PD-L1 regulates T-cell differentiation to control adaptive venous remodeling

Abstract

Objective:

Patients with end-stage renal disease depend on hemodialysis for survival. Although arteriovenous fistulae (AVF) are the preferred vascular access for hemodialysis, the primary success rate of AVF is only 30–50% within 6 months, showing an urgent need for improvement. Programmed death ligand 1 (PD-L1) is a ligand that regulates T-cell activity. Since T-cells have an important role during AVF maturation, we hypothesized that PD-L1 regulates T-cells to control venous remodeling that occurs during AVF maturation.

Approach and results:

In the mouse aortocaval fistula model, anti-PD-L1 antibody (200mg, 3 times/week intraperitoneal) was given to inhibit PD-L1 activity during AVF maturation. Inhibition of PD-L1 increased T-helper type 1 cells and T-helper type 2 cells but reduced regulatory T-cells to increase M1-type macrophages and reduce M2-type macrophages; these changes were associated with reduced vascular wall thickening and reduced AVF patency. Inhibition of PD-L1 also inhibited smooth muscle cell proliferation and increased endothelial dysfunction. The effects of anti-PD-L1 antibody on adaptive venous remodeling were diminished in nude mice; however, they were restored after T-cell transfer into nude mice, indicating the effects of anti-PD-L1 antibody on venous remodeling were dependent on T-cells.

Conclusions:

Regulation of PD-L1 activity may be a potential therapeutic target for clinical translation to improve AVF maturation.

Introduction

More than 740,000 people in United States have end-stage renal disease and are dependent on hemodialysis.¹ Arteriovenous fistulae (AVF) are the preferred vascular access for hemodialysis because of increased patency and fewer complications compared with grafts or catheters; however, the primary success rate of AVF is only 30–50% within 6 months.^2–4 Successful use of an AVF for hemodialysis requires venous maturation, that is outward remodeling and wall thickening, to support the increased flow and needle punctures required for the hemodialysis sessions. Although the factors that regulate AVF maturation without exuberant thickening are not clearly understood, the immune system and inflammation are important components;^5–8 however, molecular mechanisms by which the immune system and/or inflammation regulate venous remodeling during AVF maturation are not yet understood well.^5–8

Inflammatory cells such as T-cells and macrophages have been observed in the vascular wall during AVF maturation,^9–13 and we have previously shown that M2-type macrophages appear to be required for wall thickening during AVF maturation.^{12, 13} T-cells regulate macrophage accumulation in the remodeling AVF wall, including M2-type macrophages.¹¹ Mature T-cells are characterized by cell membrane markers such as CD4 or CD8; we have previously shown that T-cells, and particularly CD4+T-cells, induce macrophage accumulation in the AVF to regulate adaptive remodeling,¹¹ suggesting that CD4+T-cells may be a mechanism by which inflammation regulates venous adaptive remodeling.

Programmed death ligand 1 (PD-L1) is a transmembrane protein expressed in most cells including endothelial cells.^14–16 PD-L1 is a ligand that binds to programmed cell death 1 (PD-1) that is specifically expressed on T-cells, and regulates the immune checkpoint system.^14–16 In CD4+T-cells, PD-L1 induces differentiation into regulatory T-cells (Treg) that suppress other inflammatory cells including T-helper cells, cytotoxic T-cells (CTL) and inflammatory macrophages.^17–19 We have previously shown that M2-type macrophages are necessary for venous remodeling¹³ and that CD4+T-cells regulate macrophage polarization.¹¹ Although the function of endothelial PD-L1 during venous remodeling is still not understood, however, it is possible that endothelial PD-L1 may regulate CD4+T-cells during venous adaptive remodeling. Since regulation of CD4+T-cells is likely to promote AVF maturation,¹¹ and PD-L1 regulates CD4+T-cells,^14–16 we hypothesized that PD-L1 is a molecular mechanism by which T-cells accumulate in the AVF wall to regulate adaptive venous remodeling.

Methods

The authors declare that all supporting data are available within the article [and its online supplementary files].

Mouse aortocaval fistula model.

All animal experiments were performed in strict compliance with federal guidelines and with approval from the Yale University IACUC. Wild type male and female C57BL6/J and athymic male NU/J mice 9–11 weeks of age were used. Infrarenal aorto-caval fistulae were created as previously described.²⁰ Briefly, AVF were created by needle puncture from the aorta into the inferior vena cava (IVC) using a 25 G needle. Visualization of pulsatile arterial blood flow in the IVC was used to demonstrate technical success. Following surgery, all animals were monitored daily and evaluated weekly by a veterinarian for changes in health status.

PD-L1 antibody treatment.

Anti-mouse PD-L1 antibody (200 mg/mouse; BE0101, BioXcell, Lebanon, NH) was delivered via intraperitoneal injections starting on the day of the procedure and repeated 3 times/week throughout the study period. The control group received an equal volume of isotype-matched control antibody (BE0090, BioXcell).

Measurement of fistula dilation and shear stress.

Ultrasound (Vevo770 High Resolution Imaging System; Visual Sonics Inc., Toronto, Ontario, Canada; 40 MHz probe) was used to measure the diameter of the vessels as previously described.²⁰ Ultrasound was performed prior to the operation (day 0 values) and serially post-operatively. Shear stress was calculated as previously described.²⁰

Histology.

The animals were euthanized and perfused with normal saline followed by 10% formalin via the left ventricle under physiologic pressure and the AVF was extracted en bloc. The tissue was then embedded in paraffin and cut in 5-μm cross sections. Elastin Van Gieson (EVG) staining was used to measure intima-media thickness, Masson trichrome staining was used to measure collagen density, and hematoxylin eosin (HE) staining and phosphotungstic acid hematoxylin (PTAH) staining were used to detect thrombosis in 5 μm cross sections of the IVC using sections obtained 50–100 μm cranial to the fistula. Eight equidistant points around the IVC wall were averaged in each cross section to obtain the mean AVF outer wall thickness and collagen density.^{11, 13} Additional unstained cross sections in this same region were used for immunohistochemistry and immunofluorescence microscopy.

