The maize phytoene synthase gene family: overlapping roles for carotenogenesis in endosperm, photomorphogenesis, and thermal stress-tolerance

Carotenoids are essential for photosynthesis and photoprotection; they also serve as precursors to signaling molecules that influence plant development and biotic/abiotic stress responses. With potential to improve plant yield and nutritional quality, carotenoids are targets for metabolic breeding/engineering, particularly in the Poaceae (Grass family) which includes the major food crops. Depending on genetic background, maize endosperm carotenoid content varies and therefore breeding enhanced carotenoid levels has been of ongoing interest. The first committed step in the plastid-localized biosynthetic pathway is mediated by the nuclear-encoded phytoene synthase (PSY). The gene family in maize and other Grasses contains three paralogs with specialized roles that are not well understood. Maize endosperm carotenoid accumulation requires PSY1 expression. A maize antibody was used to localize PSY1 to amyloplast envelope membranes and to determine PSY1 accumulation in relation to carotenoid accumulation in developing endosperm. To test when and if PSY transcript levels correlated with carotenoid content, advantage was taken of a maize germplasm diversity collection that exhibits genetic and chemical diversity. Total carotenoid content showed statistically significant correlation with endosperm transcript levels at 20 days after pollination for PSY1 , but not PSY2 or PSY3 . Timing of PSY1 transcript abundance, previously unknown, provides critical information for choosing breeding alleles or properly controlling introduced transgenes. PSY1 was unexpectedly found to have an additional role in photosynthetic tissue, where it was required for carotenogenesis in the dark and for heat stress tolerance. Leaf carotenogenesis was shown to require phytochrome-dependent and phytochrome independent photoregulation of PSY2 plus nonphotoregulated PSY1 expression. leaf tissue of mutant inbred line. G insertion in genomic DNA and prepared from is mRNA levels of maize

antiserum also detected PSY antigen in isolated maize amyloplasts but not in isolated amyloplasts of rice endosperm which does not accumulate PSY1 transcripts (Wurtzel and Yu, unpublished).

Timing of PSY expression and carotenoid accumulation in endosperm.
To better understand the relationship between temporal regulation of PSY transcript accumulation as compared to carotenoid accumulation during endosperm development, quantitative RT-PCR was used to monitor transcript levels for PSY1, PSY2 and PSY3, in combination with spectrophotometric analysis (Kurilich and Juvik, 1999) of total carotenoid content for developing endosperms of maize inbred B73, a line which carries the dominant Y1 allele. As shown in Fig. 3A, carotenoid content increased 26.4-fold between 10 to 28 days after pollination (DAP). At 10 DAP, the levels were not statistically different from those of the maternal ovule tissue collected from unfertilized ear (UN). The period of 10 to 16 DAP represented the most significant increase in content, 15.6-fold, though carotenoid accumulation continued throughout development ( Fig. 3A).
While PSY2 levels were highest in the unfertilized ovule, compared to PSY1 and PSY3, only PSY1 transcript levels increased significantly during endosperm carotenogenesis, starting after 12 DAP. In the B73 inbred, PSY1 mRNA levels declined after 22 DAP and dropped down to significantly lower levels at 28 DAP. In contrast to maize PSY1, PSY2 mRNA levels were constant during endosperm development, except between 10 to 14 present, but no PSY1 antigen was detected. Even though transcripts levels were about the same between 10 to 12 DAP, PSY protein was first detected at 12 DAP, suggesting that PSY accumulation might be linked to endosperm development and plastid biogenesis as previously observed for post-transcriptional regulation of PSY destined for chloroplasts (Welsch et al., 2000). The period of 14-16 DAP also showed the steepest rise in accumulated carotenoids for a given time frame, following induction of PSY protein seen between 10 to 12 DAP. After 22 DAP when carotenoid accumulation was leveling off, protein levels started to be reduced, in parallel to the reduced PSY1 transcript levels.
PSY gene family transcript accumulation was also examined in maize carrying the PSY1 mutant allele, y1-602C. PSY2 transcripts increased about 3 fold between 15 and 25 DAP ( Fig. 4), in comparison to more constant levels observed for PSY2 in B73 which carries the Y1 allele (Fig. 3A); PSY3 transcripts in y1-602C were at a constant and low level throughout development, as seen in B73. These results suggest that under nonmutant conditions of the Y1 allelic background, PSY1 expression (or some carotenoid intermediate/product) may have a negative feedback effect on transcript accumulation for PSY2, but not PSY3. The inverse temporal accumulation of PSY1 relative to PSY2 and PSY3 in the Y1 endosperm of B73 (Fig. 3A) is consistent with this interpretation for which further analysis is needed to better understand the underlying mechanisms.
Endosperm transcript accumulation for PSY2 in the y1 background was later (20)(21)(22)(23)(24)(25) than seen for PSY1 in the Y1 background (12-18 DAP), which might suggest that mRNA levels at specific endosperm developmental time points) or controlling temporal expression of PSY transgenes (e.g. by choice of promoter that provides transcription in a given temporal window during endosperm development). To assess whether timing of PSY transcript level correlated with carotenoid content, advantage was taken of the genetic and chemical diversity available in maize. A subset of 10 inbred lines were chosen that exhibited a range in carotenoid content and composition (Harjes et al., 2008).
Plants were control-pollinated, endosperms dissected at 15, 20 and 25 DAP, and then PSY transcripts and carotenoid content quantified as described in Methods. Pearson correlation analysis revealed that PSY1 was the only gene whose transcripts showed significant correlation with total carotenoids (Fig. 5 A and B); this statistically significant correlation was specifically seen at 20 DAP (96% correlation [r] and p value of 0.001).
Therefore, PSY1 transcript level at 20 DAP is a good predictor of carotenoid content influenced by PSY1 expression and therefore may be used in selecting optimal PSY1 alleles for breeding or designing optimal timing of transgenes.

