Mcp1 Promotes Macrophage-Dependent Cyst Expansion in Autosomal Dominant Polycystic Kidney Disease

Abstract

Background

In patients with autosomal dominant polycystic kidney disease (ADPKD), most of whom have a mutation in PKD1 or PKD2, abnormally large numbers of macrophages accumulate around kidney cysts and promote their growth. Research by us and others has suggested that monocyte chemoattractant protein-1 (Mcp1) may be a signal for macrophage-mediated cyst growth.

Methods

To define the role of Mcp1 and macrophages in promoting cyst growth, we used mice with inducible knockout of Pkd1 alone (single knockout) or knockout of both Pkd1 and Mcp1 (double knockout) in the murine renal tubule. Levels of Mcp1 RNA expression were measured in single-knockout mice and controls.

Results

In single-knockout mice, upregulation of Mcp1 precedes macrophage infiltration. Macrophages accumulating around nascent cysts (0–2 weeks after induction) are initially proinflammatory and induce tubular cell injury with morphologic flattening, oxidative DNA damage, and proliferation-independent cystic dilation. At 2–6 weeks after induction, macrophages switch to an alternative activation phenotype and promote further cyst growth because of an additional three-fold increase in tubular cell proliferative rates. In double-knockout mice, there is a marked reduction in Mcp1 expression and macrophage numbers, resulting in less initial tubular cell injury, slower cyst growth, and improved renal function. Treatment of single-knockout mice with an inhibitor to the Mcp1 receptor Ccr2 partially reproduced the morphologic and functional improvement seen with Mcp1 knockout.

Conclusions

Mcp1 is upregulated after knockout of Pkd1 and promotes macrophage accumulation and cyst growth via both proliferation-independent and proliferation-dependent mechanisms in this orthologous mouse model of ADPKD.

Mutations in PKD1 or PKD2 cause the vast majority of cases of autosomal dominant polycystic kidney disease (ADPKD).¹ Patients with this disease can develop hundreds to thousands of renal cysts, which progressively grow and ultimately lead to loss of glomerular filtration over a 30–50 year period. In tubular cells that lose functional expression of the gene product (Pc1 or Pc2, respectively), it is proposed that abnormal proliferation, mislocalization of transporters, and net fluid secretion lead to cyst growth.²

In addition to the well described cell autonomous effects of polycystin loss on tubular cell proliferation and cyst formation,³ analysis of both murine models of ADPKD and human kidney specimens reveals that activation of the innate immune response plays an important role in the processes of cyst growth and pericystic fibrosis (reviewed in Song et al.⁴). This has been shown to result in increased numbers of interstitial macrophages in kidneys from patients with ADPKD.^5,6 Examination of murine models of polycystic kidney disease (PKD) demonstrate that macrophages progressively accumulate immediately adjacent to the Pc1 or Pc2 null tubular cells, beginning very early in the process of cyst formation, and that depletion of these macrophages slows cyst growth.^7,8 Our investigation of the potential signal for increased macrophage numbers in cystic kidneys demonstrated that monocyte chemoattractant protein-1 (Mcp1, also called Ccl2) was highly upregulated at the time of cyst initiation after knockout of Pkd1 or Pkd2.⁷ Consistent with a role for Mcp1 in promotion of macrophage-mediated cyst growth, Zheng et al. reported the presence of high levels of MCP1 in the urine of patients with ADPKD,⁹ whereas Grantham et al.¹⁰ demonstrated that levels of urinary MCP-1 correlate with kidney cyst size and are suppressed after tolvaptan treatment. Chen et al.⁶ demonstrated that macrophage migration inhibitory factor is upregulated after loss of Pkd1 and can induce Mcp1. On the basis of these findings, we generated a mouse model with simultaneous knockout of Mcp1 and Pkd1 in the renal tubule to define the functional role of Mcp1 in the macrophage response to renal tubule cyst formation.

Methods

Animal Models

Pkd1^fl/fl;Pax8-rtTA;TetO-Cre (single knockout [SKO]), Pax8-rtTA;TetO-Cre, and Pkd1^fl/fl mice were obtained from S.S. (Yale School of Medicine). Mcp1^fl/fl mice obtained from Dr. Teizo Yoshimura (National Institutes of Health) were crossed with Pkd1^fl/fl mice to generate the double homozygotes, and then crossed with Pax8-rtTA;TetO-Cre mice to generate Mcp1^fl/fl;Pkd1^fl/fl;Pax8-rtTA;TetO-Cre mice (double knockout [DKO]). To limit the total number of mice needed for these experiments, mice were bred to homozygosity for the Pax8-rtTA and TetO-Cre alleles. Genotypes of the mice were confirmed by standard PCR (Table 1). Pkd1^fl/fl;Pax8-rtTA;TetO-Cre and Mcp1^fl/fl;Pkd1^fl/fl;Pax8-rtTA;TetO-Cre mice were treated with doxycycline (2 mg/ml in drinking water) for 2 weeks (starting at 4 weeks of age) to induce SKO and DKO, respectively. Because of a modest but statistically significant difference in the rate of cyst growth between male and female mice (Supplemental Figure 1), corresponding to published results by other groups,¹¹ we used only male mice for all of the cyst growth comparison experiments. The mice were euthanized and kidneys were collected at the indicated times. Uninduced mice (UI) were used as controls.

