**Variations in Circulating Active MMP-9 Levels during Renal Replacement Therapy**

**Elena Rodríguez-Sánchez 1,**†**, José Alberto Navarro-García 1,**†**, Jennifer Aceves-Ripoll 1, Judith Abarca-Zabalía 1, Andrea Susmozas-Sánchez 1, Teresa Bada-Bosch 2, Eduardo Hernández 2, Evangelina Mérida-Herrero 2, Amado Andrés 2, Manuel Praga 2, Mario Fernández-Ruiz 3, José María Aguado 3, Julián Segura 1, Luis Miguel Ruilope 1,4,5 and Gema Ruiz-Hurtado 1,5,\***


Received: 27 February 2020; Accepted: 23 March 2020; Published: 26 March 2020

**Abstract:** Renal replacement therapy (RRT) is complicated by a chronic state of inflammation and a high mortality risk. However, different RRT modalities can have a selective impact on markers of inflammation and oxidative stress. We evaluated the levels of active matrix metalloproteinase (MMP)-9 in patients undergoing two types of dialysis (high-flux dialysis (HFD) and on-line hemodiafiltration (OL-HDF)) and in kidney transplantation (KT) recipients. Active MMP-9 was measured by zymography and ELISA before (pre-) and after (post-) one dialysis session, and at baseline and follow-up (7 and 14 days, and 1, 3, 6, and 12 months) after KT. Active MMP-9 decreased post-dialysis only in HFD patients, while the levels in OL-HDF patients were already lower before dialysis. Active MMP-9 increased at 7 and 14 days post-KT and was restored to baseline levels three months post-KT, coinciding with an improvement in renal function and plasma creatinine. Active MMP-9 correlated with pulse pressure as an indicator of arterial stiffness both in dialysis patients and KT recipients. In conclusion, active MMP-9 is better controlled in OL-HDF than in HFD and is restored to baseline levels along with stabilization of renal parameters after KT. Active MMP-9 might act as a biomarker of arterial stiffness in RRT.

**Keywords:** matrix metalloproteinase-9; dialysis; on-line hemodiafiltration; high-flux dialysis; renal replacement therapy; kidney transplantation

#### **1. Introduction**

Chronic kidney disease (CKD) is defined as abnormalities of kidney structure and/or function for a period of at least three months, with implications for health [1]. CKD is an irreversible disorder and often progresses to end-stage renal disease (ESRD), which is life-threatening and requires some form of renal replacement therapy (RRT) such as dialysis or kidney transplantation (KT). Since KT is limited by organ availability, dialysis is the most common form of RRT to remove uremic toxins in ESRD. There are two types of dialysis treatments that can be applied to the ESRD patient: peritoneal dialysis and hemodialysis. Hemodialysis can be further classified depending on the membrane modality (convective vs. diffusive). For example, high-flux dialysis (HFD) is based on the use of high diffusion and low convection, whereas on-line hemodiafiltration (OL-HDF) is based on lower diffusion and higher convection, enabling the removal of middle- and larger-molecular weight substances. However, whatever the dialysis strategy used, it is only a temporary solution until the patient can undergo KT.

Patients with CKD, especially ESRD, are regarded as being at high risk of fatal and non-fatal cardiovascular events because cardiovascular and renal disease often have similar origins and risk factors [2]. Cardiovascular risk is particularly high for patients undergoing dialysis [3] because of CKD-associated complications such as atherosclerosis and systemic inflammation [4]. Indeed, this enhanced risk of atherosclerotic cardiovascular events begins from the earliest stages of CKD. Moreover, dialysis-specific risk factors such as mineral bone disorders and excess fluid load induce vascular calcification and hypertension, which in turn induce arterial stiffness [5], a process characterized by a thickening of the arterial wall and a loss of elastin fiber integrity leading to a decrease in elasticity [6]. Arterial stiffness is a predictor of cardiovascular disease in dialysis and KT recipients [7,8], and is associated with graft dysfunction and rejection [9]. Interestingly, vascular thickening tends to diminish after KT [10] and, in fact, KT improves the longevity and quality of life of patients as compared with long-term dialysis treatment [11], although cardiovascular risk remains higher than in the general population [12].

The anomalous turnover of extracellular matrix components favors deleterious vascular remodeling, which is a major mechanism underlying arterial stiffness. Matrix metalloproteinases (MMPs) play a central role in the regulation of both physiological and pathological connective tissue turnover. In particular, the inducible MMP-9, degrades laminin, elastin, and type IV collagen [13], which are the main components of the glomerular basement membrane of the kidney. In fact, the exaggerated activation of MMP-9 has been associated with an abnormal glomerular basement membrane structure and consequent albuminuria escape in treated hypertension [14]. MMP-9 activation is triggered under stress conditions including inflammation and oxidative stress [13] and is consequently upregulated in inflammation-dependent conditions such as hypertension and associated with cardiovascular risk in patients undergoing hemodialysis [15].

Although the use of MMP-9 as a marker of cardiovascular risk has been widely described, there is a paucity of studies examining how RRT affects its activity [16]. MMP-9 can be assessed as total MMP-9 or active MMP-9 levels measured by the MMP-9/tissue inhibitor of MMP (TIMP)-1 ratio, zymography, or solid-phase immunoassays. Total MMP-9 abundance is, however, not a direct measure of its activity, as both active and inactive MMP-9 are considered. Likewise, the indirect measure of active MMP-9 using the MMP-9/TIMP-1 ratio might be inadequate because it considers that TIMP-1, the endogenous inhibitor of MMP-9, interacts completely with MMP-9 in a 1:1 stoichiometry [17]. It is known that reactive oxygen and nitrogen species can interact with both TIMP-1 and MMP-9, blocking endogenous MMP-9 inhibition and activating MMP-9 constitutively [13] and, therefore, assessment of the MMP-9/TIMP-1 ratio as an indicator of MMP-9 activation is not recommended [18], especially in conditions of oxidative stress such as CKD or ESRD.

