Fc-engineered antibodies promote neutrophil-dependent control of Mycobacterium tuberculosis

Abstract

Mounting evidence indicates that antibodies can contribute towards control of tuberculosis (TB). However, the underlying mechanisms of humoral immune protection and whether antibodies can be exploited in therapeutic strategies to combat TB are relatively understudied. Here we engineered the receptor-binding Fc (fragment crystallizable) region of an antibody recognizing the Mycobacterium tuberculosis (Mtb) capsule, to define antibody Fc-mediated mechanism(s) of Mtb restriction. We generated 52 Fc variants that either promote or inhibit specific antibody effector functions, rationally building antibodies with enhanced capacity to promote Mtb restriction in a human whole-blood model of infection. While there is likely no singular Fc profile that universally drives control of Mtb, here we found that several Fc-engineered antibodies drove Mtb restriction in a neutrophil-dependent manner. Single-cell RNA sequencing analysis showed that a restrictive Fc-engineered antibody promoted neutrophil survival and expression of cell-intrinsic antimicrobial programs. These data show the potential of Fc-engineered antibodies as therapeutics able to harness the protective functions of neutrophils to promote control of TB.

Fig. 1. Wild-type IgG1 α-glucan antibody does not drive Mtb restriction in vitro.

a, Glucan (bovine liver glycogen) antigen-binding ELISA of the α-glucan-specific antibody clone, 24c5. ELISA was run in technical duplicate. OD_450 nm, optical density at 450 nm. b, Macrophage Mtb restriction assay. The y axis shows live (GFP)/total (mCherry) Mtb burden in human MDMs normalized by the no-antibody condition for the respective donor. Each point is the triplicate average from one human macrophage donor. c, Whole-blood Mtb restriction assay. The y axis is the area under the Mtb-276 growth curve value normalized by the no-antibody condition from the respective donor. Each point represents a triplicate average from one donor. One-way ANOVA with Dunnett’s correction comparing each antibody or antibiotic with the isotype IgG1 control antibody. Adjusted P < 0.05 indicated. Error bar shows mean with standard deviation. Horizontal dashed lines indicate the no-antibody condition. Source data

Fig. 2. Fc-engineered α-glucan antibodies show a range in functional activity.

The ability of each α-glucan Fc variant to mediate (from top to bottom) NK cell degranulation (ADNKA), NK cell secretion of IFNγ (ADNKA), NK cell secretion of MIP-1β (ADNKA), complement component 3b (C3b) deposition (ADCD), phagocytosis by THP-1 monocytes (ADCP) and phagocytosis by primary human neutrophils (ADNP) was experimentally determined. Each antibody was run in duplicate across each assay, and the mean functionality is shown. Grey dashed line indicates the performance of an IgG1 isotype control antibody. See Methods for the experimental details of each assay. Source data

Fig. 3. Hierarchical clustering used for the rational down-selection of Fc-engineered α-glucan antibodies.

a, Cluster dendrogram of the α-glucan Fc-variant panel following complete-linkage hierarchical clustering of the functional profiling data. At least one variant from each cluster (boxed) was selected for antimicrobial profiling. b, Polar plots highlighting the functional profile of each α-glucan Fc variant in the down-selected panel. Each pie piece indicates the magnitude of functionality in the respective assay relative to the entire panel. The α-glucan Fc variants were max-scaled before polar plot visualization.

Fig. 4. Several Fc-engineered α-glucan antibodies drive Mtb restriction in whole blood.

a, Whole-blood Mtb restriction assay. The x axis shows the different α-glucan Fc variants (25 μg ml⁻¹), an IgG1 isotype control antibody as a negative control (25 μg ml⁻¹) and the antibiotic rifampin as a positive control (0.25 μg ml⁻¹). The y axis is the area under the Mtb-276 growth curve value normalized by the no-antibody condition from the respective donor. Each point represents a triplicate average from one donor. One-way ANOVA with Dunnett’s correction comparing each antibody or antibiotic with the isotype IgG1 control antibody. Adjusted P < 0.05 are indicated. Error bars show mean with standard deviation. Horizontal dashed line indicates the no-antibody condition. b, Spearman correlation between the normalized area under the curve in the whole-blood assay and α-glucan antigen binding determined via ELISA. Spearman correlation was two-sided. Green solid line is the linear regression line; grey shading indicates the 95% confidence interval. c,d, Cytokine Luminex using the whole-blood assay supernatant collected at 120 h. Triplicate average from donor A shown. c, Clustered heat map indicating the cytokine profile elicited by each α-glucan Fc variant. Data were z-scored before heat map visualization. d, PCA of cytokine Luminex data. Left: score plot of the first two principal components. Right: loading plot of the first two principal components. The x and y axes are principal components 1 and 2, respectively. Source data

