CD44 Promotes Inflammation and Extracellular Matrix Production During Arteriovenous Fistula Maturation

Abstract

Objective

Arteriovenous fistulae (AVF) remain the optimal conduit for hemodialysis access but continue to demonstrate poor patency and poor rates of maturation. We hypothesized that CD44, a widely expressed cellular adhesion molecule that serves as a major receptor for extracellular matrix (ECM) components, promotes wall thickening and ECM deposition during AVF maturation.

Approach and Results

AVF were created via needle puncture in wild-type (WT) C57BL/6J and CD44 knockout (KO) mice. CD44 mRNA and protein expression was increased in WT AVF. CD44 KO mice showed no increase in AVF wall thickness (8.9 μm vs. 26.8 μm; P = 0.0114), collagen density, and hyaluronic acid density, but similar elastin density when compared to control AVF. CD44 KO mice also showed no increase in VCAM-1 expression, ICAM-1 expression and MCP-1 expression in the AVF compared to controls; there were also no increased M2 macrophage markers (TGM2: 81.5 fold, P = 0.0015; IL-10: 7.6 fold, P = 0.0450) in CD44 KO mice. Delivery of MCP-1 to CD44 KO mice rescued the phenotype with thicker AVF walls (27.2 μm vs. 14.7 μm; P = 0.0306), increased collagen density (2.4 fold; P = 0.0432), and increased number of M2 macrophages (2.1 fold; P = 0.0335).

Conclusions

CD44 promotes accumulation of M2 macrophages, ECM deposition and wall thickening during AVF maturation. These data show the association of M2 macrophages with wall thickening during AVF maturation and suggest that enhancing CD44 activity may be a strategy to increase AVF maturation.

Graphical abstract

Introduction

The United States Renal Data System reported that the number of patients with end-stage renal disease in the United States continues to increase and exceeds 600,000 in 2013.¹ Patients with end-stage renal disease are typically dependent on hemodialysis as renal replacement therapy since the scarcity of organs available limits renal transplantation to less than 20,000 per year.¹ A durable vascular access is vitally important for patients undergoing hemodialysis. The autologous arteriovenous fistula (AVF) is currently the preferred choice for vascular access because of lower rates of infection, thrombosis, and access-related expenditures compared with synthetic arteriovenous grafts. However, patency rates at one year are typically less than 60% and it is not uncommon for patients to undergo repeated vascular interventions and surgical procedures.² In fact, these numbers may be overly optimistic, as they do not include fistulae that fail to mature, e.g. dilate and thicken adequately to support the increased flows necessary to support successful hemodialysis; a significant number of fistulae (28 to 53%) fail to mature adequately.³ The large number of AVF that fail to mature reflects our poor understanding of the biology of maturation, e.g. understanding the mechanisms by which the vein adapts to the unique arterial-like environment of the fistula.

We previously described a mouse AVF model that recapitulates human AVF maturation with a distinct maturation phase.⁴ We used this model to show temporal expression patterns of the extracellular matrix (ECM) proteins during AVF maturation.⁵ Beyond structural support, the ECM plays regulatory roles in numerous circumstances, including effects on cellular proliferation and differentiation, as well as creation and maintenance of three-dimensional tissue structures.⁶ ECM remodeling plays a distinct role in intimal thickening and vascular remodeling that occurs during AVF maturation.^{3, 5, 7}

CD44 is a widely expressed cellular adhesion molecule that serves as a major receptor for ECM components such as hyaluronic acid (HA); HA has been linked to vascular remodeling after injury, as it accumulates during neointimal hyperplasia.⁸ CD44 promotes adhesion of leukocytes to endothelial cells,⁹ induces macrophage chemokine secretion, and regulates vascular smooth muscle cell proliferation and migration.¹⁰ CD44 may promote atherosclerosis by mediating inflammatory cell recruitment and activation of leukocytes and vascular smooth muscle cells.^{9, 11, 12} We have previously reported that CD44 regulates vascular integrity and inflammation.^{13, 14} Since these functions are likely to be active during venous remodeling such as occurs during AVF maturation in the fistula environment, we hypothesized that CD44 plays a role during venous remodeling, increasing ECM deposition and promoting inflammatory cell recruitment during AVF maturation.

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.