PSY2, the only paralog that is up-regulated by light in photosynthetic tissue
During de-etiolation or under high light stress, carotenoid biosynthesis is induced to assist photosynthesis and to protect against photo-oxidative damage. Transcript levels of several genes encoding carotenogenic enzymes, especially PSY, have been shown to be up-regulated by light, which leads to carotenoid accumulation (Bartley and Scolnik, 1993;von Lintig et al., 1997;Simkin et al., 2003). Therefore, maize PSY gene family transcripts levels, PSY1 transcript levels in leaves were not affected by light. Maize PSY3 transcript levels remained at very low levels during de-etiolation, suggesting that PSY3 is not involved in light-induced leaf carotenogenesis.

Role of phytochrome in light-mediated induction of PSY2 transcript levels
To determine the light quality required for photoinduction of PSY2 transcript levels, darkgrown maize B73 seedlings were exposed to 100 µmol m -2 s -1 of red, far-red or blue light and transcripts measured by quantitative RT-PCR. All of the light regimes induced transcript elevation, although having specific temporal signatures and effects (Fig. 7A, C and E). In red and far-red light ( Fig. 7A and E), PSY2 mRNA levels began increasing after 2 hours, and peaked at 4 h, with a 5-fold increase in PSY2 mRNAs as compared to the unilluminated control; the far-red light response showed slower kinetics compared to the red-light response. In comparison, blue light (Fig. 7C) caused a more rapid and elevated accumulation of PSY2 mRNA levels; levels increased at 1 h and peaked after only 2 h in blue light, causing a 10.5 fold increase in PSY2 transcripts as compared to the unilluminated control. These light regimes were also tested with PSY1 and PSY3, and as predicted from the white light experiments (Fig. 6), photoinduction was not observed (Supplementary Fig. 2S and 3S).
The observed induction by red and far red light suggested that phytochrome was involved in mediating PSY2 photoinduction. To verify this, loss of the red and far-red light photoinduction was tested by blocking phytochrome activity. This was accomplished by using maize seedlings that were homozygous for the recessive elongated mesocotyl1 (elm1) allele which interferes with biosynthesis of the phytochrome chromophore, thus preventing assembly of all phytochrome types (Sawers et al., 2002). Dark-grown elm1 (Y1) seedlings were subjected to the three light regimes. No increase in PSY2 transcripts were observed after illumination by red or far-red light ( Fig. 7B and F). These results were consistent with the role of phytochrome in mediating PSY2 up-regulation by red or far-red light. When elm1 plants were subjected to blue light (Fig. 7D), photoinduction was still observed and PSY2 transcript levels peaked at 2 h as seen for B73 (which carries the dominant Elm1 allele), although the down-regulation at 4 h in B73 was delayed to 6 h in elm1 plants. The blue light result in elm1 plants was also as expected since blue light photoinduction is not phytochrome-mediated, but instead mediated by other photoreceptors such as phototrophins or cryptochromes (Briggs and Olney, 2001;Im et al., 2006). The slight temporal change in the blue light induction may be due to some cross talk between the red and blue light signaling. Again, PSY1 and PSY3 transcript levels were tested in the elm1 plants, and as expected, no change was observed as compared to B73 seedlings treated with the three light regimes (Supplementary Fig. 2S and 3S). In summary, carotenogenesis that accompanies photomorphogenesis involves the photoinduction of PSY2, which is mediated both by phytochrome and nonphytochrome receptors.

