD groups from the tibiotarsal junction. From the saphenofemoral junction, only MMP9 gene expression was significantly lower in the CEAP4-6 veins (p = 0.027). Finally, in varicose tributary veins, MMP2 (p = 0.002) and TIMP3 (p = 0.050) gene expressions were significantly increased, while TGFβR3 gene expression (p = 0.002) was significantly decreased in the CEAP4-6 group. Comparisons between the CEAP2-3 and CEAP4-6 vein groups ( Figure 3) from different anatomic harvest sites (tibiotarsal, saphenofemoral, and tributaries) were also performed. No significant differences in MMP, TIMP, and TGFβR gene expression were found between the two CVeD groups from the tibiotarsal junction. From the saphenofemoral junction, only MMP9 gene expression was significantly lower in the CEAP4-6 veins (p = 0.027). Finally, in varicose tributary veins, MMP2 (p = 0.002) and TIMP3 (p = 0.050) gene expressions were significantly increased, while TGFβR3 gene expression (p = 0.002) was significantly decreased in the CEAP4-6 group. Comparisons between the CEAP2-3 and CEAP4-6 vein groups ( Figure 3) from different anatomic harvest sites (tibiotarsal, saphenofemoral, and tributaries) were also performed. No significant differences in MMP, TIMP, and TGFβR gene expression were found between the two CVeD groups from the tibiotarsal junction. From the saphenofemoral junction, only MMP9 gene expression was significantly lower in the CEAP4-6 veins (p = 0.027). Finally, in varicose tributary veins, MMP2 (p = 0.002) and TIMP3 (p = 0.050) gene expressions were significantly increased, while TGFβR3 gene expression (p = 0.002) was significantly decreased in the CEAP4-6 group.

MMP, TIMP, and TGFβR Immunoreactivity in Healthy and Varicose Vein Walls
Qualitative results of IHC analysis are shown in Figure 4 and Table 2. Positive immunostaining for MMP2, TIMP2, and TGFβR2 was more consistently found in both intima and media layers of varicose and healthy veins. A closer look at the staining intensity revealed that MMP2 was decreased in all tunicae, TIMP2 was decreased in the intima and media, and TGFβR2 was slightly decreased in all tunicae of varicose veins when compared to controls.

MMP, TIMP, and TGFβR Immunoreactivity in Healthy and Varicose Vein Walls
Qualitative results of IHC analysis are shown in Figure 4 and Table 2. Positive immunostaining for MMP2, TIMP2, and TGFβR2 was more consistently found in both intima and media layers of varicose and healthy veins. A closer look at the staining intensity revealed that MMP2 was decreased in all tunicae, TIMP2 was decreased in the intima and media, and TGFβR2 was slightly decreased in all tunicae of varicose veins when compared to controls.  Table 2. Immunodetection of MMP2, TIMP2, and TGFβR2 in tissue sections of healthy, CEAP2-3, and CEAP4-6 veins (from three different regions). The scores were established in a blinded manner by two independent observers, according to the labeling observed with a microscope, as follows: −, negative; +, discrete; ++, moderate and +++, intense.

Region
Tunica Group MMP2 TIMP2 TGFβR2 Tibiotarsal junction From the comparisons between the CVeD groups, MMP2 was generally unchanged (except for the saphenofemoral junction samples where it was decreased in the intima and media of the CEAP4-6 group), TIMP2 was generally decreased in the media and adventitia but slightly increased in the intima (except for the saphenofemoral samples) of the CEAP4-6 group, and TGFβR2 appeared to have no relevant difference among the groups.