Results

CD44 expression increases during AVF maturation

Since AVF maturation in the arterial environment is associated with dilation and thickening of the vein wall, we performed a microarray analysis to determine whether there were global increases in expression of genes coding for the extracellular matrix (ECM).⁵ An AVF was created between the aorta and the inferior vena cava (IVC) in mice, and the venous limb of the AVF was removed on postoperative day 7 and patterns of gene expression were compared to the IVC of sham-operated mice. We previously reported 1196 genes with significantly changed expression in the AVF compared to vein; of these 1196 genes, 14 genes encoding for matrix proteins showed increased expression, including CD44 which showed a 2.1-fold increase.⁵ Since CD44 is a receptor for hyaluronic acid, and regulates inflammation, we assessed CD44 expression during AVF maturation. Quantitative real-time PCR showed significantly increased expression of CD44 mRNA in the venous limb of the AVF compared with sham-operated mice, with maximal expression of CD44 on days 1 and 7 (Figure 1A). Protein expression of CD44 was similarly elevated in the venous limb of the AVF, with elevation of the CD44s and CD44v fragments (Figures 1B, 1C). The time course of CD44 protein expression was assessed using immunohistochemistry; CD44 protein expression did not significantly increase in the AVF wall on day 1 or 7, but was increased at day 21 (Figure 1D). Immunofluorescence confirmed significantly increased CD44 immunoreactivity in the venous, but not the arterial limb, of the AVF compared with sham-operated mice at day 21 (Figures 1E-G). These results show that CD44 expression is elevated in the venous limb of the AVF at the same time as the limb adapts to the arterial environment.

Figure 1.

Increased CD44 expression in the venous limb of AVF. (A) Bar graph shows CD44 mRNA transcripts in the venous AVF limb relative to sham veins, normalized to day 0; all samples are also normalized to GAPDH. n=8. P=0.0001 (ANOVA); ***, P=0.0008, **, P=0.0057 (post hoc). (B) Representative Western blot analysis of CD44v, CD44s and GAPDH expression in venous AVF limb, day 21. Left lane, sham; right lane, AVF. (C) Bar graph shows relative densitometry of CD44s (n=5). **, P=0.0086 (unpaired t-test). (D) Bar graph shows time course expression of immunoreactive CD44 in the venous AVF limb relative to sham veins. n=2-3. (E) Representative microphotographs of immunofluorescence images of CD44 in veins; left panels, sham; right panels, AVF; day 21. Scale bar, 100μm. (F) Bar graphs show relative quantification of immunofluorescence intensity in venous AVF limb (n=3). *, P=0.0312 (unpaired t-test). (G) Bar graphs show relative quantification of immunofluorescence intensity in arterial AVF limb (n=3). P=0.2779 (unpaired t-test).

Lack of wall thickening during AVF maturation in CD44 knockout mice

To determine if CD44 plays a mechanistic role in venous remodeling during AVF maturation, AVF were made in both control and CD44 KO mice. There were no significant differences at baseline in venous diameter, wall thickness, or blood flow, as well as arterial diameter, wall thickness, or blood flow, between control and CD44 KO mice (Table 2). The patency rate of AVF performed in CD44 KO mice was 93.4% (15/16) at day 1, which was similar to the 100% (12/12) patency rate in WT mice (p>0.9999, Fisher's exact test); the patency rates at day 21 were also similar (50% (8/16) vs. 50% (6/12); p>0.9999). The diameter of the venous limb of the AVF was not different in control or CD44 KO mice (Figure 2A), although the diameter of the arterial limb increased less in CD44 KO mice compared to control mice (Figure 2B). However, by postoperative day 21, the wall thickness of the venous limb of the AVF was not increased in CD44 KO mice compared to the increase in AVF wall thickness in control mice (Figures 2C, 2D); similarly, there was less reduction in the wall thickness of the arterial limb of the AVF in CD44 KO mice compared to control mice (Figures 2C, 2E). Although there were no differences in CD31 immunoreactivity in both the venous and the arterial endothelium of the AVF between control mice and CD44 KO mice, the venous limb of the AVF in CD44 KO mice was characterized by no change in α-actin density in CD44 KO mice compared to the increased α-actin density in control mice (Figures 2C, 2L), and there was no difference in the α-actin density of the arterial limb in CD44 KO and control mice (Figures 2C, 2M).

