Cooperation of the haves and the have-nots

Upon starvation, Dictyostelium discoideum (D.d.) exhibit social behavior mediated by the chemical messenger cyclic adenosine monophosphate (cAMP). Large scale cAMP waves synchronize the population of starving cells and enable them to aggregate and form a multi-cellular organism. Here, we explore the effect of cell-to-cell variability in the production of cAMP on aggregation. We create a mixture of extreme cell-to-cell variability by mixing a few cells that produce cAMP (haves) with a majority of mutants that cannot produce cAMP (have-nots). Surprisingly, such mixtures aggregate, although each population on its own cannot aggregate. We show that (1) a lack of divalent ions kills the haves at low densities and (2) the have-nots supply the cAMP degrading enzyme, phosphodiesterase, which, in the presence of divalent ions, enables the mixture to aggregate. Our results suggest that a range of degradation rates induces optimal aggregation. The haves and the have-nots cooperate by sharing complementary resources.

These experiments were performed about 35 times with different ratios of the beads to the highdensity haves. The beads did not affect pattern formation in all cases.
Other subtler contact-dependent interactions could be in play. To completely rule these 70 out, we placed a Millipore filter between the populations of the haves and the have-nots. 71 This filter allows the exchange of chemicals, while preventing the cells from coming into 72 mechanical contact. The populations aggregated (in two trials of the experiments). Thus, 73 we have disproved our mechanical hypothesis. The effect must be chemical in origin. 74 What is this chemical interaction? Could the aggregation be caused by some yet-to-be-75 identified chemical factors secreted by the high-density cell populations into the buffer? We 76 obtained supernatants from three sources: a high-density population (2 x 10 5 cells/cm 2 ) of 77 haves, a high-density population of have-nots (2 x 10 5 cells/cm 2 ), and the mixture described 78 above (9 x 10 3 cells/cm 2 of wild type cells and 2 x 10 5 cells/cm 2 of acaA-mutants). We 79 starved these populations for six hours and then carefully collected the supernatants. Each Adding just the HMWF of the supernatant, both heat-treated and non-heat treated, caused the cells to round up. When the LMWF of the supernatant was added, the cells were healthy and polarized.
one containing chemicals > 30kDa and a fraction < 30kDa, containing small proteins, ions, 86 and other low molecular weight chemicals (see Methods). We denote these fractions as the 87 High Molecular Weight Fraction (HMWF) and Low Molecular Weight Fraction (LMWF) 88 respectively. To further fractionate the HMWF fraction, we subjected a part of it to heat 89 treatment to deactivate heat sensitive enzymes, and left the rest untreated.

91
Low densities of haves were developed in these fractions. After about 20h, we found (in 92 one experiment):

93
• Cells + LMWF: The low-density haves were viable and appeared polarized, but no 94 tight aggregates were observed. (figure 3) 95 • Cells + heat-treated, HMWF: The low-density haves rounded up and did not appear 96 viable. Furthermore, many cells had de-adhered from the substrate (figure 3).

101
• Cells + LMWF + non-heat-treated, HMWF: In this second positive control, the low-102 density haves showed clear large-scale streaming (figure 4b). 103 We also conducted a negative control, in which the low-density haves were unconditioned.

104
These cells rounded up and did not appear viable (figure 4c). Clearly, something smaller 105 than 30kDa is essential to maintain viability. This chemical is secreted by both haves and 106 have-nots, and is deficient in our buffer. Additionally, a heat-sensitive factor larger than 107 30kDa enables the cells to signal effectively over long distances.

109
What could be in the LMWF that keeps the cells alive? We hypothesize that cells carry 110 over essential ions into our low ionic strength buffer. At low densities the ionic strength 111 might not be sufficient for the cells to remain viable [15][16][17]. To test this hypothesis, we 112 supplemented our buffer with either 50µM CaCl 2 , 50µM MgCl 2 or 150µM NaCl. For the 113 latter, we used a higher concentration to match the ionic strengths of the other two buffers.

