Frontline Science: Kindlin-3 is essential for patrolling and phagocytosis functions of nonclassical monocytes during metastatic cancer surveillance

Abstract

Nonclassical monocytes maintain vascular homeostasis by patrolling the vascular endothelium, responding to inflammatory signals, and scavenging cellular debris. Nonclassical monocytes also prevent metastatic tumor cells from seeding new tissues, but whether the patrolling function of nonclassical monocytes is required for this process is unknown. To answer this question, we utilized an inducible-knockout mouse that exhibits loss of the integrin-adaptor protein Kindlin-3 specifically in nonclassical monocytes. We show that Kindlin-3-deficient nonclassical monocytes are unable to patrol the vascular endothelium in either the lungs or periphery. We also find that Kindlin-3-deficient nonclassical monocytes cannot firmly adhere to, and instead "slip" along, the vascular endothelium. Loss of patrolling activity by nonclassical monocytes was phenocopied by ablation of LFA-1, an integrin-binding partner of Kindlin-3. When B16F10 murine melanoma tumor cells were introduced into Kindlin-3-deficient mice, nonclassical monocytes showed defective patrolling towards tumor cells and failure to ingest tumor particles in vivo. Consequently, we observed a significant, 4-fold increase in lung tumor metastases in mice possessing Kindlin-3-deficient nonclassical monocytes. Thus, we conclude that the patrolling function of nonclassical monocytes is mediated by Kindlin-3 and essential for these cells to maintain vascular endothelial homeostasis and prevent tumor metastasis to the lung.

Keywords: endothelium; integrins; lungs; metastasis; nonclassical monocytes; patrolling.

Figure 1.. Kindlin-3 is necessary for patrolling behavior of nonclassical monocytes in the periphery.

A) Schematic for Kindlin-3 mouse model. B) Snapshots of CX₃CR1^gfp/+Fermt3^fllox/flox;Mx1-Cre^−/− (WT) and CX₃CR1^gfp/+Fermt3^fllox/flox;Mx1-Cre^+/− (K-KO) mouse femoral vasculature. Time-lapse images were acquired using intravital confocal microscopy (IVM) on the femoral vasculature of 12-week old female cohort mice. Accompanying videos are in Supporting Information section. Mice were injected intraperitoneal (i.p.) with 500 ng recombinant mouse TNF-α in PBS 1–2 hours before imaging. Patrolling monocytes (green cells, no red) are highlighted by white arrows. Scale bar = 30 μM. C) Number of patrolling nonclassical monocytes (GFP⁺) per 10⁵ μm surface area of blood vessel. Graph is representative of 3 biological replicates. Mice designated fl/fl cre+ without poly(I:C) and fl/fl cre- with poly(I:C) are experimental controls representing no Kindlin-3 excision, while fl/fl cre+ with poly(I:C) represents a mouse with the Kindlin-3 gene excised. Bars within the graph are the averages (means) of 5 fields of view (FOV), 18 minutes per time-lapse image, per mouse. Spider plots of WT D) and K-KO E) nonclassical monocyte tracks in mouse lung during homeostasis (GFP⁺, Ly6C⁻ or Gr-1⁻). F) Nonclassical monocyte tracks were binned into 3 groups: 1-patrolling, 2-adherence/release (arresting for more than 2 frames on the endothelium without crawling), or 3-slipping (briefly stopping along endothelium repeatedly for less than 2 frames). 1 frame = approximately 7 seconds. *p<0.05, **p<0.01.

Figure 2.. Nonclassical monocytes lacking Kindlin-3 fail to enter lung tissue and do not surround invading metastases.