Table 1.

Primer sequences used for genotyping and quantitative real-time PCR

Gene	Forward	Reverse
Genotyping
Cre	AGGTTCGTTCACTCATGGA	TCGACCAGTTTAGTTACCC
Mcp1 (Ccl2)	GATACCTGAGTGGAAGACTC	TACTACTCCTGGTAGCTCTC
Pax-8	CCATGTCTAGACTGGACAAGA	CTCCAGGCCACATATGATTAG
Pkd1	CCGCTGTGTCTCAGTGCCTG	CAAGAGGGCTTTTCTTGCTG
Quantitative PCR
Arg1	CCAGAAGAATGGAAGAGTCAGTGT	GCAGATATGCAGGGAGTCACC
Ccr2	CCTGCAAAGACCAGAAGAGG	TCCAAGCTCCAATTTGCTTC
Csf1 (M-CSF)	CCAAGCCTGATTGCAACTGCC	GTGAGTCCTCAGTGAGTCTGTC
Csf2 (GM-CSF)	CCAGTCCAAAAATGAGGAAGC	CAGCGTTTTCAGAGGGCTAT
F4/80	GGAAGGAAATGGAGAGAAAG	GAAGATCTACCCTGGTGAAT
Hprt	AGTACAGCCCCAAAATGGTTAAG	CTTAGGCTTTGTATTTGGCTTTTC
IL12	CAGAAGCTAACCATCTCCTGGTTTG	TCCGGAGTAATTTGGTGCTTCACAC
Mcp1 (Ccl2)	AGGTCCCTGTCATGCTTCTG	TCTGGACCCATTCCTTCTTG
Mrc1	CAGCGGTTGGCAGTGGA	CAGCTGATGGACTTCCTGGTAAC
Nos2	CACCAAGCTGAACTTGAGCG	CGTGGCTTTGGGCTCCTC
Sdf1	GCTCTGCATCAGTGACGGTA	TAATTTCGGGTCAATGCACA
Tnfα	GAGTGACAAGCCTGTAGCC	CTCCTGGTATGAGATAGCAAA

In Vivo Inhibition of CCR2

Pkd1^fl/fl;Pax8-rtTA;TetO-Cre mice were induced with doxycycline for 2 weeks (4–6 weeks of age) and were subsequently injected once per day from 6 to 12 weeks age with the CCR2-antagonist INCB3344 (30 mg/kg per day, administered intraperitoneally; MedChem Express) dissolved in DMSO and diluted 1:100 in PBS to a final concentration of 0.6 mg/100 µl. Induced littermates received 1% DMSO/PBS vehicle as controls. INCB3344 has an IC₅₀ for CCR2 of 10 nM, whereas the IC₅₀ for a panel of other G-protein coupled receptors, including the closely related CCR1 and CCR5, was at least 100-fold higher (>1 μm).^12,13

Quantitative Real-Time PCR

RNA from whole kidney was isolated using TRIzol (Sigma Aldrich) or from FACS sorted macrophages using RNAeasy Mini kit (Qiagen), reverse transcribed using iScript cDNA synthesis kit (Bio-Rad, Hercules, CA), and gene expression analysis performed by quantitative real-time PCR using iCycler iQ (Bio-Rad) normalized to hypoxanthine guanine phosphoribosyl transferase expression. The primer sequences used are shown in the Table 1. The results are expressed using the comparative threshold cycle (dCt) method and the mRNA-to-hypoxanthine guanine phosphoribosyl transferase expression ratios are shown as 2^−dCt.

Immunofluorescence

Kidneys were cardiac perfused with cold 1× PBS and fixed in 4% paraformaldehyde. Paraffin-embedded kidney sections were deparaffinized and stained for macrophages using rat anti-mouse F4/80 primary (1:100 dilution; Bio-Rad) and donkey anti-rat IgG-Alexa Fluor 594 (1:500 dilution; Life Technologies) secondary. Images were acquired using a NIKON Eclipse TE2000-U microscope and the F4/80⁺ macrophages quantified using ImageJ software (1.50i). The results are expressed as percentage of F4/80⁺ cells in comparison with total number of cells (Supplemental Figure 2).

Immunohistochemistry

Kidney sections fixed as above were deparaffinized, rehydrated, and subjected to antigen retrieval using either BD Retrievagen (BD Biosciences) or treatment with 2N HCl at 65°C for 10 minutes, neutralization with 0.1 M sodium borate buffer (pH 8.5) for 10 minutes, and proteinase K (20 µg/ml) for 3 minutes. The following primary antibodies were used: Ki-67 (Thermo Fisher Scientific), BrdU (Bio-Legend), and 8-OHdG (Abcam). For 8-OHdG staining, quenching of endogenous peroxidase activity was done after the primary antibody incubation. The sections were developed using Vectastain Elite ABC HRP kit (Vector Laboratories) and DAB substrate kit (Vector Laboratories). Nuclei were counterstained with hematoxylin (Thermo Fisher Scientific). Terminal deoxynucleotidyl transferase–mediated digoxigenin-deoxyuridine nick-end labeling (TUNEL) staining was performed on paraffin-embedded kidney sections using ApopTag Plus Peroxidase In Situ Apoptosis Kit, following the manufacturer’s instructions.