Immunohistochemistry.

Tissue sections were de-paraffined using xylene and a graded series of ethanol. For antigen-retrieval, sections were heated in citric acid buffer (pH 6.0) at 100°C for 15 minutes. Non-specific background staining of endogenous peroxidase was treated with 0.3% hydrogen peroxide for 30 minutes, and sections were blocked with 3% bovine serum albumin in PBS (pH 7.4) for 1 hour at room temperature. Sections were then incubated at 4°C with the primary antibody (Major Resources Table). After overnight incubation, the sections were incubated with HRP conjugated secondary antibody for 1 hour at room temperature and treated with Dako Liquid DAB+ Substrate Chromogen System (GV825, Agilent Dako) to detect the reaction products. Finally, the sections were counterstained with Dako Mayer’s Hematoxylin (Lillie’s Modification) Histological Staining Reagent (S3309, Agilent Dako). For negative controls for the antibodies, IgG isotype controls, negative tissue controls and endogenous tissue background controls were used.

Immunofluorescence.

Tissue sections were de-paraffined and then heated in citric acid buffer (pH 6.0) at 100°C for 15 minutes for antigen retrieval. The sections were blocked with 3% bovine serum albumin prior to incubation with primary antibodies (Major Resources Table) overnight at 4°C. After incubation, the sections were incubated with Alexa Fluoro secondary antibodies for 1hour and stained with 4’,6-diamidino-2-phenylindole (DAPI) (P36935, Invitrogen) to stain cellular nuclei. Positively staining cells were counted per high power fields or measured the intensity. For negative controls for the antibodies, IgG isotype controls, negative tissue controls and endogenous tissue background controls were used.

Western Blot.

The entire IVC was removed with care taken to avoid surrounding arterial and connective tissues. Proteins were extracted with RIPA lysis buffer containing protease inhibitors from single IVC.^{11, 13, 21} Protein concentrations were assessed using a colorimetric assay (Bio Rad), and equal amounts of protein were run on a sodium dodecyl sulfate polyacrylamide gel electrophoresis gel and then transferred to a polyvinylidene difluoride membrane. After blocking with 5% skim milk, membranes were probed overnight with the primary antibodies (Major Resources Table). After overnight incubation, the membranes were incubated with HRP conjugated secondary antibodies for 1 hour at room temperature and were developed with using Western Lightning Plus ECL reagent (NEL105001EA, PerkinElmer; Waltham, MA).

RNA Extraction and Real-Time Quantitative Polymerase Chain Reaction.

The entire IVC was removed with care taken to avoid surrounding arterial and connective tissue. Total RNA was isolated from these samples by using RNeasy Mini Kit (74106, Qiagen; Valencia, CA) from single IVC.^{11, 13} RNA quality was confirmed using a NanoDrop spectrophotometer (Thermo Scientific; Wilmington, DE) to determine the 260/280 nm ratio. Reverse transcription was performed using SuperScript III First-Strand Synthesis Supermix (11752250, Invitrogen). Real-time PCR was performed using iQ SYBR Green Supermix (1708880, Bio-Rad) and amplified for 40 cycles using the iQ5 Real-Time PCR Detection System (Bio-Rad). Primers are summarized in Major Resources Table. All samples were normalized to GAPDH amplification and graphed in arbitrary units.

T-cell collection and purification.

Peripheral T-cells were collected for FACS analysis to confirm T-cell phenotype after PD-L1 Ab treatment. Mouse spleens were harvested from mice and minced in PBS. The cell suspension was filtered through 50μm cell strainers and loaded on the top of the lymphocyte separation medium (1077, Promo Cell; Heidelberg, Germany) and centrifuged at 440g for 30 minutes. Lymphocytes were collected from the mononuclear cell layer. CD3+T-cells were purified using a T-cell Enrichment Column (MTCC-500, R&D systems). The purity was confirmed by flow cytometry. After washing the cell suspension with PBS, cell numbers were adjusted to 10⁶ cells/ml.

Flow cytometry.

The cell resuspension was blocked with 3% bovine serum albumin prior to incubation with primary antibodies (Major Resources Table) for 30 minutes at room temperature. After primary antibody incubation, the cells were fixed with 0.01% paraformaldehyde for 15 minutes and Incubated with Alexa Fluoro secondary antibodies for 30 min at room temperature. The LSR II flow cytometer (BD Biosciences, CA) was used for analysis.

In vitro T-cell proliferation, migration and IL-2 production assays.

Anti-mouse CD3 antibody (MAB4841, R&D systems; 1 μg/ml dilution) was loaded into each well of a 96-well plate and coated on the well bottoms overnight at 4°C. T-cells were incubated at 37°C in the CD3 antibody-coated 96-well plate in RPMI/10%FBS with PD-L1 Fc (758208, BioLegend; San Diego, CA; 0.5 μg/ml) or the same amount of PBS. Only CD3 Ab was used to stimulate T-cells to minimize potential off-target effects of additional stimulants. PD-L1 Fc was included in the medium as a stimulant. Cell numbers were counted to assess cell proliferation. After 24-hour incubation, cells and cell supernatant were collected for the migration assay and interleukin (IL)-2 measurement, respectively. The migration assay was performed using the CytoSelect 96-Well Cell Migration Assay (CBA-105, Cell Biolabs; San Diego, CA) according to the instructions for use. IL-2 in the cell supernatant were measured using IL-2 ELISA kit (M2000, R&D systems) according to the instructions for use. We collected T-cells incubated with PD-L1 Fc for FACS analysis to confirm the effects of PD-L1 on skewing of T-cell phenotypes.