Table 2.

Preoperative data in C57BL/6 (n=24) and CD44 KO (n=23) mice.

	C57BL/6 mice	CD44 KO mice	P value
Body weight (g)	26.17 ± 0.47	25.92 ± 0.64	0.7609
Aorta diameter (mm)	0.65 ± 0.08	0.63 ± 0.12	0.5017
Aorta velocity (mm/s)	268 ± 108	323 ± 140	0.1769
Aorta shear stress (dyne/cm2)	115 ± 44	145 ± 52	0.0645
Aorta wall thickness (μm)*	31.97 ± 0.50	27.56 ± 6.49	0.5669
IVC diameter (mm)	0.82 ± 0.19	0.81 ± 0.19	0.8671
IVC velocity (mm/s)	77.25 ± 26.27	86.59 ± 25.59	0.2231
IVC shear stress (dyne/cm2)	28.18 ± 12.50	33.12 ± 17.13	0.2669
IVC wall thickness (μm)*	11.38 ± 2.37	14.12 ± 6.70	0.7303

^*histological measurement

Figure 2.

Wall thickening during AVF maturation depends on CD44. (A) Line graph shows venous AVF diameter in control or CD44 KO mice, normalized to day 0 (n=13). P=0.1715 (ANOVA). (B) Line graph shows aortic diameter in control or CD44 KO mice, normalized to day 0 (n=13). ***, P=0.0008 (ANOVA). (C) Representative photomicrographs of AVF in control (left column) or CD44 KO (right column) mice, day 21; rows are hematoxylin and eosin (H&E) stain, trichrome stain, elastin stain, immunohistochemistry for hyaluronic acid, and immunohistochemistry for α-actin. Scale bar, 100 μm. Yellow star and arrowheads shows thickening of vein wall. (D) Bar graph shows wall thickness of the venous AVF limb, days 0-21. n=3. P=0.0340 (ANOVA); *, P=0.0114 (post hoc). (E) Bar graph shows wall thickness of the arterial AVF limb, days 0-21. n=3. P=0.0332 (ANOVA); **, P=0.0043 (post hoc). (F) Bar graph shows relative quantification of collagen density in the venous AVF limb, normalized to day 0 control mice. n=3. P=0.0107 (ANOVA); *, P=0.0186 (post hoc). (G) Bar graph shows relative quantification of collagen density in the arterial AVF limb, normalized to day 0 control mice. n=3. P=0.0875 (ANOVA). (H) Bar graph shows relative quantification of elastin density in the venous AVF limb, normalized to day 0 control mice. n=3. P=0.9315 (ANOVA). (I) Bar graph shows relative quantification of elastin density in the arterial AVF limb, normalized to day 0 control mice. n=3. P=0.7229 (ANOVA). (J) Bar graph shows relative quantification of hyaluronic acid density in the venous AVF limb, normalized to day 0 control mice. n=3. P=0.0036 (ANOVA); ***, P=0.0004 (post hoc). (K) Bar graph shows relative quantification of hyaluronic acid density in the arterial AVF limb, normalized to day 0 control mice. n=3. P=0.7992 (ANOVA). (L) Bar graph shows relative quantification of α-actin density in the venous AVF limb, normalized to day 0 control mice. n=3. P=0.0307 (ANOVA); *, P=0.0379 (post hoc). (M) Bar graph shows relative quantification of α-actin density in the arterial AVF limb, normalized to day 0 control mice. n=3. P=0.4845 (ANOVA). (N) Bar graph shows relative quantification of Alcian blue density in the venous AVF limb, normalized to day 0 control mice. n=3. P=0.0003 (ANOVA); ***, P=0.0006 (post hoc). (O) Bar graph shows relative number of HAS2 mRNA transcripts in control or CD44 KO mice days 0-21. n=8. P=0.0002 (ANOVA); *, P=0.0132, **, P=0.0031 (post hoc).