Stained slides were scanned in Aperio LV1 Real-time slide scanner (Leica Biosystems) and images of renal cortex were acquired using ImageScope v12.3.2.8013 (Leica Biosystems). Ki-67–, BrdU–, or 8-OHdG–positive tubular and interstitial cells were counted manually using Cell Counter plugin in ImageJ (1.50i). The results are expressed as percentage of positive-stained cells in comparison with total number of cells.

Cystic Index

The cyst index was quantified from sagittal sections of nonperfused whole kidneys, as described previously.¹⁴ Briefly, hematoxylin and eosin–stained kidney section images were acquired using Aperio LV1 Real-time slide scanner and processed using ImageScope v12.3.2.8013 software. The total kidney area and cystic and noncystic areas were measured using ImageJ (1.50i). Cystic index is expressed as the percentage cyst area versus total kidney area.

FACS

Mice were cardiac perfused with cold PBS (1×) and kidneys were harvested, pulverized, and digested in 2 ml of digestion solution (Liberase 0.5 mg/ml+DNase I 200 U/ml in 1× PBS) at 37°C for 1 hour, and filtered through a 40 μm nylon cell strainer (BD Falcon) to obtain a single-cell suspension. After blocking nonspecific Fc binding with anti-mouse CD16/32 (clone 93; eBioscience), kidney single-cell suspension was stained with anti–CD45-PE (30F11; BD Bioscience), anti-F4/80 FITC (BM8; eBioscience), and anti–CD11c-PE-Cy7 (HL3; BD Bioscience). We used unstained and isotype controls to identify autofluorescent kidney cell populations, CD45⁺F4/80⁺ cells were first selected, and then the CD11c⁻ population of the CD45⁺F4/80⁺ cells was gated for isolation on the Sony SY3200.

Electron Microscopy

Kidneys were fixed by perfusion with 2% glutaraldehyde/4% paraformaldehyde in PBS for transmission electron microscopy. The images were acquired at higher power (8000×) from three SKO mice. A total of 19 tubular nuclei were analyzed in electron microscopy images, with two representative nuclei shown.

Western Blotting

Kidney lysates from UI, SKO, and DKO mice at 8 weeks of age were prepared using Radioimmunoprecipitation assay lysis and extraction buffer (Thermo Fisher Scientific). The following primary antibodies were used: cleaved caspase-3 (Cell Signaling Technologies), γH2AX (phosphor-Histone H2AX Ser 139; Cell Signaling Technologies), and GAPDH (Santa Cruz Biotechnologies). Band densities were normalized to GAPDH and fold-change was calculated using ImageJ (v1.5i).

Isolation and Culture of Tubular Epithelial Cells

Single-cell suspensions from 8-week-old UI, SKO, and DKO kidneys were obtained as above. Cells were cultured in renal epithelial growth medium (Lonza) containing 2.5% FBS and penicillin/streptomycin. Cells were grown to confluence in six-well dishes (7 days) before being subjected to either DNA harvest for genotyping or the alkaline comet assay.

Alkaline Comet Assays

Comet assays were performed as per the manufacturer’s instructions (Trevigen, Inc., Gaithersburg, MD). Briefly, a small piece of kidney cortex was placed into 1 ml of ice cold 1× PBS with 20 mM EDTA. The tissue was minced into small pieces and let stand for 5 minutes on ice. Cell suspension was recovered, washed with 1× PBS, and replicates were embedded in LM Agarose (Trevigen). After alkaline unwinding for 1 hour at 4°C, electrophoresis was conducted at 21 V for 30 minutes, using the Comet Assay Electrophoresis System (Trevigen) at 4°C. Slides were stained with SYBR green and imaged using an EVOS FL microscope (Advanced Microscopy Group). Comet tail moment was analyzed using the OpenComet plugin in ImageJ as previously described.¹⁵ Data are presented as mean±SEM for three biologic replicates, with >100 cells analyzed per replicate.

Multiphoton Microscopy with Optical Clearing

Kidneys from 8-week-old UI, SKO, and DKO mice were harvested and divided midcoronally into 1 mm sections and fixed in 4% paraformaldehyde. In vivo perfusion was avoided to minimize artifactual changes in lumen and cyst diameters. Optical clearing and multiphoton imaging were performed as previously described.¹⁶ Briefly, the tissue was optically cleared with benzyl alcohol benzyl benzoate and then stained with DAPI and eosin for multiphoton microscopy. Images were acquired at a voxel dimension of 0.296×0.296×2 µm using a Ti-Sapphire laser and Scanimage software. All images were processed on the FIJI implementation of ImageJ v1.51n. Random numbers were generated to designate 100 glomeruli and subsequent tubule segments for tracking. S1 segments were tracked from the first transverse turn after exiting the glomerulus until the next transverse section was identified. This section was used to measure the tubular and epithelial area (Supplemental Figure 3). Tubular and epithelial area were determined by measuring the greatest tubule and lumen diameter in the x and y axis perpendicular to each other (Supplemental Figure 3B). Tubule and lumen areas were calculated using the formula πr², and epithelial area calculated by subtracting lumen area from tubule area. The data were sorted on the basis of the epithelial area from lowest to highest values. The ten tubule sections with the smallest epithelial area (S) and largest epithelial area (L) were identified and used for absolute quantification of tubular and epithelial area by masking (Supplemental Figure 3C), and for nuclear count (Supplemental Figure 3D). Comparison of calculated and measured areas for the ten smallest and largest area tubules showed tight correlation (Supplemental Figure 3E).

