Raman spectra analysis of Arabidopsis leaf blades and petioles during SAS
To investigate if Raman spectroscopy could identify metabolites that change in response to shade, we established shade conditions that were low in Red:Far-red (R:FR) light to induce SAS in Arabidopsis plants. Wild-type (WT, Col-0) Arabidopsis seedlings (10 days after germination, DAG) were grown under three different light conditions: white light (WL) for normal growth of Arabidopsis, moderate shade (MS) for partial vegetative shade, and deep shade (DS) for severe shade for 7 days (d). Figure 1a, b and Additional file 1: Fig. S1a show that MS condition reproducibly induced typical SAS in WT, including hyponasty of rosette leaves, elongation of petioles and reduction of leaf blades. SAS became more obvious in DS condition, showing that our shade setup can induce different severities of SAS in plants (Fig. 1a, b, Additional file 1: Fig. S1a). These morphological changes were accompanied by the shade-induced expression of marker genes such as A. thaliana Homeobox Protein 2/4 (AtHB2/AtHB4) and Phytochrome Interacting Factor 3-Like 1 (PIL1), which were highly induced under short durations of MS and DS treatment (Fig. 1c) [7, 22]. Likewise, the induction of the flowering time gene Flowering Locus T (FT) in shaded conditions indicated the early flowering of SAS plants (Fig. 1c) [23].
We built a tabletop Raman spectroscopy instrument with specific near-infrared (830 nm) excitation wavelength. As light signaling in plants involves perception and response to visible light, we chose the 830 nm excitation laser to avoid activating any light signaling pathways. Moreover, this wavelength of light was found to provide the largest signal-to-background of the various excitation wavelengths considered (450–830 nm). The optical background here was dominated by off-resonant chlorophyll autofluorescence excited by the infrared laser. This infrared excitation wavelength falls within a spectral window of very low optical absorption in most plant leaves [24], resulting in negligible photodamage to the plant tissues or metabolites even if the laser exposure used is 30 times longer compared to that used here.
Using this purpose-built Raman spectroscopy system (Additional file 1: Fig. S1b), we investigated changes of metabolites in each rosette leaf of WT Arabidopsis. Basically, the spectral intensities were low in the old leaves and increased in accordance to their leaf number (Additional file 1: Fig. S1c). This suggests that the metabolites in the leaf became less concentrated as the leaf aged. During the aforementioned shade treatment, 10-days-old Arabidopsis plants were subjected to shade treatment for 7 days, which exposed the third rosette leaf to the full duration of shade treatment during its development (Additional file 1: Fig. S1d). Hence, to ensure that the sampled leaf accurately represented the effect of the shade treatment, the third rosette leaf was chosen for all further Raman spectroscopy measurements.
The spectra obtained for Arabidopsis leaf blades and petioles under shade treatments showed the same spectral pattern and peak numbers (Additional file 1: Fig. S2a). We could not find any additional or missing peaks within the spectral range measured by our current Raman system. However, we noted that the intensity of most peaks above 700 cm−1 Raman shift decreased under shade conditions. This result confirms previous studies, which showed that the levels of multiple metabolites decreased under shade [10, 11].
To identify the peaks with the largest change under shade conditions, we performed a principal component analysis (PCA) plot using these data (Additional file 1: Fig. S2b). As an additional verification of metabolite changes under shade, the clustering of points were clearly separated in WL and DS, while clustering of points obtained under MS was an intermediate between WL and DS. By plotting the PCA coefficients against Raman shift, we then identified the peaks at 1150 cm−1 and 1521 cm−1 Raman shift as the largest change under shade (PC1) (Additional file 1: Fig. S2b). Based on our Raman spectrum library of chemical standards, we identified the 1150 cm−1 and 1521 cm−1 peaks to be present in all tested carotenoids (Additional file 1: Fig. S2c). Furthermore, the carotenoid standards share a third Raman peak at 1004 cm−1 Raman shift (Additional file 1: Fig. S2c). A previous study showed that shade condition reduces carotenoid biosynthesis and the total carotenoid content in Arabidopsis [25], and our results here support these previous observations. For verification, we extracted and quantified total carotenoid content in our samples using conventional UV–VIS spectroscopy and found similar results (Fig. 1d).