Since the wall thickness of the AVF venous limb did not change over time in CD44 KO mice (Figure 2C, 2D), we examined components of the ECM to determine if deposition of any of these were similarly unchanged. There was no increase in collagen deposition in the venous limb of CD44 KO mice compared to increased collagen deposition in control mice (Figure 2C, 2F, 2G); specific immunostaining for collagen-I and collagen-III in control and CD44 KO fistulae showed that both collagen-I and collagen-III deposition were less prominent at day 21 in CD44 KO fistulae compared to control fistulae (Supplemental Figure IA). However, there was no difference in elastin density in either the venous or arterial limbs between control and CD44 KO mice (Figure 2C, 2H, 2I). The expression of hyaluronic acid, the matrix component that binds CD44 as a major ligand, showed no change over time in the venous limb of the AVF in CD44 KO mice whereas it was increased in control mice (Figures 2C, 2J). This difference was not observed in the arterial limb of the AVF (Figures 2C, 2K). Similarly, the density of Alcian blue staining showed less increase in the venous limb of the AVF in CD44 KO mice compared to control mice (Figure 2N), confirming less deposition of acidic polysaccharides such as hyaluronic acid. The mRNA expression of hyaluronan synthase (HAS) 2, the isoform of hyaluronic acid synthase that is the rate limiting step for hyaluronic acid synthesis, was elevated in the venous limb of the AVF when compared to sham mice (1.366 fold increase, P = 0.0490; microarray), and showed no change in the venous limb of AVF of CD44 KO mice whereas that of control mice increased over time (Figure 2O). These results show that the venous limb of AVF in CD44 KO mice does not increase wall thickness and does not synthesize ECM components as do AVF in control mice, consistent with a mechanistic role for CD44 in AVF maturation.

Loss of inflammation during AVF maturation in CD44 knockout mice

Since CD44 may play a mechanistic role in wall thickening of the venous limb during AVF maturation, and CD44 can stimulate inflammation, we determined whether any inflammatory markers were not induced in the venous limb during AVF maturation in CD44 KO mice.¹² The mRNA expression of vascular cell adhesion molecule-1 (VCAM-1) and intercellular adhesion molecule-1 (ICAM-1), but not E-selectin, were induced on day 21 in the venous AVF limb of control mice but not CD44 KO mice (Figures 3A-C). There was also significantly decreased immunoreactivity for VCAM-1 in the venous limb of CD44 KO mice in comparison to control mice (Supplemental Figure IB, IC); however, the endothelium was still present in CD44 KO mice (Supplemental Figure ID). Interestingly, there were a reduced number of CD31-positive capillaries in the adventitia of the venous limb of CD44 KO mice compared to control mice (Supplemental Figure ID, IE). These results suggest less inflammation and angiogenesis in the venous limb of the AVF of CD44 KO mice compared to the increases in control mice.

Figure 3.

Inflammation in the venous limb of AVF depends on CD44. (A) Bar graph shows relative number of VCAM-1 mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0092 (ANOVA); *, P=0.0160 (post hoc). (B) Bar graph shows relative number of ICAM-1 mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0002 (ANOVA); **, P=0.0035 (post hoc). (C) Bar graph shows relative number of E-Selectin mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0288 (ANOVA); P=0.1063 (day 21, post hoc). (D) Bar graph shows relative number of CD45 mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.2268 (ANOVA). (E) Bar graph shows relative number of CD3 mRNA transcripts in control or CD44KO mice, days 0-21. n=8. P=0.8863 (ANOVA). (F) Bar graph shows relative number of CD68 mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0064 (ANOVA); **, P=0.0080 (post hoc). (G) Bar graph shows relative number of MCP-1 mRNA transcripts in control or CD44 KO mice, days 0-21. n=5. P=0.0006 (ANOVA); *, P=0.0172, **, P=0.0069 (post hoc). (H) Representative photomicrographs showing immunofluorescence for CD68 (green) and αSMA (red) in the venous AVF limb of control (left column) or CD44 KO (right column) mice, day 21. Orange arrowheads show CD68 positive cells. Bar graph shows number of CD68 positive cells per high-power field in control or CD44 KO mice, day 21. n=3. P=0.0448 (unpaired t-test). (I) Representative photomicrographs showing immunohistochemistry staining for CD68 in the venous AVF limb of control (left column) or CD44 KO (right column) mice, day 21. Scale bar, 25 μm. Yellow arrowheads show positive cells. Bar graph shows number of CD68 positive cells per high-power field in control or CD44 KO mice, days 0-21. n=3. P=0.0069 (ANOVA); **, P=0.0066 (post hoc). (J) Representative photomicrographs showing immunofluorescence for MCP-1 (green) and αSMA (red) in the venous AVF limb of control (left column) or CD44 KO (right column) mice, day 21. Orange arrowheads show MCP-1 positive cells. Bar graph shows number of MCP-1 positive cells per high-power field in control or CD44 KO mice, day 21. n=3. P=0.0230 (unpaired t-test). (K) Representative photomicrographs showing immunohistochemistry staining for MCP-1 in the venous AVF limb of control (left column) or CD44 KO (right column) mice, day 21. Scale bar, 25 μm. Yellow arrowheads show positive cells. Bar graph shows number of MCP-1 positive cells per high-power field in control or CD44 KO mice, days 0-21. n=3. P=0.0002 (ANOVA); *, P=0.0151. ***, P=0.0001 (post hoc).

