Humanized mice reveal a macrophage-enriched gene signature defining human lung tissue protection during SARS-CoV-2 infection

Abstract

The human immunological mechanisms defining the clinical outcome of SARS-CoV-2 infection remain elusive. This knowledge gap is mostly driven by the lack of appropriate experimental platforms recapitulating human immune responses in a controlled human lung environment. Here, we report a mouse model (i.e., HNFL mice) co-engrafted with human fetal lung xenografts (fLX) and a myeloid-enhanced human immune system to identify cellular and molecular correlates of lung protection during SARS-CoV-2 infection. Unlike mice solely engrafted with human fLX, HNFL mice are protected against infection, severe inflammation, and histopathological phenotypes. Lung tissue protection from infection and severe histopathology associates with macrophage infiltration and differentiation and the upregulation of a macrophage-enriched signature composed of 11 specific genes mainly associated with the type I interferon signaling pathway. Our work highlights the HNFL model as a transformative platform to investigate, in controlled experimental settings, human myeloid immune mechanisms governing lung tissue protection during SARS-CoV-2 infection.

Keywords: COVID-19; CP: Immunology; CP: Microbiology; antiviral responses; human immune responses to SARS-CoV-2; humanized mice; macrophage responses to SARS-CoV-2; mouse models of SARS-CoV-2; pulmonary immune responses.

Graphical abstract

Figure 1

Generation of HNFL mice (A) An NRGL mouse engrafted with pairs of human fLX (red ellipse). (B) Procedure to generate HNFL mice. Image created with BioRender.com. (C) Frequency of human CD45⁺ cells within total CD45⁺ cells (mouse + human) in naive NRGL-LX and HNFL-LX (flow cytometry analysis). n = 3–5; mean ± SEM; Welch’s t test; ^∗∗p ≤ 0.01. (D–K) t-distributed stochastic neighbor embedding (t-SNE) plots of the human (D, three fLX, 9,968 cells; F, two fLX, 5,160 cells) and mouse (H, three fLX, 5,091 cells; J, two fLX, 407 cells) compartments in NRGL-LX (D and H) and HNFL-LX (F and J). Relative representation of each species compartment within the fLX is indicated between the two t-SNE plots. Cell subset frequencies (E and G, human; I and K, mouse) are shown below the respective t-SNE plots. See also Figures S1 and S2; Tables S1 and S2; Video S1.

Figure 2

HNFL mice are effectively protected from SARS-CoV-2 infection (A–I) Representative SARS-CoV-2 N IHC (A, D, G) and five-color IHC (B, C, E, F, H, I; yellow, SARS-CoV-2 Spike; magenta, human CD31; cyan, murine CD61; red, human CD68; gray, DAPI) on naive (A–C) or inoculated NRGL-LX (D–F, 2 DPI; G, H, I, 7 DPI; 10⁶ PFU). (C, F, and I) 2× magnification of the inset located in (B, E, and H), respectively. (B, E, and H) 100×; scale bar, 200 μm; (A, D, and G) 200×; scale bar, 100 μm; (C, F, and I) 400×; scale bar, 50 μm. (J and K) SARS-CoV-2 viral RNA quantification in inoculated (blue) and contralateral NRGL-LX (red), and in NRGL lung (gray), following infection with 10⁴ (J) or 10⁶ (K) PFU (n = 4–12). Limit of detection (LOD; dotted line) is equivalent to RNA copies/mg tissues in naive fLX (n = 7). Mean ± SEM, Kruskal-Wallis test; ^∗p ≤ 0.05, ^∗∗∗p ≤ 0.001. (L) Quantification of infectious SARS-CoV-2 particles in contralateral (CL, 7 DPI) or inoculated NRGL-LX (2 and 7 DPI; 10⁶ PFU). n = 3–5 fLX. Mean ± SEM, Kruskal-Wallis test; ns, non-significant. (M) Non-linear regression between viral RNA copies and PFU per milligram of tissue in inoculated (n = 10; 2 and 7 DPI; 10⁶ PFU), contralateral (n = 3), and naive (n = 2) NRGL-LX. Yellow area represents productive infection. Productive infection threshold (PIT) equal to 10⁷ RNA copies/mg (n = 15). (N) Representative three-dimensional dorsal and profile views of a single NRGL mouse following inoculation of the right fLX with a SARS-CoV-2 NanoLuc virus (10⁶ PFU). NanoLuc signal was quantified over 12 days of infection. (O and P) Regionalized quantification of NanoLuc expression in inoculated (O; n = 4, right fLX; 10⁶ PFU) and contralateral (P; n = 4, left fLX) NRGL-LX over a 12-day course of infection. Mean signal from naive fLX (n = 3) was used to determine assay baseline (mock). Red line represents the mean signal over time. (Q) SARS-CoV-2 viral RNA quantification in inoculated HNFL-LX (10⁶ PFU) at 2 and 7 DPI (n = 3–5 fLX). LOD (dotted line) represents the mean RNA copies/mg tissues in naive fLX (n = 5). Significance between 2 and 7 DPI viral load values and area of productive infection was calculated by running a Kruskal-Wallis test (red line and asterisks) Mean ± SEM; ^∗p ≤ 0.05, ^∗∗∗∗p ≤ 0.001. PIT, productive infection threshold. (R–W) Representative SARS-CoV-2 Spike (R, S, T) and six-color IHC (U, V, W; yellow, SARS-CoV-2 Spike; cyan, hCD3e; green, hCD20; orange, hCD8; red, hCD68; gray, DAPI) on naive (R, U) or inoculated (10⁶ PFU) tissue section (S, V, 2 DPI; T, W, 7 DPI) from HNFL-LX. (R, S, T) 200×; scale bar, 100 μm; (U, V, W) 100×; scale bar, 200 μm. (X) Quantification of tissue area immunoreactive for SARS-CoV-2 Spike (percentage of analyzed tissue) in inoculated (10⁶ PFU) HNFL-LX and NRGL-LX (n = 4–5). Mean ± SD, Kolmogorov-Smirnov t test; ^∗p ≤ 0.05. See also Figure S3.