Figure 1e shows that in Arabidopsis plants, the carotenoid Raman peaks (1004 cm−1, 1150 cm−1 and 1521 cm−1) highly decreased in DS in both leaf blade and petiole. All peaks have similar decreases in intensity: in leaf blades, the 1004 cm−1 peak decreased by 4% in MS and 15% in DS, the 1150 cm−1 peak decreased by 5% in MS and 18% in DS, and the 1521 cm−1 peak decreased by 4% in MS and 19% in DS (Fig. 1e). Interestingly, shade caused a greater decrease in peak intensity for carotenoid and other peaks in the petioles than in leaf blades (Fig. 1e, Additional file 1: Fig. S2a), as the 1004 cm−1 peak decreased 22% (MS) and 37% (DS), the 1150 cm−1 peak decreased by 24% (MS) and 37% (DS), and the 1521 cm−1 peak decreased by 33% (MS) and 50% (DS), suggesting that more metabolomic changes occurred in the petioles during SAS (Fig. 1e). As all peaks showed a similar trend under shade and the 1521 cm−1 peak was less affected by autofluorescence, subsequent experiments focused on the 1521 cm−1 peak as a representative of carotenoid Raman peaks.
To further verify the decreased carotenoid content in shade conditions, we measured the expression of genes related to carotenoid biosynthesis, which are known to be down-regulated in etiolated plants [26]. Expression levels of upstream methylerythritol 4-phosphate (MEP) pathway genes (1-deoxy-D-xylulose-5-phosphate synthase, DXS; 1-deoxy-D-xylulose 5-phosphate reductoisomerase, DXR; hydroxymethylbutenyl 4-diphosphate reductase, HDR) and the first committed step of carotenoid biosynthesis (Phytoene Synthase, PSY) were down-regulated under shade conditions, with lower gene expression in DS (Fig. 1f). As such, these results show that carotenoid content decreased in both Arabidopsis leaf blades and petioles during SAS and were relative to the severity of the shade condition. Using Raman spectroscopy, we showed that carotenoid content in petioles were more responsive to shade than leaf blades.
Early diagnosis of SAS using Raman spectroscopy
After establishing that the carotenoid Raman peaks are indicative of SAS, we then asked if these metabolites would respond early to shade. To test this, WT Arabidopsis plants were subjected to different durations of DS treatment before measuring their Raman spectra (Fig. 2a). With longer exposure to shade, the plants developed more severe SAS, with morphological changes after 1–3 days of shade treatment (Fig. 2b, c). Furthermore, morphological changes affected leaf number 3 onwards, verifying the measurement of the third rosette leaf in Raman spectroscopy (Fig. 2c). Surprisingly, changes in the carotenoid peak intensities were detected before morphological changes occurred, starting from just 4 h of DS condition (Figs. 2d, Additional file 1: Fig. S3). Carotenoid peak in both leaf blades and petioles decreased between 4 h to 3 days of shade treatment, reaching a steady state level from 3 days onwards (Fig. 2d). Similar to the results thus far, there was a larger decrease in peak intensity in the petioles (18% at 4 h to 53% at 7 days) compared to the leaf blades (8% at 4 h to 24% at 7 days), confirming that petioles were more reactive to shade. To account for ontogenetic effects on carotenoid content, we designed another time-course experiment, which tracked the daily development of SAS and Raman peak changes under WL and DS (Additional file 1: Fig. S4a). Similar results verified that prolonged shade treatment caused a large decrease in the carotenoid peak intensities, especially in the petioles (Additional file 1: Fig. S4b). Together, these results highlight that the intensity of the carotenoid peaks responded quickly to shade and occurred before obvious morphological changes. These results suggest that carotenoid Raman peaks can be used as a marker for early diagnosis of SAS.
Reduced carotenoid content in phytochrome mutants
SAS is mediated by phytochrome signaling, which has photo-reversible activation/inactivation dependent on the R:FR ratio [27]. Among the five Arabidopsis phytochromes (PHYA-PHYE), PHYA and PHYB are regarded as key players in regulating SAS [27]. Under high R:FR PHYB is the predominant phytochrome that prevents SAS such as petiole elongation and reduction of leaf blade area, whereas under low R:FR PHYA blocks the excessive elongation of seedlings [27, 28].