There was increased mRNA expression of the macrophage marker CD68, but not the leukocyte markers CD45 or CD3, in the venous AVF limb of control mice but not CD44 KO mice (Figures 3D-F). Similarly, there were fewer adventitial cells that were immunoreactive for CD68 in the venous AVF limb of CD44 KO mice compared to control mice (Figure 3H, 3I). There was a similar amount of mRNA expression of the M1 markers TNF-α and iNOS in the venous AVF limb of CD44 KO and control mice (Figures 4A, 4B); however, there was no increase in mRNA expression of the M2 markers transglutaminase (TGM)-2 and IL-10 in the venous AVF limb of CD44 KO compared to the increased expression in control mice (Figures 4C, 4D). Similarly, there were fewer cells that were immunoreactive for TGM-2 and IL-10 in the venous AVF limb of CD44 KO mice compared to control mice (Figure 4E-G). There were also fewer adventitial cells immunoreactive for monocyte chemoattractant protein-1 (MCP-1), and less MCP-1 mRNA expression, in the venous AVF limb of CD44 KO mice compared to control mice (Figure 3G, 3J, and 3K). These results show that the venous limb of the AVF in CD44 KO mice does not show increased inflammatory markers, angiogenesis, and M2 macrophages as they are increased in control mice, consistent with CD44 stimulating or permissive for inflammation during AVF maturation.

Figure 4.

M2 macrophage infiltration into the venous limb of AVF depends on CD44. (A) Bar graph shows relative number of TNF-α mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0403 (ANOVA); P=0.0868 (day 21, post hoc). (B) Bar graph shows relative number of iNOS mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0160 (ANOVA); P=0.1275 (day 21, post hoc). (C) Bar graph shows relative number of TGM-2 mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P<0.0001 (ANOVA); *, P=0.0139, ***, P=0.0002 (post hoc). (D) Bar graph shows relative number of IL-10 mRNA transcripts in control or CD44 KO mice, days 0-21. n=8. P=0.0208 (ANOVA); *, P=0.0126 (post hoc). (E) Representative photomicrographs showing immunofluorescence detection of TGM-2 (upper row) or IL-10 (lower row) of the venous AVF limb in control (left column) or CD44 KO (right column) mice, day 21. L, lumen. Scale bar, 25μm. Yellow arrows show positive cells. (F) Bar graph shows number of TGM-2 positive cells per high-power field in control or CD44 KO mice, days 0-21. n=3. P=0.0001 (ANOVA); **, P=0.0018 (day 7), **, P=0.0015 (day 21; post hoc). (G) Bar graph shows number of IL-10 positive cells per high-power field in control or CD44 KO mice, days 0-21. n=3. P=0.0474 (ANOVA); *, P=0.0450 (post hoc).