PSY1 is essential for heat stress-induced carotenogenesis in photosynthetic tissue
As shown above, photoinduction of PSY2 transcript levels was shown to be linked with carotenogenesis in tissue undergoing photomorphogenesis; PSY1 and PSY3 were clearly found to be unresponsive to light. Assuming that PSY1 plays no role in leaf carotenogenesis, it was puzzling to note that in etiolated seedlings which harbor proplastids and low carotenoid levels relative to light-grown seedlings ( Table 1), PSY1 transcripts were more abundant than those for PSY2 (Fig. 6). Moreover, in fully green, de-etiolated seedlings (Fig. 6) or in light-grown seedlings (Gallagher et al., 2004), transcript levels for PSY1 and PSY2 were equivalent; in mature leaves, the level of PSY2 transcripts actually exceeded those for PSY1 (Li et al., 2008). Although the role of PSY1 in endosperm carotenogenesis is well-established, this unexpected abundance of PSY1 transcripts in green and etiolated leaves suggested that perhaps PSY1 might also have a role in green tissue carotenogenesis. To test this possibility, a PSY1 null allele of y1 was needed. Two classes of y1 alleles have been reported in maize by Robertson's earlier studies (Robertson and Anderson, 1961;1987): 1) those with white endosperm and normal green leaves; 2) those with white endosperm and pale green leaves. The y1-602C allele used in earlier experiments ( Fig. 1) is representative of Robertson's class 1 which affects carotenogenesis only in endosperm but not leaves because the mutation affects only the promoter. Therefore plants carrying y1 alleles (obtained from the Maize Genetics Stock Center, Univ. Illinois-Champaign-Urbana, IL) were screened to identify one carrying a null allele, which would likely fall into Robertson's class 2 mutants.
Sequence analysis revealed that in one of the PSY1 mutants, y1-8549, a G was inserted 377 bp downstream of the ATG initiator codon which caused a frame shift and resulted in a new stop codon at 388 bp (Fig. 8A). Previous reports of this allele (Robertson and Anderson, 1961) indicated that higher temperatures exacerbated the mutant phenotype and homozygous plants were extremely weak and generally did not survive to maturity under their field conditions, although the cause was unknown at the time. Therefore, the y1-8549 null allele was used to investigate the impact of blocking PSY1 activity on carotenoid accumulation in etiolated and mature leaves.
The Y1 nonmutant and y1-8549 mutant seedlings were grown under four conditions varying for light and heat stress, where 37 0 C is considered heat-stress and 20 0 C is within the optimal temperature range (Wilhelm et al., 1999). Under low temperature (20 0 C) in the dark, the total carotenoid levels (including both carotenes and xanthophylls) of y1-8549 seedlings were 62% of the levels found in nonmutant Y1 seedlings (Table 1A); carotenes alone were only 24% of the normal (Table 1B). When the temperature was increased to 37 0 C, Y1 seedlings grown in the dark responded with a 2.2-fold higher level of carotenoids, whereas the carotenoid level in y1-8549 seedlings was unchanged and therefore exhibited only 30% of normal levels, suggesting that the PSY1 null allele was interfering with the response to elevated temperature under dark growth conditions. Under low temperature (20 0 C) in the light, y1-8549 seedlings showed only a slight difference compared to Y1 seedlings; carotenoid and chlorophyll levels in y1-8549 were 92% and 72 % of the levels found in Y1 seedlings, respectively ( Fig. 8B and 8C, Table   1). However, when grown at 37 0 C in the light, the PSY1 null mutant was photobleached and showed only 3% carotenes, 9% xanthophylls and 21% chlorophyll, as compared to plants carrying the normal allele (Table 1B). In comparison, carotenoid levels in leaves of Y1 plants grown at 37 0 C in the light were ~6% higher than found in plants grown at 20 0 C in the light, as compared to a 2.2 fold elevation in carotenoids when plants were grown in the dark at 37 0 C. One explanation for the limited heat-stress response in the light may be due to photo-oxidation of pigments which takes place at high temperatures for plants grown in the light, and leads to carotenoid degradation thus limiting carotenoid accumulation (Britton, 1995). If the role of PSY1 is to contribute to heat-stress activation of carotenoid biosynthesis, then the observed photobleaching seen in y1-8549 plants that was exacerbated by high temperatures (Robertson and Anderson, 1961) might be due to rapid degradation of carotenoids which are insufficiently replenished due to limited biosynthetic capacity in the absence of a functional maize PSY1.
Under dark growth conditions, the mRNA levels of PSY1 in the y1-8549 mutant were 30 % and 40 % of the levels found in Y1 at 20 0 C and 37 0 C, respectively, while the mRNA levels of PSY2 were 1.5-fold greater at 20 0 C and 3-fold greater at 37 0 C in the mutant compared to Y1 (Fig. 8B). The higher accumulation of PSY2 transcripts seen in the mutant might be a result of feedback regulation of PSY1 on PSY2 expression; in the Y1 background, expression of the functional Y1 product and/or its ensuing activity somehow represses PSY2 mRNA accumulation and in the mutant this repression is absent such that PSY2 expression partially compensates for the PSY1 deficiency. This observation is reminiscent of the case seen in endosperm when either PSY1 transcripts were temporally reduced or absent, in the case of the y1-602C endosperm mutant (Fig. 4). When the seedling growth temperature was raised, the transcript levels of both PSY genes decreased in Y1 and y1-8549 seedlings. This decrease in PSY transcripts in etiolated Y1 plants was unexpected given the concurrent increase in carotenoid accumulation at high temperatures.
In summary, carotenoid levels doubled in etiolated Y1 seedlings subjected to heat stress as compared to normal temperature, whereas y1-8549 plants responded with only a slight increase in carotenoid accumulation; in the light, y1 mutants actually showed a severe decrease in carotenoids under heat stress conditions and appeared photobleached. The simplest explanation is that in the light, heat stressed plants are subjected to photooxidative damage which can be relieved by increased carotenoid synthesis that is mediated by PSY1, a protective mechanism not available in the PSY1 null mutant seedlings. Therefore these data support a role for PSY1 in leaf carotenogenesis, especially under heat stress conditions.

