Growth deregulation and interaction with host hemocytes contribute to tumor progression in a Drosophila brain tumor model

Significance How tumor-intrinsic factors and microenvironment shape brain tumor progression remains an open question. Here, we present a transcriptomic, cellular, and genetic dissection of a neural tumor generated by Notch hyperactivation in Drosophila. We have focused on the differences between the larval primary tumor and its more aggressive version, which emerges soon after transplantation to adults. This has provided insights on tumor growth strategies, like the involvement of Myc, Imp, and the insulin receptor pathway. We found that host macrophages profusely infiltrate the allografted tumor and impede its growth through phagocytosis. Surprisingly, cytokines (TNF, Jak/STAT) that often mediate macrophage-epithelial tumor signaling were not activated in this brain tumor. Our findings contribute to a better understanding of tumorigenesis strategies and tumor–microenvironment interactions.

permitted to drive the expression of UAS transgenes. Subsequently, larval CNS were dissected out and used for further experiments.
For the functional experiments with Myc, InR (constitutively active and dominant negative forms), and Imp, the respective transgenes were combined with UAS-NΔecd or UAS-LacZ (control) lines and act-F/O tumours were generated as described above. A neutral transgene, usually UAS-w-RNAi, was used as a negative control to balance the dosage of total UAS transgenes per genotype.
For quantitatively scoring hyperplasias in the larval CNS, the experimental regimens were slightly altered to create milder hyperplasias (thus allowing easier manual counting of the hyperplastic vs normal clones); for the act-F/O system, larvae of the appropriate genotypes underwent heat-shock for 35min at 37°C (72 hours AEL) and CNSs were dissected out 3 days later.
For collecting larval haemocytes, hmlΔGal4, UAS-2xEGFP larvae were used. For collecting adult haemocytes, hmlΔGal4, UAS-2xEGFP flies were crossed to appropriate responder UAS lines and their F1 progeny were bled as detailed in the "Live Imaging" section.
Host flies for the haemocyte ablation and the RNAi screen experiments were prepared as follows: female flies from hmlΔGal4, UAS-2xEGFP (BL#30140) were crossed with UAS-hid/CyO,Tb, or UAS-lacZ or the respective RNAi lines obtained from Bloomington (Materials table). Crosses were maintained at 18°C to keep Gal4 activity low during larval/ early pupal stages (hmlΔGal4 is not expressed at the embryo stage, anyway [3]). At late pupal stage (16-18 days after egg laying), progeny was shifted to 29°C to boost Gal4 activity. Adult flies circa 5-6 days after adult eclosion were injected with PBS or transplanted with 500 control brain cells or 500 grh ts >UAS-NΔecd T0 or T2 cells (haemocytes ablation experiments). Adult flies either challenged as described above or unchallenged were monitored until their death.

Transplantation procedure and survival assay
Transplantations were performed as previously described by [4] and as recently summarized in [5]- [7]. To generate brain tumours, we utilized either the act-F/O or grh ts system to either drive UAS-NΔecd, UAS-LacZ (control) or combinations of transgenes (UAS-NΔecd + UAS-RNAi lines for genes of interest) as described above. Larval CNSs were dissected and sliced into individual brain lobes. Each brain lobe was loaded into a fine glass needle and transplanted into the abdomen of adult female fly hosts (w 1118 , unless otherwise stated) using a Nanoject II Auto-Nanoliter Injector (Drummond Scientific Company; Cat# 3-000-205A). All female fly hosts were of similar age at the day of the transplantation i.e 3-4 days after adult eclosion except for the host flies used for the haemocyte ablation and the RNAi screen experiments which were all 5-6 days after adult eclosion to allow haemocytes to complete the clearance of larval fat body cells and relocate to adult tissues [8]. After the transplantation, flies were placed in fresh vials and were kept at 25°C (act-F/O system) or 29°C (grh ts system) in a horizontal position. Fly food was renewed every 2-3 days to keep injected flies clean and reduce the load of potentially harmful germs growing in the vials. Hosts were macroscopically examined on a daily basis for viability (survival assay) and GFP/RFP signal detection in the abdomen or other parts of their body (thorax, head) under an epifluorescent stereoscope. Animals that died within three days post injection were excluded from the survival analysis as mortality could be attributed to damage during the injection procedure and not by the graft. Occasionally, some host animals continued to die up to days 7 post injection.
When macroscopically examined post mortem, they had developed some tumour burden which may or may not have been the cause of their death. For the survival assays, we only scored hosts that developed detectable tumour (GFP/RFP), unless otherwise stated. For control animals bearing no insult we did not exclude any deaths from the analysis. Even in these control cases, occasional animals died during the 10 first days post eclosion in some replicates for reasons that we do not understand. The day of death of each fly host was depicted as an individual dot in Survival Scatter plot diagrams (see below the "Quantification and Statistical analysis" section).
5-15 days after transplantation, tumour (derived from the transplantation of the hyperplastic brain tissue; T0) was dissected out of the fly host into a glass slide filled with 1X sterile PBS by performing an epidermal incision in the abdominal area. The material was either used for retransplantation (serially for up to three more passages-T3), immunohistochemistry and live imaging experiments or was stored in Trizol for RNA extraction and library preparation for RNA seq analysis. To ensure that a consistent number of tumour cells was injected in the allograft passages (T1-T3), tumour pieces from the previous stage (Tn-1) were dissected out, dissociated in sterile PBS to single cell suspension (mechanically, via repeated aspiration through an insulin needle), counted on a haemocytometer and resuspended in filtered PBS at the desirable cell density. For the RNAi screen, we injected 500 tumour cells or sterile PBS at 36.8 nl per fly.

