Albino plantlets were derived from variegated leaves of E. aureum ‘Golden Pothos’ by tissue culture techniques (Hung and Xie, 2009). They were propagated in a maintenance medium containing MS salts with vitamins (Research Products International, Mount Prospect, IL, United States), 10 mg/L ascorbic acid, 25 g/L sucrose, and 5 g/L agar with pH adjusted to 5.6 and grown at 23°C under a constant light intensity of 5 μmol m^–2 s^–1. The plantlets were maintained by transferring onto fresh medium once a month. To observe the effects of light intensity on their growth, equal-sized clusters of albino plantlets were cultured on the maintenance medium at 23°C under full spectrum light with the intensities of 30, 15, and 5 μmol m^–2 s^–1, respectively for a month before they were photographed. Regenerated green transgenic plants were transplanted to soil after rooting.

The sGFP genetic cassette, CEJ899, was kindly provided by Professor Deyu Xie of North Carolina State University. The EaZIP genetic cassette, CEJ1260, was created by replacing GUS in commercial vector pBI121 (Clontech, Mountain View, CA, United States) with EaZIP coding region (Accession #: FJ666046) (Hung et al., 2010). The CHL27 genetic cassette, CEJ1264, was created by replacing GUS in pBI121 with Arabidopsis thaliana CHL27 coding region (Accession #: NM_115553) (Salanoubat et al., 2000). The sequence alignment of cDNAs of EaZIP and Arabidopsis CHL27 is presented in Supplementary Figure 1. After cloning, both inserted genes were sequenced for validation before used for transformation. Each plasmid DNA was introduced into Agrobacterium tumefaciens strain LBA4404 using the freeze–thaw method (Holsters et al., 1978).

For Agrobacterium-mediated transformation, our previous transformation system developed for green E. aureum ‘Jade’ (Zhao et al., 2013b) was modified. Briefly, petiole segments were excised from albino plantlets and infected with Agrobacterium solution (OD₆₀₀ of 0.8) containing 100 μM acetosyringone for 10 min. Then infected petiole segments were co-cultured on induction medium containing MS salts with vitamins (Research Products International), 2 mg/L N-(2-chloro-4-pyridl)-N’-phenylurea (CPPU), 0.2 mg/L α-naphthalene acetic acid (NAA), 10 mg/L ascorbic acid, 25 g/L sucrose, and 5 g/L agar (pH 5.6) in the dark. After 7-day co-culture, they were transferred onto the selection medium which is the induction medium with additional 75 mg/L kanamycin and 100 mg/L Timentin. Only one callus from each explant was isolated for the regeneration. Induced calli were transferred onto regeneration medium which is the selection medium without hormones.

Shoots maintained on regeneration medium were harvested and stored at −80°C for genomic DNA and total RNA isolation. The DNeasy Plant Mini Kit (Qiagen, Germantown, MD, United States) was used for isolating genomic DNA. The RNeasy Plant Mini Kit (Qiagen) was used for isolating total RNA. Further treatment of DNase I and first strand cDNA synthesis were performed using SuperScript^TM IV VILO^TM master mix with ezDNase^TM enzyme (Invitrogen, Carlsbad, CA, United States). For PCR and RT-PCR, the reactions were carried out with Taq DNA polymerase (Sigma Aldrich, St. Louis, MO, United States) and specific primers: NPTIIF 5′-AAGATGGATTGCACGCAGGTTC-3′ and NPTIIR 5′-ACGGGTAGCCAACGCTATGTC-3′ for the nptII; GFPF 5′-GAGCTGGACGGCGACGTAAA-3′ and GFPR 5′-GTGTCGCCCTCGAACTTCAC-3′ for the GFP; EaZIPF 5′-ACGAAGGCTAGGCAGTACAC-3′ and NosR1 5′-AAATGTATAATTGCGGGACTCT-3′ for the EaZIP; and CHL27F 5′-ACAACCAGACACATTTCGTGA-3′ and CHL27R 5′-ACGTCGACGAGCTCCTAATAGA-3′ for the CHL27. For qRT-PCR, the procedure was the same as described previously (Hung et al., 2016), while the specific primers are EaZIP-qF 5′-AGACTGAAGACATTCCCCTGGTAA-3′ and EaZIP-qR 5′-CTCAGAGACTAGTGCTGCGATGA-3′ for the EaZIP; and CHL27-qF 5′-GTGGTTCGGTTTGTCTCGAT-3′ and CHL27-qR 5′-ACGTCGACGAGCTCCTAATAGA-3′ for the CHL27. The QuantumRNA^TM 18S Internal Standards (Ambion, Austin, TX, United States) was used as an endogenous control.

