Intratumoral and extratumoral synapses are required for glioblastoma progression in Drosophila

Glioblastoma (GB) is the most aggressive, lethal and frequent primary brain tumor. It originates from glial cells and is characterized by rapid expansion through infiltration. GB cells interact with the microenvironment and healthy surrounding tissues, mostly neurons and vessels. GB cells project tumor microtubes (TMs) that contact with neurons and exchange signaling molecules related to Wingless/WNT, JNK, Insulin or Neuroligin-3 pathways. This cell to cell communication promotes GB expansion and neurodegeneration. Moreover, healthy neurons form glutamatergic functional synapses with GB cells which facilitate GB expansion and premature death in mouse GB xerograph models. Targeting signaling and synaptic components of GB progression may become a suitable strategy against glioblastoma. In a Drosophila GB model, we have determined the post-synaptic nature of GB cells with respect to neurons, and the contribution of post-synaptic genes expressed in GB cells to tumor progression. In addition, we document the presence of intratumoral synapses between GB cells, and the functional contribution of pre-synaptic genes to GB calcium dependent activity and expansion. Finally, we explore the relevance of synaptic genes in GB cells to the lifespan reduction caused by GB advance. Our results indicate that both presynaptic and postsynaptic proteins play a role in GB progression and lethality.


Glioblastoma (GB) is the most lethal and aggressive tumor of the Central Nervous
System. GB has an incidence of 3/100,000 adults per year (TAMIMI and JUWEID, 2017), and accounts for 52% of all primary brain tumors. GB originates from glial cells or glial progenitors and causes death within 16 months after diagnosis (Bi and Beroukhim, 2014) due to the low efficacy of standard treatments such as chemotherapy, radiotherapy or surgical resection.
In the last decade, Drosophila melanogaster has emerged as a reliable in vivo GB model that reproduces the features of human GB (Jarabo et al., 2021;Kegelman et al., 2017;Portela et al., 2019a;Read, 2011;Read et al., 2013). The GB condition is experimentally elicited by the expression of constitutively active forms of EGFR (Epidermal Growth Factor Receptor) and PI3K (Phosphoinositide 3-kinase) in glial cells, which are the two most common mutations in patients (Read et al., 2009). This experimental model has been previously used to study the contribution of RIO kinases (Read et al., 2013), vesicle transport (Portela et al., 2019a), the human kinase STK17A orthologue (Drak) (Chen et al., 2019;Lathia, 2019), and several metabolic pathways in GB (Chi et al., 2019). Consequently, Drosophila model of GB is well characterized and suitable to study cellular properties of GB in vivo.

Tumor microenvironment and the communication between tumoral cells and neurons
are crucial for GB progression and patient survival (Casas-Tintó and Portela, 2019;Jarabo et al., 2021;Orgazy et al., 2014;Portela et al., 2019bQu et al., 2019). In addition, neuronal activity can also stimulate GB growth. Activity-dependent release of neuroliglin-3 (NLGN3) is required for GB progression in xenograft models, and NLGN3 induces the expression of synaptic proteins in glioma cells (Venkatesh et al., 2017).
Moreover, GB samples show synaptic gene enrichment (Venkatesh et al., 2019) and glioma cells form functional glutamate synapsis with neighboring neurons, where GB cells are post-synaptic (Venkataramani et al., 2019;Venkatesh et al., 2019). These studies also demonstrated that the pharmacological or genetic inhibition of these electrical signals reduces growth and invasion of the tumor (Venkataramani et al., 2019;Venkatesh et al., 2019).
Synapses are the functional units which underlie animal behavior, memory and cognition. Chemical synapses are specialized asymmetric junctions between a presynaptic neuron and a postsynaptic target with different molecular composition, structure, and activities. Bruchpilot (Brp), Liprin alpha (Lip α ) and Synaptotagmin 1 (Syt1) are conserved proteins localized in the presynaptic side.