Given the heterogeneity in MMP-9 measurements and the gap in our knowledge on MMP-9 activity in RRT, the aim of this study was to assess the effect of different types of RRT (HFD, OL-HDF, and KT) on active MMP-9 measured in a direct manner by zymography and enzyme-linked immunosorbent assay (ELISA), as well as by the direct protein interaction between MMP-9 and TIMP-1 using AlphaLISA® technology (PerkinElmer, Waltham, MA, USA).

#### **2. Materials and Methods**

#### *2.1. Study Population*

The study included two independent cohorts of patients receiving RRT: 32 dialysis-dependent patients recruited for a cross-sectional analysis between November 2016 and March 2017 [19], and 46 KT recipients recruited for a longitudinal study between November 2014 and June 2016 [20,21]. The exclusion criterion for KT recipients was development of acute rejection in the first year post-KT. Both cohorts of patients were recruited at the Nephrology Unit of the Hospital Universitario 12 de Octubre (Madrid, Spain). Dialysis-dependent patients underwent clinical examination before dialysis, and KT patients underwent clinical examination before transplant. Pulse pressure (PP) was determined as the difference between systolic blood pressure (SBP) and diastolic (DBP) blood pressure [9]. Blood samples were drawn before (pre-) and after (post-) one dialysis session, or in the case of KT at baseline and 7 days, 14 days, 1 month, 3 months, 6 months, and 12 months post-KT. Blood samples were collected in heparin tubes and immediately centrifuged at 2000 rpm for 10 min. Plasma samples were stored at −80 ◦C until use.

All patients signed informed consent. The study was approved by the local ethics committee in compliance with the guidelines of the Declaration of Helsinki.

#### *2.2. Assessment of Active MMP-9*

Active MMP-9 was assessed by zymography and ELISA. Zymography was performed under non-reducing conditions using 10% SDS/PAGE containing 0.1% gelatin. Following electrophoresis, gels were incubated in 500 mM Tris, 6 mM CaCl2, and stained with Coomassie Brilliant Blue R-250 (Bio-Rad, Hercules, CA, USA). Digitalized gel images were analyzed with ImageJ software (NIH, Bethesda, MD, USA). Quantification of active MMP-9 was performed with a commercially available ELISA kit (QuickZyme BioSciences, Leiden, The Netherlands). Given the consistency between zymography and ELISA, follow-up of KT patients was exclusively assessed by ELISA at 7 days, 3 months, and 12 months post-KT. In the case of dialysis-dependent patients, active MMP-9 post-dialysis was corrected according to the weight loss during dialysis using the following equation:

$$\text{CC} = \frac{\text{Cpost}}{1 + \frac{BWprre - BWpost}{0.2 \ast BWpost}} \text{ \text{\textdegree}} \tag{1}$$

where *Cc* is the corrected concentration post-dialysis, *Cpost* is the concentration post-dialysis, *BWpre* is the body weight pre-dialysis, and *BWpost* is the body weight post-dialysis [22].

#### *2.3. Assessment of Total MMP-9, Total TIMP-1, and MMP-9*/*TIMP-1 Interaction*

Quantification of total MMP-9 and TIMP-1 was performed in KT patients at baseline, 7 days, 3 months, and 12 months post-KT using commercial ELISA Quantikine kits (R&D Systems, Minneapolis, MN, USA). MMP-9/TIMP-1 was calculated by dividing the levels of total MMP-9 by the levels of TIMP-1.

AlphaLISA® technology was used to detect the interaction between MMP-9 and TIMP-1 following a published protocol [18]. Briefly, AlphaLISA® acceptor beads (PerkinElmer, Waltham, MA, USA) were conjugated with an anti-MMP-9 antibody (ThermoFisher Scientific, Waltham, MA, USA) and then incubated with plasma samples from KT recipients and a biotinylated anti-TIMP-1 antibody (ThermoFisher Scientific). Streptavidin-coated donor beads were then added to bind the biotinylated anti-TIMP-1 antibody and detect the MMP-9/TIMP-1 interactions. Plates were read on an EnSpire Multimode Microplate Reader (PerkinElmer) using an excitation wavelength of 680 nm and an emission wavelength of 615 nm.

#### *2.4. Statistical Analysis*

Normality of data was determined with the Kolmogorov–Smirnov test. HFD and OL-HDF groups were compared using unpaired Student's t-test or the Mann–Whitney U test. Categorical variables were compared with Fisher's exact test. Pre- and post-dialysis groups were compared using the Wilcoxon signed-rank test, and KT follow-up groups were compared using the Friedman test. Spearman's rank-order correlation was used to analyze correlations. Results are expressed as mean ± SEM unless otherwise stated, and *p*-values < 0.05 were considered significant. Analyses were performed using GraphPad Prism 6 (GraphPad Software Inc., San Diego, CA, USA) and SPSS Statistics v22 (IBM, Armonk, NY, USA).

#### **3. Results**

#### *3.1. Clinical Characteristics*

Baseline characteristics of all hemodialysis patients stratified according to the type of dialysis applied (HDF or OL-HDF) are shown in Table 1. No differences were found in the mean age between the two groups. Likewise, there were no differences between the HFD and OL-HDF groups for hypertension, SBP, PP, diabetes mellitus, or N-terminal pro-brain natriuretic peptides. Dialysis-related parameters were also the same between the groups except for the duration of dialysis. Patients in the OL-HDF group were significantly longer on dialysis (31.4 vs. 98.4 months for HFD and OL-HDF, respectively). There were no differences between the groups for the cause of CKD or the medication used. The proportion of men was significantly higher in the OL-HFD group (65%) than in the HFD group (11%). Baseline characteristics of the KT recipients are shown in Table 2.



SBP: systolic blood pressure; PP: pulse pressure; NT-proBNP: N-terminal (NT)-pro B-type natriuretic peptide; Kt/V: where K is the dialyzer urea clearance, t is the total treatment time, and V is the total volume within the body that urea is distributed; eGFR: estimated glomerular filtration rate; CKD: chronic kidney disease; ACEi: angiotensin converting enzyme inhibitor; ARB: angiotensin receptor blocker.


**Table 2.** Baseline characteristics of kidney transplant recipients.

SBP: systolic blood pressure; PP: pulse pressure; eGFR: estimated glomerular filtration rate; CKD: chronic kidney disease; HLA: human leukocyte antigen.