The GFP fluorescence detected in transgenic plants were captured under fluorescence microscope (Keyence, Osaka, Japan). To observe the chloroplast ultra-structures in leaf tissues, TEM analysis was performed at the Center for Electron Microscopy of North Carolina State University. Fully expanded young leaves from baby jar grown plants were cut into 1 mm³ blocks and then fixed in 3% glutaraldehyde in 0.05 M KPO₄ buffer, pH 7 at 4°C. All samples were rinsed in three 30-min changes of cold 0.05 M KPO₄ buffer, pH 7, then post-fixed in 2% OsO₄ in the same buffer at 4°C in the dark. After dehydrated with graded series of ethanol (30, 50, 70, 95, and 100%), samples were infiltrated with Spurr’s resin (Ladd Research Industries, Williston, VT, United States), flat embedded and cured at 70°C overnight. Samples were sectioned with a Leica UC6rt ultramicrotome (Leica Microsystems, Wetzlar, Germany) and placed onto 200-mesh grids. The grids were then stained with 4% aqueous uranyl acetate in the dark at 25°C followed by three times of warm distilled water (40°C) washes and 1 min in Reynold’s lead citrate followed by three more warm distilled water washes. All sections were observed under a JEOL JEM 1200EX transmission electron microscope (JEOL, Peabody, MA, United States). Images were captured using a Gatan Erlangshen Model 785 ES1000W camera and Digital Micrograph software (Gatan, Pleasanton, CA, United States).

For detecting the presence of GFP protein in transgenic lines, the total proteins were first extracted from mature leaf tissues using the Plant Total Protein Extraction kit (Sigma-Aldrich). The protein extract was then subjected to SDS-PAGE. They were heated with NuPAGE^TM LDS sample buffer containing 10% reducing agent (500 mM DTT) at 70°C for 10 min, then analyzed under a NuPAGE^TM 4–12% Bis-Tris gel with a MES SDS running buffer containing antioxidant as instructed by the manufacturer. For immunoblotting, proteins were transferred onto a PVDF membrane (Bio-Rad, Hercules, CA, United States). The membrane was then blocked at 25°C for 1 h with 15% (w/v) dry milk dissolved in PBST. It was then incubated at 25°C for 1 h with 1:200 diluted primary antibody anti-GFP (B-2) (sc-9996; Santa Cruz Biotechnology, Dallas, TX, United States) followed by incubation in the secondary antibody Goat anti-Rabbit IgG horseradish peroxidase conjugate (Bio-Rad). The luminescent signals were generated after incubation with SuperSignal^® West Pico Chemiluminescent substrate (Pierce biotechnology, Rockford, IL, United States) and captured by Kodak Biomax X-ray film (PerkinElmer, Waltham, MA, United States). For staining the PVDF membrane, 0.2% (w/v) Amido Black 10B (MP Biomedicals, Santa Ana, CA, United States) in 10% (v/v) acidic acid was used.

For extracting chlorophylls, the mature leaf tissues were first ground in liquid nitrogen then resuspended in 80% acetone as described in Hung and Xie (2009). After centrifugation, the clear supernatant was collected and diluted 10x in acetone before reading their absorption wavelengths using SpectraMax Plus 384 (Molecular Devices, San Jose, CA, United States). The chlorophyll content was calculated based on the formula described by Hanfrey et al. (1996). All results were presented as the average of three biological replicates ± SD. Statistical significance was analyzed by comparing all pairs using Tukey-kramer HSD (p < 0.05).