Fig. 5. Fc-engineered α-glucan antibodies drive Mtb restriction in a neutrophil-dependent manner.

a, Spearman correlation matrix of the different functional and antimicrobial assays. The bottom left section of the correlation matrix indicates the Spearman correlation coefficient for each relationship. The upper right section of the correlation matrix contains ellipses that have their eccentricity parametrically scaled to the strength of the correlation for each relationship. Pairwise correlation between the normalized area under the curve in the whole-blood assay and the phagocytic score from the neutrophil phagocytosis assay is highlighted. Grey-shaded error band shows 95% confidence interval. Spearman correlations were two-sided. *P < 0.05; **P < 0.01; ***P < 0.001. b, Clustered heat map indicating the performance of each α-glucan Fc variant in the down-selected panel across different functional and antimicrobial assays. Data were z-scored before heat map visualization. c, Whole-blood Mtb restriction assay with neutrophil depletions. The x axis shows selected α-glucan Fc variants (25 μg ml⁻¹), an IgG1 isotype control antibody as a negative control (25 μg ml⁻¹) and the antibiotic rifampin as a positive control (0.25 μg ml⁻¹). Each antibody or antibiotic treatment was tested in whole blood (filled-in bars) and neutrophil-depleted blood (white bars). The y axis is the area under the Mtb-276 growth curve value normalized by the no-antibody condition from the respective donor. Each point represents a triplicate average from one donor. Two-tailed, unpaired t-test, comparing restriction in the neutrophil-depleted blood condition with the whole-blood condition for each treatment. Error bars show mean with standard deviation. Unadjusted P < 0.05 are indicated, and comparisons labelled as ‘NS’ (not significant) have unadjusted P > 0.05. Horizontal dashed line indicates the no-antibody condition. Source data

Fig. 6. 24c5 SEHFST LS drives the upregulation of antimicrobial gene programs in neutrophils.

a, Uniform manifold approximation and projection (UMAP) visualization of all cells and cell subsets recovered following scRNA-seq. The x and y axes represent the first and second dimensions of the UMAP embedding respectively. b, UMAP depicting fractional abundance (density) in different cell types across treatment groups. c, Neutrophil differential expression analysis. The x and y axes indicate log₂ fold change of the SEHFST LS variant compared with the no antibody and the IgG1 conditions, respectively. Genes consistently increased in the 24c5 SEHFST LS condition (red quadrant): (1) Mann–Whitney P < 0.1 and a log₂ fold change > 0.25 compared with either the 24c5 IgG1 or no Ab condition, (2) detected in a minimum fraction of 0.1 cells in either of the two conditions and (3) a log₂ fold change > 0 compared with both the 24c5 IgG1 and no Ab conditions. Genes consistently decreased in the 24c5 SEHFST LS condition (blue quadrant): (1) Mann–Whitney P < 0.1 and a log₂ fold change < −0.25 compared with either the 24c5 IgG1 or no Ab conditions, (2) detected in a minimum fraction of 0.1 cells in either of the two conditions and (3) a log₂ fold change < 0 compared with both the 24c5 IgG1 and no Ab conditions. Mann–Whitney tests were two-sided. d, Gene list enrichment analysis using GO Biological Process gene sets. The x axis indicates the adjusted P value of each GO term. Left: GO terms enriched in red quadrant genes from c. Right: GO terms enriched in blue quadrant genes from c. Vertical dashed line indicates two-sided Fisher’s exact test adjusted P value of 0.05. Numbers on each circle show the odds ratio. Top ten GO terms by adjusted P value shown.

Extended Data Fig. 1. Functional profiling of Fc-engineered α-glucan antibodies.

a, Glucan (bovine liver glycogen) antigen-binding ELISA of the α-glucan-specific Fc-variant panel. Area under the 5-point dilution curve is plotted. ELISA was run in technical duplicate. b, Heatmap of individual Fc glycan structures for each 24c5 Fc variant. c, Clustered heatmap indicating the performance of each α-glucan Fc-variant in the functional profiling assays. Data were z-scored prior to heatmap visualization. d, Spearman correlations of paired normal and half-life extended (LS) variants in the 24c5 Fc-variant panel. Spearman correlations were two-sided. Unadjusted p-values are indicated. e, Spearman correlation of functional profiling between an Fc-variant panel previously published Ebola-specific monoclonal antibody (VIC16) and the 24c5 α-glucan-specific Fc-variant panel. Spearman correlations were two-sided. Unadjusted p-values are indicated. Source data

Extended Data Fig. 2. Macrophage Mtb restriction assay of Fc-engineered α-glucan antibodies.