114
In the calcium and magnesium enriched buffers, (figures 5a,b) after 20 h, the low-density 115 haves were viable and polarized, but did not form clear aggregates. In contrast, the low-116 density haves in the sodium enriched buffer (figure 5c) rounded up, indicating that they 117 were not viable. We can conclude that divalent ions, Ca 2+ or Mg 2+ , which are secreted by 118 have-nots as well, are required to keep the cells viable. Although the addition of calcium and magnesium kept the cells viable, it did not fully 120 rescue aggregation. This is consistent with the experiment where we added the LMWF of the 121 supernatant to the low-density haves. Above, we showed that factors in the heat-sensitive 122 HMWF were necessary to rescue aggregation (figure 4a,b). In the HMWF, we expect to find We opened this manuscript with a riddle: when a low-density population of haves were 132 added to a high-density population of have-nots, the mixture aggregated, whereas each 133 population on its own fails to develop. 134 We found that the lack of divalent ions kills the low-density haves. Ions like Ca 2+ and Mg 2+ are necessary to keep the cells viable, but they alone do not enable the low-density haves to form tight aggregates. Addition of external PDE is necessary for these low-density 137 haves to aggregate. Although addition of calcium does induce the cells to secrete more PDE 138 (see Supplementary note S1 and [25-28]), the increased PDE activity due to the addition of 139 calcium is not sufficient to fully restore aggregation in a population of low-density haves. 140 We found that the addition of external PDE to the low-density haves in buffers enriched 141 with divalent ions, enabled them to form tight aggregates. Previously, it was shown that the degradation rate is about 50 min. This is much longer than the wave period. Therefore Having gained these insights, we return to the cooperation between the haves and the  We started this paper by naïvely called the wild-type cells the haves and the acaAcells 172 the have-nots, simply because the latter group is incapable of producing cAMP. But we 173 found that the mutants provide other components that are vital for survival. While investigating their cooperation, we found that (1) a critical amount of divalent ions is necessary to keep the cells viable, and (2) PDE activity within a certain range is necessary for optimal aggregation. When these two criteria are satisfied, even a small fraction of cAMP produc- The supernatant was aspirated from the Petri dishes, and then centrifuged to pellet out 196 any cells accidentally. The supernatant was then cleared using 0.2µm filters.

197
To fractionate the supernatant into the high and low molecular weight fractions, we used c) was much higher than the activity of the low density WT cells in phosphate buffer (figure 9 261 column b). As a check, we verified that our heat treatment denatures PDE, column a in figure   262 9. Finally, the degradation rate was even higher in low density populations conditioned by the characteristic decay time given by 1/γ has to be shorter than the period of the waves (6 280 -8 min), so that cAMP can be degraded before the next wave begins. Therefore, the value 281 of the degradation rate at high cell densities allows wave propagation because it is shorter 282 than the period of the wave.

283
Now, the density of our low density population is about 70 times smaller than the value 284 considered above. If we assume that the degradation scales linearly with density, the degra-285 dation rate of the low density population will be 70 times smaller, and the degradation time 286 will be 70 times larger, or about 49 min. This number is much higher than the period of the 287 waves. This larger degradation time will not allow wave propagation. As we have seen in figure 8, the extra cellular PDE decreases more rapidly with density than a linear fit would 289 predict. Therefore, for aggregation, more PDE is necessary in the low density haves.  . In our experiments, we add a much higher concentration of cAMP. So the rate of the reaction can be assumed to be independent of the substrate concentration. This rate is also V max , the maximum rate of the reaction. We observe that AMP concentration decays after the reaction is stopped by adding the stop solution. Accounting for the production and decay of AMP, we can write Here C is the concentration of AMP, k 1 is the degradation rate of PDE which creates AMP (this is the rate we are interested in finding) and k 2 is the decay rate of AMP. The solution to the equation is We stop the reaction after 20 min, the concentration of AMP after 20 min is C(20).
After these 20 min, there is only decay of AMP because PDE doesn't act anymore. The equation for the concentration of AMP after these 20 min is therefore the decaying exponential C(t) = C(20)exp(−k 2 t).
Here t is time after stopping the PDE action. We measure C(t) at different times after 299 stopping. We first subtracted the background AMP and then converted the luminosities