Comparative analysis of gene expression profiles for several migrating cell types identifies cell migration regulators

Abstract

Cell migration is an instrumental process that ensures cells are properly positioned to support the specification of distinct tissue types during development. To provide insight, we used fluorescence activated cell sorting (FACS) to isolate two migrating cell types from the Drosophila embryo: caudal visceral mesoderm (CVM) cells, precursors of longitudinal muscles of the gut, and hemocytes (HCs), the Drosophila equivalent of blood cells. ~350 genes were identified from each of the sorted samples using RNA-seq, and in situ hybridization was used to confirm expression within each cell type or, alternatively, within other interacting, co-sorted cell types. To start, the two gene expression profiling datasets were compared to identify cell migration regulators that are potentially generally-acting. 73 genes were present in both CVM cell and HC gene expression profiles, including the transcription factor zinc finger homeodomain-1 (zfh1). Comparisons with gene expression profiles of Drosophila border cells that migrate during oogenesis had a more limited overlap, with only the genes neyo (neo) and singed (sn) found to be expressed in border cells as well as CVM cells and HCs, respectively. Neo encodes a protein with Zona pellucida domain linked to cell polarity, while sn encodes an actin binding protein. Tissue specific RNAi expression coupled with live in vivo imaging was used to confirm cell-autonomous roles for zfh1 and neo in supporting CVM cell migration, whereas previous studies had demonstrated a role for Sn in supporting HC migration. In addition, comparisons were made to migrating cells from vertebrates. Seven genes were found expressed by chick neural crest cells, CVM cells, and HCs including extracellular matrix (ECM) proteins and proteases. In summary, we show that genes shared in common between CVM cells, HCs, and other migrating cell types can help identify regulators of cell migration. Our analyses show that neo in addition to zfh1 and sn studied previously impact cell migration. This study also suggests that modification of the extracellular milieu may be a fundamental requirement for cells that undergo cell streaming migratory behaviors.

Keywords: Border cells; Cell migration; Drosophila melanogaster: caudal visceral mesoderm; Fluorescence activated cell sorting (FACS); Hemocytes; Neural crest; Neyo; Singed; Zfh1.

Figure 1. Transcriptome analysis of the CVM cells and hemocytes (HC) from Drosophila embryos

(A–A”) Confocal images of CVM cells (red, HLH54F–H2A–mCherry) and HC (green, Srp-Gal4>UAS-Gap-Venus) within a live embryo. (B) Experimental scheme. (C) In order to set the threshold for Venus signal (CVM marker HLH54F-Gap-Venus), the cells from control (yw) embryos of matching stages was analyzed. (D) Flow cytometric analysis of cells from the CVM marker (HLH54F-Gap-Venus) expressing embryos shows a small but distinct population with Venus signal. The upper (R2) and lower (R3) populations are sorted as CVM marker (+) and (−) cells. (E) The CVM transcriptome analysis by comparing RNA-Seq reads from CVM marker positive (Y-axis) and negative samples (X-axis). Each dot represents a gene and plotted with the RPKMs (reads per kilobases per million reads) from CVM marker positive and negative populations. Red dots are enriched genes in CVM, blue dots are downregulated genes in CVM, and grey dots represent genes that did not show significant difference between these two populations. (E’) Differentially expressed genes between CVM and non-CVM samples are plotted with their fold change (CVM / non CVM) and enrichment significance (−log₁₀(p-value)). To plot p-values of 0 (highest significance) in log scale, the maximum was set for enrichment significance as 16. (F) The threshold for Venus signal (HC marker Srp-Gal4>UAS-Gap-Venus) was set using the cells from control (yw) embryos of matching stages. (G) Flow cytometric analysis of cells from the HC marker expressing embryos shows a distinct population with Venus signal. The upper (R2) and lower (R3) populations are sorted as HC marker (+) and (−) cells. (H) The HC transcriptome analysis by comparing RNA-Seq reads from HC marker positive (Y-axis) and negative samples (X-axis). Each dot represents a gene and plotted with the RPKMs (reads per kilobase million reads) from HC marker positive and negative populations. Red dots are enriched genes in HC; blue dots are downregulated genes in HC; and grey dots represent genes that did not show significant difference between these two populations. (H’) Differentially expressed genes between HC and non-HC samples are plotted with their fold change (HC/ non HC) and enrichment significance (−log₁₀(p-value)). To plot p-values of 0 (highest significance) in a log scale, the maximum was set for enrichment significance as 16.

Figure 2. GO term analysis of CVM enriched genes

(A) The top 10 list of each GO term category (biological process, molecular process, cellular component) in CVM short list (n=324) using DAVID database. Each term is sorted with its enrichment significance (− log₁₀(p-value)). Asterisk (*) denotes terms relevant to CVM function as muscle founder cells and ** denotes the most enriched term (see next). (B) A pie chart showing subclasses within the “integral component of plasma membrane**” (n=33). The subclass name, number of genes, and percentage within this class is labeled. (C) List of genes from enriched GO terms of interest. The term “Transcription factor activity, sequence specific DNA binding” is not shown in (A), but contains the most number of transcription factors. PM: plasma membrane, MCT: monocarboxylate transporter.

