Summary
The ecological benefits of sociality in gregarious species are widely acknowledged. However, only limited data is available on how the social environment influences non-communicable disease outcomes. For instance, despite extensive research over the past decades, the role of the social environment on cancer progression remains unclear and controversial. This is mainly because epidemiological studies suffer from the complexity of inter-correlated factors and it is still unknown whether distinct social group composition can also differentially affect tumor growth. Here, we exposed adult Drosophila with colorectal-like tumors to different social environments. We show that both cancerous flies bred in complete isolation, or in a group with non-cancerous individuals, exhibit increased tumor progression compared to those bred with other cancerous conspecifics. Based on video-tracking and social interaction analyses, we propose that this dramatic effect may be a consequence of perceived social isolation due to differential social interaction rates. We found that flies can discriminate between individuals at different stages of tumor growth; control flies actively avoid flies with cancer but only at the later stages of tumor development, whereas cancerous flies display strong social interactions with cancerous flies in the early stages of tumor growth. Our study demonstrates the reciprocal links between cancer and social interactions, as well as highlighting how sociality impacts health and fitness in animals and its potential implications for disease ecology and ecosystem functioning
Highlights
While it is well established that social life offers many fitness benefits, the influence of social environment on non-transmissible diseases has rarely been considered.
Here we used a Drosophila model to explore the reciprocal links between social environment, tumor development and behavior in individual flies.
We found that flies kept in isolation developed tumors at a faster rate than those kept in groups and more surprisingly that the identity of group members can significantly affect tumor progression.
Flies can discriminate between individual states at different stages of tumor growth; control flies actively avoid flies with cancer but only at the later stages of tumor development, whereas cancerous flies display strong social interactions with flies in the early stages of tumor growth
Our findings bring new perspectives to the importance of social structure on non-transmissible disease progression which may have important consequences for animal health.
Introduction
In gregarious species sociality not only offers important positive benefits associated with reducing predation risk [1] and increasing foraging efficiency [2], but also provides additional adaptive benefits by reducing overall metabolic demand [3], providing thermal benefits [4], decreasing stress responses [5] and increasing disease avoidance [6]. It is therefore, generally accepted that an individual’s social environment affects a large range of behavioral, psychosocial, and physiological pathways. Limited empirical evidence (mostly based on human studies) suggests that extreme social environments such as complete isolation or overcrowding of conspecifics in a group can potentially induce and accelerate pathological disorders. For example, in mammals, social isolation has been associated with faster progression of type 2 diabetes [7], cardiovascular or cerebrovascular disorders [8], and, notably, early and faster mammary cancer development [9, 10]. While, social overcrowding has been found to induce psychiatric and metabolic disorders [11]. Few human studies have attempted to explore the role of social interactions on cancer progression, and the topic remains controversial. Adverse psycho-social factors, including traumatic life events, high levels of depressive symptoms, or low levels of social support, have been related to higher rates of, for example, breast and colon cancers [12, 13]. However, these community based studies or meta-analyses often suffer from the complexity of inter-correlated factors. For example, low sample sizes, high risk behaviors associated with stress (e.g. smoking), and the heterogeneity and retrospective origins of these studies make it difficult to find a conclusive causal relationship between cancer progression and social conditions.
In addition to human studies, laboratory based experiments on gregarious species (e.g. Sprague-Dawley rats) have demonstrated an association between persistent social isolation and inflammatory responses linked to numerous disease processes, including cancer [9, 10, 14]. Despite cancer (both transmissible and non-transmissible ones), being an emerging important factor influencing life history traits [15-18], little is known regarding the reciprocal links between the social environment and the development and progression of this fitness reducing illness. Thus, it is still largely unclear for both animals and humans how specific social group composition can directly affect tumor progression, and vice versa.
Drosophila has proven to be a powerful model system to address these issues, since behavioral and physiological processes have been found to be influenced by the degree of social interaction while eliminating all other confounding variables. In Drosophila, social isolation leads to a reduced lifespan [19], increased aggression [20-22], reduced need for sleep [23, 24] and a decrease in the fiber number of the mushroom bodies in the integrative nervous center [25].