Discussion
This cross-sectional case-control study was set up in an attempt to resolve existing discrepancies and gaps in the literature regarding the role of MMP/TIMP dysregulation in CVeD pathophysiology. Specifically, the aim was to take into consideration two other variables: TGFβR expression within the vein wall and CVeD clinical progression. Moreover, unlike the majority of the studies in this field, two methodological measures were taken for the purpose of controlling additional sources of variability. Firstly, specimens were grouped and compared based on anatomic harvest site (evidence that vein source and location may be a factor in the variability has been shown previously [19]), and secondly, comparison groups were matched regarding important demographic and clinical features.
The choice of specific MMPs was based on previously published studies [29], yet only genetic data concerning MMP2, MMP9, MMP12, TIMP1, TIMP2, TIMP3, TIMP4, TGFβR2, and TGFβR3 are discussed as RT-PCR results showed no detection of MMP8 and MMP13 gene expression. Likewise, qPCR results obtained from cDNA pools of CVeD and healthy veins showed negligible MMP1 and TGFβR1 gene expression. The absence of MMP8 and MMP13 gene expression in vein samples contradicts previous findings [19,28,33]. MMP8 is frequently associated with venous ulcer healing [34,35]; however, it may not intervene in venous wall restructuring.
In our study, gene expression of MMP12, TIMP2, TIMP3, TIMP4, and TGFβR2 was decreased in CVeD veins (especially in early clinical stages) when compared to healthy veins. Elevated gene expressions of MMP2, TIMP1, and TIMP3 in varicose veins were previously described [23,25], while decreased expression of MMP2 was also reported [18,24]. Methodological differences-for example, sample size, anatomic harvest site or use/non-use of effective control samples-may partially explain the conflicting results. The disparity in genetic data among studies may also be due to the broad range of morphologic presentations of varicose vein walls (atrophic or hypertrophic segments), and the possible existence of different phases in MMP and TIMP expression/activity throughout CVeD progression. It is plausible that the imbalance of these proteins favors atrophy during a primary phase and fibrosis during a secondary phase. In view of this, vein specimens at different CVeD stages were stratified for comparison in the current study.
Significant differences were only found between healthy and CVeD veins (i.e., these differences were not present between CVeD vein groups from the tibiotarsal junction-cf. Figures 2 and 3). Despite this, we believe that in a larger sample the trends presented in Figure 3 may achieve statistical significance. We also believe that these trends give a good representation of what happens in CVeD atrophic/hypertrophic phases: a local decrease in MMP and TIMP gene expression in varicose veins from the CEAP2-3 group (during the atrophic phase) followed by a local gradual increase in MMP and TIMP gene expression in varicose veins from the CEAP4-6 group (during the hypertrophic phase). Regarding TGF-β1 receptors, decreased gene expression of the signal transducer TGFβR2 in varicose veins could suggest a counter-regulation mechanism to control chronically elevated local levels of TGF-β1, leading to a reduction in participation of this growth factor in the MMP/TIMP imbalance throughout CVeD clinical progression. This is in line with previous studies advocating a correlation between TGF-β1 enhanced expression/activity and the development of varicosities [10,12,[36][37][38]. Although TGF-β1-enhanced expression was previously found in varicose veins [36,37], its signal transducer receptor expression has not been extensively studied [39]. However, whilst our results for TGFβR gene expression may explain generally decreased gene expression of MMP and TIMP in CVeD veins (especially from early clinical stages) when compared to healthy veins, they do not explain the slightly increased gene expression of MMP and TIMP in the CEAP4-6 group (when compared to the CEAP2-3 group). It might be that another inflammatory cytokine/growth factor (e.g., interleukins, vascular endothelial growth factor, or tumor necrosis factor-α) [40,41] may play a role in linking pressure-induced leukocyte infiltration, vein wall inflammation, and alteration in MMP and TIMP expression/activity during the CVeD hypertrophic phase.
Before discussing further results regarding CVeD clinical progression, it should be noted that genetic data across anatomical vein regions were uneven ( Figure 3) and this may be important. If molecular events are not uniform in the venous system, measures of MMP and TIMP expression/activity should always be reported with reference to vein region and comparisons among vein specimens harvested from different anatomical regions might not be reliable.
Only gene expression of MMP2, MMP9, TIMP3, and TGFβR3 presented differences between CVeD groups. On the one hand, MMP2 and TIMP3 gene expressions were increased in advanced CEAP stages (from tributary veins); on the other hand, MMP9 and TGFβR3 gene expressions were decreased in advanced CEAP stages (from the saphenofemoral junction and tributary veins, respectively).
MMP2 and MMP9 have been long recognized as major contributors to proteolytic degradation of ECM [42]. Contrary to others' findings [20,23,26,43], no significant differences in gene expression of both gelatinases were found between healthy and CVeD veins, which is most probably due to the sample size. However, MMP2 and MMP9 gene expression seemed to evolve differently throughout CVeD clinical progression. This might be due to distinct response processes to different inflammatory cytokines/growth factors (other than TGF-β1) during the hypertrophic phase. Also, MMP9 (unlike MMP2) might have its preponderant proteolytic role in early CVeD stages rather than in advanced stages. This is coherent with previous studies in which an increase in plasma pro-MMP9 activity (but not MMP2) was found in response to 30 min postural blood stasis in patients with varicose veins [21]. Nevertheless, the only significant result for MMP9 was achieved in veins from the saphenofemoral region and this region might not be as reliable as the others for CVeD group comparisons (as proximal and distal segments of veins may be affected differently by the disease [19]).
The decrease in TGFβR3 gene expression throughout CVeD clinical progression is consistent with our previous supposition. If a counter-regulation mechanism to control chronically elevated local levels of TGF-β1 was in place, this coreceptor (whose main function is to regulate TGF-β1 binding and signaling through its corresponding receptors) [5,44,45] may be part of the mechanism.
With respect to IHC results, it should be explained that the selection of proteins submitted to this technique was a consequence of previous qPCR results obtained from CVeD and healthy veins: it was assumed that the proteins with higher gene expression levels (MMP2, TIMP2, and TGFβR2) were most likely to present immunostaining differences. Our results showed that MMP2, TIMP2, and TGFβR2 can be detected mainly in the tunica intima and media of healthy and varicose vein walls, although in a lower quantity in the latter. Lower levels of MMP2, TIMP2, and TGFβR2 in varicose veins, when compared to controls, have been partially described by others [18,24] and are in accordance (particularly regarding tunica intima) with our qPCR results for similar comparisons (CVeD groups vs. control group).
Regarding IHC results between the CVeD groups, it is worth mentioning that these were not always coherent among anatomical regions, suggesting once more that comparisons among vein specimens harvested from different locations might not be reliable. Nevertheless, intima layer results (from all anatomical regions) were, in general, consistent with our qPCR results. MMP2 presence in all tunicae was mainly unchanged between CVeD groups with one exception: at the saphenofemoral junction where a slightly lower presence was found in the CEAP4-6 group. This is coherent with the trend revealed with the qPCR results ( Figure 3). In tributary veins, MMP2 presence was generally unchanged between the two groups, while the qPCR results showed an increase in the CEAP4-6 group. This may be explained by the subjective nature of IHC results. With regard to TIMP2 presence, the results are generally in line with Figure 3 trends, particularly for the tunica intima. Finally, the equally low presence of TGFβR2 across CVeD groups was consistent with our qPCR findings, reinforcing the idea of a counter-regulation mechanism to reduce local TGF-β1 expression or signaling throughout CVeD clinical progression.
Whilst it has been suggested that MMP/TIMP imbalance could potentially work through elements (especially smooth muscles cells) in the tunica media [46], we highlight the importance of the tunica intima in CVeD pathophysiology, as shown by the coherence found between qPCR and IHC results for this tunica (in vein specimens from all anatomical regions). It was also noted that among the three anatomical vein regions, varicose tributary veins showed more evident differences (especially in qPCR results) between the CVeD groups, probably due to its thinner media. A thinner media tunica may make the venous walls more vulnerable to homeostatic upset induced by local hypertension, and therefore more prone to varicosity and premature morphologic changes.

