Summary
In light-limiting conditions, aerial organs of most plants reorient their growth towards the light to improve photosynthesis, through a process known as phototropism1-3. The blue light receptors phototropin control phototropic responses through light-induced protein kinase activity4. Current models posit that asymmetric activation of these sensory receptors across a unilaterally illuminated organ leads to asymmetric distribution of the growth hormone auxin ultimately leading to growth re-orientation4,5. However, the tissue properties required to generate a light gradient across the stem triggering phototropism remain unclear1. Here we show that inter-cellular air channels6,7 are required for an efficient phototropic response. These channels enhance light scattering (refraction and reflection) in Arabidopsis hypocotyls thereby enhancing the light gradient across the photo-stimulated organ. We identify an embryonically expressed ABC transporter that is required to keep air in inter-cellular spaces in seedlings and for efficient phototropism. Our work suggests that this transporter shapes cell wall properties to maintain air between cells. Moreover, we establish the functional importance of inter-cellular air channels in the hypocotyl for phototropism.
Main text
In flowering plants light direction is sensed by the blue-light (BL) absorbing phototropin photoreceptors (phot1 and phot2 in Arabidopsis)5. This typically leads to growth towards the light or positive phototropism in aerial organs such as hypocotyls and stems4. This response is believed to contribute to the optimization of photosynthesis particularly in limiting light conditions2,3,8. In dicotyledons like Arabidopsis thaliana, the upper hypocotyl is both the site of light perception and the site of differential growth ultimately leading to organ repositioning9,10. Upon unilateral BL irradiation differential phot activation between the lit and the shaded side of the seedling is considered as the first step triggering phototropism5,11. Substantial progress was made in elucidating the downstream steps linking phot activation and the differential growth response5. However, the optical features of light-sensing tissues enabling the formation of a light gradient that are required for a phototropic response remain poorly characterized1.
Transparency of hypocotyls causes phototropic defects
In a screen for Arabidopsis seedlings with reduced phototropism we identified a mutant with transparent hypocotyls (Fig. 1). The causal gene was mapped to ATP-BINDING CASETTE G5 (ABCG5), which was confirmed by comparing the phenotype of multiple alleles and by complementation (Extended data Fig. 1). We hypothesized that the phototropic defect in the mutant was caused by enhanced light transmission and a shallower light gradient in the upper part of the hypocotyl. To test this hypothesis, we compared abcg5-5 (from now on abcg5) and two previously identified cristal mutants (cri7 and cri8), which also have transparent hypocotyls12. The defective gene in these mutants is not known12, but they are not allelic12 and we found that the ABCG5 gene was unaltered (see Supplementary Methods). The three mutants had similarly enhanced light transmission in the hypocotyl (Fig. 1a). In response to low unilateral BL abcg5, cri7 and cri8 growth orientation was random, while the wild type (WT) aligned with the light source and phot1 was unresponsive (Fig. 1b). To determine whether these mutants have a specific hypocotyl tropism defect, we analyzed their gravitropic response, which also relies on asymmetric growth caused by redistribution of the growth hormone auxin 13. An auxin transporter mutant deficient in three PIN-FORMED genes (pin3pin4pin7) showed the expected inability to re-orient its hypocotyl upon gravistimulation (Fig. 1c). In contrast, the response of abcg5 and cri7 was similar to the WT, while cri8 showed a reduced response to gravity (Fig. 1c). Both phototropism and gravitropism depend on growth; we therefore measured hypocotyl elongation during the phototropic experiment and found that while abcg5 hypocotyls grew like the WT, cristal mutants showed a growth defect (Extended data Fig. 2). We conclude that having a transparent hypocotyl correlates with an altered ability to respond to light direction and pursued our study with the abcg5 mutant due to the pleiotropic nature of cristal mutants12.