To investigate if the decreased intensity of carotenoid Raman peaks during SAS is associated with phytochrome signaling, we used Arabidopsis phytochrome mutants (phyB-9BC and phyA-211) and measured their Raman spectra under shade conditions. Consistent with previous studies, phyB-9BC displayed constitutive SAS, whereas phyA-211 showed no SAS under WL but more severe SAS than WT when under shade (Fig. 3a, Additional file 1: Fig. S5a) [1, 28,29,30]. Under all growth conditions, phyB-9BC displayed similar and very low intensities for carotenoid Raman peaks, reflecting its constitutive SAS (Figs. 3b, Additional file 1: Fig. S5b). In phyA-211, the carotenoid peaks in leaf blades and petioles decreased in MS and DS, following a similar trend as WT in shade (Fig. 3b, Additional file 1: Fig. S5b). However, phyA-211 has lower intensities than WT under all conditions and phyA-211 petioles have a larger decrease in intensity (45% in MS, 70% in DS) than WT petioles (22% in MS, 50% in DS) (Fig. 3b). This is consistent with the knowledge that PHYA senses low R:FR light and reduces the severity of SAS, thus the loss of PHYA causes a greater sensitivity to SAS [28]. Extraction of total carotenoids further verified the changes observed in the carotenoid Raman peaks (Fig. 3c). Therefore, the results show that the reduction in carotenoid content is caused by phytochrome-mediated SAS.
Raman spectra analysis at high density planting
Besides vegetative shade, high density planting also causes low R:FR light and induces SAS [31]. To further verify our results obtained from the shade experiment, we planted Arabidopsis plants under low to high densities and measured their Raman spectra. Figure 3d shows that high plant densities (9–25 plants/pot) resulted in increasingly severe SAS in Arabidopsis plants. The measured Raman spectra reflected this change in SAS, as the carotenoid Raman peak had lower peak intensity at higher planting densities (Fig. 3e, Additional file 1: Fig. S6). Similar to the shade treatment results, the peak intensities in the petioles showed more significant decrease (55% decrease in 25 plants/pot) than those in the leaf blades (14% decrease in 25 plants/pot) (Fig. 3e). These results show that the carotenoid peak in a Raman spectrum can also be used as a marker to detect SAS caused by high density planting conditions in Arabidopsis.
Raman spectra analysis of leafy vegetables under shade conditions
To further validate the use of Raman spectroscopy in SAS, we investigated its application in Brassica species. Two species of leafy vegetables, Kai Lan (Brassica oleracea var. alboglabra) and Choy Sum (Brassica rapa var. parachinensis) were treated under shade for 14 days. While Kai Lan reacted to the shade conditions with both reduction of leaf blades and elongation of petioles, Choy Sum under shade developed smaller leaf blades but no significant elongation of the petioles (Fig. 4a, b, Additional file 1: Fig. S7b).
Raman spectroscopy was applied to the first true leaf of each vegetable as it received the full duration of shade treatment (Additional file 1: Fig. S7a). Figure 4c shows that the carotenoid peak of Kai Lan leaf blade and petiole decreased under both MS and DS (leaf blades, 17% and 37%; petioles, 7% and 41%, respectively). While Choy Sum leaf blades decreased their carotenoid peak intensity by 23% in DS, its petioles had no significant change regardless of the shade condition (Fig. 4c). This result may explain the lack of petiole elongation in Choy Sum compared to Kai Lan, especially in the true leaves (Fig. 4a, b, Additional file 1: Fig. S7b). The reduction in extracted total carotenoids in Kai Lan and Choy Sum under shade was verified by measuring total carotenoid content in homogenized plant tissues (Fig. 4d). Similar to Raman peak intensities, carotenoid content in petioles of Choy Sum was very low compared to that of Kai Lan (Fig. 4c, d). Together, these results show that SAS in Kai Lan is similar to Arabidopsis, whereas only Choy Sum leaf blades but not petioles responded to shade. Importantly, the differences between leaf blades and petioles can be clearly detected by measuring their respective carotenoid Raman peaks, making it a useful diagnostic tool for SAS in these Brassica species.
Interestingly, the 1045 cm−1 Raman peak in Kai Lan and Choy Sum petioles showed a distinctively opposite trend under shade. While most peaks in the Raman spectra decreased under shade conditions, the 1045 cm−1 Raman peak increased in only DS (Additional file 1: Fig. S7c). This suggests that Kai Lan and Choy Sum petioles specifically accumulate a different metabolite under DS, which was not observed in Arabidopsis.