Here, to explore the reciprocal relationship between social environment and cancer progression, we first exposed adult Drosophila females with colorectal-like tumors to various social environments for 21 days and measured tumor growth and social interactions. Subsequently, we tested control and cancerous flies for their social environment preferences predicting that cancer flies should presumably prefer the social environment which limits cancer progression.
Results
Biological model
The flies used contained heat shock-induced MARCM clones [26] induced in 3-day old adult virgin females intestinal progenitor cells. The clones were mutant for both Drosophila APC genes, Apc and Apc2, and expressed the oncogenic form of Ras, RasV12 and the GFP marker (Apc-Ras clones) [27]. It has been shown that these compound Apc-Ras clones, but not clones either expressing RasV12 or mutated for the APC genes, expand as aggressive intestinal tumor-like overgrowths that reproduce many hallmarks of human colorectal cancer [27]. The number of GFP-positive gut cells was monitored with age every 7-days by flow cytometry from flies bearing either Apc-Ras or neutral clones, hereafter referred to as cancerous and control flies, respectively. As expected, a clear increase in the number of GFP-positive tumor cells was observed three weeks after clone induction (Supplementary Fig. 1 F1,33=8.6; P=0.006). Gut dissections confirmed the presence of tumors in the gut (Supplementary Fig. 2). The presence of tumor cells (Apc-Ras clones) had little impact on fly performance and survival over the three weeks of the experimental study [27] (Supplementary Fig. 3)
Cancer progression and social environment
To investigate the impact of social environment on tumor progression we exposed adult cancerous females for 21 days, post induction, to various social environments in 40ml food tubes. Individual virgin cancerous females were either bred in tubes alone (social isolation), in groups composed of seven other cancerous flies (homogeneous groups) or in groups with seven non-cancerous control females (heterogeneous groups). Groups of eight control flies were used as a reference (homogeneous group). Tumor growth was significantly affected by the breeding social environment (Wald , P=0.031) After 21 days we observed that tumor growth was dramatically higher in cancerous flies kept in isolation than in cancerous flies kept in homogeneous groups (Fig. 1). More surprisingly, we also observed that cancerous individual flies kept within a group of control flies showed an increased number of tumor cells compared to cancerous flies grouped together (Fig. 1).
Social interactions
we then analyzed how social interactions were affected by tumor progression and group composition. Using a video tracking setup, we followed the locomotion and interactions of groups of flies (3 weeks post induction) placed in an arena for 1 hour. For social interaction measures we used homogenous groups of eight control or eight cancerous flies, and a heterogeneous group consisting of seven control and one cancerous fly which were kept together for 21 days post induction. Social interaction analyses confirmed that control and cancerous flies had similar locomotor activity independent of their social environment (Fig. 2A; log (trail length): social environment: F1,84 = 2.64, P = 0.1; fly state: F1,84 = 0.13, P = 0.7; fly state x social environment: F1,84 = 3.8, P = 0.061). However, the length of interaction a fly had with another strongly diverged according to group composition. In homogeneous groups, cancerous flies had longer interactions compared to homogenous control groups (Fig. 2B; F1,34 = 31.10; P<10−3). Control flies showed the same contact duration whether in homogeneous or heterogeneous groups (Fig. 2B; F1,42=1.1; P = 0.3), whereas cancerous flies showed a strong decrease in contact duration in the presence of control flies (Fig. 2B F1,42=25.3; P<10-3). Flies in the control group had the same number of contacts when in the homogenous and heterogeneous groups, while cancerous flies, once again, showed a decrease in the number of contacts when placed with control flies (heterogeneous group) compared to when in a group with other cancerous flies (Fig. 2C control flies: F1,42=2.3; P = 0.13; cancerous flies: : F1,42=14.08; P<10−3). We thus concluded that, for a cancerous fly, the type of social group composition strongly affects the level of social interactions. However, our measure of social contact was constrained by the small size of the arena and therefore did not allow us to disentangle the direction of the social contact i.e which fly showed avoidance and which fly showed attraction.