MCP-1 increases venous AVF wall thickness and M2 macrophages

Since loss of CD44 is associated with no increase in AVF venous wall thickness (Figure 2) and no increased MCP-1 expression (Figure 3) as well as diminished numbers of M2 macrophages (Figure 4), we determined whether delivery of MCP-1 to the AVF venous wall increased the number of M2 macrophages and/or thickness of the AVF venous wall. Pluronic gel-embedded microparticles containing MCP-1 released MCP-1 continually in vitro, with most MCP-1 released by 30 days (Figure 5A). Microparticles, either with or without MCP-1, in pluronic gel were delivered to the adventitia of the venous AVF wall of WT mice at the time of surgery; the AVF wall contained more immunoreactive MCP-1 on day 21 when treated with MCP-1 eluting microparticles compared to control microparticles (Supplemental Figure IIA). After 21 days the diameter of the venous and arterial AVF limbs in WT and CD44 KO mice were not different whether treated by control or MCP-1 eluting microparticles (Figure 5B, 5C; Supplemental Figures IIB, IIC). However, the venous AVF limb was significantly thicker when treated with MCP-1 eluting microparticles compared to control microparticles, in both WT (Figure 5D) and CD44 KO mice (Figure 5E), although there was no significant increase in the wall thickness of the arterial AVF limb (Supplemental Figures IID, IIE). AVF wall thickness was similar in WT mice treated with control microparticles and CD44 KO mice treated with MCP-1 eluting microparticles (P=0.612). There was also significantly increased collagen and elastin immunoreactivity (Figure 5F), as well as increased density of α-actin (Supplemental Figure IIA), in the venous AVF limb of WT and CD44 KO mice treated with MCP-1 eluting microparticles compared to control microparticles. Similarly there were more CD68, TGM-2, and IL-10 positive cells, but not iNOS positive cells, in the venous AVF limb of WT and CD44 KO mice treated with MCP-1 eluting microparticles compared to control microparticles (Figure 5F). MCP-1 delivery also increased colocalization of TGM-2 in CD68-positive cells, e.g. there were increased M2 macrophages in the AVF wall (Figure 5G). At day 21, the AVF in WT and CD44 KO mice showed similar numbers of PCNA/α-actin- and Ki67/α-actin-dual positive cells in AVF treated with control or MCP-1 eluting microparticles (Supplemental Figure III), consistent with lack of effect of MCP-1 delivery on smooth muscle cell proliferation. In toto, these results are consistent with thicker AVF walls, increased collagen density, and increased number of M2 macrophages in AVF treated with MCP-1, e.g. delivery of MCP-1 rescued the CD44 KO phenotype.

Figure 5.

Adventitial delivery of MCP-1 increases AVF M2 macrophages and wall thickness. (A) Line graph shows in vitro MCP-1 protein release from alginate microparticles (1 mg) in 22% pluronic F-127 gel (n = 3). (B) Line graph shows infra-renal venous AVF diameter of WT KO mice treated with control (black line) or MCP-1 (red line) microparticles, normalized to day 0. n=3. P=0.6551 (ANOVA). (C) Line graph shows infra-renal venous AVF diameter of CD44 KO mice treated with control (black line) or MCP-1 (red line) microparticles, normalized to day 0. n=3. P=0.0698 (ANOVA). (D) Bar graph shows wall thickness (day 21) of venous AVF in WT mice treated with control or MCP-1 microparticles, day 21. n=3. *, P=0.0383 (unpaired t test). (E) Bar graph shows wall thickness (day 21) of venous AVF in CD44 KO mice treated with control or MCP-1 microparticles, day 21. n=3. *, P=0.0306 (unpaired t test). (F) Representative photomicrographs of the venous AVF wall (day 21) in WT (left columns) or CD44 KO (right columns) AVF treated with control or MCP-1 microparticles; rows are trichrome stain, elastin stain, and immunohistochemistry for CD68, iNOS, TGM-2, or IL-10. Scale bar, 100μm. Black arrowheads show thickening of vein wall and arrows (black or yellow) show positive cells. Bar graphs show quantification of the corresponding row (left graph, in WT mice; right graph, in CD44 KO mice). *, P<0.05. P-values (unpaired t test; WT, CD44 KO respectively): collagen, 0.0184, 0.0432; elastin, 0.0103, 0.0482; CD68, 0.0241, 0.0272; iNOS, 0.3605, 0.7888; TGM-2, 0.0344, 0.0359; IL-10, 0.0187, 0.0335. (G) Representative photomicrographs showing immunofluorescence for CD68 (green) and TGM-2 (red) in WT (left columns) or CD44 KO (right columns) venous AVF limb treated with control or MCP-1 microparticles, day 21. Bar graph shows ratio of single or dual positive cells per high-power field after MCP-1 delivery to those of control, normalized by ratio of single positive cells. Day 21. n=3-5. *, P=0.0058; **, P=0.0001 (unpaired t-test).

