C. elegans detect the color of pigmented food sources to guide foraging decisions.

Here we establish that contrary to expectations, Caenorhabditis elegans nematode worms possess a color discrimination system despite lacking any opsin or other photoreceptor genes. We found that simulated daylight guides C. elegans foraging decisions with respect to harmful bacteria that secrete a blue pigment toxin. By absorbing yellow-orange light, this blue pigment toxin alters the color of light sensed by the worm, and thereby triggers an increase in avoidance of harmful bacteria. These studies thus establish the existence of a color detection system that is distinct from those of other animals. In addition, these studies reveal an unexpected contribution of microbial color display to visual ecology.


6
ASH/ASI/PVQ, or ASK. Restoration of lite-1 expression in ASJ or ASH/ASI/PVQ are each 94 sufficient to rescue rapid light-dependent avoidance of OP50 + 2.5 mM pyocyanin, while 95 restoration in ASK is not (Fig. 2D). These results localize lite-1 function in rapid light-and 96 pyocyanin-dependent bacteria avoidance to a small number of primary sensory neurons also 97 involved in long-term avoidance of PA14, and which only partially overlap with those 98 responsible for escape responses to short-wavelength light (Fig. 2E). 99 To test whether the blue color of pyocyanin independent of its toxic chemistry is 100 sufficient to induce rapid avoidance of non-toxic OP50 lawns, we employed OP50 lawns doped 101 with non-toxic inert blue food dye (Fig. 3A). Neither wild-type nor lite-1 null-mutant worms 102 avoid OP50 lawns doped with blue food dye, whether in dark or light, indicating that it is not 103 solely the blue color of pyocyanin that drives light-potentiated rapid avoidance but also its 104 chemical reactivity (Fig. 3B). One of the chemical features of pyocyanin that confers toxicity is 105 that it enters eukaryotic cells and generates reactive oxygen species (ROS) through various 106 mechanisms (7)(8)(9)(10)22). To determine if pyocyanin's combination of blue color and ROS-107 generating toxic chemistry underlies light-potentiated avoidance, we employed OP50 lawns 108 doped with the colorless ROS-generating toxin paraquat (24) and non-toxic inert blue food dye 109 (Fig. 3C). Wild-type worms rapidly avoid OP50 doped with 30 mM paraquat and blue dye, but 110 only in the presence of light (Fig. 3D). This light-potentiated avoidance is abolished in lite-1 111 null-mutant worms (Fig. 3D). As with OP50 + pyocyanin, higher concentrations of paraquat 112 mediate rapid avoidance independent of both incident light and lite-1, while lower concentrations 113 of paraquat with blue dye are insufficient to mediate avoidance (Fig. 3D). Doping OP50 lawns 114 with 30 mM paraquat without any dye or with inert non-toxic red dye (Fig. 3D) are each 115 insufficient to confer light-potentiated avoidance (Fig. 3F). These results indicate that rapid 7 light-and lite-1-dependent avoidance of pyocyanin-containing bacterial lawns relies both on its 117 chemical reactivity as a generator of ROS and on its blue color. They also indicate that 118 avoidance conferred by higher concentrations of pyocyanin relies solely on its ROS-generating 119 capacity and is independent of lite-1 function. This integration of color and chemical information 120 to guide avoidance of food sources could enable more accurate discrimination of toxic from non-121 toxic lawns. 122 We hypothesized that blue pigment confers rapid light-potentiated avoidance of bacterial 123 lawns containing ROS-generating toxins by absorbing long-wavelength light and thereby altering 124 the spectral composition of light sensed by the worm. To test this possibility, we employed a 125 series of shortpass and longpass optical filters to alter the spectra of incident light (corresponding 126 photographs and spectra of filtered light are shown in fig. S2). Consistent with the previously 127 determined action spectrum for lite-1-dependent photophobic responses (4-6), longpass filtered 128 light lacking blue light fails to potentiate avoidance of OP50 + 2.5 mM pyocyanin (Fig. 4A). 129 Surprisingly, however, light shortpass filtered with cut-offs of <500 nm or <550 nm also failed to 130 potentiate avoidance (Fig. 4A). Only light shortpass filtered with <600 nm cut-off, which still 131 includes the yellow-orange peak of 6500 K simulated daylight, was sufficient to potentiate 132 avoidance of OP50 + 2.5 mM pyocyanin (Fig. 4A). This establishes that light-potentiated 133 avoidance of bacterial lawns containing ROS-generating toxins and blue pigment requires not 134 only short-wavelength blue light, but also long-wavelength yellow-orange light. This suggests 135 the existence of a yellow-orange sensing pathway in addition to the lite-1-dependent blue light 136 pathway.

137
Completely eliminating yellow-orange light abolishes light-dependent potentiation of 138 avoidance of OP50 + 2.5 mM pyocyanin (Fig. 4A). Thus, we hypothesized that the presence of 8 blue pigment in the lawn reduces, but without eliminating, the yellow-orange content of light 140 sensed by the worm, and that it is this decreased ratio of yellow-orange to blue light that 141 potentiates avoidance of bacterial lawns containing ROS-generating toxins. It is uncertain 142 exactly how blue pigment in its microenvironment alters the spectrum of light sensed by the 143 worm. To circumvent this uncertainty, we eliminated blue pigment from the lawn and directly 144 modified the spectral composition of light sensed by the worm with a "blue vinyl" filter that 145 reduces without eliminating yellow-orange content to mimic the hypothesized effect of blue 146 pigment ( Fig. 4B, fig. S2, A and B). Remarkably, simulated daylight filtered through this blue 147 vinyl filter recapitulates white light-potentiated rapid avoidance of OP50 + 30 mM paraquat, but 148 now in the absence of blue pigment in the lawn (Fig. 4C). These results indicate that blue 149 pigments confer light-potentiated avoidance of toxic bacterial lawns by absorbing yellow-orange 150 light and thereby altering the spectral composition of light sensed by the worm. They also imply 151 the existence of a yellow-orange light-sensing pathway that is required to be activated to 152 potentiate avoidance of toxic lawns, but with activation of this pathway beyond some threshold 153 preventing such potentiation.

154
As a further test of the hypothesis that it is the altered color of light, specifically the blue-155 to-yellow-orange ratio, that potentiates avoidance and not altered illuminance, we altered the that it is the blue-to-yellow-orange ratio, and not illuminance, driving light-potentiated 159 avoidance. While further increases to 50 klx or 100 klx did mediate increased avoidance of OP50 160 + 30 mM paraquat, these very high light intensities closer to that of direct sunlight (100-120      Lawn avoidance assay 376 P. aeruginosa lawn avoidance assays were performed on slow killing assay (SKA) plates as 377 described (8-10, 16, 18, 20). P. aeruginosa cultures were grown in LB liquid media at 37 o C for 378 16-22 hours. 7 µL spots were added to the center of a 3.5 cm plate filled with 5 mL of SKA agar.

379
Plates were first incubated at 37 o C for 22-24 hours, and then at room temperature for 16-20 380 hours. Worms were maintained in regular room lighting or incubator lighting conditions.
Growing worms in the presence or absence of light had no effect on assay results (data not 382 shown). Thirty 18-24 hours post-L4 staged adult worms were transferred to the lawn for each 383 assay. The fraction of worms out of those thirty that were off the lawn after the specified time 384 interval was recorded as fraction avoiding.

385
Solutions of specified concentrations of pyocyanin (Cayman Chemical) and paraquat (ULTRA 386 Scientific) were added to E. coli OP50 lawns on plates prepared exactly as described above.