Discussion
We recently established that three maize PSY paralogs arose prior to evolution of the Grasses (Gallagher et al., 2004;Li et al., 2008). There remain open questions about the specific roles of these three genes that encode a key enzyme operating at the pathway entry point; the six downstream steps all appear to involve single-copy genes in maize (Li et al., 1996;Matthews et al., 2003;Li et al., 2007;Harjes et al., 2008). Whereas some species contain a single PSY gene, the PSY gene duplication in the Grasses could potentially provide Grasses with a fine tune control of carotenoid biosynthesis for different regulatory scenarios.

PSY1 timing in endosperm carotenogenesis
From early genetic analysis to more recent association studies, the PSY1 locus has been associated with maize endosperm carotenoids (Randolph and Hand, 1940;Palaisa et al., 2003). Additionally, in maize and wheat, a major quantitative trait locus (QTL) for endosperm carotenoids maps to the PSY1 locus (Wong et al., 2004;Pozniak et al., 2007).
However, it was unclear what was the temporal relationship between gene expression during endosperm development and kernel carotenoid content. Efforts to develop strategies that provide predictable changes in plant chemistry, or "predictive metabolic engineering", are dependent on availability of such data. In the absence of such information, use of a tissue-specific promoter provides tissue-specificity but not necessarily the optimal window of expression for maximizing pathway end-product accumulation. Therefore, endosperm carotenoid accumulation was determined for developing kernels, together with PSY transcripts and protein. Carotenoid accumulation was observed from 10 days after pollination to maturity. This pattern of continuous accumulation is in contrast to that seen in sorghum varieties which exhibited a peak of accumulation mid-development followed by a steep drop (Kean et al., 2007) transcript level at 20 DAP was found to be a good predictor of carotenoid accumulation and therefore may be useful in selecting optimal PSY1 alleles for breeding or designing optimal timing of transgenes.
The maize diversity lines are a powerful resource for gaining insight into the regulation of plant chemistry across genetically diverse germplasm. When used in association analysis, the diversity lines may provide information on polymorphisms that can be used for breeding, as in the case of improving provitamin A carotenoid content (Harjes et al., 2008). As applied here, these lines provided insight into the timing of candidate gene expression. There will likely be other steps and/or regulatory factors that may affect carotenoid accumulation and for which expression can be examined using these diversity lines. Germplasm diversity collections of other species, for which metabolic breeding holds interest, represent untapped value to examine candidate gene expression in relation to metabolite composition.

Amyloplast envelope membrane localization of PSY1
Maize PSY1 was shown to be resident on endosperm amyloplast envelope membranes, suggesting that the envelope membrane is the site of the carotenoid biosynthetic metabolon in amyloplasts. An envelope location for the pathway may not be limited to amyloplasts. Early studies provided evidence that an envelope-localized carotenoid biosynthetic pathway might also exist in chloroplasts. Chloroplast envelope membranes were shown to contain a unique carotenoid profile compared to thylakoid membranes (Jeffrey et al., 1974;Siefermann-Harms et al., 1978); chloroplast envelope membrane extracts also exhibited enzyme activity for conversion of zeaxanthin to violaxanthin (Costes et al., 1979). More recent reports have localized PSY and other pathway enzymes to thylakoid membranes in chloroplasts (Linden et al., 1993;Bonk et al., 1997;Welsch et al., 2000). Therefore, the carotenoid pathway may actually be two pathways from a topological