Repo is expressed exclusively in glial cells (Xiong et al., 1994). A repo-Gal4 driver that recapitulates Repo expression was used to express UAS-Ras^V12 (Ogienko et al., 2020; Sepp et al., 2001). This led to robust production of primary cultures however these failed to survive beyond early passages (Supplementary file 1). To counter potential cell death or modulate growth signaling, additional genotypes were tested including co-expression of UAS-transgenes encoding the P35 baculovirus cell survival factor, dsRNAs targeting tumor suppressors, or the Gal4 inhibitor Gal80^ts (Supplementary file 1). Co-expression of a UAS-brat^dsRNA or expression of tub-Gal80^ts each produced a single line of cells that could be propagated for extended passages however the latter line was difficult to maintain and eventually lost (Supplementary file 1). The repo-Gal4: UAS-brat^dsRNA; UAS-Ras^V12 (Rbr6) line has been passaged more than 50 times. The parental Rbr6 line and three clonal derivatives (Rbr6-2, Rbr6-4, and Rbr6-F9) have an elongated morphology and stained positive for Repo (Table 1; Figures 1 and 3, and Figure 3—figure supplement 1). A few cells expressed neuronal markers (Figure 3—figure supplement 1; Supplementary file 2). To induce differentiation, we gave cells two 24 hr ecdysone treatments separated by 24 hr to approximate the pulses of ecdysone during the larval to pupal transition. Cells from each of the clones survived treatment with ecdysone suggesting they are of adult type, two clones showed morphological changes and formed a network, and all continued to express Repo (Figure 3 and Figure 3—figure supplement 2).

The results of RNAseq analysis revealed that the three Rbr6 clones have very similar expression patterns (Figure 2—figure supplement 2). In addition, their DE gene signatures are also a close match to gene signatures of glial cells as identified by single-cell RNAseq (Table 2) and to glial-associated genes reported in the literature. For example, zydeco (zyd), which encodes a potassium-dependent sodium/calcium exchanger, is upregulated in all three clones, consistent with the literature (Zwarts et al., 2015; Featherstone, 2011), and gcm2, a transcription factor, is upregulated in two clones (Figure 2—figure supplement 3). These data suggest the Rbr6 clones will be a useful in vitro source of glial cells.

Breathless is expressed in the tracheal epithelium and a btl-Gal4 driver was used to express UAS-Ras^V12 (Shiga et al., 1996). Patches of cells with epithelial morphology proliferated in primary cultures and several continuous lines were generated (Table 1, Supplementary file 1). We were unable to derive clones of these using dilution or selection methods, which were successful for other cell types. Correspondingly, three parental lines were examined: Btl3, Btl7, and Btl8 (Table 1). All showed expression of the epithelial marker Shotgun/E-Cadherin (Shg/Ecad) and two grew in a squamous epithelial sheet with Ecad expression at the cell periphery (Figures 1 and 4, and Figure 4—figure supplement 1). In comparison S2 did not show peripheral expression of Ecad (Figure 4). Treatment of the squamous epithelial cells (Btl3 and Btl7) with ecdysone caused aggregation and formation of large multicellular clusters (Figure 4, Figure 4—figure supplement 2).

RNAseq data analysis comparing the top upregulated genes in the Btl cell lines with scRNAseq datasets revealed that the lines closely match the signatures of the adult trachea, a network of epithelial tubules (Table 2) and Btl3 expresses trachealess (trh) a master regulator of tracheal identity (Wilk et al., 1996; Figure 2—figure supplement 3). Overall, the morphological and molecular characteristics of the lines are consistent with an epithelial cell type of tracheal origin.