#### *3.2. Dialysis Reduces Active MMP-9 Levels*

We first compared the plasma levels of active MMP-9 in all dialysis patients pre- and post-dialysis. Active MMP-9 estimated by zymography (Figure 1A) or quantified by ELISA (Figure 1B) was lower post-dialysis than pre-dialysis, and this was significant for the ELISA analysis. Baseline plasma levels of active MMP-9 were significantly lower in the OL-HDF group than in the HFD group, both by zymography (Figure 1C) and ELISA (Figure 1D), and active MMP-9 levels were reduced by dialysis only in the HFD group while OL-HDF patients' active MMP-9 levels remained at the same level after the dialysis process. Active MMP-9 negatively correlated with the interdialytic urine volume (Figure 1E), in the sense that less urine volume was associated with higher MMP-9 activity.

**Figure 1.** Matrix metalloproteinase (MMP)-9 activity in pre- and post-dialysis according to the type of dialysis applied. (**A**) Representative zymography gel showing plasma gelatinase MMP-9 activity (upper panel) and quantification (bottom panel). (**B**) Quantification of active MMP-9 by ELISA in plasma of dialysis patients before (pre-dialysis) and after (post-dialysis) one session of dialysis. (**C**) Representative zymography gel showing plasma gelatinase MMP-9 activity (upper panel) and quantification (bottom panel). (**D**) Quantification of active MMP-9 formed by ELISA in plasma of dialysis patients stratified as those on high-flux dialysis (HFD) (pre- and post-dialysis) and on-line hemodiafiltration (OL-HDF) (preand post-dialysis). (**E**) Negative correlation between interdialytic urine volume and pre-dialysis active MMP-9 in dialysis patients with residual urine volume. Correlation was performed using Spearman's test and a linear regression of the data is displayed. \* *p* < 0.05 vs. pre-dialysis; # *p* < 0.05 and ## *p* < 0.01 vs. HFD pre-dialysis.

#### *3.3. Kidney Transplantation Increases Total and Active MMP-9 and Total TIMP-1 Levels, but Not MMP-9:TIMP-1 Protein Interactions*

Representative gel zymography of active MMP-9 in KT recipients before KT and at follow-up is shown in Figure 2A. Active MMP-9, measured by gel zymography (Figure 2B) or quantified by ELISA (Figure 2C), increased significantly 7 days after KT. Active MMP-9 levels remained high 14 days after KT (Figure 2A,B), but thereafter decreased significantly at one month after KT, reaching levels not significantly different to baseline at 3 and 12 months after KT (Figure 2A–C). Active MMP-9 levels were related to renal function, decreasing as estimated glomerular filtration rate (eGFR) increased and plasma creatinine decreased (Figure 2D).

**Figure 2.** MMP-9 activity profile before and after kidney transplantation. (**A**) Representative zymography gel showing plasma gelatinase MMP-9 activity. (**B**) Quantification of active MMP-9 by zymography in plasma of kidney transplantation (KT) patients at baseline (before KT) and after 7 and 14 days, and 1, 3, 6, and 12 months. (**C**) Quantification of active MMP-9 by ELISA in plasma of KT patients at baseline and after 7 days, 3 months, and 12 months. (**D**) Evolution of eGFR, plasma creatinine, and active MMP-9 at baseline and 7 days, 3 months, and 12 months after KT. \*\*\* *p* < 0.001 vs. baseline and ### *p* < 0.001 vs. 7 days after KT.

To evaluate whether the changes in active MMP-9 abundance were due to changes in its expression or, alternatively, to its endogenous inhibition by TIMP-1, we measured total MMP-9 and TIMP-1 levels in KT recipients. Total MMP-9 levels increased significantly at 7 days after KT relative to baseline levels, but significantly decreased at 3 and 12 months after KT (Figure 3A). A similar pattern was observed for TIMP-1 levels, with a significant increase at 7 days after KT and a return to baseline levels at 3 months (Figure 3B). However, TIMP-1 levels continued to decrease at 12 months after KT and were significantly lower than baseline levels at this time (Figure 3B). To indirectly estimate MMP-9 activity and following the line of the majority of clinical studies, we calculated the MMP-9/TIMP-1 ratio, which revealed a significant increase in the MMP-9/TIMP-1 ratio over baseline levels at 7 days after KT (Figure 3C). The ratio decreased to baseline levels 3 months after KT but increased 12 months after KT, reaching an intermediate level (Figure 3C). To confirm the interaction between both molecules, we used the AlphaLISA protocol to measure MMP-9:TIMP-1 protein interactions [18]. No changes in MMP-9:TIMP-1 interactions were found until 12 months after KT, when the interaction was significantly higher than at baseline (Figure 3D).

**Figure 3.** Total protein MMP-9 and tissue inhibitor of MMP (TIMP)-1 levels, MMP-9/TIMP-1 ratio, and MMP-9:TIMP-1 interactions before and after KT. (**A**) Total MMP-9 protein levels (**B**) and total TIMP-1 protein levels quantified by ELISA at baseline and at 7 days, 3 months, and 12 months after KT. (**C**) MMP-9/TIMP-1 ratio estimation. (**D**) AlphaLISA® MMP-9:TIMP-1 interaction immunoassay expressed as binding relative luminescence units (RLUs). \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 vs. baseline; # *p* < 0.05, ## *p* < 0.01 and ### *p* < 0.001 vs. 7 days after KT; and <sup>ϕ</sup> *p* < 0.05, ϕϕ *p* < 0.01, ϕϕϕ *p* < 0.001 vs. 3 months after KT.

#### *3.4. Active MMP-9 Positively Correlates with Arterial Sti*ff*ness in RRT Patients, but Not with Systolic Blood Pressure*

We examined for correlations between active MMP-9 levels, SBP, and the arterial stiffness PP parameters in patients undergoing RRT. No correlation was found between active MMP-9 and SBP in the dialysis or KT cohorts (Figure 4A,B). However, we found a positive significant correlation between active MMP-9 levels and PP in both dialysis and KT cohorts (Figure 4A,B).