standpoint, potentially existing on envelope and thylakoid membranes. In amyloplasts, the pathway is likely associated with the envelope membrane while in chloroplasts, the pathway may be associated with both envelope and thylakoid membranes. In assembly of the biosynthetic metabolons, the nuclear-encoded pathway enzymes are recruited to membrane locations which may differ from one plastid type to another. Since many of the pathway enzymes are encoded by single copy genes (Li et al., 1996;Matthews et al., 2003), specificity of targeting must be controlled by factors beyond the transit sequence. Optimization of metabolic engineering/breeding of carotenoids is predicated on elucidating the mechanisms that control recruitment of pathway enzymes to metabolons on different plastid membranes in various plastid types.

PSY2 is associated with photomorphogenesis
Given the involvement of carotenoids in photosynthetic tissue, it was expected that the PSY genes were photoregulated. For example, the single PSY gene in Arabidopsis was shown to be regulated by continuous red, far-red and blue light regimes (von Lintig et al., 1997). However, in a Y1 genetic background only PSY2 was photoregulated by light, mediated by phytochrome-dependent and phytochrome-independent mechanisms which involved blue light signals. Maize PSY1 and PSY3 did not respond to light of any spectral type. It was previously demonstrated that PSY3 was associated with carotenogenesis leading to drought and salt-induced ABA biosynthesis in roots (Li et al., 2008); recent work on rice PSY3 showed the rice homolog to have a similar role in rice roots (Welsch et al., 2008). Unlike maize, rice PSY1 and PSY2 are both photoregulated, although blue light was not tested (Welsch et al., 2008), and rice PSY1 is not expressed in endosperm (Gallagher et al., 2004). Therefore, the ancestral Poaceae PSY was a photoregulated gene; the duplications analyzed in Y1 maize represent losses in the photoregulatory mechanisms and evolution of other features, including the gain-of-function endospermspecific expression seen for maize PSY1 in Y1 maize, but absent in its evolutionary ancestor teosinte and other subfamilies of the Grasses including species such as rice.
A second role for PSY1 in photosynthetic tissue and thermal tolerance Examination of leaf carotenogenesis revealed involvement for both maize PSY1 and PSY2, indicating that PSY1 is not exclusively associated with endosperm carotenogenesis.
In addition to endosperm function, PSY1 was shown to be required for leaf carotenogenesis, in the dark and under conditions of heat stress. When PSY1 expression was eliminated in the maize y1-8549 PSY1 null mutant, the plant had reduced carotenoids in the dark and suffered severe photobleaching at elevated temperatures in the light. Dual roles for a supposedly "nonphotosynthetic tissue" PSY paralog are not unique to maize.
The loss of the "fruit-specific" PSY1 in the tomato r,r mutant showed reduced carotenoid accumulation not only in fruit but also in leaf tissue (Fraser et al., 1999). In that case, it is not known whether the tomato PSY1 plays a role in stress resistance. While PSY1 (y1) has been labeled as the gene needed for endosperm carotenogenesis, the data presented indicate that PSY1 is essential in controlling carotenogenesis in photosynthetic tissue. Its absence has severe effects on plant vigor that are exacerbated by high temperature. It is unclear at present how PSY1 -controlled carotenogenesis mediates stress-resistance in leaf tissue. PSY1 is not impacting ABA biosynthesis as carotenoids are not limiting in photosynthetic tissue as they are in roots (Li et al., 2008). Since carotenoids are known to protect against heat stress-induced lipid peroxidation, it appears that PSY1 is the maize paralog that is essential for heat-stress induced biosynthesis of carotenoid antioxidants that protect plastid membranes (Davison, 2002;Havaux et al., 2007).