Statistics.

Data are represented as mean value ± standard error of the mean. All data were analyzed using Prism 8 software (GraphPad Software, Inc., La Jolla, CA). Equal variance was confirmed using the Shapiro-Wilk test prior to performing parametric analyses.. Statistical significance was determined using Student’s t-test or ANOVA with Sidak’s post hoc correction. We used the Mann-Whitney U test or the Kruskal-Wallis test with Dunn’s post hoc correction if the sample size was smaller than 6. Patency and survival were analyzed using Kaplan–Meier analysis. P values < 0.05 were considered significant.

Results

Decreased PD-L1 expression and increased PD-1+T-cells during venous remodeling in wild type mice

To determine whether PD-L1 expression is regulated during venous remodeling that occurs during AVF maturation, we created AVF in male wild type mice and assessed the time course of PD-L1 expression. Western blot and qPCR showed significantly decreased PD-L1 expression and mRNA transcripts in the AVF as early as day 3 (Figure 1A, 1B), and immunohistochemistry showed significantly decreased PD-L1 immunoreactivity in the intima at 3 days after surgery that returned to baseline by 21 days (Figure 1C). These findings are consistent with diminished PD-L1 expression in the remodeling venous wall.

Figure 1. Decreased PD-L1 expression is associated with PD-1+T-cell accumulation during venous remodeling.

(A) Representative Western blot analysis of PD-L1 and GAPDH immunoreactivity in AVF at day 0, 3, 7, 14 and 21. Bar graph shows relative densitometry of PD-L1. n=3–6, P=0.0238 (Kruskal Wallis test). P=0.0150 (post-hoc). (B) Bar graphs show relative number of PDL1 mRNA transcripts in AVF at day 0, 3, 7 and 21. n=3, P=0.0234 (Kruskal Wallis test). P=0.0197 (post-hoc). (C) Representative photomicrographs of AVF at day 0, 3, 7 and 21 stained with anti-PD-L1 antibody. Bar graphs show quantification of PD-L1 positive area. n=3, P=0.0014 (Kruskal Wallis test). P=0.0197 (post-hoc). (D) Representative photomicrographs showing CD3 (red) and PD-1 (green) in the AVF wall at days 0, 3, 7 or 21. Scale bar, 50 μm. Bar graph shows quantification of CD3+PD-1+ cells in the AVF at day 0, 3, 7 or 21. n=3, P=0.0034 (ANOVA), P=0.0012 (Kruskal-Wallis). P=0.0094 and P=0.0031 (post hoc). (E) Bar graph shows relative numbers of migrating mouse lymphocytes treated with vehicle (control) or PD-L1 Fc. The values are normalized to the control group. n=4. P=0.0272 (t-test), P=0.0429 (Mann-Whitney U test). (F) Line graph shows relative cell numbers of mouse lymphocytes treated with vehicle (control) or PD-L1 Fc. n=4. P=0.2660 (ANOVA). (G) Bar graph shows IL-2 concentration in cell culture supernatants of mouse lymphocytes treated with vehicle (control) or PD-L1 Fc. n=4. P=0.0178 (t-test), P=0.0286 (Mann-Whitney U test).

Since PD-L1 inhibits PD-1+T-cell activation, we assessed PD-1+T-cell accumulation in the remodeling venous wall (Figure 1D); PD-1+T-cells in the remodeling venous wall initially increased and then were reduced at day 21, in inverse proportion to PD-L1 expression (Figure 1A–C). Similar trends were observed in the remodeling venous wall of female wild type mice (Supplemental figure I). These findings suggest that PD-L1 may regulate PD-1+T-cell accumulation in the remodeling venous wall. Since these findings suggest that decreased PD-L1 expression was associated with PD-1+T-cell accumulation in the remodeling venous wall, we confirmed that PD-L1 regulates PD-1+T-cell function in vitro. Stimulation of PD-L1 using PD-L1 Fc reduced T-cell migration (Figure 1E), but not T-cell proliferation (Figure 1F); in addition, stimulation of PD-L1 reduced T-cell IL-2 secretion (Figure 1G). There was no significant difference in migration between control Ab- or PD-L1 Ab-treated T-cells (Supplemental figure IC). These data suggest that PD-L1 ligand regulates T-cell functions such as migration that could enable T-cell accumulation in the remodeling venous wall.

PD-L1 regulates CD4+ T-cell differentiation and macrophage polarization during venous remodeling in wild type mice

Since T-cell accumulation in the remodeling venous wall is also associated with macrophage accumulation,¹¹ and PD-L1 is associated with T-cell accumulation (Figure 1), we next determined if PD-L1 regulation of T-cell accumulation is a mechanism of macrophage accumulation using an antibody against PD-L1 that inhibits the PD-L1-PD-1 association. There were no adverse effects of PD-L1 antibody treatment including no change in postoperative survival (Supplemental figure IIA) or body weight (Supplemental figure IIB).