Cancer progression and social environment choice
Based on the results described above we tested whether cancerous and/or control flies would show variation in social environment choice depending on the level of tumor progression. Using a similar protocol to Saltz [28], we assessed social preference by putting two small mesh cages, each containing 8 “stimulus flies” (cancerous or control) in a plastic, transparent box. The small mesh cages were placed on top of a small petri-dish containing standard food. We introduced a “target fly” (cancerous or control) in the enclosed box and recorded their position over 7h i.e. whether the fly was found on one of the two mesh cages. Target and stimulus flies were tested at different ages post heat-shock induction.
When target flies were given the choice between a cancerous or a control stimulus group of the same age, cancerous and control flies tended to show difference choice behaviors (Fig 3; target fly: Wald , P=0.04; age: Wald , P=0.09; age x target fly: Wald , P=0.1). Cancerous flies appeared, on average, more attracted than control flies to other cancerous individuals. This was especially pronounced when flies were young i.e. at the very beginning of the tumor development. However, at later stages of tumor growth, cancerous flies showed stronger preference towards control individuals (age: Wald , P<10-3). On the other hand, control flies showed no preference for either group until the third week of cancer growth, where a strong preference for other control flies was observed (Fig 3).
To understand whether these preferences, in a dual choice, were due to avoidance or attraction, young (7 days post heat shock) target flies were given a choice between a stimulus group in a mesh cage (8 flies) and an empty mesh cage using a similar experimental design. Cancerous flies showed, on average, attraction for the social group, independent, of the age or the state of the stimulus flies (Fig 5; stimulus: Wald , P=0.79; stimulus age: Wald , P=0.23; stimulus x stimulus age: Wald , P=0.44). While control flies showed, on average, no clear attraction for the social group, they clearly avoided 3-week-old cancer flies (Fig 4; stimulus: Wald , P=0.1; stimulus age: Wald , P=0.066; stimulus x stimulus age: Wald , P=0.01).
Discussion
Here, we show that social environment can significantly shape the development of intestinal-like cancer in Drosophila. Consistent with previous studies on mammals [10, 29], cancerous flies bred in isolation exhibit faster tumor progression than flies kept in groups of other cancerous individuals. However, more importantly is the finding that variation in group composition also leads to increased proliferation of tumor cells, thus highlighting how subtle variations in social structure may have dramatic effects on the progression of non-transmissible diseases.
Despite the opportunity to interact with others, individual cancerous flies, kept in groups with control flies, developed tumors at similar rates to when cancerous flies were bred in isolation. Social interaction analyses revealed that despite similar locomotor activities, cancerous flies interact considerably less with control flies compared to when they are housed with other cancerous conspecifics. This reduction of social contact may potentially be perceived as a form of social isolation by cancerous flies, which could result in increased tumor growth, analogous to when flies are kept in true isolation. In humans, it has been proposed that the subjective perceived feeling of social isolation may impact psychological and physiological traits as much as real social isolation [30]. Furthermore, the social environment choice experiment suggests that control flies may actively recognize and avoid cancerous ones especially when the tumor growth is significant. Potentially, this may simply reflect common infection avoidance [31, 32], a behavior that has also recently been observed in Drosophila [33]. Even if not contagious, cancerous flies may show particular behaviors, or produce chemical cues, which are generally associated with being sick.
These findings offer new perspectives on the reciprocal relationship between disease and social behavior. While we observe that social structure has profound effects on disease progression, our study also suggest that disease might play a fundamental role in influencing group composition. We found that cancer growth developed at a slower rate when flies were with other cancerous flies. Moreover, we observed that cancerous flies exhibit strong social attraction towards each other, especially at the beginning of tumor development. This raises questions on the very early impact of internal oncogenic process on individual behavior and natural selection pressures on this process [34]. For example, such attraction could be the result of an adaptive process, which reduces cancer progression. Further studies would be necessary to determine the exact proximate factors responsible for this effect, as well as the extent to which generalizations can be made across other cancer types and animal species.