To further investigate the difference between leaf blades and petioles of Kai Lan and Choy Sum during SAS, we measured the expression of homologues of Arabidopsis shade-induced marker genes (PIL1; PAR1, Phytochrome Rapidly Regulated 1; IAA29, Indole-3-acetic acid Inducible 29; XTH33, Xyloglucan:xyloglucosyl Transferase 33), which were up-regulated under shade treatment [32, 33]. Generally, gene expression fold-change in Kai Lan were either similar or higher than Choy Sum, which may explain the more severe SAS seen in Kai Lan (Fig. 4e). In both leaf blades and petioles of Kai Lan, the marker genes were highly up-regulated after short durations of shade treatment but gradually decreased over time (Fig. 4e). However, their expression patterns were slightly different in Choy Sum. For instance, the expression of XTH33 homologue was down-regulated in Choy Sum leaf blades under shade and PIL1 homologue was not detectable in Choy Sum petioles regardless of the shade duration (Fig. 4e). Moreover, while the expression of PAR1 homologue in Kai Lan returned to the baseline over time, the PAR1 homologues in Choy Sum (especially petioles) remained induced after prolonged shade treatment. In Arabidopsis, PAR1 inhibits phytochrome interacting factors (PIFs) from reducing carotenoid biosynthesis in shaded conditions [25]. Therefore, the prolonged induction of PAR1 homologue in Choy Sum may explain the smaller decrease in carotenoid levels during SAS. These results may correlate the differences in SAS response and respective carotenoid Raman peaks between Kai Lan and Choy Sum petioles.
Early diagnosis of SAS in leafy vegetables using Raman spectroscopy
We then investigated if carotenoid Raman peaks can also be used in the early diagnosis of SAS in leafy vegetables as it was applicable in Arabidopsis. Both vegetables were subjected to a 14 days time-course experiment under DS condition, similar to the time-course shade experiment performed with Arabidopsis plants (Fig. 2a). Both Kai Lan and Choy Sum displayed increasingly severe SAS as the duration of shade treatment increased (Fig. 5a). Morphological changes in both Kai Lan and Choy Sum started after 3 days of shade treatment, as petioles of Kai Lan were elongated and leaf blades of both vegetables were reduced in size (Fig. 5b). Raman spectra of Kai Lan and Choy Sum in the time-course shade treatment showed that the carotenoid peak reduced in all samples during SAS except Choy Sum petioles (Fig. 5c, Additional file 1: Fig. S8), which is consistent with our findings in Fig. 4c. After 14 days of shade, the carotenoid peak of Kai Lan leaf blades and petioles decreased by 50% and 49% respectively, while Choy Sum leaf blades showed a 29% decrease at 14 days shade and Choy Sum petioles displayed no significant changes (Fig. 5c). Carotenoid peak intensity decreased within 1 to 3 days of shade treatment in Kai Lan and Choy Sum (Fig. 5c, Additional file 1: Fig. S8). This decrease preceded clear morphological changes of leafy vegetables under shade condition (Fig. 5b). Therefore, these results show the relative shade tolerance in Choy Sum and demonstrated that the decrease in carotenoid Raman peaks can be used in early identification of SAS in Brassica vegetables.
Raman spectra analysis of leafy vegetables grown at high density
Next, Raman spectra analysis was performed in Kai Lan and Choy Sum grown at low to high plant densities. At high density, both Kai Lan and Choy Sum developed SAS that were similar to those under shade conditions (Fig. 6a, b, Additional file 1: Fig. S9a). The carotenoid Raman peak also followed a similar trend to those observed in the shade conditions (Figs. 4c, 6c), thereby showing that this finding is likewise applicable in high density growing of non-Arabidopsis plants.
Detection of SAS across various plant species using Raman spectroscopy
To demonstrate the general utility of Raman spectroscopy for the early diagnosis of SAS in plants, we tested other types of vegetables and plants, including Bok Choy cultivars (Brassica rapa var. chinensis), Romaine Lettuce (Lactuca sativa L. var. longifolia), and two tobacco species, Nicotiana benthamiana and Nicotiana tabacum. Figure 7a shows that all tested plants displayed different degrees of SAS when subjected to shade treatment. Baby Bok Choy, Bok Choy ‘Purple King’, and Romaine lettuce were more sensitive to shade compared to the two tobacco species (Fig. 7a). These were detected by Raman spectroscopy, as the carotenoid peak intensities decreased in both leaf blades and petioles of two Bok Choy cultivars and Romaine lettuce in MS and DS (Fig. 7b). The 1045 cm−1 Raman peak also clearly increased in Romaine lettuce petioles under shade (Additional file 1: Fig. S10), which is similar to Kai Lan and Choy Sum petioles in DS (Additional file 1: Fig. S7c). Plants of N. benthamiana showed clear SAS only in DS, while N. tabacum did not show any elongation of petioles under shade (Fig. 7a). This greater tolerance to shade corresponds to the smaller decrease in carotenoid peak intensities of leaf blades under shade and no clear trend in the petiole carotenoid peak (Fig. 7b). Overall, our results demonstrate that carotenoid levels and its Raman peaks are widely applicable as a biomarker of SAS regardless of the plant species.