We first confirmed the effects of PD-L1 antibody treatment on T-cell accumulation in the remodeling venous wall (day 7; Figure 2A); PD-L1 inhibition was associated with increased CD8+T-cell (Figure 2B) and naïve CD8+ T-cell accumulation (Figure 2C), with few CTL observed and without a significant difference between the groups (Figure 2D). These findings suggest that PD-L1 antibody treatment increased only inactivated CD8+T-cell accumulation in the AVF. We next confirmed the effects of PD-L1 antibody treatment on CD4+T-cell accumulation in the remodeling venous wall; PD-L1 antibody treatment reduced CD4+T-cell accumulation (Figure 2E). Accumulation of T-cells and macrophages in the remodeling venous wall was observed at day 7 and significantly decreased at day 21 (Figure 2F–M). To identify which phenotypes of CD4+T-cells were present, we used immunofluorescence to assess CD4+IFN-γ (T-helper 1 (Th1)), CD4+IL-4 (T-helper 2 (Th2)) or CD4+Foxp3 (Treg) cells. PD-L1 antibody treatment increased Th1 (Figure 2F) and Th2 (Figure 2G) cell accumulation; however, PD-L1 antibody treatment reduced Treg accumulation (Figure 2H). However, there was no accumulation of Th17-positive cells in the remodeling venous wall (Supplemental figure ID); FACS analysis of peripheral T-cells in vivo (Supplemental figure IE) as well as T-cells incubated with PD-L1 Fc in vitro (Supplemental figure IF) confirmed lack of significant differences compared with control, suggesting that PD-1/PD-L1 interactions locally affect venous remodeling. Since Th1 cells induce M1-type macrophage polarization, and Th2 and Treg induce M2-type macrophage polarization, we determined if PD-L1 antibody treatment increased M1-type and reduced M2-type macrophage accumulation in the remodeling venous wall (Figure 2I). As expected, PD-L1 antibody treatment increased M1-type macrophages (Figure 2J, 2K) but reduced M2-type macrophages (Figure 2L, 2M). These data suggest that PD-L1 antibody treatment increases Th1 and Th2, but reduces Treg cells, thereby inducing M1-type macrophage polarization in the remodeling venous wall.

Figure 2. PD-L1 Ab is associated with T-cell and macrophage accumulation in wild type mouse AVF.

(A) Representative photomicrographs showing CD3 (green) and CD8 (red) or CCR7 (green) and CD8 (red) in the AVF wall (day 7) in mice treated with control antibody or PDL1 antibody (Ab). Scale bar, 50 μm. (B) Bar graph shows quantification of CD3+CD8+ cells in the AVF (day 7) treated with control or PD-L1 Ab. n=3. P=0.0005 (t-test), P=0.1000 (Mann-Whitney U test). (C) Bar graph shows quantification of CCR7+CD8+ cells in the AVF wall (day 7) in mice treated with control or PD-L1 Ab. n=3. P=0.0009 (t-test), P=0.1000 (Mann-Whitney U test). (D) Bar graph shows quantification of CCR7-CD8+ cells in the AVF wall (day 7) in mice treated with control or PD-L1 Ab. n=3. P=0.8774 (t-test), P=0.9999 (Mann-Whitney U test). (E) Representative photomicrographs showing CD3 (green) and CD4 (red) in the AVF wall (day 7) in mice treated with control or PD-L1 Ab. Scale bar, 50 μm. Bar graph shows quantification of CD3+CD4+ cells in the AVF. n=6. P<0.0001 (t-test). (F) Representative photomicrographs showing CD4 (green) and IFN-γ (red) in mouse AVF treated with control or PD-L1 Ab (day 7 or 21). Scale bar, 50 μm. Bar graph shows quantification of CD4+IFN-γ+ cells in the AVF at day 7 or 21. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (G) Representative photomicrographs showing CD4 (green) and IL-4 (red) in mouse AVF treated with control or PD-L1 Ab (day 7 or 21). Scale bar, 50 μm. Bar graph shows quantification of CD4+IL-4+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (H) Representative photomicrographs showing CD4 (green) and Foxp3 (red) in mouse AVF treated with control or PD-L1 Ab (day 7 or 21). Scale bar, 50 μm. Bar graph shows quantification of CD4+Foxp3+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc), P=0.0055 (post-hoc). (I) Representative photomicrographs showing CD68 (green) and TNF-α (red), CD68 (green) and iNOS (red), CD68 (green) and TGM2 (red) or CD68 (green) and CD206 (red) in mouse AVF treated with control or PD-L1 Ab (day 7 or 21). Scale bar, 50 μm. (J) Bar graph shows quantification of CD68+TNF-α+ cells in the AVF. n=4–5. P=0.0225 (ANOVA). P=0.0372 (post-hoc). (K) Bar graph shows quantification of CD68+iNOS+ cells in the AVF. n=4–5. P=0.0372 (ANOVA). P=0.0134 (post-hoc). (L) Bar graph shows quantification of CD68+TGM2+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (M) Bar graph shows quantification of CD68+CD206+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc).

PD-L1 regulates vascular wall thickening and patency in wild type mice

Since M2-type macrophages promote adaptive wall thickening during venous remodeling,^{8, 11, 13} we next determined whether reduced M2-type macrophages using PD-L1 antibody treatment (Figure 2L, 2M) was associated with reduced wall thickening during venous remodeling in wild type mice. PD-L1 antibody treatment reduced wall thickening in the remodeling venous wall at day 21 (Figure 3A) without any differences in outward remodeling (Supplemental figure IIC) and shear stress (Supplemental figure IID). Reduced wall thickening with PD-L1 antibody treatment was also associated with reduced smooth muscle cell (SMC) proliferation (Figure 3B) without any increases in apoptosis (Supplemental figure IIE). Similarly, reduced wall thickening with PD-L1 antibody treatment was not associated with any differences in collagen density (Supplemental figure IIF).

Figure 3. PD-L1 Ab inhibits vascular wall thickening and reduces AVF patency.