The 24B-Gal4 driver is an insertion in held out wings (how) and is expressed in mesoderm and muscle cells (Brand and Perrimon, 1993; Zaffran et al., 1997). Expression of UAS-Ras^V12 with 24B-Gal4 readily produced continuous lines (Table 1, Supplementary file 1). Four clones (24B5-B8, 24B5-D8, 24BG1-F3, and 24BG1-G1) derived from two parental lines (24B5 and 25BG1) were analyzed in more detail (Table 1). The cells had a bipolar shape and expressed mesoderm markers including Twist and Mef2 (Figures 1 and 5, and Figure 5—figure supplement 1). When treated with ecdysone, cells from both parental lines and clones 24B5-B8 and 24B5-D8 elongated, fused as indicated by multinucleate cells, formed a network, and expressed Myosin heavy chain (Mhc) (Figure 5 and Figure 5—figure supplement 2). There was also extensive cell lysis. Beginning 2 days after the second ecdysone treatment, the cells began to contract spontaneously. Contraction of cells from the 24B5 parental line and the two derivative clones (24B5-B8 and 24B5-D8) was visible in real time (Videos 1 and 2), whereas contraction of parental line 24BG1 cells was much slower and visualized more clearly in time-lapse (Videos 3 and 4). The clones 24BG1-F3 and 24BG1-G1 underwent morphological change but did not express Mhc or contract (Figure 5—figure supplements 2 and 3). In later passages, the 24BG1 parental line also lost expression of Mhc and the ability to contract (Figure 5—figure supplement 2). This highlights the importance of using early passage cells and avoiding extended passaging that could alter the phenotypic (and genotypic) characteristics of the cells.

We also attempted to derive lines from Mef2-Gal4 because Mef2 regulates muscle development and is expressed in muscle progenitors and differentiated muscle suggesting Mef2-Gal4 would be a good candidate for deriving cell lines (Bour et al., 1995; Gossett et al., 1989; Lilly et al., 1995; Ranganayakulu et al., 1995). However, only rare primary cultures had some proliferating cell patches, and none progressed to continuous lines (Supplementary file 1; Figure 5—figure supplement 4). Analysis of larvae from the cross (Mef2-Gal4/+; UAS-GFP/UAS-Ras^V12) and control larvae (Mef2-Gal4/+; UAS-GFP/+) showed that Ras^V12 expression disrupted muscle development, suggesting that the prevalent amorphous GFP-positive cells observed in primary cultures were abnormal muscle cells (Figure 5—figure supplement 4).

The RNAseq analysis for 24B-Gal4-derived cell lines, identified the cells as muscle (Table 2). 24B5-B8 cells express high levels of the transcription factors nautilus (nau) and twist (twi) (Figure 2—figure supplement 3; Figure 5—figure supplement 1; Table 2), and high levels of myoblast city (mbo), which encodes an unconventional bipartite GEF with a role in myoblast fusion (Erickson et al., 1997). The capacity of these mesoderm-derived cell lines to differentiate into active muscle shows that the cells are muscle precursors and thus should be a useful reagent to analyze muscle physiology and development.

To target neuronal cells, we expressed UAS-Ras^V12 with the pan-neural drivers scratch-Gal4 and elav-Gal4, however none of the primary cultures resulted in continuous cell lines (Supplementary file 1; Figure 6—figure supplement 1). In previous work, we made primary cultures from embryos with ubiquitous expression of UAS-Ras^V12 using the Act5C-Gal4 driver (Simcox et al., 2008b). The cells growing in these cultures included neuronal cells (Simcox et al., 2008b). Here, we used an Act5C-GeneSwitch-Gal4 driver to express UAS-Ras^V12. GeneSwitch-Gal4 is only active in the presence of the drug, RU486/mifepristone, which provides the advantage of being able to regulate Ras^V12 expression (Nicholson et al., 2008; Osterwalder et al., 2001). Several continuous lines were generated (Supplementary file 1). Clones derived from two of these (ActGSB-6 and ActGSI-2) (Table 1) were positive for the neuronal marker, HRP (horseradish peroxidase) (Figure 6, Figure 6—figure supplements 2 and 3). After differentiation with ecdysone, expression of Futsch/MAPB1 (Hummel et al., 2000) and Fas2 (Mao and Freeman, 2009) was enhanced and revealed axonal-like outgrowths from the cells (Figure 6 and Figure 6—figure supplement 3). Differentiated cells also showed enhanced expression of Elav, which is commonly used as a marker for postmitotic neurons (Figure 6 and Figure 6—figure supplement 3; Robinow and White, 1991). Elav is also expressed transiently in glial cells and proliferating neuroblasts Berger et al., 2007; however, the cells were negative for the glial marker Repo (Supplementary file 2).