Our findings highlight the importance of social structure on disease progression, beyond the context of transmission. This is the first time that a direct link between social environment, specifically group composition, and cancer progression has been shown, while removing all other confounding psycho-sociological parameters that are frequently encountered in human studies. More generally, this study brings new light to how sociality impacts health and fitness in animals and its potential implication in both conservation and human disease therapy. Moreover, we provide essential data to the emerging topic of evolutionary ecology of cancer, and demonstrate the importance of neoplasia as a fitness limiting factor that influences life history adaptations and strategies.
Author Contributions
E.H.D, T.B, C.M., F.T, J.M and F.M designed the experiments. E.H.D., T.B., J.D.S., C.M., M.S. and J.M. performed the experiments. C.S, J.M and F.M analyzed the data. E.H.D., J.M.., F.T., A.C., C.S., B.U. and F.M. wrote the manuscript.
Methods
Drosophila stocks
yw,HS-flp;esg-gal4,UAS-GFP;FRT82B,Tub-Gal80, yw,HS-flp;UAS-RasV12;FRT82B,Apc2N175K,ApcQ8 and yw,HS-flp;;FRT82B [27] flies were balanced over co-segregating SM5-TM6B balancers. MARCM clones were generated by a 1 hr heat shock at 37°C of 3 days old females and were marked by the progenitor cell marker escargot-gal4 (esg-gal4) Gal4 line driving the expression of UAS-GFP. In all experiments cancerous flies were HS-flp;esg-gal4,UAS-GFP/UAS-RasV12;FRT82B,Tub-Gal80/FRT82B,Apc2N175K,ApcQ8; whereas controls were HS-flp;esg-gal4,UAS-GFP;FRT82B,Tub-Gal80/FRT82B. Although several attempts were made to use non-induced (no heat shock) HS-flp;esg-gal4, UAS-GFP/UAS-RasV12;FRT82B,Tub-Gal80/FRT82B,Apc2N175K,ApcQ8 as controls, due to unpredictable cancer penetrance (unpredictable percentage of flies developing cancer without heat shock) the lineage was declared not suitable as reliable control.
Flow cytometry
Prior to the quantification of GFP-positive cells as an estimate of tumor progression, flies were starved overnight, provided only with water. The entire midgut and the Malpighian tubules, that also exhibit tumor-like structures, were dissected in PBS (Phosphate buffer saline). Samples of 5 guts were digested by collagenase (125μg in 60μl PBS) (Sigma-Aldrich) for 2h at 27°C with gentle agitation. Samples were then complemented with 10 μl Trypsin 10X (Sigma-Aldrich) and nuclei were stained by Hoechst 33342 (0.5μg/ml) for at least 1hr. Samples were filtered and analyzed on a Partec PAS III (Figure S1).
Social breeding
flies were sexed at emergence and control or cancerous females were kept in groups until the third day post emergence. Control and cancerous virgin females were heat shocked at 37°C during 1h. Flies were then introduced into new 40ml food tubes according to their treatment. Control and cancerous flies were partially wing-clipped on the right or left wing to distinguish their genotype. Flies were then kept at 25°C on standard food (renewed every 3 days) until day 21 post induction. Note that the size of the breeding tubes was small (40ml) enough to limit any effect of complete social isolation. Tumour size was estimated with flow cytometry. Data were analyzed with generalized linear model (binomial distribution, Pearson correction for over-dispersion) on the number of tumor cells vs the total number of cells counted.