(A) Representative photomicrographs of the AVF wall of mice treated with control or PD-L1 Ab, stained with Elastin van Gieson (EVG); upper row, day 7; lower row, day 21. Scale bar, 25μm. Arrowheads show intima-media thickness. Bar graph shows intima-media thickness of the AVF wall. n=4–6. P=0.0030 (ANOVA). P<0.0001 (post hoc). (B) Representative photomicrographs showing α-SMA (green) and ki67 (red) in the AVF wall (day 7) in mice treated with control or PD-L1 Ab. Scale bar, 50 μm. Bar graph shows percentage of α-SMA+ki67+ cells in the AVF (day 7). n=6. P<0.0001 (t-test) ,P=0.0079 (Mann-Whitney U test). (C) Patency curves after AVF creation in wild type mice treated with control or PD-L1 Ab. n=16–17. P=0.0356 (log-rank test). (D) Representative photomicrographs of the patent or failed AVF, stained with Hematoxylin Eosin (HE) or phosphotungstic acid hematoxylin (PTAH); day 42. Scale bar, 25μm. * The fistula between aorta and IVC. (E) Representative photomicrographs showing vWF (green) and VCAM1 (red) or Thrombomodulin (TM) and Tissue factor (TF) in the AVF wall (day 7 and 21) in mice treated with control or PD-L1 Ab. Scale bar, 50 μm. (F) Bar graph shows intensity of VCAM1 in the AVF. n=4–5. P=0.0005 (ANOVA). P=0.0021 (post-hoc). (G) Bar graph shows intensity of vWF in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (H) Bar graph shows intensity of TM in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (I) Bar graph shows intensity of TF in the AVF. n=4–5. P=0.1041 (ANOVA).

Because PD-L1 antibody treatment is associated with reduced adaptive wall thickening during venous remodeling (days 0–21 in the mouse model; Figure 3A), we determined if PD-L1 antibody treatment affects AVF patency (days 21–42). PD-L1 antibody treatment was associated with reduced AVF patency in wild type mice, although loss of patency occurred as early as day 14 (Figure 3C) suggesting that AVF failure with PD-L1 antibody treatment may be due to thrombosis rather than neointimal hyperplasia.²⁰ Staining the AVF with both hematoxylin and eosin as well as phosphotungstic acid hematoxylin stains confirmed the presence of thrombosis in failed fistulae at the fistula site; there was no thrombosis in patent AVF (Figure 3D). These data show that PD-L1 antibody treatment is associated with AVF thrombosis that reduces patency in wild type mice.

Since inflammation can cause endothelial cell dysfunction with thrombus formation,²² we determined whether PD-L1 antibody treatment was associated with markers of vascular inflammation (Figure 3E). PD-L1 antibody treatment was associated with increased vascular cell adhesion molecule 1 (VCAM1) expression (Figure 3F), reduced von Willebrand factor (vWF) expression (Figure 3G) and reduced thrombomodulin expression (Figure 3H), without any significant difference in tissue factor expression between the groups (Figure 3I). These data suggest that PD-L1 antibody treatment is associated with inflammation and reduced anticoagulant function. In toto, PD-L1 antibody treatment is associated with reduced SMC proliferation and less wall thickening in the remodeling venous wall, as well as inflammation and endothelial dysfunction that may reduce AVF patency in wild type mice, suggesting that PD-L1 has a regulatory function during venous adaptive remodeling.

PD-L1 regulation of venous remodeling is T-cell dependent

To determine whether the reduced wall thickening and patency with PD-L1 antibody treatment are mediated by T-cells, we performed AVF in nude mice. There were no adverse effects of PD-L1 antibody treatment in nude mice after AVF creation, including no change in survival (Supplemental figure IIIA) or body weight (Supplemental figure IIIB). Immunofluorescence for CD3+CD4 or CD3+CD8 confirmed the absence of T-cells in the venous walls of nude mice (Supplemental figure IIIC). PD-L1 antibody treatment was not associated with any significant differences in either CD68+TNF-α+ dual-positive cells or CD68+iNOS+ dual-positive cells (Figure 4A–C), that is M1-type macrophages; similarly, there were no significant differences in either CD68+TGM2+ dual-positive cells or CD68+CD206+ dual-positive cells (Figure 4D, 4E), that is M2-type macrophages.

Figure 4. Effects of PD-L1 Ab on venous remodeling are lost in nude mice.

(A) Representative photomicrographs showing CD68 (green) and TNF-α (red), CD68 and iNOS, CD68 (green) and TGM2 (red) or CD68 (green) and CD206 (red) in the AVF wall (day 7 and 21) in nude mice treated with control or PD-L1 Ab. Arrowheads show intima-media thickness. Scale bar, 50 μm. (B) Bar graph shows quantification of CD68+TNF-α+ cells in the AVF at day 7 or 21. n=3–4. P=0.7856 (ANOVA). (C) Bar graph shows quantification of CD68+iNOS+ cells in the AVF at day 7 or 21. n=3–4. P=0.8586 (ANOVA). (D) Bar graph shows quantification of CD68+TGM2+ cells in the AVF at day 7 or 21. n=3–4. P=0.7153 (ANOVA). (E) Bar graph shows quantification of CD68+CD206+ cells in the AVF at day 7 or 21. n=3–4. P=0.8005 (ANOVA). (F) Representative photomicrographs of the AVF wall of nude mice treated with control or PD-L1 Ab, stained with Elastin van Gieson (EVG); day 7 or 21. Scale bar, 25μm. Bar graph shows intima-media thickness of the AVF wall. n=3–4. P=0.6844 (ANOVA). (G) Representative photomicrographs showing α-SMA (green) and ki67 (red) in the AVF wall (day 7) in nude mice treated with control or PD-L1 Ab. Scale bar, 50 μm. Bar graph shows percentage of α-SMA+ki67+ cells in the AVF. n=4. P=0.6183 (t-test) ,P=0.7429 (Mann-Whitney U test). (H) Patency curves after AVF creation in nude mice treated with control or PD-L1 Ab. n=17–19. P=0.7746 (log-rank test). (I) Representative photomicrographs showing vWF (green) and VCAM1 (red) or Thrombomodulin (TM) and Tissue factor (TF) in the AVF wall (day 7 and 21) in nude mice treated with control or PD-L1 Ab. Scale bar, 50 μm. (J) Bar graph shows intensity of VCAM1 in the AVF wall. n=3–4. P=0.6924 (ANOVA). (K) Bar graph shows intensity of vWF in the AVF wall. n=3–4. P=0.6860 (ANOVA). (L) Bar graph shows intensity of TM in the AVF wall. n=3–4. P=0.3535 (ANOVA). (M) Bar graph shows intensity of TF in the AVF wall. n=3–4. P=0.6744 (ANOVA).