BUN and Creatinine Measurement

BUN and creatinine were measured by the Yale O’Brien Kidney Center using blood collected at the indicated times. Creatinine was measured by HPLC.

Statistical Analyses

Two-tailed t test with Welch correction assuming unequal variances and two-way ANOVA was used to compare data between groups. Significance was determined at P<0.05. Data were presented as mean±SEM.

Approval of Animal Studies

All experiments involving mice were conducted under an approved protocol overseen by the Yale University Institutional Animal Care and Use Committee guidelines and procedures.

Results

Tubular Cell Mcp1 Promotes Macrophage Accumulation and Cyst Growth in PKD

To identify the signal most likely to be responsible for increased macrophage numbers in cystic kidneys, we examined the early stages of cyst development in an orthologous, postdevelopmental model of PKD. Pkd1^fl/fl;Pax8-rtTA;TetO-Cre mice, in which Pkd1 is selectively deleted in tubular cells after doxycycline treatment (SKO), were induced with doxycycline for 2 weeks, beginning at 4 weeks of age. Quantification of F4/80⁺ cells (as shown in Supplemental Figure 2) was performed in UI kidneys (8 weeks age), at the end of induction (SKO, 6 weeks age), and 2 weeks after the end of induction (SKO, 8 weeks age). These results confirmed that macrophage numbers are significantly increased at the beginning of cyst formation (8 weeks age; Figure 1A). Comparison of macrophage homing chemokine mRNA expression at 8 weeks age in SKO and UI kidneys revealed a marked increase in the level of Mcp1 with no change in the level of Sdf1 (Figure 1B). Levels of fractalkine, macrophage inflammatory protein-2 (MIP2), and macrophage inhibitory factor ranged from approximately 1000-fold lower than Mcp1 to undetectable (data not shown). mRNA from CD45⁺F4/80⁺Cd11c⁻ macrophages demonstrated upregulation of the Mcp1 receptor Ccr2 on macrophages from the SKO kidneys, with no upregulation of Ccr1 or Ccr5, and undetectable levels of Ccr4, Cxcr3, and Cxcr6 (Figure 1C and data not shown). Whole kidney mRNA quantitative PCR analysis of Pkd1 and Mcp1 from UI kidneys (8 weeks age), at the end of induction (SKO, 6 weeks age), and 2 weeks after the end of induction (SKO, 8 weeks age) demonstrated that Mcp1 levels were not significantly elevated until the 8 week time point, coincident with the nadir of Pkd1 mRNA and the increased macrophage numbers (Figure 1D).

Figure 1.

Conditional knockout of Mcp1 in tubular epithelial cells prevents infiltration of macrophages. (A) Quantification of F4/80⁺ macrophages in kidney sections from 6 to 8 week old SKO compared with UI mice. **P<0.01 relative to UI. (B) Renal expression of the mRNA for Mcp1 and Sdf1 relative to Hprt1 in 8-week-old UI and SKO kidneys. ***P<0.001 relative to UI. (C) Expression levels of chemokine receptors Ccr1 and Ccr2 mRNA in flow-sorted CD45⁺F4/80⁺Cd11c⁻ macrophages from 12-week-old UI and SKO mice kidneys. *P<0.05 related to UI. (D) Quantitative real-time PCR of whole kidney RNA from UI (8 weeks old) and SKO (6 and 8 weeks old) mice for Mcp1 and Pkd1. *P<0.05 relative to UI Pkd1 levels; ***P<0.001 relative to UI Mcp1 levels. (E and F) Quantitative real-time PCR of Mcp1 (E) and F4/80 (F) in whole kidney RNA from UI, SKO, and DKO mice at the indicated ages (weeks). P<0.001 by two-way ANOVA for both genes comparing SKO with UI and SKO with DKO. (G) Quantitation of F4/80+ macrophages in sections from UI, SKO, and DKO kidneys at the indicated ages (weeks). P<0.001 by two-way ANOVA comparing SKO with UI and SKO with DKO. (H) Representative images for immunofluorescence staining for F4/80 in 8-week-old UI, SKO, and DKO kidneys (red, F4/80). The n listed under each bar represents the number of individual mice analyzed at that time point.

Mcp1^fl/fl;Pkd1^fl/fl;Pax8-rtTA;TetO-Cre mice, in which Mcp1 and Pkd1 are simultaneously deleted in tubular cells, were generated (DKO) and compared with SKO mice. Quantitation of Mcp1 mRNA, F4/80 mRNA, and F4/80⁺ cells from whole kidney specimens revealed that tubular Mcp1 knockout prevented the majority of the increase in whole kidney Mcp1 levels and macrophage numbers seen after Pkd1 knockout (Figure 1, E–H). Analysis of the phenotypic effect of loss of both Mcp1 and Pkd1, as opposed to Pkd1 alone, over the 14 weeks after the start of induction showed a significant slowing in kidney enlargement (Figure 2A), kidney weight-to-body weight ratio (Figure 2B), and cystic dilation of tubules (Figure 2C, quantified in Figure 2D) in the DKO mice. This reduced rate of cyst growth was accompanied by improved renal function (Figure 2, E and F) and survival (Figure 2G) in the DKO mice.

Figure 2.