Social interactions
Each group (composed of flies taken from different food tubes i.e. had never previously interacted together) was introduced into a semi opaque white polyoxymethylene (Delrin) arena (diameter 100mm; height 5mm) covered with a transparent Plexiglas for 1h. Our experimental design allowed us to simultaneously track 4 groups of 8 flies over the 4h. The tracking apparatus consisted of four synchronized firewire cameras (Guppy pro, Allied vision technologies), each filming one interaction arena that was backlighted by a 150X150 mm IR backlight (R&D vision). We used Vision software to analyze spatial data (open-source C-trax 0.3.7 [35] that allowed us to collect 10 positions per second for each fly over 1h video experiments. Tracking corrections were made post C-trax analysis with fixerrors matlab toolbox (Ctrax-allmatlab version 0.2.11) using Matlab software 7.11.0 to suppress swaps between individuals. We then calculated, for each fly, the total length of the path, the distance to other flies, the number of contacts with other flies (a contact was considered when the distance between the centers of two individuals was smaller than, or equal to, one mean body length of the individuals for 1s or more) and the duration of each contact.
Social environment choice
flies were sexed at emergence and control or cancerous females were kept in groups until the third day post emergence. Control and cancerous virgin females were heat shocked at 37°C during 1h and kept in group until the day of the experiment. The experimental setup consisted of a 17x12x5cm plastic cage in which 2 small 2x2x2cm mesh cages were introduced and each placed on a 3cm diameter petri dish containing standard food. Groups of eight flies were introduced in the mesh cages. In the dual choice experiment one mesh cages contained control flies whereas the other one contained cancerous flies. In the attraction vs avoidance experiment only one of the two cages contained stimulus flies. A target fly (control or cancerous) was introduced in the cage 15h before starting the experiment. The position of the fly was then visually recorded every 30min between 10am and 5pm. We only considered the cases where the fly was on a mesh cage or the associated petri dish. Data (number of choices over the 7h) were then analyzed with a general linear model and a binary logistic regression. For the dual choice experiment, target and stimulus flies were of the same age (7, 14 or 21 days post induction) whereas for the attraction vs aversion experiment the target fly was always 7 days old post induction and the stimulus flies were 7 or 21 days post induction.
Fly physical performance
We compared physical and behavioral performances between control and cancerous flies at different ages by exposing them to a ‘negative geotaxis’ behavior test. This involved repeatedly tapping a tube of flies, which causes them to fall, and then measuring the number of individuals that show an escape response (ascending the walls of the tube) over time. When repeatedly performed, this escape response tends to diminish due to exhaustion. This test has been extensively used as a robust proxy to estimate physical and behavioral performance and locomotor. We followed a protocol based on the one initially developed by JW Gargano and co-workers [36] and modified by MJ Tinkerhess [37]. Groups of 10 cancerous or control females (7, 14 or 21 days post induction) were introduced into 40ml tubes and placed vertically on a platform which automatically taps the tubes every 4 seconds causing all flies to fall to the bottom of the tube. Every 15 min we visually recorded the number of flies in each tube showing negative geotaxis (climbing at least 2/3 of the tube between two taps). The experiment lasted 2 hours. The data were analyzed with repeated measures ANOVA including fly state (cancerous or control) and age as factors. The presence of tumor cells (Apc-Ras clones) has little impact on fly performance and survival over the three weeks of the experimental study. Cancerous and control flies subjected to repeated taps over two hours showed a progressive decline in negative geotaxis response (Repeated measure ANOVA: time: F1,50 = 319,9 P<10−3; Fig. S3) which was stronger as flies got older (time x age: F1,50 = 8.8; P = 0.004). However, no difference between cancerous or control flies could be observed at any age (P>0.5 for each age tested i.e. 7,14 or 21 days post induction) suggesting that despite the general observation of age related impairments, Apc-Ras-induced tumors do not affect physical performance and locomotion 3 weeks after induction.
Acknowledgements
This work was supported by the ANR (Blanc project EVOCAN to FT and project DROSONET to FM and CS), the CNRS (INEE and INSB), Fondation ARC (1555286 to JM and FM), The French league against Cancer (M27218 to JM), IDEEV program (to FM), by an International Associated Laboratory Project France/Australia, by the French-Australian Science Innovation Collaboration Program Early Career Fellowship (BU), by André Hoffmann (Fondation MAVA), Fyssen Foundation (to FM and EHD) and the French Government (fellowship 2015-155 to MD). We thank F. Bastin, J.C. Sandoz and A. Couto for their help with confocal imaging.