RNAseq analysis revealed that many neuronal genes are upregulated in these cell lines, including Glutamic acid decarboxylase 1 (Gad1), slowpoke (slo), 5-hydroxytryptamine (serotonin) receptor 1A (5-HT1A), Protein C kinase 53E (Pkc53E), Diuretic hormone 31 Receptor (Dh31-R), and straightjacket (stj). In addition, comparison of the top upregulated genes in these cells to marker genes from scRNAseq data identifies a cell type of neuronal origin as the best match (Table 2). The cells should be a useful source of neuronal cells.

Cells of clone ActGSI-3 derived from the ActGSI parental line (UAS-Ras^V12 expression with Act5C-GeneSwitch-Gal4; Table 1, Supplementary file 1) show characteristics of hemocytes and express the hemocyte marker Hemese (Figure 7; Kurucz et al., 2003). They are also positive for HRP, but not other neuronal markers (Figure 7—figure supplement 1). ActGSI-3 cells divide in floating clusters, contrasting with S2 cells, which are also thought to be hemocytes, that grow as single cells (Figures 1 and 7).

RNAseq analysis demonstrated that many hemocyte genes are upregulated in these cells, including serpent (srp), Hemese (He), eater, u-shaped (ush), Cecropin A2 (CecA2), and Cecropin C (CecC). Comparison of top upregulated genes with scRNAseq data showed that the cells have a strong match to the top marker genes of hemocytes (Table 2).

We determined the cell density at confluence for the cell lines (Table 3). The cells in each line grow to confluence attached to the tissue-culture surface, except ActGSI-3, which grow as floating cell clusters (Figure 8). The cells are not contact inhibited and cell clusters are formed allowing cells to grow to higher density (Figure 8). We determined the doubling time of 13 cell lines and clones using growth curves (Table 3; Figure 8—figure supplement 1). Most had doubling times within a range of approximately 20–40 hr (Table 3). The hemocyte-like clone ActGSI-3 was an outlier with a longer doubling time of 70 hr (Table 3). In cells from clones ActGSB-6, ActGSI-2, and ActGSI-3, expression of Ras^V12 is dependent on GeneSwitch Gal4, which is active only in the presence of mifepristone. In the absence of the drug the cells become quiescent (Figure 8—figure supplement 1).

We determined the gross karyotype of 13 cell lines and clones. In keeping with previous findings for Ras^V12 expressing cell lines, most (8) were diploid, or near diploid (Simcox et al., 2008b; Table 3; Figure 3—figure supplement 3; Figure 4—figure supplement 3; Figure 5—figure supplement 5; Figure 6—figure supplement 4; Figure 7—figure supplement 2). Related clones had similar karyotypes, which likely indicates that parental lines may also be clonal as a result of selective pressure for cells that grow well in culture. Some lines were polyploid and common aneuploid conditions include loss of an X chromosome and varying numbers of chromosome 4 (Table 3).

Nine parental and clonal lines were transfected with an Act5C-EGFP plasmid and the fraction of GFP-positive cells was determined after 48 hr. Cells from all lines tested could be transfected. The range of efficiency was from 16% to 34% with most lines showing transfection of approximately one quarter of the cells (Table 3). Similarly treated, cells from the S2 line showed an efficiency of 53%.

The following fly stocks were used to create primary cell lines: Gal4 drivers: 24B/how-Gal4, w[*]; P(w[+mW.hs]=GawB)how[24B] (BL 1767); repo-Gal4, P(GAL4)repo (BL 7415); btl-Gal4, P(GAL4-btl.S)3-2 (BL 78328); Act5C-GeneSwitch-Gal4, P(UAS-GFP.S65T)Myo31DF[T2]; P(Act5C(-FRT)GAL4.Switch.PR)3 (BL 9431). Transgenes: UAS-Ras^V12 (3), P(w[+mC]=UAS-Ras85D.V12)TL1 (BL 64195); UAS-Ras^V12 (2), P(w[+mC]=UAS-Ras85D.V12)2 (BL 64196); UAS-Ras^V12 with RMCE site (3), P(w[+mC]=UAS-Ras85D.V12)TL1, P(w[+mC]=attP.w[+].attP)JB89B (BL 64197); UAS-GFP nuclear, P(UAS-GFP.nls)14 (BL 4775); brat^dsRNA, P(y[+t7.7] v[+t1.8]=TRiP.HMS01121)attP2 (BL 34646); UAS-p35 baculovirus death inhibitor, P(w[+mC]=UAS-p35.H)BH1 (BL 5072) and Gal80ts, w[*]; P(w[+mC]=tubP-GAL80[ts])20 (BL 7019).

