CCR5 drives NK cell-associated airway damage in pulmonary ischemia-reperfusion injury

Abstract

Primary graft dysfunction (PGD) limits clinical benefit after lung transplantation, a life-prolonging therapy for patients with end-stage disease. PGD is the clinical syndrome resulting from pulmonary ischemia-reperfusion injury (IRI), driven by innate immune inflammation. We recently demonstrated a key role for NK cells in the airways of mouse models and human tissue samples of IRI. Here, we used 2 mouse models paired with human lung transplant samples to investigate the mechanisms whereby NK cells migrate to the airways to mediate lung injury. We demonstrate that chemokine receptor ligand transcripts and proteins are increased in mouse and human disease. CCR5 ligand transcripts were correlated with NK cell gene signatures independently of NK cell CCR5 ligand secretion. NK cells expressing CCR5 were increased in the lung and airways during IRI and had increased markers of tissue residency and maturation. Allosteric CCR5 drug blockade reduced the migration of NK cells to the site of injury. CCR5 blockade also blunted quantitative measures of experimental IRI. Additionally, in human lung transplant bronchoalveolar lavage samples, we found that CCR5 ligand was associated with increased patient morbidity and that the CCR5 receptor was increased in expression on human NK cells following PGD. These data support a potential mechanism for NK cell migration during lung injury and identify a plausible preventative treatment for PGD.

Keywords: Chemokines; Innate immunity; NK cells; Pulmonology; Transplantation.

Figure 1. RNA sequencing following mouse orthotopic lung transplantation with prolonged cold ischemia (OLT-PCI) identifies NK cell–independent upregulation of chemokine ligands.

(A) OLT-PCI was followed by RNA sequencing of right (Ctrl) and left (IRI) lungs from mice treated with a control antibody (n = 8) and left lungs (Dep) from mice treated with an NK cell–depleting antibody (n = 7). Data were collected 6 hours after transplantation. (B) Volcano plot highlighting the top 25 differentially expressed genes by log(fold change). (C) Differential gene transcription analyses showing KEGG and Gene Ontology pathways were enriched for those containing chemokine-associated transcripts (red). (D) Heatmap and hierarchical clustering of the top 11 chemokine ligand transcripts. Differences across the 3 conditions are shown for transcript counts of the following genes: (E) Ccl2, (F) Ccl3, (G) Ccl4, (H) Ccl5, (I) Cxcl1, (J) Cxcl2, (K) Cxcl3, (L) Cxcl9, (M) Cxcl10, and (N) Cxcl16. Box-and-whisker plots display individual data points bound by boxes at 25th and 75th percentiles and medians depicted with bisecting lines. Differences were assessed using the Mann-Whitney U test. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 2. Mouse and human lung NK cell metagenes are correlated with CCR5 receptor ligand transcripts during IRI.

Chemokine and NK cell metagene scores were derived from OLT-PCI RNA sequencing data shown in Figure 1. Mouse data were collected 6 hours after transplantation. (A) NK cell metagene score compared between OLT-PCI left lung (IRI), contralateral control (Ctrl), and OLT-PCI with NK cell depletion (Dep). (B) Chemokine metagene score across these 3 conditions. (C) Correlation matrix relating this NK cell metagene with the chemokine metagene and subcomponent chemokines, showing greatest correlation with Ccl2, Ccl4, and Cxcl10. (D) Correlation dot plot showing that NK cell gene score and Ccl2 distinguish control from injured lungs. (E) Mouse IRI gene score is increased in human BAL samples with PGD. RNA sequencing was performed on human bronchoalveolar lavage collected on postoperative day 1. (F) Human chemokine gene score derived from most differentially expressed mouse chemokine ligands is increased in human PGD. (G) Human correlation matrix between NK cell gene transcripts and individual chemokines, showing greatest correlation with CCL5. (H) Correlation dot plot showing NK cell gene score and CCL5 transcripts colored by PGD status. Box-and-whisker plots display individual data points bound by boxes at 25th and 75th percentiles and medians depicted with bisecting lines. Differences between 3 groups were assessed using the Kruskal-Wallis test. Post hoc testing and comparisons between 2 groups employed the Mann-Whitney U test with Benjamini-Hochberg corrections for multiple comparisons. P values are shown or *P < 0.05; **P < 0.01.

Figure 3. Mouse and human CCR5 ligand proteins are increased in BAL during mouse and human IRI and predict clinical outcomes.

We collected left lung BAL 4 hours after hilar suture removal in the mice subjected to HC (n = 8) compared to sham (S, n = 8) procedures and measured BAL concentrations for CCR5 ligand proteins (A) CCL3/MIP1α, (B) CCL4/MIP1β, and (C) CCL5/RANTES. In humans, we collected BAL on the first postoperative day following lung transplantation in 34 recipients with severe PGD and 77 recipients without PGD (grade 0 or 1). We measured CCR5 ligand proteins (D) CCL3/MIP1α, (E) CCL4/MIP1β, and (F) CCL5/RANTES. (G) BAL CCL5/RANTES is inversely correlated with PaO₂/FiO₂ ratio on day 1. (H) Kaplan-Meier plot of mechanical ventilation time stratified by BAL CCL5/RANTES concentration and PGD. Summary data are displayed with box-and-whisker plots illustrating individual data points, bound by boxes at 25th and 75th percentiles, and with medians depicted with bisecting lines. Individual P values as assessed with Mann-Whitney U test (A–C), generalized linear models adjusted for recipient baseline characteristics (D–G), or log-rank test for Kaplan-Meier plot (H).