If the abcg5 phototropic defect were due to impaired light direction sensing, we would expect that it requires active phototropins. Consistent with this hypothesis, the phot1abcg5 double mutant behaved like a phot1 mutant in response to low BL (Extended data Fig. 3a), while the abcg5 mutants showed many seedlings growing in the opposite direction (Fig. 1, Extended data Fig. 3a). These experiments were performed with seedlings growing on vertical plates in contact with the media. Given that agar scatters light and that plants were growing at the media-air interface, the light environment in this experimental setup was probably relatively complex. Thus, the following experiments were performed in free standing etiolated seedlings to obtain a simpler directional light cue (see Supplementary Methods). To better characterize the phototropic phenotype of the abcg5 mutant, we analyzed pulse-induced first positive phototropism, which corresponds to the conditions where light fluence (μmol m-2) is proportional to the phototropic response14. We irradiated etiolated seedlings with a 1-minute BL pulse of various intensities (1.7, 0.17, 0.017, and 0.0017 μmol m-2 s-1). In agreement with previous reports15,16, WT plants showed a bell-shaped fluence-response curve, whereas abcg5 mutants showed a severe phototropic impairment (Extended data Fig. 3b). Next, we examined time-dependent phototropism14, by irradiating etiolated seedlings continuously with different BL fluences (0.025, 0.125, and 2.5 μmol m-2 s-1) and recorded growth re-orientation over 6 hours. We found that WT plants took longer to reach maximum curvature with increasing light intensity, confirming earlier studies16. Strikingly, the abcg5 mutant took much longer than the WT to reach maximum curvature at all tested light intensities (Extended data Fig. 3c). We used a bi-directional light treatment to further test the ability of seedlings to respond to complex light environments. In such conditions WT plants grow towards the stronger light source1. In our test area seedlings received a similar light intensity but different light gradients depending on their position (P) (steep (P1, P6), medium (P2, P5) and shallow (P3, P4)) (Extended data Fig. 3d). The ability of abcg5 mutants to grow towards stronger light was reduced particularly when the gradient was shallow (Extended data Fig. 3e). Moreover, their response was slower in this situation (Extended data Fig. 3f). Collectively our phototropism experiments indicated that abcg5 mutants showed a phototropic defect in all tested conditions. The phenotype was particularly pronounced in response to very low light fluences, high fluence rates, and in complex light environments as on the surface of plates or with bi-directional irradiation.
Despite these obvious phototropic defects, we found that phot1-mediated phosphorylation events occurring within minutes of light perception were not altered in the abcg5 mutant (Extended data Fig. 4). Indeed, the BL-induced mobility shifts of phot1 and its targets NPH3 and PKS4 observed on SDS-PAGE gels were normal in abcg5. This is consistent with the abcg5 phototropic defect being due to a reduced ability to establish a light gradient across the hypocotyl rather than in downstream phot signaling (Extended data Fig. 4)17-19. Moreover, the defective BL response of abcg5 was specific to phototropism as BL-induced inhibition of hypocotyl elongation was unaltered in abcg5 (Extended data Fig. 5a). Transparency of the abcg5 mutant was restricted to the embryonic phase, while true leaves and other plant organs developed similarly to the WT (Extended data Fig. 5b). This allowed us to test the phototropic response in petioles. Interestingly, abcg5 mutant petioles showed a WT response (Extended data Fig. 5c), hence we conclude that the phototropic defect of the mutant is restricted to transparent organs. Analysis of ABCG5 gene expression showed that the gene was particularly strongly expressed in developing embryos, with strongest expression in the cortex (Extended data Fig. 5d), while we did not observe expression with a reporter line in seedlings. Of note, while we could complement the mutant by expressing the GFP-ABCG5 transgene from the ABCG5 promoter, complementation was unsuccessful when the construct was driven by the viral 35S promoter (Extended data Fig. 1c). Given that the 35S promoter does not drive gene expression during the early stages of embryogenesis in other species20,21, our complementation assays suggest that embryonic expression of ABCG5 is functionally important. This may also explain the seedling-specific phenotype of abcg5 mutants, while the mutant showed no obvious phenotype later in development (Extended data Fig. 5). We conclude that abcg5 has a specific phototropic defect in seedlings that is most likely due to hypocotyl transparency.