Tubular epithelial deletion of Mcp1 slows cyst growth and improves renal function. (A and B) Kidney weight (A) and kidney weight-to-body weight ratio (B) of SKO and DKO mice compared with UI littermate controls at the indicated ages (weeks). P=0.009 by two-way ANOVA comparing SKO with DKO. (C) Representative images of sagittal kidney sections from 18-week-old UI, SKO, and DKO mice. Scale bar, 3 mm. (D) Cystic index in UI (n=4–6 per time point), SKO (n=5 per time point), and DKO (n=5–10 per time point) mice at the indicated ages (weeks). P=0.002 by two-way ANOVA comparing SKO with DKO. (E and F) BUN and creatinine values from UI, SKO, and DKO mice at the indicated time points. P<0.001 by two-way ANOVA for both BUN and creatinine comparing SKO with DKO. (G) Kaplan–Meier survival curve comparing SKO and DKO mice at 18 weeks of age (P=0.005). The n listed under each bar represents the number of individual mice analyzed at that time point.

Pharmacologic Inhibition of the Mcp1 Receptor Ccr2 Slows Cyst Growth

The presence of high levels of the Mcp1 receptor Ccr2 on macrophages in SKO kidneys led us to determine if inhibition of Ccr2 could also decrease macrophage accumulation and slow cyst growth. Pkd1^fl/fl;Pax8-rtTA;TetO-Cre mice were induced with doxycycline beginning at 4 weeks and then given either the Ccr2 inhibitor INCB 3344 (30 mg/kg per day, administered intraperitoneally) or DMSO vehicle beginning at the completion of induction (6 weeks age), and were euthanized at 12 weeks age for analysis. Quantification of macrophage numbers revealed that inhibition of Ccr2 led to a significant decrease in macrophage numbers in the SKO kidneys (Figure 3A), resulting in a significant decrease in kidney weight-to-body weight ratio (Figure 3B) and cyst index (Figure 3C). Renal function, as judged by BUN and creatinine, was improved in the setting of pharmacologic Ccr2 inhibition (Figure 3D), although not to the extent seen with tubular Mcp1 knockout.

Figure 3.

CCR2 inhibition reduces macrophage infiltration and cyst index in SKO mice. (A) Quantification of F4/80⁺ macrophages in kidney sections from 12-week-old SKO mice treated with either DMSO vehicle (SKO-V) or CCR2 antagonist INCB3344 (SKO-INCB). n=number of individual mice analyzed. ***P<0.001 versus vehicle treatment. (B) Kidney-to-body weight ratio of UI mice compared with SKO mice treated with vehicle (V) or INCB. **P<0.01 versus vehicle treatment. (C) Representative images of sagittal kidney sections from 12 weeks age SKO mice treated with either vehicle (V) or INCB3344 (INCB). Scale bar, 4 mm. (D) Cystic index of UI mice compared with SKO mice treated with vehicle (V) or INCB. **P<0.01 versus vehicle treatment. (E) BUN values of SKO mice treated with vehicle (V) or INCB. P=NS versus vehicle treatment. (F) Creatinine values of SKO mice treated with vehicle (V) or INCB. *P<0.05 versus vehicle treatment.

Bimodal Activation of Macrophages after Pc1 Loss

To identify the mechanism by which macrophages promote cyst growth, we compared rates of kidney cell proliferation in UI, SKO, and DKO mice at 6, 8, 12, and 18 weeks of age (2, 4, 8, and 14 weeks after start of induction). Immunostaining for the nuclear protein Ki-67 demonstrated that both SKO and DKO kidneys had significantly more Ki-67–positive tubular cells at all time points as compared with UI controls (Figure 4A arrows, quantified in Figure 4B), but with diverging effects over time. In both SKO and DKO mice, the absolute increase in Ki-67⁺ tubular cells compared with UI control at 6–8 weeks age was approximately 1.0%–2.4% (P<0.05 versus UI control; P=NS for SKO versus DKO). After the 8 week time point, tubular cell proliferative rates decreased slightly in the DKO mice (paralleling the decrease in the UI mice), whereas SKO kidneys exhibited a further absolute increase in Ki-67⁺ tubular cells that peaked at >6% by 18 weeks. To determine whether these changes in Ki-67⁺ cell numbers reflected a true increase in tubular cell proliferation, we included BrdU in the drinking water from the time of induction to 8 weeks of age (Supplemental Figure 4). These data confirm a modest increase in the cumulative number of BrdU⁺ tubular cells in both SKO and DKO kidneys as compared with UI at 8 weeks, with only DKO tubular cells reaching statistical significance. Of note, there was no increase in BrdU⁺ interstitial cells in SKO kidneys compared with UI or DKO kidneys, arguing against in situ macrophage proliferation as the source of the increased macrophage numbers in the SKO kidneys.

Figure 4.