Brp is a well studied component of the presynaptic component of the synapses in
Drosophila that accumulates in mature active zones (AZ). Brp is the orthologue of human AZ protein ELKS/CAST/ERC, and it is required for synapse formation (Wagh et al., 2006). Lip α is a presynaptic scaffolding protein, orthologue to several human genes including PPFIA1 (PTPRF interacting protein alpha 1) and PPFIA2 (PTPRF interacting protein alpha 2). Lip α directly interacts with tyrosine phosphatase receptors and it is involved in synapse formation, anterograde synaptic vesicle transport, neuron development, synapse organization and axon guidance (Astigarraga et al., 2010;Fouquet et al., 2009;Kaufmann et al., 2002). Finally, Syt 1 is a pre-synaptic vesicle calcium binding protein that functions as the fast calcium sensor for neurotransmitter release at synapses (Yoshihara and Montana, 2016).
Synapses elicit neurotransmission by mediating the clustering and fusion to the plasma membrane of neurotransmitter containing vesicles which release into the synaptic space (Chou et al., 2020). The postsynaptic side is characterized by the accumulation of neurotransmitter receptors, including Glutamate receptors (GluR), the protein discs large (Dlg), orthologue of human PSD95 protein which mediates the clustering of postsynaptic molecules (Koh et al., 1999), and Synaptotagmin 4 (Syt 4), a vesicular calcium binding protein, directly implicated in retrograde signaling at synapses. Syt 4 is proposed to regulate calcium-dependent cargo trafficking within the postsynaptic compartment (Harris et al., 2016).
Benefitting from the conserved nature of most synaptic components, we set out to dissect the pre-versus post-synaptic contributions to GB progression using a Drosophila model of the human disease, in which the pathological condition of each cell type can be genetically manipulated. Thus, in addition to demonstrate that neuronglioblastoma synaptogenesis is a conserved mechanism in GB progression, we show that synapses are also formed intratumoral and identify several synaptic genes required for GB expansion and premature death.

Fly Stocks
Flies were raised in standard fly food at 25ºC, otherwise indicated.

Inmunohistochemistry
Third-instar larval brains, were dissected in phosphate-buffered saline (PBS), fixed in 4% formaldehyde for 30min, washed in PBS + 0.1 or 0.3% Triton X-100 (PBT), and blocked in PBT + 5% BSA for 1 hour. Samples were incubated overnight with primary antibodies diluted in block solution, washed in, incubated with secondary antibodies diluted in block solution for 2 hours and washed in PBT. Fluorescent labeled samples were mounted in Vectashield mounting media with DAPI (Vector Laboratories).

TEM
Transmission electron microscopy (TEM) was performed in CNS of 3rd instar larvae with horseradish peroxidase (HRP) genetically driven to glial cells (repo-Gal4:UAS-HRP CD2). Brains were fixed in 4% formaldehyde in PBS for 30 min at room temperature, and washed in PBS, followed by an amplification of HRP signal using the ABC kit (Vector Laboratories) at room temperature. After developing with DAB, brains were washed with PBS and fixed with 2% glutaraldehyde, 4% formaldehyde in PBS for 1h at room temperature. After washing in a phosphate buffer the samples were postfixed with OsO4 1% in 0.1 M 7phosphate buffer, 1% K3[Fe(CN)6] 1h at 4ºC. After washing in dH2O, Brains were incubated with tannic acid in PBS for 1min at room temperature then washed in PBS for 5min and dH2O 2x5min. Then the samples were stained with 2% uranyl acetate in H2O for 1h at room temperature in darkness followed by 3 washes in H2O2d. Brains dehydrated in ethanol series (30%, 50%, 70%, 95%, 3x100% 10 min each at 4ºC). Infiltration: samples were incubated in EtOH:propylene's OXID (1:1;V.V) for 5 min, propylene's OXID 2x10min, propylene's OXID:Epon (1:1) for 45 min, Epon 100% in agitation for 1 h and Epon 100% in agitation overnight. Then change to Epon 100% for 2-3 h. After, the samples were encapsulated in BEEM and incubated 48h at 60ºC for polymerization. Finally, the samples were cut in ultra-fine slices for TEM imaging (Martín-Peña et al., 2014).

Imaging
Fluorescent images were acquired by confocal microscopy (LEICA TCS SP5) and were processed using Fiji (Image J 1.50e). These images were quantified with Fiji (Image J 1.50e) or Imaris 6.3.1 (Bitplane) software. The images of the ultra-fine slices were taken with a Transmission electron microscopy JEM1010 (Jeol) with a CMOS TemCam the number of synaptic active sites was quantified by using the spots tool. The tumor volume was quantified using the surface tool. We selected a minimum size and threshold for the punctae or surface in the control samples of each experiment to establish the conditions. Then we applied the same conditions to the analysis of each corresponding experimental sample.
Fiji quantification: -CaLexA expression: We used the NFAT-based neural tracing method-CaLexA (calcium-dependent nuclear import of LexA)-for labeling active neurons in behaving animals. CaLexA (green) signal intensity where determined using ImageJ to calculate the mean grey value of each brain lobe.
-Syt1-GFP expression: Syt1-GFP (green) signal and glia membrane myrRFP (red) signal intensity were determined using ImageJ (mean grey value) in three single slices at the middle of each brain lobe to calculate the ratio GFP/RFP. All samples were treated, acquired and measured under the same conditions and in parallel -GRASP: We used a modified version of this system to specifically detect synaptic contacts. It is based on the fusion of synaptobrevin protein (Syb) to the 1-10 fragment of GFP (Syb-GFP 1-10 ), and the expression of a membrane bound form of the 11 fragment of GFP (CD4-GFP 11 ).