Specimen Collection
Samples (2 cm length) of healthy great saphenous veins were harvested from the tibiotarsal junction of 13 patients undergoing coronary bypass surgery (control group). Samples of pathologic refluxing great saphenous veins (from the saphenofemoral and tibiotarsal junctions) and varicose tributaries (from veins showing tortuosity and significant diameter increase with blood filling at the thigh or leg), including the adventitia, were obtained from 31 patients submitted to surgical ablation of the great saphenous vein (CVeD group). The methods of harvesting, storage, and processing samples were identical in both groups.
Before collection, preoperative venous duplex ultrasonography was performed (to confirm venous reflux in CVeD samples and its non-existence in controls) and CVeD patients were evaluated according to the CEAP (Clinical, Etiologic, Anatomic and Pathophysiologic) classification [47] and then regrouped (CEAP2-3/CEAP4-6 groups). Subjects with the following conditions were excluded: surgery within the previous six weeks; steroids or intravenous drug use; deep vein thrombosis or thrombophlebitis; active infection; and collagen diseases and conditions that could modify leukocyte activity (e.g., diabetes mellitus, neoplasia, rheumatoid arthritis, vasculitis). After collection, all vein samples were aseptically washed free of blood using a physiological salt solution, immersed in RNA-Later (Ambion ® , Carlsbad, CA, USA), refrigerated at 4 • C for 24 h, then snap frozen in liquid nitrogen and stored at −80 • C until use.
In compliance with the Declaration of Helsinki, all the procedures carried out with human samples were approved by the Ethics Committee of "Cova da Beira University Hospital Centre, Covilhã, Portugal" (protocol No. 28/200928/ , 26 February 2009. Informed consent was obtained from all participants.