**Figure 4.** Correlation between MMP-9 activity and systolic blood pressure (SBP) and pulse pressure (PP) at pre-dialysis and before KT. (**A**) No association of MMP-9 activity with SBP (in green) and a positive significant correlation with PP (in blue) in dialysis patients in pre-dialysis. (**B**) No association of MMP-9 activity with SBP (in green) and a positive significant correlation with PP (in blue) in KT recipients before KT (baseline). Correlation was performed using Spearman's test and a linear regression of the data is displayed.

#### **4. Discussion**

Our study shows that: (i) patients undergoing OL-HDF have lower levels of active MMP-9 compared with those undergoing HFD; (ii) active MMP-9 increases early post-KT at 7 and 14 days and stabilizes in parallel with renal markers at 3 months after KT; and (iii) active MMP-9 is associated with arterial stiffness measured by PP in RRT (pre-dialysis and before KT).

Renal dysfunction is characterized by a uremic state that aggravates as renal function declines, reaching its maximum in ESRD. This state is intimately associated with all-cause mortality and especially with cardiovascular morbidity and mortality. Undoubtedly, RRT improves the lifespan and the quality of life of patients with ESRD. Among RRT patients, KT significantly improves the quality of life compared with dialysis, but the access to renal transplants is limited and not all ESRD patients can undergo a surgical procedure. Although dialysis is the cornerstone of RRT, it is also associated with a state of chronic inflammation and oxidative stress [23]. Dialysis *per se* triggers an increase in oxidative stress and inflammation because of incompatibilities with the dialysis membrane, which activates circulating leukocytes, and also because low-molecular weight antioxidant systems are filtered and therefore eliminated during the procedure. In addition, the type of dialysis used directly influences the oxidative status of the patients. Indeed, recent studies demonstrate that patients under OL-HDF have less inflammation, endothelial damage, and oxidative stress [19,24,25], which could be associated with the improved prognosis in this group [26,27]. The reduction in middle-molecules is associated with the preservation of residual renal function [28], which decreases the mortality risk in dialysis [29] and is improved in OL-HDF [30]. The systemic activation of MMP-9 is a marker of a lasting inflammatory state that is inherent to renal disease, as has been demonstrated even with adequate pharmacological treatment in patients with CKD [14,31]. In the case of ESRD, and especially dialysis, there are conflicting results on the levels of MMP-9 in the pre- and post-dialysis states [32,33]. This could be due to the different types of dialysis membranes [34,35], or to the type of dialysis itself [19,25]. We demonstrate herein that active MMP-9 is cleared during dialysis when the type of dialysis is not considered, but a decrease in active MMP-9 is only seen in HFD. Other authors have described a decrease in total MMP-9 during OL-HDF, but not during other hemodialysis [36]. Nevertheless, total MMP-9 does not necessarily correlate with active MMP-9, and the measurement of active MMP-9 is a more direct assessment of its pathophysiological activity [14]. The fact that only HFD effectively clears active MMP-9 might suggest that HFD is more efficient than OL-HDF; however, pre-dialysis active MMP-9 was significantly lower in the OL-HDF group than in HFD patients, indicating that the levels of active MMP-9 are better controlled in patients under OL-HDF. Moreover, residual renal function is associated with lower inflammation in OL-HDF [37,38]. Supporting this, we also observed that active MMP-9 is lower in patients with greater interdialytic urine volume, which is superior to eGFR as an estimate of residual renal function in dialysis [28].

KT is the only curative treatment for ESRD and long-term survival is significantly greater after KT than after dialysis. Despite its benefits, however, KT is not without risks, such as the inevitable ischemia and reperfusion (I/R) injury, which triggers several processes such as immune system activation, endothelial dysfunction, and cell death. Likewise, I/R induces morphological changes in the kidney that lead to fibrosis, as shown by the elevation of fibrosis-related biomarkers in the kidney of patients with chronic allograft nephropathy [39]. By contrast, a well-known benefit of KT is the normalization of inflammatory and oxidative stress markers [40]. For example, the C-terminal agrin fragment, a biomarker for kidney function and a breakdown product of agrin, the major proteoglycan of the glomerular basement membrane, strongly correlates with creatinine and eGFR and is stabilized at 1–3 months after KT [41]. In the same line, interleukin 6, a well-known precursor of MMP-9 activation, shows a burst of activity one week after KT before stabilizing [42]. In the present study, we demonstrate that systemic active MMP-9 levels increase one week after KT, but rapidly and significantly decrease from the second week after surgery. We speculate that the increase in active MMP-9 evident one week after KT likely originates from neutrophils in response to leukocyte activation inherent to I/R processes. This is supported by experimental studies demonstrating that MMP-9 promotes mononuclear cell infiltration in a rat model of early allograft nephropathy [43].

The peak in active MMP-9 observed early post-KT (after 7 and 14 days) might be due to an increase in total MMP-9 or an enhancement in the enzyme activation by the inflammatory environment. MMP-9 is synthesized as a zymogen with a covered zinc-containing active site. Other MMPs and MMP-9 itself cleave the peptide chain that covers the active site in physiological conditions. However, reactive oxygen and nitrogen species can pathologically expose the active site by binding to the zinc ion. Reactive oxygen and nitrogen species can also bind to TIMP-1 and hamper the physiological inhibition of MMP-9. Indeed, MMP-9:TIMP-1 interactions are reduced under conditions of increased oxidative stress and can therefore mask the results obtained from the surrogate marker of MMP-9 activity, the MMP-9/TIMP-1 ratio [18]. Given the increase in oxidative stress associated with I/R, we aimed to investigate whether the peak in active MMP-9 was due to the pathological activation of MMP-9 or to an increase in MMP-9 protein levels. Our results indicate that the peak in active MMP-9 was in fact due to an increase in total MMP-9 expression, and the results of the MMP-9/TIMP-1 ratio indicate the same. However, the reduction in TIMP-1 levels at 12 months post-KT leads to an increase in the MMP-9/TIMP-1 ratio, erroneously suggesting increased MMP-9 activity. Indeed, active MMP-9 is unaltered 12 months post-KT because the MMP-9:TIMP-1 interaction is increased, supporting the importance of measuring active MMP-9 rather than the MMP-9/TIMP-1 ratio. The suggested increase in active MMP-9 by the MMP-9/TIMP-1 ratio indicates an increase in mortality risk [44]. However, total and real active MMP-9 do not vary, and TIMP-1, which is also associated with mortality [45], is reduced. Therefore, the patients might actually have a reduced risk of mortality as the KT becomes more stable.