To explore the potential applications of regenerated albino plants, we optimized their growth conditions in order to have healthy plants for downstream studies. They were propagated and maintained on Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) with 25 g/L sucrose as a carbon source and addition of 10 mg/L ascorbic acid. Ascorbic acid was used as an antioxidant agent, which is due to the notion that albino plants lacking functional chloroplasts are unable to carry out photosynthesis and more sensitive to the photo-oxidation (Spoehr, 1942; Hess et al., 1994; García-Alcázar et al., 2017). Initially, regenerated plants from color defective leaf sectors were pale yellow and had rooting capacity (Hung and Xie, 2009). However, they gradually lost pale yellow color and rooting capacity, and became albino plants without roots (Figure 1A). We found that they grew slowly but well under low light (5 μmol m^–2 s^–1) on MS medium with addition of 10 mg/L ascorbic acid. Although growths were much faster under high light conditions (15 and 30 μmol m^–2 s^–1), there were signs of cell damage on the tip of leaves (Figure 1C). Because of the high light associated cell damage, low light conditions were chosen for growth and maintenance. Since then they have grown and been successfully propagated for more than 11 years and could be considered as long-lived albino plants in comparison with other known albino plants (Hess et al., 1994; Ruppel et al., 2011; Liu et al., 2016; Zhang et al., 2016; García-Alcázar et al., 2017).

After establishing the optimal growth conditions for the albino plants, a transformation system to produce stable albino transgenic plants was developed. We adopted our previous Agrobacterium-mediated transformation system developed for green pothos plants (Zhao et al., 2013b) with three major changes. Briefly, nptII was used to replace hpt as a selection gene (Figure 2A), 75 mg/L kanamycin was used for screening transgenic plants, and ascorbic acid was supplemented into induction, selection and regeneration media to prevent oxidative damage.

Transformation results showed that visible kanamycin resistant calli first appeared after 4 weeks on selection medium (Figure 2B). The average callus induction rate observed after 8 weeks of subculture was ∼52% (Table 1), which was higher than green pothos plants (Zhao et al., 2013b). After growing for an additional 4–6 weeks, calli were transferred onto a regeneration medium. After about 2 weeks, newly regenerated white shoots appeared (Figure 2B). To confirm the presence of transgenes sGFP and nptII in these regenerated white shoots, seven independent kanamycin-resistant lines (Figure 2C) were examined by PCR and the results showed that both genes were detected in all lines (Figure 2D). To confirm the expression of sGFP, RT-PCR and immunoblotting analyses were performed. The expression of sGFP was confirmed in all seven lines by RT-PCR (Figure 2E). Immunoblotting results based on equal loading of total proteins further confirmed the presence of GFP protein in five out of seven lines which had high transcript levels (Figure 2F). The GFP fluorescence was also observed in regenerated plants (Figure 2G). These results indicate that albino plants can be stably transformed with a high transformation efficiency.

Our previous study suggested that no expression of EaZIP could be responsible for the loss of green color in albino plants (Hung et al., 2010). We found that the albino plants accumulated MPE, the substrate of the MPE cyclase, when they were fed with 5-aminolevulinic acid, the first committed intermediate of the porphyrin synthesis pathway. These results indicate that the albino plants lack functional MPE cyclase. Our previous cloning effort obtained an EaZIP from green pothos (Hung et al., 2010), which lacks a sequence region encoding a N-terminal chloroplast transit peptide (cTP). When overexpressing the cloned EaZIP in wild type tobacco plants which carry the NtZIP also lacking 5′-end sequence coding for a cTP (Liu et al., 2004), surprisingly we found transgenic plants showing variegated phenotype (Guan et al., 2017). However, whether the cloned EaZIP is a functional version or not has not been tested. To determine if overexpression of the EaZIP could restore green color, we first overexpressed EaZIP in albino plants (Figure 3A).

Following EaZIP overexpression, no green shoots were observed among all regenerated plantlets. These regenerants were derived from 31 independent kanamycin-resistant calli grown with or without light exposure (Table 1), even though the growth of kanamycin-resistant calli and regenerated shoots was robust (Figure 3B) with a high transformation rate of 69% (Table 1). To confirm the presence and expression of transgenes in these regenerated plantlets, PCR and RT-PCR were performed. The presence and expression of EaZIP were confirmed in all selected seven lines (Figures 3C,D). The finding that overexpressing EaZIP was unable to restore the chlorophyll production complements with our previous study that overexpressing the EaZIP caused normal green tobacco plants to become variegated (Guan et al., 2017). The results from these two studies implied that the cloned EaZIP might be defective. One possible reason might be the lack of cTP. For further testing, using a functional Arabidopsis CHL27 with a cTP (Tottey et al., 2003) could be an alternative.