Tubule cell proliferation after Pc1 loss involves a sustained macrophage-independent component along with late acceleration dependent on a switch to alternative macrophage activation. (A) Representative images of 8-week-old UI, SKO, and DKO kidneys showing cortical Ki-67⁺ tubular cells (arrows). Scale bar, 200 µm. (B) Quantification of Ki-67⁺ tubular cells as shown in (A). UI (n=6 mice per time point), SKO (n=6 mice per time point), and DKO (n=5 mice per time point). P=0.004 for SKO versus DKO mice by two-way ANOVA. (C–I) Quantitative real-time PCR of Nos2 (C), Tnfα (D), IL12 (E), Arg1 (F), Mrc1 (G), Csf1 (H), and Csf2 (I) using whole kidney RNA from UI, SKO, and DKO mice at the indicated ages (weeks). P<0.05 for Tnfα, Mrc1, and Csf1, and P<0.01 for Nos2, IL12, Arg1, and Csf2 using two-way ANOVA comparing SKO with DKO. The n listed under each bar represents the number of individual mice analyzed at that time point.

Analysis of the whole kidney mRNA expression of proinflammatory (M1) and alternative (M2) macrophage gene activation at 8, 12, and 18 weeks revealed a significant upregulation of the Stat1 and NFκB-induced proinflammatory genes inducible nitric oxygen synthase (Nos2), Tnfα, and Il12 at the time of early cyst initiation (8 weeks age), with no expression of the Stat5- and Stat6-induced alternative activation markers arginase 1 (Arg1) or mannose receptor (Mrc1; Figure 4, C–G). At 12 weeks there was a mixed phenotype with expression of both proinflammatory genes (Tnfα, Nos2) and alternative activation genes (Arg1), whereas 18-week-old SKO mice exhibited significant upregulation of both Arg1 and Mrc1 but proinflammatory gene expression returned to baseline. Sorting of macrophages at week 12 in UI and SKO kidneys confirmed the mixed phenotype, with significant upregulation of both Tnfα (ten-fold) and Arg1 (>1000 fold) in macrophages from SKO kidneys. Consistent with the late switch to alternative macrophage activation, SKO kidneys exhibit a significant increase in the renal-associated M2 activation chemokines Csf1 and Csf2^17,18 at 12 and 18 weeks (Figure 4, H and I). The marked increase in SKO tubular cell proliferation at 12–18 weeks correlates closely with this switch to alternative macrophage activation (Figure 4B).

Initial Macrophage Proinflammatory Activation Induces Oxidative DNA Injury

Proinflammatory macrophages are known to activate NADPH oxidase and induce the expression of reactive oxygen species that promote oxidative cell damage.¹⁹ Consistent with this, expression of kidney injury molecule-1 (Kim1) was increased in 8-week-old SKO kidneys but not DKO kidneys (Figure 5A), and cyst lining cells from SKO kidneys exhibited a significant increase in oxidized DNA as detected by anti–8-hydroxyguanosine (8-OHdG²⁰; Figure 5B arrowheads, quantified in 5C). TUNEL staining revealed increased TUNEL positivity (Figure 5D arrows, quantified in Figure 5E); however, nuclear morphology of tubular cells was normal both by immunohistochemistry (Figure 5D) and electron microscopy (Figure 5F), arguing against double-strand breaks and nuclear condensation/fragmentation. Consistent with this, levels of cleaved caspase-3 were not increased in SKO kidneys (Figure 5, G and H) and staining for γH2AX failed to demonstrate evidence for double-strand DNA breaks (Figure 5, G and I).

Figure 5.

Proinflammatory macrophages induce early oxidative DNA damage and single-strand DNA breaks. (A) Quantitative real-time PCR of Kim1 using whole kidney RNA from UI, SKO, and DKO mice at 8 weeks age. **P<0.01 for SKO versus UI and SKO versus DKO mice. (B) Representative images of immunohistochemistry for 8-hydroxy-2'-deoxyguanosine (8-OHdG) in cortex from 8-week-old kidneys of the indicated genotypes. Arrows depict the 8-OHdG⁺ cells from cystic and dilated tubules. Scale bar, 50 µm. (C) Quantification of 8-OHdG⁺ tubular cells in renal cortex as in (B). **P<0.01 for SKO versus UI and SKO versus DKO mice. (D) Representative images of cortical TUNEL staining in 8-week-old mice of the indicated genotypes. Arrows depict the TUNEL-positive cells. (E) Quantification of TUNEL-positive tubular cells in renal cortex as shown in (D). ***P<0.001 for SKO versus UI and SKO versus DKO mice. (F) Representative electron microscopy image of cortex from an 8-week-old SKO mouse kidney demonstrating normal morphology of both tubular nuclei (tn) and adjacent macrophage nucleus (mn). (G) Western blotting for cleaved caspase-3 (CC3, upper panel) and γH2AX (lower panel) in lysates from 8-week-old mouse kidneys. Each lane represents a separate kidney. GAPDH was used as the loading control. (H and I) Quantification of CC3 and γH2AX Western blotting analysis normalized to GAPDH. P=NS between all groups. (J) Representative images of alkaline comet assay using freshly isolated cells from 8-week-old UI, SKO, and DKO mouse kidneys. Arrows indicate nuclei with a comet tail. Scale bar, 400 µm. (K) Quantification of comet tail moment from freshly isolated tubular cells [at isolation, as shown in (J)] and after 7 days in culture [cultured, as shown in (M)]. **P<0.01 for SKO versus UI mice; P<0.05 for SKO versus DKO mice. (L) PCR analysis of genomic DNA from cultured epithelial cells isolated from UI, SKO, and DKO mice. Each lane represents a separate culture. (M) Representative images of alkaline comet assay using cultured cells from 8-week-old UI, SKO, and DKO mouse kidneys. Scale bar, 400 µm. The n listed under each bar represents the mice analyzed at that time point.