Finally, arterial stiffness increases SBP by wave reflection and decreases DBP, which increases PP, a marker of large artery stiffness [46]. Arterial stiffness is attenuated after KT as compared with hemodialysis [9,47–50], and is subject to donor age, living kidney donation, and mean blood pressure [5]. Our study shows for the first time, to our knowledge, that active MMP-9 correlates with PP in patients treated with RRT. We found that, independently of the type of dialysis, PP is lower as the levels of active MMP-9 decline. A decrease in active MMP-9 suggests that arterial stiffness might be reduced in patients undergoing OL-HDF. However, we failed to find differences in PP between groups of patients on the different types of dialysis, in accord with results from other groups describing that arterial stiffness is not different between hemodialysis and HDF [5]. We also found that baseline active MMP-9 correlates with PP before KT, supporting the idea that a reduction in active MMP-9 levels in ESRD patients is needed to ameliorate arterial stiffness.

The main limitation of the present study was the small population of dialysis patients. More studies are needed in order to confirm our results, but this pilot study is the first step in assessing the insights of arterial stiffness in different types of dialysis.

#### **5. Conclusions**

Active MMP-9 is lower in OL-HDF than in HFD and shows a peak in KT recipients after I/R in the early post-KT stages (7 and 14 days). However, the peak in active MMP-9 is resolved in parallel with the restoration of renal function, indicating that MMP-9 activity is likely a surrogate of I/R injury. In both contexts—dialysis and KT—active MMP-9 is associated with PP in pre-dialysis and just before KT, indicating that MMP-9 is a marker of vascular dysfunction in patients undergoing RRT.

**Author Contributions:** E.R.-S., J.A.N.-G., J.A.-R., J.A.-Z. and A.S.-S. performed the experiments and the data analysis; T.B.-B., E.H., E.M.-H. and M.P. collected samples and followed dialysis patients; A.A., M.F.-R. and J.M.A. recruited and followed KT patients. L.M.R., J.S., J.M.A. and G.R.-H. contributed to the funding acquisition. J.A.N.-G., E.R.-S. and G.R.-H. wrote the manuscript; G.R.-H. conceived the idea and designed the study. All authors reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was mainly supported by projects from the Instituto de Salud Carlos III (PIE13/00045, PI17/01093, PI17/01193, ayuda confinanciada por el Fondo Europeo de Desarrollo (FEDER) and CP15/00129, CP18/00073, FI18/00261 ayuda cofinanciada por el Fondo Social Europeo (FSE)) and partially supported by Sociedad Española de Nefrología/Fundación SENEFRO and Fundación Renal Íñigo Álvarez de Toledo (FRIAT), and co-funded by the European Regional Development Fund.

**Acknowledgments:** The authors thank Tamara Ruiz and Patricia Parra Aragón from Hospital Universitario 12 de Octubre.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **CD147 is a Novel Interaction Partner of Integrin** α**M**β**2 Mediating Leukocyte and Platelet Adhesion**

**David Heinzmann 1,\*,**†**, Moritz Noethel 1,**†**, Saskia von Ungern-Sternberg 1, Ioannis Mitroulis 2, Meinrad Gawaz 1, Triantafyllos Chavakis 2, Andreas E. May <sup>3</sup> and Peter Seizer <sup>1</sup>**


Received: 23 February 2020; Accepted: 31 March 2020; Published: 2 April 2020

**Abstract:** Surface receptor-mediated adhesion is a fundamental step in the recruitment of leukocytes and platelets, as well as platelet–leukocyte interactions. The surface receptor CD147 is crucially involved in host defense against self-derived and invading targets, as well as in thrombosis. In the current study, we describe the previously unknown interaction of CD147 with integrin αMβ2 (Mac-1) in this context. Using binding assays, we were able to show a stable interaction of CD147 with Mac-1 in vitro. Leukocytes from Mac-1−/<sup>−</sup> and CD147+/<sup>−</sup> mice showed a markedly reduced static adhesion to CD147- and Mac-1-coated surfaces, respectively, compared to wild-type mice. Similarly, we observed reduced rolling and adhesion of monocytes under flow conditions when cells were pre-treated with antibodies against Mac-1 or CD147. Additionally, as assessed by antibody inhibition experiments, CD147 mediated the dynamic adhesion of platelets to Mac-1-coated surfaces. The interaction of CD147 with Mac-1 is a previously undescribed mechanism facilitating the adhesion of leukocytes and platelets.

**Keywords:** Mac-1; CD147; leukocytes; platelets; adhesion; integrin αMβ2

#### **1. Introduction**

The recruitment of leukocytes and platelets to activated endothelium as well as platelet-leukocyte interactions are of fundamental significance for innate and adaptive immunity, as well as thrombosis [1]. The surface receptor CD147 (basigin, extracellular matrix-metalloproteinase inducer; EMMPRIN) has been shown to be important in the host defense from self-derived as well as invading targets and is a major factor regulating the expression of matrix metalloproteinases (MMPs). CD147 is a pathophysiologically important multi-ligand receptor of the immunoglobulin superfamily. It is expressed in various tissues and cell types, including leukocytes, endothelial cells, and platelets. A range of different proteins, including cyclophilins, monocarboxylate transporters (MCTs) 1-4, CD43, CD44, CD98, NOD2, galectin-3, γ-catenin, apolipoprotein-D, members of the S100 protein family, and integrins, such as α3β<sup>1</sup> and α6β1, have been reported to interact with CD147 [2–6].