This follows a detailed method, which has additional information (Simcox, 2013), except that no yeast paste is used on the egg collection plates. Yeast paste, even when sterilized, promotes contamination in the cultures. Crosses were made between the Gal-4 driver lines and UAS-Ras^V12 lines. Some Ras^V12 stocks had additional alleles as noted in Supplementary file 1. Approximately 200 males and 200 females of a cross were transferred into a laying cage, with a fluted Whatman 3MM paper insert to increase surface area, and eggs were collected using 60-mm Petri dishes containing egg laying medium. Egg collections were made during the day for 8 hr at room temperature or 16 hr overnight at 17°C. After collection, approximately 3 ml of TXN (NaCl [0.7%], Triton X [0.02%] in water) was added to the plate. Any hatched larvae, which rise to the surface, were removed and the unhatched embryos were dislodged using a large soft paint brush to gently release them from the surface. Embryos were tipped off with the liquid into a sieve. Additional rinsing and brushing were used to ensure most embryos were dislodged and collected in the sieve. After thorough rinsing of the embryos with TXN from a squirt bottle, the sieve was upended over a 15-ml Falcon tube and a stream of TXN was used to transfer the embryos into the tube. Once the embryos settled, the TXN was removed and replaced with 3 ml of 50% bleach (Clorox) in water. The tube was capped and inverted three to five times and subsequently the embryos were treated using sterile techniques. The embryos were allowed to settle at the bottom of the tube and the bleach was removed after 3–5 min. The bleach dechorionates and surface sterilizes the embryos. The embryos were rinsed 2× with 4 ml of sterile TXN and transferred to a fresh tube of TXN to minimize bleach contamination. After two additional TXN rinses the embryos were transferred to TXN in a 5-ml glass homogenizer (with Teflon pestle). Embryos were rinsed in 3 ml of water followed by a rinse in 1 ml of Schneider’s S2 medium (supplemented with 10% heat inactivated fetal bovine serum and 1× Pen-strep solution). Embryos tend to clump in the Schneider’s S2 medium and stick to the sides of the homogenizer and pipette and care is needed to remove the medium without disturbing the embryos. 3 ml of fresh Schneider’s S2 medium was added to the homogenizer and the embryos were disrupted by three gentle strokes with the pestle. Care was taken to minimize bubbles by not withdrawing the pestle beyond the surface of the liquid. The homogenate was allowed to settle for 2 min and the supernatant was transferred to a 15-ml Falcon tube leaving the large cell clumps and any whole embryos in the bottom of the homogenizer. 3 ml of fresh Schneider’s S2 medium was added to the homogenizer and three more strokes, with a twist at the bottom, were used to disrupt remaining tissue and embryos. The second homogenate was added to the Falcon tube. The tube was centrifuged in a benchtop centrifuge at 1400 × g. The supernatant was discarded, and the pellet was resuspended in 3-ml Schneider’s S2 medium and centrifugation step and washing with Schneider’s S2 medium was repeated twice more. The final pellet size was estimated and plated in 1 or more 12.5 cm² T-flasks with 2–3 ml Schneider’s S2 medium. The number of flasks needed for a given pellet size can also be estimated from the volume of packed embryos with approximately 30 µl of packed embryos being sufficient for one flask.