ABCG5 is required for inter-cellular air channels formation
Light absorbing pigments were proposed to contribute to light gradient formation across photo-stimulated plant tissues thereby enabling phototropism 22,23. We therefore analyzed the absorption spectrum of soluble crude extracts from etiolated seedlings and found that abcg5 extracts showed an absorption spectrum comparable to the WT (Extended data Fig. 6). A recent study showed that ABCG5 is required for cuticle development in cotyledons with the mutant showing higher permeability of cotyledons in light-grown seedlings24. Thus, we evaluated the hypothesis that a defect in cuticle development in the hypocotyl explains the phototropic defect and hypocotyl transparency in etiolated seedlings. First, we assessed the cuticle structure by transmission electron microscopy (TEM). We did not find large differences between WT and abcg5, although as reported for the cotyledons, the cuticle in abcg5 was slightly less compact than in the WT (Extended data Fig. 7a). However it was still functional, since the toluidine blue cuticle permeability test showed that abcg5 and WT had a similar permeability, which contrasted with a severe cuticular deficient mutant (long-chain acyl-CoA synthetase2, lacs2) 25, (Extended data Fig. 7b). Moreover, despite cuticular defects, hypocotyls of lacs2 mutants were not transparent and had a robust phototropic response indicating that hypocotyl transparency and the phototropic defect in abcg5 are presumably unrelated to the cuticle (Extended data Fig. 7c). Light-grown abcg5 seedlings sink in water suggesting that they contain less air than the WT24. We performed a floating assay with dark-grown seedlings and found that dissected roots and hypocotyls as well as whole seedlings of the abcg5 mutant sank more frequently than the WT, indicative of reduced air content in the abcg5 mutant (Fig. 2a). Air channels were previously observed in embryos and hypocotyls of several species6,7. They are present at the tricellular junctions between cortex cells or between cortex and epidermal cells6. Thus, we analyzed transverse cuts of etiolated hypocotyls using cryo-scanning electron microscopy (cryo-SEM) and TEM. Both in WT and abcg5 hypocotyls we observed intercellular spaces at the tricellular junctions formed by epidermal and cortex cells. However, those spaces looked empty in the WT, while they were filled in the abcg5 mutant (Fig. 2b, Fig. 2c, Extended data Fig. 8). Moreover, in TEM images the WT showed a well-defined electron dense layer in the outer side of the cell wall surrounding the intercellular spaces. Interestingly, in abcg5 mutants this layer was diffuse, heterogeneous, and sometimes absent (Fig. 2b, Extended data Fig. 8). This structure may correspond to the “splitting layer”26, that was hypothesized to contain hydrophobic material sealing the intercellular space and allowing the formation of air channels. To further investigate whether the difference between WT and abcg5 was the presence of air in intercellular spaces, we used 3D-non-destructive X-ray microtomography7,27. We detected air channels in the longitudinal direction in the WT but not in abcg5 mutants (Fig. 2d). Collectively, our data suggest that the difference in light transmission between the WT and abcg5 may be explained by the presence of air in the intercellular spaces of the WT but not in the mutant.
Air channels contribute to the formation of directional light cues
To better characterize the optical properties of etiolated seedlings we used an integrating sphere allowing measurements of total transmittance and reflectance along with light scattering, i.e., diffused transmittance and reflectance. As observed with our light microscopy measurements (Fig. 1), abcg5 seedlings transmitted more light than the WT (Fig. 3a). Moreover, we filled the air spaces in WT seedlings by water infiltration and found that infiltrated WT samples showed optical properties similar to the abcg5 mutant. In contrast, infiltration of abcg5 samples did not lead to any significant changes in optical properties (Fig. 3a). Our data showed that abcg5 mutants and infiltrated seedlings showed reduced diffused transmitted light, reflected light and diffused reflected light (Fig. 3a). This data is consistent with air channels enhancing light scattering in plant tissues due to the strong difference in the refractive index of air compared to water, cellular fluid and cell walls28. To understand how changes in optical properties affect the light microenvironment within the hypocotyl, we used confocal microscopy. We reconstructed hypocotyl transverse cuts using Z-stacks of transmitted light images. These images showed a similar pattern in all samples, but the contrast was higher in the WT compared to the abcg5, cristal mutants and infiltrated samples indicating stronger light scattering in the WT (Fig. 3b, Extended data Fig. 9a). Moreover, combining these images with images of membrane-associated fluorescent proteins we could determine that the areas of strong scattering coincided with the intercellular spaces at tricellular junctions (Fig. 3c, Extended data Fig. 9b,c). Collectively, our optical characterization of etiolated hypocotyls showed that intercellular air channels contribute to light scattering thereby limiting light transmittance across the hypocotyl.