Figure 3

HNFL mice are protected from severe histopathology (A) Cumulative histopathologic score of inoculated NRGL-LX (10⁴ or 10⁶ PFU) at 2 and 7 DPI (n = 5–10). Mean ± SEM, two-way ANOVA; ^∗p ≤ 0.05, ^∗∗p ≤ 0.01. (B) Cumulative histopathologic score of inoculated (10⁶ PFU; 2 and 7 DPI) and naive/contralateral (CL) NRGL-LX (n = 8–12). Mean ± SEM, Kruskal-Wallis test; ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001. (C–K) Representative histopathologic phenotypes of inoculated NRGL-LX (10⁶ PFU). (C) Neutrophil accumulation within air spaces (within black dashed lines). (D) Syncytial epithelial cells (black arrows) and interstitial edema (asterisks). (E) Fibrin thrombi occluding intermediate-sized blood vessels (black arrows), with interstitial edema (asterisks). (F) Denuded epithelium (black arrows). (G) Coagulative necrosis (left of dashed line), adjacent to viable fLX (right of dashed line). (H) Intra-air-space hemorrhage. (I) Fibrin thrombi occluding interstitial capillaries (black arrows). (J) Accumulation of necrotic debris within air spaces (asterisks). (K) Hyaline membrane (black arrows). Hematoxylin and eosin (H&E) staining, 400×, scale bar, 50 μm. (L–Q) Ultrastructural analysis of inoculated NRGL-L (10⁶ PFU). (L) SARS-CoV-2 viral particles in double-membrane-bound vesicle (DMV, asterisk) in AT2 pneumocytes (lamellar body, LB) at 2 DPI. Top right corner is a magnification of the top left inset (dotted box). (M) Infected pneumocyte with viral particles around the peripheral extracellular area (left of dashed line) at 2 DPI. (N) Magnification (3.75×) of the inset from (M) with event of viral particle maturation in DMV. (O) Air space filled with necrotic cellular debris, including lamellar bodies and denuded AT2 pneumocytes undergoing apoptosis at 7 DPI. (P) Virus particles at variable stages of maturation within the cytoplasm of AT2 cells at 7 DPI (lamellar bodies, LB). Free and DMV-associated virus particles and electron-dense viral replication centers (asterisk). (Q) Cluster of mature virus particles with radiating spikes and aggregates of nucleocapsid protein contained within a DMV at 7 DPI. (R–T) Representative H&E staining on naive (R) or inoculated (S, 2 DPI; T, 7 DPI; 10⁶ PFU) HNFL-LX tissue section; 200×, scale bar, 100 μm. (U) Cumulative histopathologic score of inoculated HNFL-LX (10⁶ PFU) at 2 and 7 DPI. Mean ± SEM (n = 4). Statistically significant differences with mean cumulative histopathologic score of NRGL-LX (B) at the same time point are indicated with asterisks (two-way ANOVA; ^∗p ≤ 0.05, ^∗∗∗p ≤ 0.001). See also Figure S4.