To determine if the TUNEL-positive cells indicated single-strand DNA breaks, we performed alkaline comet assays using tubular cells freshly isolated from 8-week-old kidneys. This revealed a significant increase in tail moment in the SKO cells indicative of DNA breaks, but not in cells isolated from UI or DKO kidneys (Figure 5J, quantified in Figure 5K). To determine whether the DNA breaks occurred in response to loss of Pc1 expression, or were induced by the presence of proinflammatory macrophages, tubular epithelial cells from 8-week-old UI, SKO, and DKO kidneys were cultured for 7 days and then subjected to alkaline comet assay. PCR confirmed that the SKO and DKO cultures were comprised of Pkd1^−/− cells (Figure 5L), whereas the alkaline comet assay revealed complete resolution of the DNA strand breaks (Figure 5M), consistent with a requirement for macrophages as the inducer of the in vivo oxidative DNA damage.

Macrophages Promote Proliferation-Independent and Proliferation-Dependent Tubule Enlargement

High-resolution multiphoton microscopy was used to define the effects of macrophages on cyst development and growth. Because of the heterogeneity in tubule morphology in the SKO and DKO kidneys (Figure 6, A–C), we calculated the cross-sectional tubule area and epithelial area in the first transverse section identified immediately after the first bend of the S1 nephron segment (Supplemental Figure 3B) in all three kidney types. Analysis of 100 randomly chosen S1 segments from three to four mice of each genotype revealed that UI kidneys exhibited a narrow range of tubule and epithelial areas, with a tubular-to-epithelial area ratio tightly maintained at 1.02–1.06 (Figure 6, D and H). In contrast, SKO kidneys exhibited two distinct morphologic phenotypes: tubules in which the epithelial areas is less than that seen in UI kidneys but the tubule area is increased (Figure 6E, left side of graph with representative tubule shown above); and tubules with a proportional increase in both tubular and epithelial area such that the ratio remained near 1 (Figure 6E, right side of graph with representative tubule shown above). Analysis of S1 segments from DKO kidneys showed that these were intermediate in size between the UI and SKO tubules and overall maintained a tubular-to-epithelial ratio close to 1 (Figure 6F).

Figure 6.

Proinflammatory macrophages promote proliferation-independent tubule dilation. (A–C) Representative multiphoton microscopic images of the indicated genotypes. (D–F) Graphs showing calculated tubular and epithelial area from 100 randomly analyzed S1 segments from kidneys of 8-week-old mice of the indicated genotypes. n=3 separate mice for the UI analysis, and four separate mice for the SKO and DKO analyses. Each bar represents an individual tubule area, with the overlying dot representing the epithelial area for that same tubule cross section. Drop down arrows and the red bar identify the specific isolated tubule cross section shown in the accompanying images. The ten tubule sections with smallest epithelial areas (S) and largest epithelial areas (L) were used for absolute tubule and epithelial area quantification. (G) Average tubular and epithelial area determined by absolute quantification of optically isolated tubule cross sections from 8-week-old kidneys of the indicated genotypes (n=10 per group). (H) Ratio of tubular-to-epithelial area obtained from absolute quantification shown in (G). ***P<0.001 for SKO “S” versus UI “S,” P<0.001 for SKO “S” versus DKO “S.” (I) Quantification of nuclei in isolated tubule cross-sections from the same sections analyzed in (G). **P<0.01 for SKO “L” versus UI “L” and DKO “L” versus UI “L.”

Quantitative analysis of the average tubular area, epithelial area, and tubular cell number for the ten tubules from each genotype with either the smallest calculated epithelial area (labeled S) or the largest calculated epithelial area (labeled L) was determined (Figure 6, G–I). The tubules from SKO kidneys with reduced epithelial area exhibited a disproportionately increased tubular area, leading to a tubular-to-epithelial area ratio of approximately 4 and the appearance of a true cyst (Figure 6E). These tubules had normal numbers of epithelial cells (Figure 6I). In contrast, the SKO and DKO tubules with the largest epithelial area exhibited a significant and equal increase in cell numbers compared with UI. Taken together, these results demonstrate two separate components of early cyst growth: proliferation-dependent, macrophage-independent tubule enlargement coincident with the loss of Pc1 (the large tubules with increased tubular cell numbers and near normal tubular-to-epithelial ratio in both SKO and DKO); and proliferation-independent, macrophage-dependent tubule dilation that accompanies proinflammatory macrophage-mediated tubular cell injury (the dilated tubules with normal tubular cell numbers and reduced epithelial area seen predominantly in SKO kidneys).