Figure 3. In situ hybridizations confirm gene expression in the CVM

The top panel is a representative antibody staining specifically marking CVM cells. All embryos are oriented with the anterior to the left; those marked with a ‘D’ are shown from the dorsal view, while all others are seen from the lateral view. Arrowheads refer to CVM-specific staining, in those cases where this staining could be deemed ambiguous due to additional staining in other tissues, levels of expression, or viewpoint. Gene expression patterns are shown in (A) wild-type embryos, stages 10–13 (B) HLH54F^Δ598 mutant embryos lacking CVM, stages 11, 12, or 13 and (C) htl^AB42 mutant embryos with CVM displaying various migratory defects, stages 11 or 12.

Figure 4. Unexpected trends in our transcriptional profiling data highlight potentially meaningful tissue interactions

(A) Diagram and cross-section of a stage 12 Drosophila embryo that illustrate the main tissues that potentially interact with the CVM. mRNA expression patterns in the TVM (B), gut primordium (C), yolk (D), germ cells (E), and the less common anal pad precursors (F), which reside above the CVM before they begin their migration in stage 10.

Figure 5. GO term analysis of hemocyte enriched genes

(A) The top 10 list of each GO term category (biological process, molecular process, cellular component) in hemocyte short list (n=386) using DAVID database. Each term is sorted with its enrichment significance (− log₁₀(p-value)). * denotes terms relevant to HC function as immune cells and processing reactive oxygen species. ** denotes the most enriched term “plasma membrane”. (B) A pie chart showing subclasses within the “plasma membrane” (n=51). The subclass name, number of genes, and percentage for each subclass is shown. (C) List of genes from enriched GO terms of interest. MCT: monocarboxylate transporter, ECM: extracellular matrix.

Figure 6. In situ hybridizations confirm gene expression in the hemocytes

Embryos shown from dorsal view. Top panels depict immunostained embryos expressing the Venus reporter under the control of the HC-specific srp-GAL4 driver. In situ hybridization using indicated riboprobes to detect expression in embryos spanning stages 9–13.

Figure 7. Comparative analysis of gene expression profiles of migratory cell populations within and across species

(A) A Venn diagram illustrating overlaps between four Drosophila datasets: CVM, hemocyte, and two border cell (BCwang, BCborghese) gene expression profiles. For CVM and hemocyte profiles, the short lists described previously were used. For BCwang list, gene names were taken from Table S1 (Wang et al., 2006) “Genes enriched in the migratory border cells”. The BCborghese list was extracted from Table S1 (Borghese et al., 2006) “Genes significantly up-regulated in the WT border cells compared to follicle cells”. (B) The list genes in overlaps. (C) A Venn diagram showing overlaps in gene expression profiles across species: CVM, hemocyte, and chick neural crest cells. (D) The list of genes that are common in all three dataset (CVM, hemocyte, and chick neural crest cells), which are categorized into three functional groups.

Figure 8. Tissue-specific RNAi against zfh1 and neo results in CVM and BC migration defects

(A–E) Immunostained embryos expressing the GV2 reporter. In WT embryos, the CVM cells undergo synchronous migration as two closely associated yet dynamic clusters (A,D). RNAi-mediated ablation of zfh1 specifically in the CVM results in a loss of directionality and synchrony between the two migrating groups of cells (B). Tissue-specific knockdown of neo via RNAi results in a similar asynchronous migration phenotype (E). In situ hybridization confirms expression of zfh1 (C) and neo (F) in the CVM. (G,H) Immunostained egg chambers stained against GFP (green), Lamin (blue), and Phalloidin (red) to mark the BCs, nuclear membrane, and cell membranes, respectively. In stage 10 control egg chambers, BCs migrate through the nurse cells to the periphery of the developing oocyte (G). In contrast, tissue-specific expression of a neo RNAi construct in the BCs via the slbo-GAL4 driver often results in a failure to migrate to the oocyte periphery (H; Sup. Figure 4). (I–K) Temporal color-coded projections of CVM migration in control (A), 5053-GAL4>zfh1 RNAi (B), and 5053-GAL4>neo RNAi (C) embryos expressing the HC2 reporter. Each projection is compiled from 80 movie stills of time points taken every 3 minutes over a 4-hour span (A’,B’C’), with each still assigned a unique color code corresponding to a specific time point. (I,I’) In WT embryos, CVM cells migrate in a closely associated yet dynamic fashion. In contrast, tissue-specific knockdown of zfh1 in the CVM via expression of a hairpin construct using the 5053-GAL4 driver results in reduced cohesion within each migrating cohort, such that individual cells wander and approach each other more closely at the midline (J,J’), while RNAi-mediated knockdown of neo in the CVM results in stalling, dysregulated cell division, and concomitant asynchronous migration of the two groups of cells (K,K’).