Figure 4. CCR5⁺ NK cells are increased during mouse HC and express markers of maturity and tissue residence.

We performed HC (n = 5) and sham (S, n = 5) procedures and quantified NK cells and their phenotypes via spectral flow cytometry across blood, spleen, thoracic lymph node (LN), and lung tissues collected 4 hours after reperfusion. (A) Contour plot of CCR1 on NK (CD45⁺CD3^–F480^–CD19^–NK1.1⁺NKp46⁺) cells in the lung. (B) CCR1 frequency of total NK cells during HC or S procedures. (C) Histograms of CCR1 on NK cells with fluorescence minus one (FMO) control, HC, and S. (D) CCR1 on NK cells by median fluorescence intensity (MFI). (E) Contour plot of CCR5 on NK cells in the lung. (F) CCR5 frequency of total NK cells during HC or S procedures. (G) Histograms of CCR5 on NK cells with FMO control, HC, and S. (H) CCR5 on NK cells by MFI. (I–L) We quantified maturation states of CCR5⁺ and CCR5^– NK cells. (M) Frequencies of CD49a on CCR5⁺ and CCR5^– NK cells. (N) Heatmap of MFIs of additional markers of NK cell activation. Summary data are displayed with box-and-whisker plots illustrating individual data points, bound by boxes at 25th and 75th percentiles, and with medians depicted with bisecting lines. Differences were assessed using the Mann-Whitney U test with Benjamini-Hochberg corrections for multiple comparisons. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 5. CCR5 blockade reduces mouse NK cell airway inflammation.

(A) Schematic of maraviroc (allosteric CCR5 antagonist) administration 24 hours and 1 hour before left lung HC and reperfusion (n = 7) compared to vehicle control and HC (vehicle, n = 8) and sham procedures (n = 8). Samples were collected 4 hours after hilar suture removal (reperfusion). (B) Absolute number of NK cells within left lung BAL samples per mL. (C) NK cells as a percentage of total BAL lymphocytes. (D) Percentage of NK cells with CCR1 in the BAL. (E) Percentage of NK cells expressing CCR5 in the BAL. (F) Percentage of NK cells expressing CD49a in the BAL. (G) NK cells as a percentage of total lymphocytes in the lung. (H) Percentage of CCR5⁺ NK cells in the lung tissue. (I) Percentage of CD49a⁺ NK cells in the lung tissue. Summary data are displayed with box-and-whisker plots illustrating individual data points, bound by boxes at 25th and 75th percentiles, and with medians depicted with bisecting lines. Differences were assessed using the Kruskal-Wallis test. Post hoc testing between groups employed the Mann-Whitney U test with Benjamini-Hochberg corrections for multiple comparisons. P values are directly shown.

Figure 6. CCR5 blockade reduces lung damage during mouse pulmonary IRI.

(A) Schematic of maraviroc (allosteric CCR5 antagonist) administration 24 hours and 1 hour before left lung HC and reperfusion compared to vehicle control with HC and sham procedures. Injury was assessed 4 hours after left hilar suture removal. (B) Representative H&E staining. Scale bars: 800 μm (top row), 100 μm (middle), and 50 μm (bottom). We also performed quantitative measures of lung injury: (C) partial pressure of oxygen in mmHg (PaO₂), (D) extravascular lung water (μL), and (E) percentage of endothelial permeability. Experiments studied at least n = 6 animals per condition. Summary data are displayed with box-and-whisker plots illustrating individual data points, bound by boxes at 25th and 75th percentiles, and with medians depicted with bisecting lines. Differences were assessed using the Kruskal-Wallis test. Post hoc testing between groups employed the Mann-Whitney U test with Benjamini-Hochberg corrections for multiple comparisons. P values are directly shown.

Figure 7. CCR5⁺ NK cells are increased in human BAL during PGD.

Human BAL samples were prospectively collected from lung transplant patients with severe PGD (n = 4) and those without PGD (n = 6) at 2 weeks after transplant. (A) Histogram demonstrating CCR5 staining versus unstained control on NK cells from BAL samples. (B) Percentage of CCR5⁺ NK cells in BAL. (C) Heatmap of surface marker MFI differences between CCR5⁺ and CCR5^– NK cells with individual plots of surface markers on NK cells stratified by CCR5 shown for (D) NKG2D, (E) NKp46, (F) NKG2A, (G) NKG2C, (H) FCER1G (FCRG), (I) CD62L, (J) CD16, and (K) CD57. Summary data are displayed with individual data points alone (D–K) or bound by boxes at 25th and 75th percentiles, and with medians depicted with bisecting lines (B). Differences were assessed with generalized linear models adjusted for recipient baseline characteristics. *P < 0.05, **P < 0.01.