Viability assays
Flies were crossed at restricted temperature (17 °C, to inactivate the UAS/Gal4 system with tub-Gal80ts) for 4 days then progeny was transfer at 29°C (when the UAS/Gal4 system is active and the glioblastoma develops). The number of adult flies emerged from the pupae were counted for each genotype. The number of control flies was considered 100% viability and all genotypes are represented relative to controls.
Experiments were performed in triplicates.

Survival assay
For survival analyses of adult flies, males and females were analyzed separately. 0-5 day old adult flies raised at restricted temperature were put at 29°C in groups of 10 animals per vial and were monitored blinded every 2-3 days; each experiment was done at least three times.

Quantifications and Statistical Analysis
All experiments including different genotypes were done in parallel under the same experimental conditions, with the exception of viability analysis where each genotype was normalized with their parallel control. Data were analysed and plotted using

Results
We performed a Drosophila biased genetic screening to search for relevant genes related to GB progression. We selected 2000 genes involved in cell to cell communication, and we used VDRC UAS-RNAi lines to knockdown the expression of such genes encoding transmembrane, secreted and cell to cell communication proteins. In addition, we used the previously validated EGFR/PI3K model Portela et al., 2019bRead et al., 2009). GB induction in larvae causes premature death and animals do not reach adulthood. We took advantage of this unequivocal phenotype as a read-out, quantifying the number of adult flies that emerged from each experiment. We obtained 25 RNAi lines that rescued the lethality caused by the GB. Among the suppressors, we found well known mediators of GB progression such as Frizzled1 (Fz1) or Gryzun (Gry) receptors and PI3K signaling pathway members (Portela et al., 2019b(Portela et al., , 2019a. These genes validate the experiment as positive controls. Most RNAi lines, as well as negative controls UASyellow RNAi or UAS-beta-galactosidase (lacZ), did not rescue GB-induced pupal lethality however, we found genes encoding synaptic proteins, such as liprin α (lip α ) and synaptotagmin1 (Syt 1) that rescue GB-induced lethality ( Figure 1A). These results motivated this study to determine the contribution of synaptic components to GB progression.

Neurons produce synaptic contacts with glioma cells.
It was recently described that neurons establish functional synapses with glioblastoma cells in mouse xenografts (Venkataramani et al., 2019;Venkatesh et al., 2019). In these studies GB cells are postsynaptic, however our results from the screening indicate that presynaptic proteins are also involved in GB-induced lethality ( Figure 1A).
Therefore, we wondered if GB cells were pre-or postsynaptic in the Drosophila GB model. We used the GFP reconstitution across synaptic partners (GRASP) technique (Macpherson et al., 2015) to determine synaptic contacts between GB cells and Cell proliferation in GB is associated with calcium-mediated activity (Venkataramani et al., 2019), thus we analyzed the contribution of dlg or GluRIIA to calcium activity in GB cells. To monitor calcium activity, we used the CaLexA system (see Materials and Methods). Quantification of CaLexA signal showed a significant increase of calcium signal in GB samples, but dlg knockdown in GB cells maintained calcium levels as controls (Figure 2 G). However, we did not find significant differences upon GluRIIA downregulation. These results indicate that GB cells show enhanced calciumdependent activity, in line with previous data from other GB models (Venkatesh et al., 2019). Moreover, our data indicate that this enhanced calcium activity is dependent on dlg expression, while independent on that of GluRIIA.

Vesicle calcium binding proteins are required for GB progression
Vesicle calcium binding, vesicle transport and neurotransmitter release are cellular mechanisms related to synaptic function (Kavalali, 2015;Südhof, 2013). We have found that GB has an enhanced calcium activity that can be reduced by downregulating the expression of dlg. Moreover, our screening results indicate that downregulation of Syt 1, which encodes a presynaptic Calcium-binding protein, partially rescued the lethality caused by the GB (Figure 1 A). This motivated the study of These data unveil a role of these synaptic proteins in the biology of normal glial cells that was unknown hereto.