Figure 4

Infiltration of human hematopoietic lineages in HNFL-LX upon SARS-CoV-2 inoculation (A) Frequencies of human CD45⁺ cells within total CD45⁺ cells (mouse + human) in HNFL-LX at 2 and 7 DPI (flow cytometric analysis). Dotted line represents mean frequency of CD45⁺ cells in naive HNFL-LX (Figure 1C) (n = 3–7). Mean ± SEM, one-way ANOVA; ^∗∗∗p ≤ 0.001, ^∗∗∗∗p ≤ 0.0001 over naive HNFL-LX. (B–F) Normalized cell count (number of cells analyzed in subset × [total fLX cell count/total cells analyzed]) of hCD3⁺ T cells (B), hCD3⁺ hCD4⁺ T cells (C), hCD3⁺ hCD8⁺ T cells (D), hCD20⁺ cells (E), and hCD3⁻ hCD20⁻ hCD33⁻ hCD56⁺ cells (F) in naive or inoculated HNFL-LX (2 and 7 DPI) (n = 3–7). Mean ± SEM, one-way ANOVA; ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, ^∗∗∗p ≤ 0.001. (G–J) t-SNE plots of the human compartment in HNFL-LX at 2 (G, three fLX, 6,736 cells) and 7 DPI (I, three fLX, 11,269 cells). Cell subset frequencies (H, 2 DPI; J, 7 DPI) are shown below the respective t-SNE plots. (K) Frequencies of the human hematopoietic fraction (immune) within the human cellular compartment in naive (two fLX, 5,160 cells) or inoculated (2 DPI, three fLX, 6,736 cells; 7 DPI, three fLX, 11,269 cells) HNFL-LX. (L) Frequencies of different human hematopoietic lineages within the human cellular compartment (two fLX, 5,160 cells) in naive or inoculated (2 DPI, three fLX, 6,736 cells; 7 DPI, three fLX, 11,269 cells) HNFL-LX. (M) Frequencies of AT1 (green), AT2 (purple), ciliated cells (blue), and club cells (red) within the human epithelial compartment in naive (two fLX, 2,159 cells) or inoculated (2 DPI, three fLX, 937 cells; 7 DPI, three fLX, 1,825 cells) HNFL-LX. A dotted line symbolizes the variation in the size of the AT2 compartment upon infection. (N, O) t-SNE plots displaying clustered and scaled expression of several transcripts coding for myeloid, inflammatory, and regulatory markers in the human compartment of HNFL-LX at 2 DPI. Clusters of interest (dotted circles) are labeled in the top left plot of (N) (AIM, activated inflammatory macrophages; ARM, activated regulatory macrophages; SSM, steady-state macrophages; DC, dendritic cells). Cluster-defining genes are marked by an asterisk, and log2FC values are indicated near the corresponding cluster(s). n = 3 fLX, 6,736 cells. See also Figure S5 and Table S1.