A) Snapshots of lung intravital imaging (Leica SP8, 25× 0.8 NA objective) in CX₃CR1^gfp/+Kindlin-3^fl/flMx1-cre- (WT) or cre+ (K-KO) mice 30 minutes after i.v. injection of 3×10⁵ B16F10-RFP cells. Scale bar = 50 uM. B) Analysis of intravital microscopy imaging from A), using Imaris software to detect the average number of GFP+ monocytes per for every frame of the recording within 30 μM of B16F10-RFP cells. n = 3 mice per group, independent experiments. C) Number of patrolling nonclassical monocytes per FOV from A). D) Flow cytometry gating for selection of vascular (CD45v) or interstitial (CD45i) leukocytes, followed by monocyte gating using CD115 and CD11b markers. Lungs were mechanically dissociated so as not to lose CD115 expression. E) Ly6C⁻ monocyte frequencies within interstitial lung tissues out of total CD45+ live cells. n = 6–7 per group. F) Ly6C⁻ monocyte frequencies within the vascular niches of the lung tissue out of total CD45+ live cells. n=6–7 mice per group. G) 3×10⁵ B16F10 cells were labeled with CellTrace Violet and injected i.v. into DsRed:K-KO BMT mice. 16 hours later, classical (Ly6C⁺) and nonclassical (Ly6C⁻) monocytes from WT (DsRed) donors or K-KO donors within the same mice were assessed for uptake of B16F10-violet particles per 10⁵ CD45⁺ cells in the lung. n = 8 mice per group. Kolmogorov-Smirnov nonparametric test for single comparisons, Kruskal-Wallis nonparametric test for multiple comparisons. ns = not significant, *p<0.05, **p<0.01, ****p<0.0001.

Figure 3.. Patrolling monocytes increase their expression of LFA-1 on their surface in the lung environment and require integrins for sensing tumor cells.

A) MFI of beta integrins ß1 and ß2) present on Ly6C- monocytes from blood and lung in healthy WT mice by flow cytometry. n = 4 mice per group. B) Frequency of CD115⁺ monocytes in the lungs of E2:integrin KO BMT and C) the specific frequency of Ly6C⁻ monocytes in the lungs of E2:integrin KO BMT mice. n = 4 mice per group. Numbers of Ly6C⁻ D) or Ly6C⁺ E) monocytes that are positive for B16F10 cell particles (CellTrace Violet-labeled) 16 hours after i.v. injection in E2:integrin KO BMT mice, n = 4 mice per group. Multiple t tests with Holm-Sidak method for grouped analysis, Kruskal-Wallis nonparametric test for multiple comparisons. ns = not significant, *p<0.05.

Figure 4.. Dysregulation of endothelial cell mRNA transcripts when nonclassical monocytes fail to patrol.

A) Scheme of bone marrow transplantation for integrin knockout chimeras with E2 bone marrow donors at a 50:50 ratio. B-M) Endothelial cells were digested out of lungs harvested from E2:WT and E2:K-KO BMT mice, with and without i.v. injection of 5×10⁵ B16F10 cells, and endothelial cells were then sorted and purified for mRNA. B-M) qPCR for inflammatory cytokines, integrin ligands, and other signaling cytokines were run in duplicate with Hprt used as a normalization control. n = 3–4 mice per group. N) Heat map of transcript signatures for each group of mixed BMT mice with and without B16F10 cancer cells injected. Each gene expression value was rescaled to reflect the range of expression for each individual gene. Blue = Homeostasis, Pink = Metastasis. ns = not significant, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Figure 5.. Metastatic cancer is better controlled when nonclassical monocytes are able to patrol.

A) Hematoxylin and eosin (H&E) stained lungs were prepared from E2:WT and E2:K-KO mice i.v. injected with 5×10⁵ B16F10 cells 2 weeks before harvesting. Images represent one biological triplicate. Scale bar 1000 μM. Error bars = 95 % CI. B) The area of total tumor mass over total lung area for each H&E slide analyzed. C) Number of B16F10 metastases quantified per section of H&E slide analyzed. Each point is an individual section from a total of 3 biological replicates. D) E2:WT and E2:K-KO BMT mice were sacrificed 16 hours after B16F10-RFP injection and the frequency of B16F10-RFP cells out of all lung cells was measured between E2:WT and E2:K-KO BMT mice. n = 6–7 mice per group. E) frequency of NK1.1⁺ cells in lungs of E2:WT compared to E2:K-KO mice. n = 6–7 mice per group. F) frequency of Ly6G⁺ neutrophils in the vasculature of lungs of E2:WT compared to E2:K-KO mice. n = 6–7 mice per group. G) Frequency of CD11c⁺ cells in lungs of E2:WT compared to E2:K-KO mice. n = 6–7 mice per group. H) Frequency of migratory dendritic cells (DCs) in lungs of E2:WT versus E2:K-KO mice. Kolmogorov-Smirnov nonparametric test for single comparisons. ns = not significant, *p<0.05, **p<0.01, ***p<0.001.