In the context of thromboinflammation, CD147 acts as a pro-inflammatory and pro-thrombotic receptor, eliciting leukocyte chemotaxis and adhesion, as well as platelet activation and subsequent thrombus formation through the binding of various interaction partners [4,7–12]. Notably, its interaction with cyclophilin A (CyPA) contributes significantly to various inflammatory diseases. In the context of cardiovascular diseases, CyPA induces leukocyte chemotaxis and adhesion, and myocardial MMP expression, facilitating subsequent myocardial remodeling. Inhibition of extracellular

CyPA significantly decreases platelet activation and thrombus formation as well as the formation of monocyte–platelet aggregates [4,7,8,13,14]. Furthermore, our group identified glycoprotein VI (GPVI) to be an adhesion-mediating partner for CD147 on the platelet surface, which is the first time it has been demonstrated that CD147 plays a direct role in cell adhesive events, apart from mediating adhesion via intracellular signaling, leading to the expression of adhesion molecules [11].

Mac-1 (integrin αMβ2, complement receptor 3, CD11b/CD18) is a well-characterized heterodimeric integrin, mostly found on polymorphonuclear leukocytes [15]. As a member of the β2-integrin family, it is known to be involved in the leukocyte adhesion cascade. After the initial contact of the leukocyte with endothelial cells lining the vessel wall, a complex signaling cascade is initiated, leading to a selectin-dependent rolling motion of the leukocyte [16]. To establish a firm contact on the luminal side of the endothelium necessary for extravasation, chemokine-induced inside-out activation of integrins on the surface is necessary [17]. Especially on neutrophils, Mac-1 plays an important role in adhesion to the endothelium upon activation [18–20]. In this context, Mac-1-dependent adhesive interactions enable the cells to crawl on the surface of endothelial cells prior to extravasation [16,21]. Numerous cell surface receptors have been identified to interact with Mac-1 in its high-affinity state, including ICAM1-4, JAM-C, Thy-1, RAGE, DC-SIGN, and CD40L [15,22,23]. A plethora of soluble ligands for Mac-1 have been identified, amongst which are fibrin, fibrinogen, plasminogen, factor Xa, heparin, polysaccharides, ssDNA, and dsRNA [15,24]. Mac-1 can also interact with matrix proteins, including vitronectin, collagen, Cyr61, and fibronectin [15]. Mac-1 is also involved in complement-mediated immune responses by recognizing complement C3-opsonized pathogens as well as immune complexes [25].

In this study, we describe for the first time the interaction of integrin Mac-1 with CD147. We provide evidence that CD147 is a novel and relevant binding partner for Mac-1 on leukocytes and platelets.

#### **2. Materials and Methods**

#### *2.1. Mac-1-CD147 Binding ELISA*

Binding between Mac-1 and CD147 was evaluated using a modified enzyme-linked immunosorbent assay (ELISA). In a 96-well plate, wells were coated with recombinant Mac-1 (R&D Systems, Minneapolis, MN, USA) or bovine serum albumin (BSA) as the control. Recombinant CD147 was added in increasing concentrations (0–20 μg/mL) for 1 h. After removing CD147 and gentle washing, the wells were incubated with an anti-CD147 antibody (mouse-anti CD147, Abcam, Cambridge, UK) followed by a biotinylated secondary antibody (polyclonal rabbit anti mouse, Dako, Glostrup, Denmark) and a streptavidin/HRP complex (Life technologies, Carlsbad, CA, USA). The binding was detected using 3,3 ,5,5 -tetramethylbenzidine (Serva, Heidelberg, Germany). The reaction was stopped using 1 M H2SO4. The absorption was measured using an ELISA plate reader at 450 nm with a reference value of 570 nm. Measurements from 6 independent experiments were analyzed.

#### *2.2. Murine Leukocyte Isolation and Static Adhesion*

Murine leukocytes were isolated from bone marrow. CD11b−/<sup>−</sup> mice were from Jackson Laboratories, and CD147+/<sup>−</sup> mice were a kind gift from Professor R. A. Nowak (Urbana, IL, USA). The femur and tibia were removed from sacrificed CD11b+/+ and CD11b−/<sup>−</sup> (hereafter designated Mac-1+/+ and Mac-1−/−, respectively, n = 5), as well as from CD147+/<sup>−</sup> and CD147+/+ (n = 12, respectively) mice (with a C57Bl/6 background). Throughout this, all efforts were made to minimize animal suffering. The bone marrow was washed out of the bones using RPMI 1640 medium (Life technologies, Carlsbad, CA, USA) supplemented with 10% FCS (Life technologies, Carlsbad, CA, USA), 1% Pen/Strep (Sigma-Aldrich, St. Louis, MO, USA), 1% HEPES (Sigma-Aldrich, St. Louis, MO, USA), and β-mercaptoethanol. The harvested cells were strained through a 70-μm strainer and were centrifuged (300× *g* for 5 min) and resuspended in ammonium chloride to lyse erythrocytes. For the leukocyte culture, cells were washed and resuspended in RPMI 1640 medium supplemented with 0.0001% GMCSF.

For the static adhesion assay, a 96-well plate was coated overnight with Mac-1 (10 μg/mL), CD147 (20 <sup>μ</sup>g/mL), or BSA (1%). Hereafter, the coated wells were blocked with 4% BSA for 1 h. Then, 2 <sup>×</sup> 104 isolated leukocytes from Mac-1+/+, Mac-1−/−, CD147+/+, or CD147+/<sup>−</sup> mice were added into each well. After allowing the cells to adhere to the coating for one hour, the plate was washed twice with medium to remove non-adherent cells. The number of adherent cells was analyzed using photo documentation.

#### *2.3. Isolation of Human Monocytes*

Human monocytes were isolated as described before [26]. Briefly, mononuclear cells were isolated form venous blood drawn from the antecubital vein of healthy volunteers. The blood was collected in citrate-phosphate-dextrose-adenine (CPDA) buffer and was centrifuged over a Ficoll gradient at 920× *g* for 20 min. Leukocytes were cultured overnight in RPMI 1640 (Life technologies, Carlsbad, CA, USA) supplemented with 10% FCS (Life technologies, Carlsbad, CA, USA) and 1% Pen/Strep (Sigma-Aldrich, St. Louis, MO, USA) in plastic dishes. Non-adherent cells were removed by gentle washing. The adherent cells were detached using trypsin (Life technologies, Carlsbad, CA, USA) and were resuspended in RPMI 1640 supplemented with 10% FCS and 1% Pen/Strep for further use.