Post-transcriptional regulation
Examination of gene expression in some of the mutants suggested that the various PSY genes may be subject to feedback regulation. In different tissues, absence of PSY1 transcripts was associated with elevation of PSY2 transcripts. During endosperm development there was an inverse relationship seen between transcripts for PSY1 relative to transcripts for PSY2 or PSY3. In leaf tissue, where PSY1 was absent due to a null mutation in y1, there appeared a concomitant increase in transcripts for PSY2 under heat stress conditions. PSY1 may also be subject to post-transcriptional regulation as suggested by the absence of endosperm PSY1 protein at early developmental time points where transcripts were detected; perhaps translation is linked to amyloplast biogenesis in a similar fashion whereby there appears to be a link between chloroplast biogenesis and translation of PSY mRNAs in Arabidopsis (von Lintig et al., 1997).

Summary
Each of the maize PSY paralogs evolved unique roles to control carotenogenesis in different tissues and in response to stress. The ongoing maize studies are greatly facilitated by the mutant and germplasm diversity collections which generally are unavailable tools for research on other agronomically important grasses. However, phylogenetic identification of orthologous genes may suggest potential targets for plant improvement whether it is to enhance plant yield through stress resistance or in nutritional improvement.

Total carotenoid content measurement
The carotenoid extraction procedure used was slightly modified from Kurilich and Juvik (1999). Tissue samples of 500 mg were ground in 6ml of ethanol containing 0.1% butylated hydroxytoluene (BHT) and were incubated in an 85 0 C water bath for 6 min prior to a 10 min saponification with 120 µl (1g/ml) of KOH. All samples were vortexed once during saponification, after which they were immediately placed in ice to which 4 ml of cold deionized distilled water were added to each sample, followed by 3 ml of petroleum ether:diethyl ether (2:1 v/v), vortexed and centrifuged for 10 min at 3500 rpm.
The upper layer was transferred to a fresh tube and the aqueous phase was twice subjected to centrifugation with 3 ml of PE:DE each time and combined the fractions.
The combined fractions were topped to 10 ml, and 1ml of aliquot was used to measure total carotenoids at OD 450nm using a Lambda 25 UV/VIS spectrometer (PerkinElmer Life Sciences, Boston).