Discussion

Multiple studies have shown that polycystic kidneys contain abnormally large numbers of macrophages.^5,7,8 In this study we identified Mcp1 as the primary factor that promotes increased macrophage numbers in the kidney after loss of Pc1, and show that the tubular cell itself is the predominant source of Mcp1. Analysis of the kinetics of Mcp1 upregulation after Pkd1 loss shows that more than a 50% reduction in Pkd1 mRNA levels occurs before Mcp1 is upregulated. Interestingly, this level of Pkd1 mRNA reduction is not seen until after doxycycline induction is completed at the end of week 6. The further decline in Pkd1 mRNA levels between 6 and 8 weeks may be because of prolonged mRNA stability despite successful knockout, or by a selective loss of any remaining Pc1-expressing cells.²¹

Our data are consistent with the model that the Mcp1 upregulation in response to Pkd1 loss acts primarily to recruit circulating monocytes to the kidney, although a small component of in situ macrophage proliferation cannot be ruled out. Furthermore, we demonstrate that prevention of Mcp1-dependent macrophage recruitment to the kidney using tubule-specific knockout of both Pkd1 and Mcp1 significantly decreases the rate of cyst growth and improves kidney function and mouse survival. In a proof-of-principle experiment, we also demonstrated that pharmacologic inhibition of the Mcp1 receptor Ccr2 decreases macrophage numbers in Pkd1 knockout mice, again resulting in decreased cyst growth and improvement in serum creatinine. Analysis of the mechanism by which macrophages promote cyst growth led us to several additional findings that provide new insights into why cysts progressively enlarge.

The specific events that lead to the disruption of normal tubule architecture after the loss of Pc1 function have been intensely studied. The pathway that has received the greatest focus is dysregulation of tubular cell proliferation, leading to enlargement of the tubule in response to greater cell numbers.^22,23 The studies of autologous, adult murine models reveal tubular proliferative rates that are generally modest (typically 1%–4%^14,24), despite the fact that all or most of the tubular cells have undergone knockout of the Pkd1 or Pkd2 gene. Our results examining tubular cell Ki67 expression (identifying actively dividing cells as well as cells that have entered the cell cycle) and BrdU positivity (identifying cells that have replicated their DNA) concur with those previous murine studies. Our analysis reveals that knockout of either Pkd1 alone or Pkd1 and Mcp1 results in an absolute tubular cell proliferative rate of no more than 2.5%–3% at 6 and 8 weeks, an increase of approximately 1.2% and 2.2% over baseline, respectively. Despite this modest increase in tubular cell proliferation, cystic index in the Pkd1 knockout mice is already increased to 14% at 8 weeks, as compared with <1% for UI control kidneys. Although cyst index and cell proliferation are distinct measures, this suggests that cyst growth is likely to be driven by both proliferation-dependent and proliferation-independent factors. While this conclusion appears to be valid for our rapidly progressive murine model, it is unclear how critical the proliferation-dependent and proliferation-independent components of cyst growth are in humans where it is believed that a single cell undergoing a second-hit mutation proliferates over many years to form the final cyst.

Analysis of the highly cystic kidneys from 18-week-old Pkd1^fl/fl;Pax8-rtTA;TetO-Cre SKO mice in this study demonstrated large numbers of interstitial macrophages and associated high levels of Arg1 and Mrc1, as well as the macrophage survival and alternative activation factors Csf1 and Csf2.^17,18 This phenotype is similar to what we observed in macrophages isolated from kidneys after ischemia-reperfusion injury, where these alternatively activated macrophages can directly promote tubular cell proliferation.²⁵ Consistent with this, our analysis reveals an increase in the total number of Ki-67⁺ tubular cells from 2.7% at 8 weeks to 6.4% at 18 weeks in SKO kidneys, correlating with the timing of the expression of alternative macrophage activation and the greatest increase in cystic index. In contrast, in the absence of alternatively activated macrophages, tubular cell Ki67 positivity in the DKO mice remains relatively stable at 1.5%–2% greater than that in normal UI mice over the 12 weeks evaluated. These data support a model in which alternatively activated macrophages promote a late acceleration in Pc1 null tubular cell proliferation and cyst growth.

In contrast, during the phase of cyst initiation, macrophages exhibit a proinflammatory activation state and appear to promote cyst expansion in a proliferation-independent manner. The use of high-resolution multiphoton imaging during this early period demonstrates that the SKO tubules with the most pronounced cystic phenotype have normal tubular cell numbers, but exhibit a flattened epithelium reminiscent of tubular injury, accompanied by marked expansion of the tubule area. Multiple investigators have shown that proinflammatory macrophages cause nitrosylation and oxidative injury, including DNA damage, to nearby cells.^25–32 Li et al.³³ demonstrated that Ccr2-expressing monocytes traffic to renal tubules after ischemia-reperfusion injury and express a proinflammatory phenotype,²⁸ and at least one of those proinflammatory mediators, Tnfα, has been shown to promote cyst enlargement in murine models of PKD. Consistent with this, analysis of 8-week-old SKO kidneys revealed multiple features of tubular injury, including an increase in the proximal tubule injury marker Kim1, as well as an increase in oxidized guanosine and single-strand DNA nicking (Figure 5), that are absent or reduced in DKO kidneys.

An important question is whether macrophage accumulation and activation can be therapeutically targeted to slow cyst growth in patients, especially in light of the existence of an oral CCR2 inhibitor currently in clinical trials for treatment of organ fibrosis.³⁴ Our results using the Ccr2 antagonist INCB3344 are promising, but a study by Zoja et al.³⁵ found that bindarit, an inhibitor of Mcp1 synthesis, failed to slow cyst growth in a nonorthologous model of PKD in rats. Furthermore, although ischemia-reperfusion kidney injury has been found to increase cyst growth in rodent models of PKD,^36–39 it is currently unknown whether tubular injury, and the inflammatory signaling that follows, is an important regulator of cyst growth in human ADPKD. Clearly additional studies are required before consideration of this approach in patients.

Disclosures

None.