Y-axis shows live (GFP) / total (mCherry) Mtb burden in human monocyte-derived macrophages normalized by the no antibody condition for the respective donor. Each point is the triplicate average from 1 human macrophage donor. Source data

Extended Data Fig. 3. Cytokine Luminex of whole-blood restriction assay.

Cytokine Luminex using the whole-blood assay supernatant collected at 120 hours. a,d, Luminex MFI across selected cytokines in a, Donor A, and d, Donor B. One-way ANOVA with Dunnett’s correction comparing each antibody with the isotype IgG1 control antibody. Green (Restrictive); grey (Non-restrictive). Adjusted p-values < 0.05 are shown. Error bars indicate mean with standard deviation. Run in technical triplicate. b, Clustered heatmap indicating the cytokine profile elicited by each α-glucan Fc-variant in Donor B. Data were z-scored prior to heatmap visualization. c, Principal component analysis of cytokine Luminex data from Donor B. Left, score plot of the first two principal components. Right, loading plot of the first two principal components. e, Cytokine blockade experiments. 24c5 variants were tested for their ability promote Mtb restriction in whole-blood (grey), whole-blood with an IL-1β blocking antibody (pink), or whole-blood with an IFNγ blocking antibody (blue). One-way ANOVA with Dunnett’s correction comparing restriction in the IL-1β and IFNγ blockade conditions, with the whole-blood condition for each treatment. Error bars indicate mean with standard deviation of 3 biological replicates from distinct donors. Source data

Extended Data Fig. 4. Whole-blood Mtb restriction assay with monocyte depletions and complement inhibition.

a, Monocyte depletions. Each antibody or antibiotic treatment was tested in whole-blood (filled in bars) and monocyte depleted blood (white bars). Grey dotted line indicates the no antibody condition. b, Complement inhibition using cobra venom factor (CVF). Each antibody or antibiotic treatment was tested in whole-blood (filled in bars) and CVF-treated blood (white bars). Grey dotted line indicates the no antibody condition. Black dotted line indicates the CVF-treated no antibody condition. a,b, X-axis shows selected α-glucan Fc-variants (25 μg/mL), an IgG1 isotype control antibody as a negative control (25 μg/mL), and the antibiotic rifampin as a positive control (0.25 μg/mL). Y-axis is the area under the Mtb-276 growth curve value normalized by the no antibody condition from the respective donor. Each point represents a triplicate average from one donor. Two-tailed, unpaired t test, comparing restriction in the monocyte depleted or CVF-treated blood condition with the whole-blood condition. Unadjusted p-values < 0.05 are indicated are comparisons labeled as “ns” (not significant) have unadjusted p-values > 0.05. Error bars show mean with standard deviation. Source data

Extended Data Fig. 5. CD14 monocyte differential expression analysis.

scRNAseq analysis of CD14 monocytes. a, Genes consistently increased in the 24c5 SEHFST LS condition (red quadrant): (i) Mann-Whitney p-value < 0.1 and a log2 fold change > 0.25 compared to either the 24c5 IgG1 or no Ab condition, (ii) detected in a minimum fraction of 0.1 cells in either of the two conditions, and (iii) a log2 fold change > 0 compared to both the 24c5 IgG1 and no Ab conditions. Genes consistently decreased in the 24c5 SEHFST LS condition (blue quadrant): (i) Mann-Whitney p-value < 0.1 and a log2 fold change < −0.25 compared to either the 24c5 IgG1 or no Ab conditions, (ii) detected in a minimum fraction of 0.1 cells in either of the two conditions, and (iii) a log2 fold change < 0 compared to both the 24c5 IgG1 and no Ab conditions. Mann-Whitney tests were two-sided. b, Gene list enrichment analysis using GO Biological Process gene sets. Left, GO terms enriched in red quadrant genes from panel C. Right, GO terms enriched in blue quadrant genes from panel C. Vertical dashed line indicates two-sided Fisher’s exact test adjusted p-value of 0.05. Numbers on each circle show the odds ratio. Top ten GO terms by Benjamini-Hochberg adjusted p-value shown.