Discussion

We show increased CD44 expression in the maturing venous limb of the AVF (Figure 1) and that venous wall thickening during AVF maturation depends on CD44 function (Figure 2), suggesting a mechanistic role for CD44 in AVF maturation. CD44 also stimulates inflammation and macrophage accumulation in the adventitia of the maturing venous AVF limb (Figure 3), with specific enhancement of M2 macrophage accumulation (Figure 4). Delivery of MCP-1 to CD44 KO mice rescues M2 macrophage accumulation and wall thickness (Figure 5). These results suggest that CD44 regulates venous wall thickening during AVF maturation by promoting M2 macrophage accumulation in the adapting venous wall.

The primary finding of our study is that CD44 promotes accumulation of M2 macrophages, ECM deposition, and inflammation during AVF maturation, contributing to AVF venous wall thickening. Multiple studies have previously shown that inflammation regulates vascular remodeling.^15-17 CD44 is also known to have a role in inflammation as well as vascular biology;^{12, 18-20} however, our linkage of CD44 with AVF maturation has not been previously demonstrated. This novel association is consistent with the known roles of CD44 in regulation of cell–cell interactions, adhesion, migration, lymphocyte activation, hematopoiesis, and tumor metastasis.^{8, 12, 18-20} The complex temporal expression pattern of CD44 during AVF maturation is consistent with the patterns of other regulators of AVF maturation,⁵ and likely reflects the diversity of CD44's multiple cellular functions during venous remodeling.¹³ In addition, the association of CD44 function with inflammation during AVF maturation occurs in the absence of significant differences in the mean blood pressure between CD44 KO and C57BL/6J mice.²¹ Interestingly, genetic deletion of CD44 shows different consequences between the venous and arterial sides of the AVF, with different degrees of outward remodeling and changes in matrix components (Figure 2), suggesting differential roles for CD44 in veins and arteries in response to the distinct hemodynamics of the AVF. It is possible that the slightly different genetic backgrounds of C57BL/6 and CD44 KO mice may affect the immune or inflammatory responses present during AVF maturation, despite >99.9% genetic homology after 11 backcross generations. However, C57BL/6 mice show similar arterial remodeling compared to other commonly used mouse strains.²² In addition, C57BL/6 and CD44 KO mice have equivalent preoperative vascular measurements (Table 2), as well as similar postoperative AVF patency rates, consistent with previous reports that showed C57BL/6 mice are appropriate controls for CD44 KO mice, both in vitro^{13, 14} and in vivo.^{14, 23-26}

The finding of M2 macrophages in the maturing AVF suggests a role for CD44 as a specific inflammatory mediator during AVF maturation. Although there are no previous reports describing M2 macrophages during AVF maturation, we have previously shown that M2 macrophage function may be an important mechanism regulating vein graft adaptation.²⁷ The presence of M2 macrophages is consistent with the early phase of AVF maturation being an adaptive remodeling response of the vein to the arterial environment, rather than a pathological response that necessarily leads to AVF failure. However, since there is considerable heterogeneity of tissue-resident macrophages,²⁸ it is not surprising that specific subpopulations of macrophages that are responsible for venous remodeling are not well described.

