Molecular traces of Drosophila hemocytes reveal transcriptomic conservation with vertebrate myeloid cells

Abstract

Drosophila hemocytes serve as the primary defense system against harmful threats, allowing the animals to thrive. Hemocytes are often compared to vertebrate innate immune system cells due to the observed functional similarities between the two. However, the similarities have primarily been established based on a limited number of genes and their functional homologies. Thus, a systematic analysis using transcriptomic data could offer novel insights into Drosophila hemocyte function and provide new perspectives on the evolution of the immune system. Here, we performed cross-species comparative analyses using single-cell RNA sequencing data from Drosophila and vertebrate immune cells. We found several conserved markers for the cluster of differentiation (CD) genes in Drosophila hemocytes and validated the role of CG8501 (CD59) in phagocytosis by plasmatocytes, which function much like macrophages in vertebrates. By comparing whole transcriptome profiles in both supervised and unsupervised analyses, we showed that Drosophila hemocytes are largely homologous to vertebrate myeloid cells, especially plasmatocytes to monocytes/macrophages and prohemocyte 1 (PH1) to hematopoietic stem cells. Furthermore, a small subset of prohemocytes with hematopoietic potential displayed homology with hematopoietic progenitor populations in vertebrates. Overall, our results provide a deeper understanding of molecular conservation in the Drosophila immune system.

Fig 1. Integration of Drosophila larval hemocyte Drop-seq datasets.

(A) A UMAP plot of the nine major hemocyte types identified in Drosophila. The cell count for each cell type is indicated in parentheses. (B) UMAP plots showing the tissue origins (top) and experimental conditions (bottom) of hemocytes. (C) The proportion (left) or count (right) of tissue origins (top) and experimental conditions (bottom) of hemocytes for each cell type. (D) The proportion of cell types for each sampling time point and condition, wild type (WT) or wasp-infected (inf). (E) A dot plot presenting the expression of the top 5 cell type markers in the lymph gland (top) and circulation (bottom). The dot color indicates the average level of expression, and the dot size represents the percentage of cells expressing the gene in each cell type.

Fig 2. Characteristics of public Drosophila larval hemocyte scRNA-seq datasets.

(A) A UMAP plot of hemocyte clusters newly identified in the integrated Drosophila scRNA-seq dataset. (B) The proportion of each cell cluster represented by each dataset. The cell count for each cell type is indicated in parentheses. (C) Dot plots presenting the expression of the top three markers for each cell type. The dot color indicates the average level of expression, and the dot size represents the percentage of cells expressing the gene in each cell type. (D) Proportions of broad cell types for each cell type/state defined in the integrative analysis (left) and categorized by experimental condition (middle) and tissue origin (right).

Fig 3. Identification of cell type clusters using orthologous genes.

(A) The workflow of the analysis comparing immune cell types between species. (B) A summary of the expressed orthologous genes between each species used in this study. (C) The t-SNE plots of Drosophila hemocytes and zebrafish immune cells using 5739 orthologous genes. The Drosophila data were downsampled to one-tenth (4389 cells). Data from all 3301 zebrafish cells were used. (D) The t-SNE plots of zebrafish and mouse immune cells using 8714 orthologous genes. The mouse data were randomly downsampled to one-fifth (3034 cells). (E) The t-SNE plots of mouse and human immune cells using 10,379 orthologous genes. The human data were randomly downsampled to one-fiftieth (5253 cells). (F and G) Bar plots showing the fold enrichment of biological processes as identified by gene ontology. Only genes conserved between zebrafish and mice (F) or mice and humans (G) were tested.

Fig 4. Drosophila CG8501, an orthologous gene of human CD59.

(A) Schematic illustration of the orthologous gene selection process. (B) Expression of CD orthologs in Drosophila hemocyte sub-populations. The dot color indicates the average level of expression, and the dot size represents the percentage of cells expressing the gene in each cell type. (C) Expression of protein CG8501 in the hemocyte detected by antibody staining against human CD59 protein. Protein CG8501 (magenta) was expressed in the cytosol and did not overlap with NimC1 (green) or phalloidin (white). Nuclei were stained by DAPI (blue). (D) Decrease in Hml+ hemocyte numbers in CG8501 RNAi expressing mutants. Compared to wild-type hemocytes (HmlΔ-Gal4 UAS-GFP Oregon R), knockdown CG8501 hemocytes (HmlΔ-Gal4 UAS-GFP CG8501 RNAi) show low Hml (green) and NimC1 (white) expressions. However, PPO1 (magenta)-positive mature crystal cells or the number of total hemocytes (DAPI, blue) did not change. (E) Quantification of Hml+ or NimC1+ hemocyte numbers in wild-type hemocytes (Oregon R) and knockdown CG8501 hemocytes (CG8501RNAi) (**p < 0.001). Horizontal bars indicate median values. (F) Whole mount images of wild-type larvae (Oregon R) and larvae with Hml+ blood cell (HmlΔ-Gal4 UAS-GFP CG8501 RNAi). Magnified images are on the right. (G) A visualization of the phagocytic ability of Drosophila hemocytes. Hemocytes (green) showed reduced phagocytotic ability against E. coli (magenta, top) and S. aureus (magenta, bottom) in CG8501 RNAi-expressing mutants (HLT-Gal4 UAS-GFP CG8501 RNAi). (H) Quantifications of the phagocytotic abilities of hemocytes against bacteria in panel G (***p < 0.0001). Horizontal bars indicate median values.

Fig 5. Unsupervised cross-species analysis using MetaNeighbor.

MetaNeighbor AUROC values calculated using (A) Drosophila and zebrafish, (B) zebrafish and mouse, and (C) mouse and human immune cells. The MetaNeighbor analysis was performed using the pseudo-cell transformed expression data of the orthologous genes.

Fig 6. Conservation map of immune cells across species.

A conservation score heatmap predicted by integrating MetaNeighbor predictions and GSVA scores. Conservation scores were calculated by averaging MetaNeighbor AUROC and scaled GSVA scores for each cell type pair. Only cell type pairs assigned with reciprocal best hits by MetaNeighbor or conservation scores above 0.8 were included.