#### *2.4. Isolation of Human Platelets*

Human platelets were isolated as described before [27]. In brief, venous blood was drawn from the antecubital vein of healthy volunteers in adenosine citrate dextrose (ACD) buffer. The blood was centrifuged at 210× *g* for 20 min. Hereafter, the platelet-rich plasma (PRP) was collected and Tyrodes-HEPES buffer (HEPES 2.5 mM; NaCl, 150 mM; KCl, 1 mM; NaHCO3, 2.5 mM; glucose, 6 mM; BSA, 1 mg/mL; pH 7.4) was added and centrifuged at 836× *g* for 10 min. The pellet was resuspended in Tyrodes-HEPES buffer supplemented with 1 mM CaCl2 and 1 mM MgCl2 for further use.

#### *2.5. Cell Adhesion under Flow Conditions*

Using a flow chamber assay, human monocytes (2 <sup>×</sup> 105/mL, stimulated with 1 <sup>μ</sup>g/mL LPS (Sigma-Aldrich, St. Louis, MO, USA) for 2 h (n <sup>≥</sup> 7) or platelets (1 <sup>×</sup> <sup>10</sup>8/mL, n <sup>=</sup> 7) were pre-incubated with anti-Mac-1 (clone #2LPM19c, monoclonal mouse-anti-Mac-1 Santa Cruz, Dallas, TX, USA), anti-CD147 antibody (clone #UM-8D6, monoclonal mouse-anti-CD147, Ancell, Bayport, MN, USA), or IgG control (all 20 μg/mL) for 2 h (monocytes) or 30 min (platelets). The cells were then perfused over Mac-1-, CD147-, or BSA-coated coverslips with arterial shear rates (2000 s−1). All experiments were recorded in real time and were analyzed for adherent and rolling monocytes/platelets, respectively.

#### *2.6. Ethics Statement*

All experiments were conducted in accordance with the German animal protection law and according to the Declaration of Helsinki. All protocols were conducted according to current animal protection laws and were approved by Regierungspräsidium Tübingen and the local ethics committee (Anzeige nach §8a TierSchG).

#### *2.7. Statistical Analysis*

Data are shown as mean ± SEM. Differences between groups with cardinal values were evaluated using Student's *t*-test for two groups or ANOVA with subsequent post hoc analyses. Differences were considered statistically significant if *p* < 0.05. Analysis was performed using Prism 8 (GraphPad Software, La Jolla, CA, USA).

#### **3. Results**

#### *3.1. Mac-1 Interaction with CD147*

We used a binding ELISA to assess whether Mac-1 and CD147 can interact in a stable manner. As a negative control, coating with bovine serum albumin showed no significant interaction. When Mac-1 was used for coating the respective capture signal of CD147 was significantly enhanced, demonstrating a direct and stable interaction between the two molecules (Figure 1).

**Figure 1.** Mac-1 and CD147 interaction. Mac-1/CD147 interaction was assessed using an ELISA with recombinant Mac-1 or BSA (control) coating. Recombinant CD147 was added in ascending concentrations for 1 h. Evaluation was facilitated using a biotinylated secondary anti-CD147 antibody with a streptavidin/HRP complex. Changes in absorption of 3,3 ,5,5 -tetramethylbenzidine were measured at 450 nm. Measurements from 6 independent experiments were analyzed, \* indicates *p* < 0.05.

#### *3.2. Mac-1*/*CD147 Interaction under Static Conditions*

Based on the initial interaction study, we hypothesized that the interaction of these molecules could play a role in leukocyte adhesive events, as both have been described in this context separately. To this end, we performed a static adhesion assay with leukocytes from Mac-1-deficient and -sufficient mice. Mac-1−/<sup>−</sup> leukocytes showed a markedly reduced adhesion to immobilized CD147 compared to wild-type mice, indicating a relevant binding strength. With heterozygous leukocytes from CD147+/<sup>−</sup> mice, the effect was expected to be less pronounced. CD147+/<sup>−</sup> cells showed a higher adhesion under basal conditions with a slight non-significant increase of adhesion to the Mac-1-coated surface (Figure 2).

**Figure 2.** Mac-1/CD147 interaction in static leukocyte adhesion. A static adhesion assay with leukocytes form wild-type, Mac−/<sup>−</sup> (n = 5) (**A**), and CD147+/<sup>−</sup> (n = 12) (**B**) mice was used. Leukocytes were allowed to adhere to CD147- or BSA-coated surfaces (**A**) or to Mac-1- or BSA-coated surfaces (**B**) for 1 h. \* indicates *p* < 0.05, n.s. indicates *p* > 0.05 compared to BSA control, respectively.

#### *3.3. Mac-1*/*CD147 Interaction under Dynamic Conditions*

After establishing a possibly relevant interaction strength for pathophysiological mechanisms, we studied LPS-stimulated human monocytes, as prototypical members of innate immunity under dynamic flow conditions. LPS-stimulated leukocytes were perfused over coated surfaces to evaluate the relative ability of the Mac-1/CD147 interaction to induce adhesion. Using a flow chamber assay, we observed a stable number of rolling and adherent cells on CD147- as well as on Mac-1-coated surfaces, compared to BSA under arterial shear rates (n = 9). In both cases, the effect could be reduced by the addition of either anti-CD147 or anti-Mac-1 antibody to the setup. The number of adherent monocytes over CD147 failed to reach statistical significance when treated with anti-Mac-1 (*p* = 0.06) (Figure 3).

**Figure 3.** Mac-1/CD147 interaction facilitates leukocyte adhesion under dynamic flow conditions. To evaluate the relevance of the Mac-1/CD147 interaction for adhesion under arterial shear conditions, a flow chamber assay was used. LPS-stimulated human monocytes were perfused over BSA- or CD147-coated surfaces (**A** rolling cells, **B** adherent cells), or over BSA- or Mac-1-coated surfaces (**C** rolling cells, **D** adherent cells). Inhibitory antibodies against CD147 or Mac-1, as well as IgG control were added to determine the relevance of the respective protein for adhesion to its proposed counterpart. Results of ≥7 independent experiments are shown, \* indicates *p* < 0.05 compared to uninhibited control.