Immunohistochemistry
Fixation and immunohistochemistry of larval tissues were performed according to standard protocols [9]. Allograft tumours were fixed and stained as previously described [5]. After separating the abdomen from the rest of the fly body into a glass plate filled with 1X PBS, a slight cut was performed ventrally and fixation into 4% formaldehyde followed for 35 min at RT. After blocking in PBT (1X PBS, 5% BSA, 0.1% Triton) for at least 2h or O/N, tissues were incubated in primary antibody O/N at 4°C in a 96-well plate. After three 10-minute washes with 1X PT (1X PBS, Triton 0.1%), an O/N incubation with secondary antibodies and Hoechst followed. After breaking up the abdomen to fragments of internal organs mixed with tumour, we mounted the samples in glass microscopy slides in mounting medium (80% glycerol with 0.5% N-propyl gallate).
Primary and secondary antibodies used are described in the Materials table. Samples were imaged using a Leica SP8 confocal microscope at the FORTH-IMBB confocal imaging facility.

Live imaging
For larval brain co-cultures with larval haemocytes: Larval brains were dissected in Complete

For tumour explant cultures and co-cultures with adult haemocytes:
Tumour explants were dissected out on a glass slide in 300μl culture medium and were gently dissociated to clumps by pipetting up and down using a yellow pipette tip. Explants were either imaged immediately after isolation (to observe the tumour-resident haemocytes) or were co-cultured with freshly isolated adult haemocytes. Adult haemocytes were collected with perfusion as previously described [10].
An insulin syringe was loaded with culture medium and a fine glass capillary needle (same needle used for the nanoinjector) was attached to the needle. A small incision was made at the posterior ventral side of the abdomen of a cold anaesthetized female and the tip of the glass capillary was inserted into the lateral thorax. Culture medium was gently perfused in the fly and 5 drops/fly coming out from the abdominal cut were collected on a Mattek dish. A total of ten adult flies were bled for each experiment. Tumour suspension was subsequently (10-15 min later) added on top of the haemocyte population prior to imaging. Samples were immediately imaged for 4-6 hours on a Leica TCS SP8 confocal microscope (FORTH-IMBB).
Whole fly extract was prepared in 1X Passive Lysis Buffer using Kontes pestles and Eppendorf tubes. The extracts were centrifuged twice at 10.000g for 15 minutes at 4°C and the supernatant was stored at -20°C. Luciferase activity was measured using a Luminometer (Turner Designs; Cat# TD-20210) by adding Luciferin substrate (LARII; provided in the kit) to each sample.
Normalization of Luciferase levels was performed by measuring total protein content in the samples by Bradford assay.

Cryosections
Adult tumour-bearing flies were anesthetized on ice and quickly washed in EtOH on a glass slide.
To create an opening for the fixative, flies were placed on 1X PBS and the proboscis was removed.

FACS purification, RNA prep, RNA-seq library
Neural stem cell-like cells from larval brains were dissociated and isolated by FACS according to published protocols [13] and as previously described in [5]. NGS libraries were generated using total RNA as input with polyA mRNA magnetic isolation kit (NEB) and the NEB Ultra II RNA library kit for Illumina kit according to manufacturer's protocol, using 13 cycles of amplification. Libraries were sequenced on Illumina Nextseq 500 on 1 x 75 High flowcell.

Quantification and statistical analysis
Quality control checks on raw fastq sequence data was performed by FastQC [14] (Version 0.11.9).
Generated fastq files were aligned to the Drosophila melanogaster reference genome (Genome assembly: BDGP6.32) using STAR [15](Version 2.7.5), while raw counts were created using the For Figure S3, differentially expressed genes with |log2FC| ≥ 0.5, padj ≤ 0.05 were used as an input to generate heatmaps to search for similarities with microarray transcriptome data from other allograft brain derived tumours originating from genetic insults in the asymmetric cell division machinery [20]. They were also used as an input to generate venn diagrams to look for similarities with genes enriched in normal NSCs and neurons [21].

Figure S1: NΔecd tumours have lower number of differentiating cells upon serial transplantations and the reduction in host lifespan is proportional to the amount of
NΔecd tumourigenic cells injected and temperature dependent.    Note that wt lineages occasionally contain two Dpn-positive cells, as early GMCs have not downregulated Dpn yet [27]. We did not observe any NB loss upon the Myc RNAi with the actF/O driver, however we did notice that the NBs often became smaller, in agreement with [28].     Deposited data [20] Microarray Data Sup. Material info Table S2 [2] ChIP data GSE68614 [6] ChIP data GSE141794 [5] RNA seq data GSE179507

Movie Legends
Movie S1: grh ts >NΔecd tumour cells (magenta) raised in an hml>GFP+lacZ host (green haemocytes) for 10 days prior to explanting and imaging live. Timelapses were captured every 2min for a duration of circa 4 hours. Top: fluorescent image superimposed on brightfield to visualize cell morphology. Scale bar 20μm. Movie is related to Fig.5E.
Movie S2: grh ts >NΔecd allograft tumour cells (magenta) co-cultured with naïve isolated adult haemocytes (green) from hmlΔ>GFP+lacZ (control). Timelapses were captured every 3min for a duration of circa 4-6 hours. Top: fluorescent image superimposed on brightfield to visualize cell morphology. Note the highly motile filopodia emanating from the haemocytes. Scale bar 25μm. Movie is related to Fig.7A.