MCP-1 has been found in human AVF,²⁹ although the significance of MCP-1 during venous remodeling is not clearly understood. Using a jugular-to-carotid AVF model in aged mice, Juncos et al. previously reported that MCP-1 contributes to AVF failure, with MCP-1 KO mice having thinner walls and increased lumen diameter.³⁰ With a similar jugular-to-carotid AVF model in mice with chronic renal failure, Misra reported lex-1 regulates MCP-1, with lex-1 KO mice having reduced MCP-1 and neointimal hyperplasia.³¹ Our data shows no increase in MCP-1 RNA and protein expression, as well as fewer MCP-1-positive cells, in the AVF of CD44 KO mice (Figure 3), suggesting that one function of CD44 is to promote macrophage accumulation in the maturing AVF limb. However, it is also noteworthy that the increase in macrophages at day 21 is preceded by increased MCP-1 expression at day 7 (Figure 3). Together with the observation that MCP-1 delivery stimulates wall thickening, ECM synthesis, and M2 macrophage accumulation in the AVF wall even in the absence of CD44 (Figure 5), it is possible that inflammation that occurs during AVF maturation is stimulated by both CD44-dependent and CD44-independent pathways. The unexpected finding that delivery of MCP-1 similarly increases wall thickness, ECM synthesis, and M2 macrophage accumulation in control mice suggests that MCP-1 can drive these pathways in the presence of CD44. In addition to the scientific importance of this alternative pathway, the significance of this finding lies in its potential clinical application to (CD44-wild type) human patients needing improved AVF maturation; thus stimulation of MCP-1 activity may be a therapeutic approach translatable to human patients. Since we have previously shown that our mouse aorto-caval AVF model recapitulates human AVF maturation,⁴ we believe that our data suggests a role for MCP-1 during AVF maturation. However, the mechanism by which MCP-1 promotes specific accumulation of M2 macrophages during venous remodeling remains to be determined.

In conclusion, CD44 promotes accumulation of M2 macrophages, ECM deposition and inflammation, enhancing AVF maturation. These data suggest that promoting CD44 activity may be a strategy to enhance AVF maturation, and also show the importance of inflammation to enable wall thickening, e.g. adaptive remodeling, during AVF maturation. The use of anti-inflammatory modulators such as pharmaceutical compounds or mesenchymal stem cells may possibly be used as adjuvant therapy to decrease AVF failure rates; this work also suggests that microparticles are a potential mechanism to deliver these small molecule anti-inflammatory modulators.

Table 1. Primer sequences.

Gene	Forward	Reverse
GAPDH	AATGTGTCCGTCGTGGATCTGA	AGTGTAGCCCAAGATGCCCTTC
CD68	TGTCTGATCTTGCTAGGACCG	GAGAGTAACGGCCTTTTTGTGA
CD45	GAGCAGACCCGAGATCCAC	GCAGCACTACCAGAAAAGGCA
CD3	GGTCCAAGTCTCCGGCTCTA	AGCACATGACTCAATCCTACAGT
E-Selectin	GACCTGGAACCCTACATGGAT	TTATGCAAACACTTCTCGGCT
VCAM-1	ATGTCAACGTTGCCCCCAA	CAGGACTGCCCTCCTCTAGT
ICAM-1	GTGATGCTCAGGTATCCATCCA	CACAGTTCTCAAAGCACAGCG
CD44	TATCCTCGTCACGTCCAACACC	TGTAGCTTTCTGGGGTGCTCTT
TGM2	ATCTTGGTCAGCCTCAGTGC	CATATTCCCGTCGCTCCTCC
IL-10	GGTGAGAAGCTGAAGACCCTC	GGCCTTGTAGACACCTTGGTC
TNF-α	ACAGAAAGCATGATCCGCGA	GCTCCTCCACTTGGTGGTTT
HAS2	GGGCGAAGCGTGGATTATGT	ACCGCTTATGCACTGGACAC
MCP-1	TTAAAAACCTGGATCGGAACCAA	GCATTAGCTTCAGATTTACGGGT
iNOS	GTTCTCAGCCCAACAATACAAGA	GTGGACGGGTCGATGTCAC

Highlights.

• CD44 expression increases in the maturing venous limb of the arteriovenous fistula (AVF).

• Venous wall thickening during AVF maturation depends on CD44 function.

• CD44 promotes accumulation of M2 macrophages, ECM deposition, and inflammation during AVF maturation, contributing to AVF venous wall thickening.

• Delivery of MCP-1 to CD44 KO mice rescues M2 macrophage accumulation and wall thickness.

• Promoting CD44 activity may be a strategy to enhance AVF maturation.

Nonstandard Abbreviations and Acronyms

AVF

arteriovenous fistula

ECM

extracellular matrix

HA

hyaluronic acid

IVC

inferior vena cava

MCP-1

monocyte chemoattractant protein-1

iNOS

inducible nitric oxide synthase

HAS2

hyaluronan synthase 2

VCAM-1

vascular cell adhesion molecule-1

ICAM-1

intracellular adhesion molecule-1

TNF

tumor necrosis factor

TGM

transglutaminase

IL

interleukin