Real-time PCR
RNA isolation and cDNA synthesis were carried out as described by Gallagher (2004) . Real-time PCR was performed using iQ TM SYBR green supermix (BioRad, Hercules, CA) with 10 ng cDNA. Primers and PCR conditions for maize PSY1, PSY2, PSY3 and the internal control actin were described previously (Li et al., 2008). Specificity of amplification was confirmed via melt curve analysis of final PCR products by ramping the temperature from 50°C to 90°C with fluorescence acquired after every 0.5°C increase.
The fold change of transcript abundance of target genes was first calculated as 2 -∆Ct , where ∆ Ct is the number of PCR cycles required to reach the log phase of amplification for the target gene minus the same measure for actin. Transcript abundance of maize fusion protein [including 12 aa, 1.32 kD of T7-Tag encoded by the vector and Y1 protein (aa#124--aa#410, 31.57 kD), was separated on by 12% SDS-PAGE, the fusion protein band (about 50% yield) excised, and immersed in 5 ml of 125 mM NaCl. Rabbits, chosen for absence of preexisting immune response to maize proteins, were injected with 100 µ g Y1 fusion protein preparation at week one, two, three, and five (Lampire Biological Laboratory Inc., Ottsville, PA). For western analysis and localization experiments, bleed from rabbit #3445 (five weeks after immunization) was used and diluted 1:10,000 in TBST for westerns and 1:25 for confocal experiments. Specificity of the antiserum was verified by testing for presence of the antigen in normal but not homozygous y1 endosperm, using the y1 Maize COOP alleles 602C (shown in Fig 1) and 607C (data not shown), which condition reduced endosperm carotenoids, but do not affect leaf carotenoids.

Expression of maize PSY2 and PSY3 T7.tag fusion proteins
To test the specificity of anti-maize PSY1antiserum, maize PSY2 and maize PSY3 T7.tag fusion proteins were also expressed using the pET23 vector (Novagen). pEMPSY2-1 contained a partial B73 maize PSY2 cDNA (GenBank AY450646), encoding aa#109 to aa#402 (33.46 kD) of PSY2 (Gallagher et al., 2004) fused to 16 residues (1.67 kD) of the T7 tag, resulting in a PSY2 fusion product of 35.13 kD. pEMPSY3 was constructed by inserting a full-length maize PSY3 cDNA (encoding residues 1-426, 47.32 kD, GenBank DQ36430) into the pET23C (+) vector (in frame fusion with 16 residues of a T7 tag, encoding 1.67 kD) to produce a PSY3 fusion protein of 48.99 kD. Expression of the fusion proteins was induced by addition of 0.5 mM IPTG for 4 hours when the E. coli cells reached an A 600 of 0.6 at 37°C. To detect the fusion protein by immunoblot analysis, Maize proteins from endosperms were extracted according to the method described by Wurtzel (1987). Protein samples were separated on 12% SDS-PAGE in a Criterion TM Cell system (BioRad) and transferred to nitrocellulose membrane for western blot in a Trans-Blot Cell (BioRad). The first antibody, anti-maize PSY1, and the second antibody, goat anti-rabbit IgG, were diluted at 1:10,000 for use in Western analysis according to the manufacturer's protocol (Novagen). To control for protein loading, a duplicate western was probed with a polyclonal antiserum raised against maize Shrunken1 protein (Wurtzel et al., 1987)

Immunolocalization of PSY in maize endosperm
Maize B73 endosperms at 17 DAP were dissected and immersed in 50 mM sodium phosphate buffer (pH 7.2) and fixed according to Zhang et al. (1992) with modifications.
Several slices (2 mm x 4 mm x 1 mm) for each tissue were fixed in 3% glutaraldehyde in sodium phosphate buffer either for 2 hours at room temperature or overnight at 4°C.
Fixed slices were washed with sodium phosphate buffer at room temperature three times for 15 minutes each, dehydrated at room temperature through a series of ethyl alcohol 10%, 25%, 50%, 70%, and 95% (v/v) for 10 minutes each, then twice in 100% ethyl alcohol for 10 minutes each time. Slices were infiltrated in 25%, 50%, and 75% (v/v) LR white (Polysciences, Warrington, PA) (v/v) in ethyl alcohol at room temperature for 2 hours each, and then in 100% LR white at room temperature over night. Slices were transferred to 100% LR white in beam capsules (Polysciences, Warrington, PA) and polymerized at 65°C for 24 hours. One µ m semi-thin sections were prepared using glass knives on an Ultra Microtome (Leica). Sections were collected onto precoated glass slides (Polysciences, Inc., Warrington, PA) and air dried. All steps of immunofluorescent labeling were carried out at room temperature. Slides were rehydrated in PBS (137 mM        far-red light (100 µmol m -2 s -1 ) (panels E, F) and quantitative real time PCR was used to determine transcript levels for maize PSY2. All quantifications were normalized to β actin and shown relative to maize PSY1 transcript levels at 0 h using the same conditions.
Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants. Chlorophyll content; C, carotenoid content. The mRNA levels of maize PSY1 (D) and PSY2 (E) in leaf tissue were normalized to levels of ß-actin transcripts measured in the same samples and are shown relative to maize PSY1 mRNA levels from seedlings grown in the dark at 20 0 C. Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants. Table 1. Chlorophyll, carotene and xanthophyll analysis of Y1 and y1-8549 mutant seedlings. Figure 1S. Western analysis to test specificity of the anti-maize PSY1 antiserum against maize PSY2 and PSY3 fusion proteins. PSY proteins were expressed as fusion constructs with T7 tags and extracted from E. coli before (-) and after (+) IPTG induction.