Cells were grown in 25 cm² T-flasks at 25°C in Schneider’s S2 medium and were passaged at between 90% and full confluence (Figure 8) using trypsin to release cells from the tissue-culture surface. Trypsin is needed as cells in all the lines are adherent except ActGSI3 cells that float freely (Figures 1 and 8). Cells were pelleted and approximately 20–25% of the cells were plated in a new flask. Cells were checked using an inverted microscope approximately every 5 days. The medium was changed on cultures showing signs of poor cell health (extended processes, little growth). This was sometimes necessary for cell types that are more metabolically active and acidify the medium, including the mesodermal lines. Cells were passaged every 5–7 days. Cell freezing (Schneider’s S2 medium with 20% heat inactivated fetal bovine serum and 10% DMSO (Dimethyl sulfoxide)) was used to keep a supply of frozen aliquots so that cells with similar passage numbers were used in experiments.

For puromycin selection, 2–6 × 10⁵ cells in a 35-mM well were transfected with 0.4 µg of DNA encoding a puro resistance plasmid (pCoPURO, Addgene #17533) using Effectene Transfection Reagent (QIAGEN). After 24 hr, cells were selected with puromycin at 0.5–2.5 µg/ml for 5 days. After 2–4 weeks, colonies were isolated and expanded. For dilution cloning, cells were seeded into a 96-well plate at a concentration of 0.5–1 cell/well in 100 µl conditioned media (Housden et al., 2015).

To simulate the major pulse of ecdysone at the larval to pupal transition, cells were treated with two 24 hr doses of β-ecdysone (Sigma 5289-74-7) at 1 µg/ml separated by 24 hr in non-supplemented medium.

Cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences) for 15 min or 3.5% formaldehyde (Sigma) for 30 min at room temperature, and then rinsed twice with 0.1% Tween-20 in phosphate-buffered saline (PBS-T). Cells were permeabilized (0.2% Triton X-100 in PBS) for 10 min at room temperature. Cells were blocked (5% bovine serum albumin in PBS-T) for 30 min at room temperature and incubated with diluted primary antibodies overnight at 4°C. Cells were washed three times with PBS-T and incubated with diluted secondary antibodies in blocking buffer for 1 hr at room temperature or overnight at 4°C. Cells were washed three times with PBS-T and mounted in VectaShield with DAPI (Vector Laboratories). For the Dcad2 antibody, cells were fixed and processed as described in Oda et al., 1994. The following primary antibodies and dilutions were used: HRP (rabbit polyclonal, Jackson ImmunoResearch 323-005-021, 1:500), 22C10 (mouse monoclonal anti-Futsch, Developmental Studies Hybridoma Bank, DSHB, 1:100), ELAV (rat monoclonal, DSHB 7E8A10, 1:100), Repo (mouse monoclonal, DSHB 8D12, 1:100), FasII (mouse monoclonal, DSHB 1D4, 1:100), Twist (a gift from M. Levine, UC Berkeley, CA, guinea pig 1:500), MHC (mouse monoclonal, DSHB 3E8-3D3, 1:100), Dcad2 (rat monoclonal, DSHB, 1:100), and DMef2 (a gift from J. R. Jacobs [Vanderploeg et al., 2012], rabbit polyclonal, 1:500), H2 (mouse monoclonal, [Kurucz et al., 2003], 1:10). Cells were incubated with the following secondary antibodies at the indicated dilutions: Cy3-conjugated goat anti-mouse (Jackson ImmunoResearch 115-165-003, 1:1000), Cy3-conjugated goat anti-rat (Jackson ImmunoResearch 112-165-003, 1:1000), Cy3-conjugated goat anti-guinea pig (Jackson ImmunoResearch 106-165-003, 1:1000), Cy3-conjugated goat anti-rabbit (Jackson ImmunoResearch 111-165-045, 1:1000), and Alexa Fluor 488-conjugated donkey anti-rabbit (Invitrogen A-21206, 1:1000).

1–2 × 10⁵ cells were plated in a 12-well plate. Cells were counted from triplicate wells every 3 days over a 9-day period. Doubling time was calculated using log2 cell numbers (Roth, 2006).

Cells were grown to 50–90% confluence and incubated with 0.05 µg/ml KaryoMAX (Gibco-Thermo Fisher 15212012) for 3–18 hr. Cells were processed for analysis using the method in Lee et al., 2014, which uses 0.5% sodium citrate as a hypotonic solution and a 3:1 ice cold mix of methanol and acetic acid as a fix. After dropping fixed cells, slides were air dried and mounted in VectaShield with DAPI (Vector Laboratories) and viewed with an Olympus BX41 microscope.

