Autotoxin-mediated voluntary triage in starved yeast community

When organisms face crises, such as starvation, every individual should adapt to environmental changes (1, 2), or the community alters their behaviour (3–5). Because a stressful environment reduces the carrying capacity (6), the population size of unicellular organisms shrinks in such conditions (7, 8). However, the uniform stress response of the cell community may lead to overall extinction or severely damage their entire fitness. How microbial communities accommodate this dilemma remains poorly understood. Here, we demonstrate an elaborate strategy of the yeast community against glucose starvation, named the voluntary triage. During starvation, yeast cells release some autotoxins, such as leucic acid and L-2keto-3methylvalerate, which can even kill the cells producing them. Although it may look like mass suicide at first glance, cells use epigenetic “tags” to adapt to the autotoxin inheritably. If non-tagged latecomers, regardless of whether they are closely related, try to invade the habitat, autotoxins kill them and inhibit their growth, but the tagged cells can selectively survive. Phylogenetically distant fission and budding yeast (9) share this strategy using the same autotoxins, which implies that the universal system of voluntary triage may be relevant to the major evolutional transition from unicellular to multicellular organisms (10).

Introduction 20 D R A F T even the clonal cells of the toxin-producing cells when they are transferred from glucose-rich conditions. This may look like mass suicide at first glance. However, cells precultured in starved conditions continue 31 to grow even in the conditioned medium, as they tag themselves to adapt to the toxins through starvation, 32 and such adapted state is inheritable. In other words, cells autonomously differentiate into two types, 33 adapted and non-adapted ones, and the cellular community selectively saves the former, just like triage 34 in emergency medical care. Such voluntary triage works as a competitive strategy against both closely 35 related and distant species. Starved yeast cells release toxins, which prevent an invasion of latecomers 36 by killing them, as the Greek philosopher argued: the plank of Carneades (14). Surprisingly, the same 37 strategy was seen in budding yeasts, which are phylogenetically distant relatives to the fission yeast (9). 38 Indeed, we identified the same toxic molecules from both media conditioned by fission and budding yeast. 39 Therefore, we hypothesised that the voluntary triage is evolutionarily conserved in fungal microbes.   introduce such a phase (Fig. 1B). This indicated that in the early growth phase, cells released inhibitors 49 for growth or depleted some of the nutrients required for such a phase.

50
To determine whether cells release inhibitors or deplete essential nutrients, we constructed a conditioned 51 medium using a 1,6 bis-phosphatase deletion mutant (fbp1∆), which did not have a functional gluconeo-52 genetic pathway (15). Such a mutant strain could not grow without glucose ( Fig. S1 and (16,17)) and was 53 expected not to consume the nutrients required for growth. The CM made using fbp1∆ cells (fbp1∆ CM) 54 also caused the delay phase ( Fig. 1C), as shown with the WT CM. This suggested that the delay phase 55 resulted from the release of inhibitory molecules by cells rather than the depletion of nutrients.

56
In addition, when we administered a sufficient concentration of glucose to the CM (Fig. 1D) to recover 57 the carrying capacity, cellular growth was not disrupted, and the delay phase was not observed, i.e., the 58 growth curve of cells in such media was almost the same as that of those in MM with glucose ( Fig. 1E).

59
This indicated that inhibitory molecules in the CM worked only in the absence of glucose.

60
After the delay phase, the growth rate in the CM returned to almost the same level as in the MM. This  Fig. 2C). Note that some of the molecules in the candidate list were 93 difficult to obtain commercially and could not be tested.

94
The two inhibitory molecules had similar characteristics. When the concentration of these molecules was 95 not sufficient, we never observed the delay phase (Figs. 2D and E). Then, the more we administrated the 96 inhibitors, the longer the delay phase was. Finally, if the concentration was higher than the critical con-     Such a strategy, appearing as a suicide at first glance, helps the yeast to select an appropriate offspring 146 that produces toxins and selfishly purify their genome from closely related species. Moreover, voluntary 147 triage overcomes the problems of the toxin-antitoxin system. In such a system, toxin producers should 148 continuously produce antitoxins to protect themselves, and the maintenance of this state is a heavy burden 149 for them (27). Thus, the toxin producer is lost to a cheater, which only has the antitoxin system, whereas 150 the cheater loses to cells having neither any toxin nor immunity (28). In contrast, the voluntary triage does 151 not cost much because the adaptation mechanism is usually offed without the toxin. This suggests that 152 voluntary triage is resistant to cheaters. 153 We found that distant yeast species universally conserved voluntary triage, even at the molecular level.

154
This might be because the toxins in the reported system are simple molecules, while toxins in the toxin-155 immunity systems in bacteria and yeast are highly evolved proteins. In the bacteriocin system of Es-156 cherichia coli, toxins and receptors set off an arms race between the diversification of toxins and en-D R A F T multiple species of unicellular fungi have the former as quorum sensing (35). However, the latter has 172 not been reported. The mechanism we found here meets the criteria required for growth inhibition for 173 multicellularity (36,37), that is, the toxins cause cell death depending on the cell state and smoothly dif- is key to solving the enigma of major transitions in evolution (10), and our study provides a significant 179 milestone.