To visualize the light gradient across an etiolated hypocotyl we used a pPHOT1:PHOT1-GFP line either in phot1phot2 29 (ABCG5) or in a phot1phot2abcg5 (abcg5) background. We made Z stacks across the entire width of the hypocotyl using a confocal microscope equipped with a BL laser. As observed previously9,29, the GFP signal was strongest in cortex cells due to the pPHOT1 expression pattern (Fig. 3c,d). In the ABCG5 background we noticed GFP signal gaps along the plasma-membrane, which correspond to the position of intercellular air spaces formed between the epidermis and cortex or cortex-cortex junctions (Fig. 3c, Extended data Fig. 9b, c). These fluorescence gaps were not observed in the abcg5 background (Fig. 3c, d, Extended data Fig. 9b). Similar gaps were observed when using a line expressing 35S promoter-driven plasma-membrane associated GFP (35S:myri-GFP) in the WT, indicating that the effect of air channels on fluorescence visualization is not specific to PHOT1 (Extended data Fig. 9c). To determine the effect of light scattering on the light gradient across the hypocotyl, we compared the PHOT1-GFP fluorescent signal on the lit versus shaded sides of the hypocotyl (Fig. 3d). Our quantifications showed that the light gradient was significantly steeper in the WT than in the abcg5 mutant or infiltrated seedlings (Fig. 3d, Extended data Fig. 9d). Taken together our results show that air channels present in the WT enhance light scattering thereby leading to a stronger light gradient across a unilaterally illuminated hypocotyl. The characterization of the abcg5 mutant, which lacks these air channels, shows that this is functionally relevant for the detection of directional light cues.
Conclusion
The photoproduct-gradient model of phototropism states that the difference in the levels of photoproduct between the lit and shaded side regulates phototropic bending1. Some experimental evidence supports the importance of light absorbing pigments in the establishment of a light gradient22,30. However, it was also noted that light scattering (in the sense of light diffusion by refraction and reflection) is likely to be important given that the light path inside plant tissues traverses media with different refractive indices (RI, air RI= 1, cell wall RI= 1.42, and cellular fluid RI=1.33)28. Our data using the abcg5 mutant and water-infiltrated WT hypocotyls is consistent with earlier work in other species in showing that air channels in hypocotyls contribute to such a light gradient (Fig. 3)28. The shallower gradient in abcg5 presumably explains why the mutant has more difficulty growing towards the light maximum in complex light environments and shows a delay in responding to unilateral light (Fig. 1, Extended data Fig. 3). The residual phototropic response in the mutant can be explained by the presence of a shallow gradient (Fig. 3). This gradient is presumably due to the difference in refractive indices between cellular fluids and the cell wall, as it could be largely eliminated by infiltrating sunflower hypocotyls with cedarwood oil, which created a medium with a homogeneous RI28. Our work shows that ABCG5 is important to seal air channels which were reported in the hypocotyl of several species6,7,26,27. These channels form following the separation of a poorly characterized cell wall layer called the “splitting layer”26. How this process occurs is poorly understood but interestingly it was proposed that this layer comprises a lipophilic film analogous to suberin or cutin26. ABCG5 is an ABC transporter related to transporters implicated in delivering precursors of the cuticular layer to the extracellular space 31. Strong expression of ABCG5 in embryonic cortex cells (Extended data Fig. 5d) and its localization at the plasma-membrane24 suggest that it may be implicated in the deposition of lipidic precursors required to seal air channels in embryonic tissues. Consistent with this idea, we described an electron-dense layer visible by TEM surrounding the intercellular space in the WT, which is reduced or absent in abcg5 mutants (Fig. 2b, Extended data Fig. 8). Finally, our work shows that these air channels which play a role in gas exchange 6,7,27 also shape the optical properties of translucent tissues enabling them to provide directional light information to plants.
Author contributions
G.M.N, M.L., A.G. and C.F. conceived the original research plans. G.M.N, M.L. and A.G. performed the experiments and analyzed the data. E.S. collaborated with data analysis for the identification of the causal mutation in abcg5-5. J.F and A.S. collaborated with the integrating sphere measurements and with discussions about optics. A.M. and D.D.B. performed the Cryo-SEM and TEM experiments, respectively. G.M.N., M.L. and C.F. wrote the article with contributions of all the authors.
Competing interests
The authors declare no competing interests.
Acknowledgements
We thank Prof. Sara Simonini and Dr. Célia Baroux for their help imaging embryos and for fruitful discussions. The X-ray microtomography experiments were done at the EPFL Platform for X-Ray Radioscopy and Tomography (EPFL PIXE) with the help of Albert Taureg. Confocal microscopy was done at the Cellular Imaging Facility (CIF) at the University of Lausanne. The identification of ABCG5 as the causal mutation of the transparent hypocotyl was done at the Genomic Technologies Facility (GTF) at the University of Lausanne. Dr. Johanna Krahmer developed the script to plot phototropism and gravitropism data. This work was supported by the University of Lausanne, the Swiss National Science Foundation (grants no. 310030B_179558 and 310030_200318 to C.F.), Human Frontiers Science Program (LT000829/2018-L to M.L.), European Commission Marie Curie fellowship (grant no. H2020–MSCA–IF–2017–796443 to M.L. and grant no. H2020-MSCA-IF-2018-843247 to G.M.N).