Cells in a 6-well plate (approximately 70% confluent) were transfected with 0.4 µg of an Actin5C-EGFP plasmid (pAc5.1B-EGFP, Addgene #21181) using Effectene Transfection Reagent (QIAGEN). The fraction of GFP-positive cells was scored after 48 hr.

Cell cultures were grown and expanded in their respective media. All cell lines were cultured in Schneiders Drosophila Medium (Gibco Cat # 21720001), supplemented with 10% fetal bovine serum (Cytiva Hyclone Cat SH30070.03). For Act5C-GS>Ras attP-GFP-LI-Clone 2, Act5C-GS>Ras attP-GFP-LI-Clone 3, and Act5C-GS>Ras attP-GFP-LB-Clone 6 cultures were grown in the same basal media supplemented with 10 nM of Mifepristone (Thermo Fisher Cat# H11001). Cultures were allowed to grow in T-25 flasks to become confluent before treatment with trypsin (Gibco Cat# 12604013) for 4 min to dislodge the cell monolayer from the growth surface. The cells were resuspended in 4 ml of their respective media and 1 ml of the cell suspension was collected for pelleting, followed by washing in 1× PBS, and then flash-freezing in liquid nitrogen. All cell samples were processed in triplicates.

Total RNA was isolated from the pellets using the TRIzol reagent (Life Technologies [Ambion], Cat#:15596018) as per the manufacturer’s instructions. The isolated total RNA was subjected to further purification using the RNeasy Mini Kit (QIAGEN, Cat#74104) and the RNA post-cleanup was eluted in RNase-free water. The eluted total RNA was confirmed to have a A₂₆₀/A₂₈₀ ratio >1.8 and RIN >7.

Upon passing the quality control parameters, Illumina TruSeq libraries were constructed using TruSeq stranded mRNA HT kit (Illumina, Cat# RS-122-2103). Paired end sequencing was performed on an Illumina NextSeq 500 with a 150-cycle high output kits (Illumina, Cat# FC-404-2002).

Raw data processing was performed using the STAR sequence aligner (https://github.com/alexdobin/STAR; Dobin et al., 2013). Reads were aligned to the Drosophila genome and featureCounts were used to get gene counts from all samples into a count matrix for downstream analysis. A principal component analysis plot was produced using heatmaply. FPKM values were calculated using fpkm(DEseq2) using gene length output by featureCounts. The reference genome used was FB2022_05, dmel_r6.48 (FlyBase) (Jenkins et al., 2022). Both raw sequencing reads and the count matrix were deposited in the NCBI Gene Expression Omnibus (GEO) database under the accession number GSE219105. The processed dataset has also been imported into DGET database for user to mine gene(s) of interest or search for genes with similar expression pattern (https://www.flyrnai.org/tools/dget/web/).

Each sample was compared against all other samples by using DESeq2 ( Love et al., 2014) to determine differentially expressed genes (DE calling). The set of top DE genes for each cell line was compared with the top 100 markers in single-cell RNAseq datasets corresponding to cell types in the Fly Cell Atlas 10× datasets (Li et al., 2022). Enrichment analysis was conducted using the DRscDB tool to identify the Fly Cell Atlas cell type that matched closely to each cell line (Hu et al., 2021). We also compared the DE genes with the genes identify in various tissues in embryo and larval based on in situ data (PMID: 24359758, 17645804, 12537577) and majority of the best matching tissues are consistent with the analysis using scRNAseq datasets (Table 2).

The RNAseq data for the cell lines described in this work were also compared with RNAseq datasets determined previously for 24 other Drosophila cell lines (Cherbas et al., 2011). The comparison was conducted by hierarchical clustering analysis using Pearson correlation coefficient scores. To survey the activities of major signaling pathways in the cell lines, we specifically selected the ligands and receptors annotated at FlyPhoneDB (PMID: 35100387) to plot their expression levels using heatmap.

All cell lines described here have been deposited to the Drosophila Genomics Resource Center (DGRC) at Indiana University. The lines are available for distribution to the research community.