Since macrophage accumulation is associated with wall thickening,^{8, 11–13} we assessed wall thickness in the remodeling venous wall of nude mice. The vascular wall thickness was approximately 10 μm at day 21 after AVF creation, however, there was no significant difference in venous wall thickness between nude mice treated with control or PD-L1 antibody (Figure 4F). Similarly, there was no significant difference in SMC proliferation in the venous wall of nude mice treated with control or PD-L1 antibody (Figure 4G), unlike the association in wild type mice (Figure 3B). There were also no significant differences in outward remodeling (Supplemental figure IIID) or shear stress (Supplemental figure IIIE). We also assessed the patency of AVF in nude mice treated with control or PD-L1 antibody; although PD-L1 antibody treatment was associated with reduced AVF patency in wild type mice (Figure 3C), there was no significant difference in AVF patency among control and PD-L1 antibody treated nude mice (Figure 4H). As expected, there were no significant differences in VCAM1 immunoreactivity (Figure 4I, 4J), vWF immunoreactivity (Figure 4K), thrombomodulin immunoreactivity (Figure 4L) and tissue factor immunoreactivity (Figure 4M) in the venous walls of control and PD-L1 antibody treated nude mice. These data show that the effects of PD-L1 antibody treatment on the remodeling venous wall that were observed in wild type mice were diminished in the remodeling venous wall in nude mice, suggesting that the effects of PD-L1 antibody treatment are T-cell dependent.

T-cell transfer restores PD-L1 regulation of macrophage accumulation in the remodeling venous wall

To confirm the effects of PD-L1 antibody treatment are T-cell-dependent, we assessed the effects of PD-L1 antibody treatment in the remodeling venous wall of nude mice after T-cell transfer. T-cells were collected from wild type mouse spleens and purified; 93.3% purity was confirmed by flow cytometry (Supplemental figure IVA). T-cells were then transferred into nude mice and confirmed to be present (Supplemental figure IVB). There were no adverse effects of PD-L1 antibody treatment on nude mice after T-cell transfer including no reduction in either postoperative survival (Supplemental figure IVC) or body weight (Supplemental figure IVD).

We assessed T-cell accumulation in the remodeling venous wall of nude mice after T-cell transfer (Figure 5A). Similar to wild type mice, PD-L1 antibody treatment was associated with fewer numbers of CD4+T-cells (Figure 5B) and increased numbers of CD8+T-cells (Figure 5C) in the venous wall of nude mice after T-cell transfer. Similarly, PD-L1 antibody treatment was associated with increased numbers of Th1 cells (Figure 5D) and reduced numbers of Treg cells (Figure 5E) without any change in Th2 cells (Figure 5F) in the remodeling venous walls of nude mice after T-cell transfer. Furthermore, PD-L1 antibody treatment increased the numbers of M1-type macrophages (Figure 5G–I) and reduced the numbers of M2-type macrophages (Figure 5J, 5K) in the remodeling venous walls of nude mice after T-cell transfer. These data show that the effects of PD-L1 antibody treatment on macrophage accumulation in the remodeling venous wall were restored in nude mice after T-cell transfer.

Figure 5. Restored effects of PD-L1 Ab on T-cell and macrophage accumulation in nude mouse AVF after T-cell transfer.

(A) Representative photomicrographs showing CD3 (green) and CD4 (red) or CD3 (green) and CD8 (red) in the AVF wall (day 7) in nude mice after T-cell transfer treated with control or PD-L1 Ab. Scale bar, 50 μm. (B) Bar graph shows quantification of CD3+CD4+ cells in the AVF. n=5. P<0.0001 (t-test). P=0.0079 (Mann-Whitney U test). (C) Bar graph shows quantification of CD3+CD8+ cells in the AVF. n=5. P<0.0001 (t-test), P=0.0079 (Mann-Whitney U test). (D) Representative photomicrographs showing CD4 (green) and IFN-γ (red) in the AVF wall in nude mice after T-cell transfer treated with control or PD-L1 Ab. Day 7 or 21. Scale bar, 50 μm. Bar graph shows quantification of CD4+IFN-γ+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (E) Representative photomicrographs showing CD4 (green) and Foxp3 (red) in the AVF wall of nude mice after T-cell transfer treated with control or PD-L1 Ab. Day 7 or 21. Scale bar, 50 μm. Bar graph shows quantification of CD4+Foxp3+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (F) Representative photomicrographs showing CD4 (green) and IL-4 (red) in the AVF wall in nude mice after T-cell transfer treated with control or PD-L1 Ab. Day 7 or 21. Scale bar, 50 μm. Bar graph shows quantification of CD4+IL-4+ cells in the AVF. n=4–5. P=0.2819 (ANOVA). (G) Representative photomicrographs showing CD68 (green) and TNF-α (red), CD68 (green) and iNOS (red), CD68 (green) and TGM2 (red) or CD68 (green) and CD206 (red) in the AVF in nude mice after T-cell transfer treated with control or PD-L1 Ab (day 7 or 21). Scale bar, 50 μm. (H) Bar graph shows quantification of CD68+TNF-α+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (I) Bar graph shows quantification of CD68+iNOS+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (J) Bar graph shows quantification of CD68+TGM2+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (K) Bar graph shows quantification of CD68+CD206+ cells in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc).