Figure 5

HNFL mice recapitulate a potent and balanced human antiviral response against SARS-CoV-2 infection (A) Cluster heatmap representing the top 33 proteins significantly (p ≤ 0.05) up- (Z > 0) and downregulated (Z < 0) in HNFL-LX at 2 DPI (10⁶ PFU, n = 4) in comparison with naive (n = 4) HNFL-LX. (B) Relative differential expression of the set of 33 selected proteins in HNFL-LX (n = 4) and NRGL-LX (n = 4) at 2 DPI (10⁶ PFU) represented through a semi-cluster heatmap. Proteins significantly (p ≤ 0.05) up- (Z > 0) and downregulated (Z < 0) are labeled in red. (C and D) Differentially expressed proteins in HNFL-LX (C) or NRGL-LX (D) at 2 DPI. Proteins with p ≤ 0.05 (horizontal dashed line) and with logFC ≥ 1 or ≤ −1 (vertical dashed lines) are considered significantly up- or downregulated, respectively. Naive n = 4; 2 DPI n = 4. (E and F) Differentially phosphorylated proteins at 2 DPI in HNFL-LX (E) and NRGL-LX (F). Proteins with p ≤ 0.05 (horizontal dashed line) and with logFC ≥ 1 or ≤ −1 (vertical dashed lines) are considered significantly up- or downregulated, respectively. Naive n = 4; 2 DPI n = 4. (G and H) Significantly (p ≤ 0.05) differentially expressed genes (upregulated, red; downregulated, blue) in HNFL-LX at 2 (G) and 7 DPI (H) following infection (10⁶ PFU) in comparison with naive HNFL-LX. Fold changes were computed using MAST (Model-based Analysis of Single-Cell Transcriptomics) from pooled scRNA-seq clusters. Transcripts with p ≤ 10⁻²⁰⁰ (horizontal dotted line) and with log2 fold change ≥ 0.2 or ≤ −0.2 (vertical dotted lines) are highlighted. Naive n = 2; 2 DPI n = 3. (I) List of PDGs found to be upregulated by both proteomics and transcriptomic approaches in inoculated HNFL-LX (YES) or solely via the proteomic approach (NO). Only PDGs found to be upregulated through both approaches were considered as definitive PDGs. (J–L) Differentially expressed transcripts in inoculated (J, 2 DPI; K, 7 DPI; 10⁶ PFU) and contralateral non-inoculated NRGL-LX (L, 7 DPI) in comparison with naive NRGL-LX. Transcripts with p_adj ≤ 0.05 and with log2 fold change ≥ 2 are considered significantly up- (red) or downregulated (blue). Naive n = 3; 2 DPI n = 4; 7 DPI n = 6; CL/contralateral n = 3. (M) Number of differentially up- (red) or downregulated (blue) genes per time point (2 and 7 DPI) and infection settings (inoculated/CL) in NRGL-LX. Naive n = 3; 2 DPI n = 4; 7 DP n = 6; CL/contralateral n = 3. See also Figure S6; Tables S2, S3, and S4.

Figure 6

Macrophages are the dominant mediators of the antiviral and anti-inflammatory response in HNFL mice (A) t-SNE plots displaying clustered and scaled expression of PDG transcripts (OAS1–3, DDX58, IFI44L, IFIT2–3, HLA-B, IFH1, USP18, and ISG20) in the human compartment of HNFL-LX at 2 DPI. Clusters of interest (dotted circles) are labeled in the top left plot. Cluster-defining genes are marked by an asterisk, and log2FC values are indicated near the corresponding cluster; n = 3 fLX, 6,736 cells. (B) Scaled mean expression of all PDGs in each cluster of the human compartment of HNFL-LX at 2 DPI. For each cluster, the median of the scaled mean expression of all PDGs is shown (red line). Statistically significant differences between combined PDG expression across clusters are reported as a heatmap at the right (RM one-way ANOVA with Geisser-Greenhouse correction). Non-significant differences (p ≥ −log10[0.05]) are shown in gray. (C) t-SNE plot displaying clustered and scaled expression of ISG15 in the human compartment of HNFL-LX at 2 DPI. Clusters of interest (dotted circles) are labeled in the top left plot of (A). Asterisk indicates that ISG15 is an activated macrophage-defining gene, and log2FC values are indicated near the corresponding cluster; n = 3 fLX, 6,736 cells. (D) Differential co-expression of the USP18-ISG15 axis across human lineages in inoculated HNFL-LX. Cumulative scaled expression of USP18 and ISG15 was calculated for all human clusters regrouped as specific lineages in HNFL-LX at 2 DPI and plotted on a x/y axis with x and y corresponding to ISG15 and USP18 expression, respectively. Four categories of co-expression were identified and are delineated by dashed lines. (E) Viral gene count per cluster, segregated between activated macrophage clusters (AM) and all others (Others) in inoculated HNFL-LX (2 and 7 DPI) (n = 29 clusters). Mean ± SD, Mann-Whitney t test, ^∗∗∗p ≤ 0.001. AIM, activated inflammatory macrophages; ARM, activated regulatory macrophages; Epi, epithelial cells; Endo, endothelial cells; SSM, steady-state macrophages; DC, dendritic cells; Fib/Fibro, fibroblasts.

Figure 7

The HNFL mouse model to uncover myeloid-mediated mechanisms of lung tissue protection during SARS-CoV-2 infection (A and B) Working model depicting the cellular and molecular mechanisms that may drive pathology and protection in inoculated HNFL-LX and NRGL-LX, respectively. Image created with BioRender.com.