Intratumoral synapses in GB
The results obtained so far indicate the presence of presynaptic proteins in GB cells.  (Figure 6 A). Also, the knockdown of Syt1 also prevents GB-induced lethality (Figure 6 B) suggesting that the expression of these pre-synaptic genes is required in GB cells to cause premature death.
In addition, we analyzed the contribution of GluRII, dlg and Syt 4 to life span in GB. The results show that GluRIIA or Syt4 knockdown in GB cells, expands the life span of animals with GB, but the expression of dlg does not modify the premature death caused by GB progression (Figure 6 B-C).

Discussion
In addressing the mechanisms that facilitate cell to cell communication in GB progression, we found genes that encode for synaptic proteins and are required for GB However, dlg knockdown in GB cells shows a particular phenotype, dlg RNAi prevents brain volume expansion and GB cells number increase, and also attenuates the synapse loss caused by GB. However it is not sufficient to prevent premature death caused in GB. It is tempting to discuss the contribution of Dlg to GB progression, and furthermore, the requirements of Dlg in glial cells for the normal function. Dlg protein is involved in post-synaptic structures, but also in cell polarity, neuronal differentiation and organization, and septate junctions in cellular growth control during larval development.
In addition, Dlg contains a guanylate kinase domain that suggests a role in cell adhesion and signal transduction to control cell proliferation (Albertson and Doe, 2003;Koh et al., 1999;Li et al., 2009;Maiya et al., 2012;Ohshiro et al., 2000;Zhang et al., 2007). In consequence, dlg knockdown could affect a number of cellular functions that reduces life span, independently of the prevention of GB progression.
The multiple functions of many proteins has recently emerged as a novel point of view in biology, and reconciles the complex mechanisms behind cellular physiology, and the limited number of known genes. For example, Troponin-I was described as a central player in muscle formation, but recent discoveries show that Troponin-I is also involved in apico-basal polarity, chromosome stability and tumorigenesis (Casas-Tintó and Ferrús, 2019;Casas-Tintó et al., 2016;Sahota et al., 2009). Moreover, Caspases are another example of multi-functional proteins involved not only in apoptosis, but also in cancer progression and quiescence (Arthurton et al., 2020;Baena-Lopez, 2018;Baena-Lopez et al., 2018). Therefore, the impact of Dlg in GB and viability goes in line with this multiple functions of proteins, and brings a complex scenario that will require further study.

Pre-synaptic genes are required for GB progression
In addition, we investigated the contribution of synaptic genes that encode for presynaptic proteins, such as Lip α , Syt 1 and Brp. Lip α and Syt 1 appeared as hits in the unbiased genetic screening that we performed to search for anti-GB strategies, suggesting a contribution for pre-synaptic genes in GB. However, GB does not form any pre-synaptic structure with regards to the synapses established with neurons according to GRASP results. Therefore, we discarded that the pre-synaptic components in GB are related to GB-neuron synapses. In consequence, we propose for the first time that GB cells establish intratumoral synapses and these are required for GB aggressiveness.
In particular, we have evaluated the contribution of Syt 1, Lip α and Brp pre-synaptic proteins. We found that all of them contribute to calcium signaling enhancement in GB.
Additionally we demonstrated that GB cells upregulate In spite of all these results, we cannot determine if intratumoral synapses function as glutamatergic synapses described in the nervous system. Our results indicate that synaptic proteins are relevant for calcium waves, which are associated to activity in glial cells, GB progression and negative consequences, and we have observed the accumulation of proteins comparable to functional synapses, but the precise mechanisms that underlie intratumoral synapses will have to be determined in the future. It is proposed that synaptic proteins have ancestral functions related to secretion of neurotransmitters in chonoflagellates, and conserved through metazoans.
Moreover, studies on postsynaptic proteins in choanoflagellates revealed unexpected localization patterns and new binding partners, both which are conserved in metazoans (Alié and Manuël, 2010;Burkhardt, 2015). Thus it is possible that by downregulating synaptic proteins we were disrupting cytoneme formation and hence preventing GB growth.
On the other hand, breast to brain metastasis is driven by activation of N-methil-Daspartate receptors (NMDAR) through glutamate ligands. Metastatic tumor cells do not produce sufficient glutamate ligands to induce signaling, which is achieved by the formation of tripartite synapses between cancer cells and neurons (Zeng et al., 2019).
Besides it has been shown that samples of human cancers such as