#### *3.4. Mac-1*/*CD147 Interaction in Platelets*

To establish whether this mechanism is specific to leukocytes, we decided to perform a similar dynamic adhesion assay using isolated human platelets. When isolated platelets were perfused over immobilized Mac-1, adhesion was greatly induced compared to perfusion over BSA, which was used as a control. When anti-Mac-1 or anti-CD147 antibody was added, adhesion to Mac-1 was significantly reduced, almost reaching control levels (n = 7). The addition of IgG antibody had no modifying effect (Figure 4).

**Figure 4.** Mac-1/CD147 interaction facilitates platelet adhesion under dynamic flow conditions. The adhesion of platelets was studied in a similar flow chamber assay with arterial shear conditions. Isolated human platelets were perfused over Mac-1- or BSA-coated surfaces. Inhibiting antibodies against CD147 or Mac-1, as well as IgG control were added to determine the relevance of the protein for adhesion. Video analysis was performed to assess rolling (**A**) and adherent cells (**B**). Results of 7 independent experiment are shown, \* indicates *p* < 0.05 compared to uninhibited control.

#### **4. Discussion**

The recruitment of leukocytes is a vital mechanism for innate and adaptive immunological response. Especially, integrins have been found to play a central role for the adhesion of activated leukocytes to the endothelium [28].

CD147 is a promiscuous surface receptor expressed on a variety of cell types. Ligand binding to CD147 has been shown to elicit many pro-inflammatory and pro-fibrotic aspects, such as induction of matrix-metalloproteinases, induction of leukocyte chemotaxis, platelet activation, and adhesion [4,7,8,29].

Thus far, especially, homodimeric signaling of CD147 and the interaction with extracellular cyclophilin A has been characterized in the inflammatory and pro-fibrotic context [30,31]. In this study, we proposed Mac-1 as a previously unknown binding partner for CD147 on leukocytes and platelets.

Initial interaction studies showed a stable binding of the two molecules, making detection via an ELISA-based interaction assay feasible. As both molecules have been found to be involved in leukocyte recruitment, we used static and dynamic adhesion assays to establish whether the stability of the interaction could be of pathophysiological significance. We found that inhibition of either binding partner strongly reduced the ability of leukocytes and especially monocytes to adhere to CD147- or Mac-1-coated surfaces under static and dynamic conditions.

With this in mind, we were interested in identifying the mechanism by using leukocytes from Mac-1 knock-out mice and CD147 heterozygous mice. Both genotypes showed little adhesion to the immobilized binding partner, whereas leukocytes from wild-type mice showed strong adhesion. Interestingly, this interaction also seems to be of importance for the adhesion of platelets. While platelets play a role in inflammation by releasing pro-inflammatory and pro-fibrotic chemokines, in clinical practice, their pro-thrombotic properties are especially of great interest, as the initial contact of the platelets with the activated endothelium and associated factors triggers a wide range of responses, including degranulation, activation by the presentation of adhesion molecules, and shape changes. Under flow conditions, platelet adhesion to Mac-1 was abolished by the addition of anti-Mac-1 or anti-CD147 antibodies. This finding suggests that the binding between Mac-1 and CD147 may mediate platelet–leukocyte interactions as well.

CD147 has been identified to play a prominent role in the induction of pro-inflammatory and pro-thrombotic effects in various disease models and across species. Our group has previously described platelet glycoprotein VI (GPVI) as a novel receptor for CD147, mediating platelet rolling and monocyte–platelet aggregates [11]. Furthermore, the interaction of CD147 with cyclophilin A (CyPA) has been intensively investigated in the context of thromboinflammation. CyPA can be released from the cytoplasm by various pro-inflammatory stimuli. The binding of CyPA to CD147 on the surface

of platelets induces rapid degranulation and activation, leading to shape changes and the release of pro-inflammatory chemokines [4]. Furthermore, the CD147–CyPA interaction induces leukocyte chemotaxis and activation, expression of matrix-metalloproteinases, promotes atherosclerotic plaque formation, and plays critical roles in several inflammatory and cardiovascular diseases [14,31–33]. CD147 has also been associated with other integrins, including α3β<sup>1</sup> and α6β<sup>1</sup> [2]. S100A9, belonging to the S100 family of proteins, has been identified to induce leukocyte migration, thrombosis, and cytokine release via CD147 [5,34]. Mac-1, predominantly expressed on myeloid cells, has also been extensively implicated in various thromboinflammatory mechanisms, including leukocyte migration, phagocytosis, and the facilitation of monocyte–platelet aggregates [15,35].

The presented data therefore links two molecules, both deeply embedded in thromboinflammatory mechanisms, complimenting the existing literature. However, the pathophysiological importance in vivo remains to be elucidated. This study was primarily designed to establish a plausible link between CD147 and Mac-1 in the context of thromboinflammation using in vitro experiments. Further investigations are necessary to characterize the underlying signaling mechanisms and to subsequently evaluate its pathophysiological relevance.

#### **5. Conclusions**

In this study, we described for the first time the interaction of integrin Mac-1 with CD147. We furthermore provide evidence that CD147 is a novel and relevant binding partner for Mac-1 on leukocytes and platelets.

**Author Contributions:** Conceptualization, P.S., A.E.M., M.G., D.H., I.M., and T.C., Methodology, P.S., M.G., D.H., I.M., M.N., and S.v.U.-S.; Investigation, D.H., M.N., S.v.U.-S.; Resources, D.H., P.S., M.G.; Writing—Original Draft Preparation, D.H., and P.S.; Writing—Review and Editing, D.H., M.N., S.v.U.-S., I.M., M.G., A.E.M., T.C., P.S.; Visualization, D.H., M.N., S.v.U.-S.; Supervision, P.S., A.E.M., M.G., and P.S.; Funding Acquisition, M.G., P.S., and D.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by grants from the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG #374031971-TRR 240-CRU-240), and the German Heart Foundation (Deutsche Herzstiftung).

**Acknowledgments:** The authors thank K. Posavec for providing outstanding technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