The function of Arabidopsis CHL27 has been fully characterized (Tottey et al., 2003). We reasoned that expressing CHL27 in albino plants might restore chlorophyll biosynthesis. Positive results would demonstrate that the impaired expression of EaZIP could be responsible for albino phenotype. To test this hypothesis, we created a genetic cassette by replacing EaZIP with CHL27 (Figure 4A). The transformation rate was high at ∼88% (Table 1). When 22 kanamycin-resistant calli with 0.5–1 cm diameters were exposed to light for approximately 2 weeks, cells with green color started to appear (Figure 4B), and subsequently green shoots emerged (Figure 4B) in 17 individual calli (Table 1). Rooting capacity was also recovered in CHL27 transgenic plants, which allowed them to grow well in soil (Figure 4D). To confirm these green shoots (Figure 4C) expressing CHL27, PCR, and RT-PCR were performed in seven lines. The PCR and RT-PCR results showed that all green shoots carried (Figure 4E) and expressed (Figure 4F) CHL27. These results confirm that the cTP is required and a functional form of EaZIP with cTP may be present in green pothos cells.

We observed that the majority of regenerated shoots derived from the same transgenic callus had three types of leaves: pure green (PG), variegated (V), and pale yellow (PY) (Figure 4C). Different leaf colors appeared in regenerated plants were relatively stable when they grew under the same growth conditions (Figure 5). This phenomenon, however, did not appear in tissue culture regenerated pothos green and albino plants (Figure 6) where calli derived from green or albino explants generated uniform green or white color shoots. To determine whether the leaf color correlates with the expression of CHL27, three types (PG, V, and PY) of transgenic plants from the same transgenic line (Figure 7) were used to measure the contents of chlorophylls and quantify the expression levels of CHL27. Their ultra-structures of leaf cells were also examined since chlorophylls are required for chloroplast biogenesis, such as for assembly and maintenance of photosynthetic apparatus and stacking of thylakoid membranes (Eckhardt et al., 2004; Block et al., 2007).

The results showed that the chlorophyll a and b levels in PG transgenic plants were 3.4 and 1.25 mg/g FW, respectively (Table 2). They were elevated from undetectable in initial wild-type albino plants to 64% of levels found in pothos green plants regenerated from tissue culture. Although the total contents of chlorophylls were less, the chlorophyll a/b ratio was the same as in pothos green plants (Table 2). The chlorophyll a and b levels of V (1.72 and 0.86 mg/g FW) and PY (0.28 and 0.03 mg/g FW) transgenic plants were lower than those of pothos green and PG transgenic plants, but higher than those of wild-type albino plants. In addition, the chlorophyll a/b ratios of PY transgenic plants were in average of 8.14 which was abnormal compared to green plants ranging around 2.7 (Table 2). The stable chlorophyll a/b ratio is important and thought to be very critical to acclimate to changes in light conditions (Rudiger, 2002). When the expression levels of CHL27 were quantified by qRT-PCR in three types of transgenic plants as well as wild-type green and albino plants, the levels of PG transgenic plants were 3.7- and 6.2-fold higher than those of V and PY plants, respectively (Table 3). As for pothos endogenous EaZIP, qRT-PCR result showed that comparing to pothos green plants, none of the wild type albino and CHL27 PG, V and PY transgenic plants had detectable EaZIP (Table 3). These results further indicate that transgenic plants with different colors correlate well the restored levels of chlorophylls, which were indeed contributed from the level of CHL27 expression.

When ultra-structures were examined under the TEM, leaf cells from PG transgenic plants exhibited normal chloroplasts (Figure 7) but had fewer grana and more unstacked thylakoids compared to those of pothos green plants (Figure 6A). Leaf cells of V transgenic plants had completely developed chloroplasts, but only a few developed thylakoids. As for leaf cells of PY transgenic plants, they contained partially developed chloroplasts (Figure 7), which were different from those in wild-type albino leaf cells with only plastids (Figure 6B). Both partially developed chloroplasts in PY transgenic plants and plastids in wild-type albino plants did not have developed thylakoids. The numbers of thylakoids and the degrees of their stacking correlate with the chlorophyll a and b contents in these three types of CHL27 transgenic plants, which agrees with the important roles of chlorophyll a and b in chloroplast development. These results prove that both defective color and dysfunctional chloroplast development in albino plants can be restored by compensating the loss of MPE cyclase.