Conventional and Quantitative Real-Time Polymerase Chain Reaction
Total ribonucleic acid (RNA) was isolated from 50 to 100 mg of tissue sample using the TRI Reagent (Ambion ® , Carlsbad, CA, USA) and following the manufacturer's instructions. For complementary deoxyribonucleic acid (cDNA) synthesis, 500 ng of total RNA was reverse transcribed using the M-MuLV Reverse Transcriptase kit (NZYTech ® , Lisbon, Portugal) in a final volume of 20 mL. Both procedures have been described elsewhere [30].

TGFβR3
Sense: CTG TTC ACC CGA CCT GAA AT 502 Antisense: CGT CAG GAG GCA CAC ACT TA The gene expression of positively-confirmed MMP, TIMP, and TGFβR was determined by quantitative real-time PCR (qPCR) using gene-specific primers (STABVIDA ® , Lisbon, Portugal; Table 3) and SYBR-Green/Fluorescein qPCR Master Mix (Fermentas Life Sciences ® , Vilnius, Lithuania). β-actin housekeeping gene was used to normalize gene expression levels. The efficiency of the amplifications was determined for all primer sets using serial dilutions (1, 1:5 and 1:25) of cDNA. Primer concentrations and annealing temperatures were optimized, and the specificity of amplicons was determined by melting curve analysis.
The qPCR was performed as described elsewhere [30]. Both conventional and quantitative real-time PCR was carried out for a pool of specimen cDNA and then for each specimen separately.

Immunohistochemistry
Vein specimens were fixed in 2% paraformaldehyde/0.2% glutaraldehyde for 24 h, transferred to a 70% alcohol solution and paraffin embedded the following day. Paraffin-embedded vein blocks were cut into 6 mm sections and mounted onto poly-L-lysine-coated slides.
Staining was developed for the same period of time for each antibody, for both control and CVeD specimens, and was scored in a blinded manner by two independent observers. The final results took into account the staining intensity and relative difference between different groups of vein specimens (control, CEAP2-3, and CEAP4-6).

Statistical Analysis
Statistical analysis was performed by means of IBM SPSS Statistics, (v.22.0, Armonk, NY, USA). Using data from subsamples (selected by quota sampling) of the 44 participants, statistically significant differences in gene expression among three (controls vs. CEAP2-3 vs. CEAP4-6) or two (CEAP2-3 vs. CEAP4-6) groups of participants were tested for. ANOVA (followed with Bonferroni tests) or the Student test was used to compare the means for independent groups. Equivalence between the groups, regarding participants' demographic and clinical features, was assessed using Fisher's exact test or its Freeman-Halton extension (for 2 × 3 contingency table) and Student test or ANOVA, as appropriate. To meet parametric assumptions, data were transformed using log 10 (x) when necessary and outliers were controlled. Before violation of those assumptions, non-parametric tests (Kruskal-Wallis test, followed by the Mann-Whitney test) were performed. Data descriptive statistics are presented as (absolute and relative) frequencies, mean values ± standard error of the mean (SEM) and ranges. All tests were two-tailed and the significance was set at p ≤ 0.05.

Conclusions
Whilst further studies about molecular events in varicose vein walls are required, our results have contributed more evidence on MMP/TIMP imbalance in venous walls throughout CVeD clinical progression, as well as on the role of TGF-β1 in this event. Differences in MMP and TIMP expression should be expected not only among healthy, atrophic, and hypertrophic varicose veins, but also across anatomical vein regions. The full functional role of TGF-β1 remains to be defined but our results regarding TGFβR expression may suggest a counter-regulation mechanism to control chronically elevated local levels of TGF-β1, leading to a reduction in participation of this growth factor on MMP/TIMP imbalance throughout CVeD clinical progression.
These findings represent another step towards the understanding of CVeD pathophysiology and may provide some cues for therapeutic approaches targeting TGF-β1 signaling.