T-cell transfer restores PD-L1 regulation of venous remodeling and AVF patency

Since the effects of PD-L1 antibody treatment on macrophage accumulation in the AVF were restored in nude mice after T-cell transfer (Figure 5), and reduced M2-type macrophages were associated with reduced venous wall thickening in wild type mice (Figure 3A), we next determined if reduced vascular wall thickening by PD-L1 antibody treatment was restored in the remodeling venous wall of nude mice after T-cell transfer. After T-cell transfer into nude mice, the vascular wall thickness was approximately 15 μm at day 21 (Figure 6A) which is thicker than the wall of AVF in nude mice without T-cell transfer (Figure 4F). These findings suggest that T-cells have an important role regulating vascular wall thickening during venous remodeling. PD-L1 antibody treatment was associated with reduced venous wall thickening at day 21 (Figure 6A) without differences in outward remodeling (Supplemental figure IVE) or shear stress (Supplemental figure IVF). There was also reduced SMC proliferation with PD-L1 antibody treatment (Figure 6B). In addition, T-cell transfer restored the reduced AVF patency that was associated with PD-L1 antibody treatment (Figure 6C). T-cell transfer also restored the increased endothelial inflammation and reduced anticoagulant phenotype associated with PD-L1 antibody treatment (Figure 6D–H). In toto, the effects of PD-L1 antibody on venous remodeling were diminished in nude mice but were restored after T-cell transfer, suggesting that the effects of PD-L1 antibody treatment on venous remodeling were T-cell-dependent.

Figure 6. Restored effects of PD-L1 Ab on venous remodeling and patency of nude mouse AVF after T-cell transfer.

(A) Representative photomicrographs of the AVF wall of nude mice after T-cell transfer treated with control or PD-L1 Ab, stained with Elastin van Gieson (EVG); upper row, day 7; lower row, day 21. Scale bar, 25μm. Arrowheads show intima-media thickness. Bar graph shows intima-media thickness of the AVF wall. n=4–5. P=0.0009 (ANOVA). P<0.0001 (post hoc). (B) Representative photomicrographs showing α-SMA (green) and ki67 (red) in the AVF wall (day 7) in nude mice after T-cell transfer treated with control or PD-L1 Ab. Scale bar, 50 μm. Bar graph shows percentage of α-SMA+ki67+ cells in the AVF. n=5. P=0.0004 (t-test) ,P=0.0079 (Mann-Whitney U test). (C) Patency curves after AVF creation in nude mice after T-cell transfer treated with control or PD-L1 Ab. n=22–29. P=0.0248 (log-rank test). (D) Representative photomicrographs showing vWF (green) and VCAM1 (red) or Thrombomodulin (TM) and Tissue factor (TF) in the AVF wall (day 7 and 21) in nude mice after T-cell transfer treated with control or PD-L1 Ab. Scale bar, 50 μm. (E) Bar graph shows intensity of VCAM1 in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 (post-hoc). (F) Bar graph shows intensity of vWF in the AVF. n=4–5. P<0.0001 (ANOVA). P<0.0001 and P=0.0002 (post-hoc). (G) Bar graph shows intensity of TM in the AVF. n=4–5. P=0.0006 (ANOVA). P=0.0065 and P=0.0339 (post hoc). (H) Bar graph shows intensity of TF in the AVF. n=4–5. P=0.3553 (ANOVA).

Discussion

During venous remodeling, decreased PD-L1 immunoreactivity was associated with PD-1 positive cell accumulation in the sub-endothelium consistent with PD-L1 suppression of T-cell migration and IL-2 production (Figure 1). Anti-PD-L1 antibody increased Th1 and Th2 cells, but decreased Treg cells, which was associated with increased M1-type and decreased M2-type macrophage in the remodeling venous wall (Figure 2). Increased M1-type macrophages and decreased M2-type macrophages were associated with reduced vascular wall thickening and patency of the AVF (Figure 3). The effects of the anti-PD-L1 antibody were diminished in T-cell deficient nude mice (Figure 4) and were restored after T-cell transfer (Figure 5, 6). These data suggest that PD-L1 specifically regulates T-cell accumulation in the venous wall to regulate macrophage accumulation, wall thickening and patency during venous remodeling.

The primary finding in this study is that inhibition of PD-L1 is associated with reduced vascular thickening and reduced patency during adaptive venous remodeling (Figure 3). Similar to our previous studies,^{12, 13} this study showed the association of reduced M2-type macrophage accumulation with reduced SMC proliferation and reduced vascular wall thickening (Figures 2, 3). Since PD-L1 induces T-cells to differentiate into Treg to induce polarization of M2-type macrophages,^{8, 14–17, 23} our data is consistent with PD-L1 inhibition decreasing Treg cell accumulation to decrease M2-type macrophages (Figure 2) and thereby reduce venous wall thickening. However, this study also shows that inhibition of PD-L1 reduces AVF patency due to increased thrombosis (Figure 3), consistent with the potential for inflammation to cause endothelial dysfunction with thrombus formation²² and our detection of markers of endothelial dysfunction (Figure 3). Although we do not show a specific role for M1-type macrophages, increased M1-type macrophages by inhibition of PD-L1 might be associated with increased inflammation and reduced AVF patency, since inflammation caused by M1-type macrophages is associated with vein graft failure.^24–26 In toto, these data show that inhibition of PD-L1 is associated with reduced Treg accumulation, increased M1-type macrophages, decreased M2-type macrophages, reduced venous wall thickening and reduced AVF patency, suggesting that PD-L1 regulates adaptive venous remodeling.