Supplementary Data
Blots were probed with anti-maize PSY1 antiserum (left) or anti-T7-Tag Antibody, AP Conjugated (right). The control "C" for anti-maize PSY1 antiserum and anti-T7 western blots were total protein extracted from 20 DAP B73 endosperm and T7.tag positive control (Novagen), respectively. M, molecular weight marker with sizes indicated in kDa. Figure 2S. PSY1 transcript levels in de-etiolating maize B73 (Elm1) and elm1 mutant seedlings treated with light of various spectral qualities. Dark-grown B73 (panels on left A,C, and E) and elm1 (panels on right, B, D, and F) seedlings were illuminated with red light (100 µmol m -2 s -1 ) (panels A, B), blue light (100 µmol m -2 s -1 ) (panels C, D), or far-red light (100 µmol m -2 s -1 ) (panels E, F) and quantitative real time PCR was used to determine transcript levels for maize PSY1. All quantifications were normalized to β actin and shown relative to maize PSY1 transcript levels at 0 h using the same conditions.
Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants. red light (100 µmol m -2 s -1 ) (panels A, B), blue light (100 µmol m -2 s -1 ) (panels C, D), or far-red light (100 µmol m -2 s -1 ) (panels E, F) and quantitative real time PCR was used to determine transcript levels for maize PSY3. All quantifications were normalized to β actin and shown relative to maize PSY1 transcript levels at 0 h using the same conditions.
Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants.      left A,C, and E) and elm1 (panels on right, B, D, and F) seedlings were illuminated with red light (100 µmol m -2 s -1 ) (panels A, B), blue light (100 µmol m -2 s -1 ) (panels C, D), or far-red light (100 µmol m -2 s -1 ) (panels E, F) and quantitative real time PCR was used to determine transcript levels for maize PSY2. All quantifications were normalized to β actin and shown relative to maize PSY1 transcript levels at 0 h using the same conditions.
Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants.   Figure 1S. Western analysis to test specificity of the anti-maize PSY1 antiserum against maize PSY2 and PSY3 fusion proteins. PSY proteins were expressed as fusion constructs with T7 tags and extracted from E. coli before (-) and after (+) IPTG induction.

Supplementary materials
Blots were probed with anti-maize PSY1 antiserum (left) or anti-T7-Tag Antibody, AP Conjugated (right). The control "C" for anti-maize PSY1 antiserum and anti-T7 western blots were total protein extracted from 20 DAP B73 endosperm and T7.tag positive control (Novagen), respectively. M, molecular weight marker with sizes indicated in kDa. left A,C, and E) and elm1 (panels on right, B, D, and F) seedlings were illuminated with red light (100 µmol m -2 s -1 ) (panels A, B), blue light (100 µmol m -2 s -1 ) (panels C, D), or far-red light (100 µmol m -2 s -1 ) (panels E, F) and quantitative real time PCR was used to determine transcript levels for maize PSY1. All quantifications were normalized to β actin and shown relative to maize PSY1 transcript levels at 0 h using the same conditions.
Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants.  left A,C, and E) and elm1 (panels on right, B, D, and F) seedlings were illuminated with red light (100 µmol m -2 s -1 ) (panels A, B), blue light (100 µmol m -2 s -1 ) (panels C, D), or far-red light (100 µmol m -2 s -1 ) (panels E, F) and quantitative real time PCR was used to determine transcript levels for maize PSY3. All quantifications were normalized to β actin and shown relative to maize PSY1 transcript levels at 0 h using the same conditions.
Values represent the mean of 3 RT-PCR replicates +/-SD from five pooled plants.