PD-L1 is a regulator of the immune system with specific activity to regulate multiple T-cell functions.^{14–17, 23} Anti-PD-L1 antibodies are in clinical use in the treatment of cancer to activate T-cell immunity against cancer cells.²⁷ PD-1/PD-L1 interactions on T-cells decrease the initiation of the secondary T-cell immune response. As we have previously shown that TGF-β, MCP-1, LIX and IFN-γ are increased in the remodeling AVF wall,¹¹ it is possible that PD-L1 may inhibit T-cell activation caused by TGF-β, MCP-1, LIX and IFN-γ. PD-L1 is normally expressed in endothelial cells and regulates T-cell differentiation, proliferation and cytokine production;^{16, 28} however, the roles of endothelial PD-L1 during venous remodeling are not well understood. Our data shows that decreased PD-L1 expression in the remodeling venous wall is associated with accumulation of sub-endothelial PD-1 positive cells, consistent with our in vitro data that PD-L1 suppresses T-cell migration and IL-2 production (Figure 1) and suggesting that PD-L1 may regulate T-cell migration and cytokine production in the remodeling venous wall. Our data show that PD-L1 regulates specific subsets of T-cells during venous remodeling, with differential effects on CD8+T-cells and CD4+T-cells (Figure 2); interestingly, most of the CD8+T-cells in the venous wall were inactivated CD8+T-cells, suggesting that CD8+T-cells may not be a major regulator of venous remodeling. On the other hand, CD4+T-cells, such as Th1, Th2 and Treg, may be important regulators of venous remodeling.⁸

Since PD-L1 induces Treg differentiation to suppress some types of inflammatory cells,^{14–17, 23} inhibition of PD-L1 activity during venous remodeling may be a mechanism by which reduced Treg accumulation increases Th1 and Th2 cells in the remodeling venous wall. Decreased Treg (Figure 2H) and increased Th1 (Figure 2F) and Th2 (Figure 2G) T-cells may result in decreased numbers of M2-type macrophages (Figure 2L–M) and increased numbers of M1-type macrophages (Figure 2J–K). This is consistent with our data showing that the effects of anti-PD-L1 antibody were diminished in nude mice (Figure 4) and restored after T-cell transfer (Figure 5, 6), that is they are dependent on T-cells. AVF in wild type mice showed increased Th2 cells but decreased M2-type macrophage accumulation; this finding is not consistent with data showing that Th2 cells induce M2-type macrophage polarization²⁹. AVF in nude mice showed few Th2 cells (Figure 5F) and normal venous remodeling (Figure 6A), suggesting that Th2 cells may not a mechanism by which PD-L1 regulates venous remodeling. We hypothesize that Treg have greater effects on M2 polarization than Th2, consistent with our observation that AVF in nude mice showed restored effects on M2 polarization after T-cell transfer that necessarily include a few number of Th2. Thus, additional studies are needed to demonstrate specific roles for each phenotype of T-cells. Macrophages and B cells also express PD-1 that may interact with PD-L1 during venous remodeling; however, lack of effects of PD-L1 Ab in nude mouse AVF (Figure 4) and restored effects of PD-L1 Ab in nude mouse AVF after T-cell transfer (Figure 5) suggest that the effects of PD-L1 are independent of macrophages and B cells. As such, PD-L1 may play a specific regulatory role during adaptive venous remodeling by regulating specific subsets of T-cells, a potential mechanism by which the venous wall thickens in response to the fistula environment but without excessive wall thickening and neointimal hyperplasia.

There are some limitations in this study. Although we show that inhibition of PD-L1 affects adaptive venous remodeling, we did not assess the effects of increased PD-L1 activity; since PD-L1 is decreased in the remodeling venous wall and inhibition of PD-L1 is associated with AVF failure, it is possible that increased PD-L1 activity may improve AVF maturation. In addition, although we confirmed that female mice show similar changes in PD-L1 expression as male mice during venous remodeling, our mechanistic studies were only performed in male mice; however, female mice show similar early venous adaptive remodeling as do male mice, despite sex differences in hemodynamics.^{30, 31} We assessed subtypes of T-cells and macrophages with immunofluorescence; however, this technique cannot show the relationships between the distinct subtypes of T-cells and macrophages. Flow cytometry may be able to analyze these relationships, but this technique remains challenging since collecting intact T-cells and macrophages from the wall of a mouse AVF is currently not possible.

In conclusion, we show that PD-L1 regulates T-cell differentiation with subsequent macrophage polarization during venous remodeling; these effects are associated with wall thickening and fistula patency. Regulation of PD-L1 activity may be a therapeutic target for clinical translation to improve AVF maturation. In addition, use of checkpoint inhibitor therapy with anti-PD-L1 antibodies may warrant additional study in patients with AVF on hemodialysis, as this therapy may be associated with increased AVF thrombosis.

Highlights.

• PD-L1 regulates T cell and macrophage accumulation in the remodeling venous wall.

• Inhibition of PD-L1 reduces arteriovenous fistula wall thickening but also reduces fistula patency due to thrombosis.

• The effects of PD-L1 inhibition are dependent on T cells.

• PD-L1 activity may be a potential therapeutic target to improve venous remodeling.

Abbreviations

AVF

arteriovenous fistula

CTL

cytotoxic T-cells

DAPI

4’,6-diamidino-2-phenylindole

EVG

elastica van gieson stains

HE

hematoxylin eosin stains

IL

interleukin

IVC

inferior vena cava

PD-1

programmed death receptor 1

PD-L1

programmed death ligand 1

PTAH

phosphotungstic acid hematoxylin stains

SMC

smooth muscle cell

Th1

T-helper 1 cells

Th2

T-helper 2 cells

Treg

regulatory T-cells

VCAM-1

vascular cell adhesion molecule 1

vWF

von Willebrand factor