The seeds of S. lurida were taken from the 4000-m-high northern slope of the Shergyla Mountain of Nyingchi (Tibet, China), in September 2013. The seedlings germinated from these seeds grew in pots containing a substrate mixture (pearlite: vermiculite: Pindstrup Moss Peat = 1:1:2) in a greenhouse (16-h light). When the seedlings had grown 8–10 cm in height, their leaves, stems, and roots were harvested and immediately immersed in liquid nitrogen for later use. These seedlings were bacteria-free, which were established following the method we recently reported (Qin et al., 2014).

Total RNAs were, respectively, isolated from the frozen leaves, stems, roots, and hairy roots by using the Plant RNA kit (TianGen, Beijing, China). RNA from the roots was used as the initial material for gene cloning. First-strand cDNAs were synthesized with a PrimeScript RT Kit (TaKaRa, Dalian, China). A pair of degenerate primers, fdSlTRI and rdSlTRI, were used for amplifying the SlTRI fragment. Based on the ensuing amplified fragments, the gene-specific primers were then used to amplify the SlTRI. For SlTRI, both SlTRI3-1 and the M 13 Primer M4 were used to amplify the 3′-end; then, the two primers SlTRI3-2 and M 13 Primer M4 were used to nest the amplified 3′-end. The corresponding fragments were assembled to generate the putative cDNA sequences. Finally, the physical sequences of SlTRI were amplified, by using two gene-specific primers, fSlTRI and rSlTRI. Three clones for SlTRI were sequenced to confirm the gene sequences. The amino acid sequences were deduced with help of ORFinder and their corresponding theoretical molecular weights were calculated by Vector NTI software. BLASTP was used to analyze the sequence similarity¹; ClustalX (Larkin et al., 2007) and MEGA (Tamura et al., 2011) were used for the phylogenetic analysis. The relative gene expression levels were detected by real-time quantitative PCR (qPCR) in the leaves, stems, roots, transgenic, and non-transgenic hairy root cultures of S. lurida. The phosphoglycerate kinase gene (PGK) served as the internal reference. For the qPCR analysis, the kits purchased from Bio-Rad were used; the qPCR system was Bio-Rad IQ5 (Bio-Rad, Hercules, CA, United States). We used our previously reported method for the gene expression analysis (Qin et al., 2014). The primers for the gene expression analysis are listed in Table 1. The data were collected from at least three biological repeats.

To obtain the recombinant protein, the coding sequence of SlTRI was inserted into the protein expression vector, pET-28a, within the restriction enzymes of BamHI and SacI. The primers used for constructing the protein recombination vector are listed in Table 1. The Escherichia coli strain Rosetta was used as the host cells for expressing the recombinant protein. The methods of producing and purifying the recombinant proteins followed those of Qiang et al. (2016). The tropinone feeding assay was used to identify the function of SlTRI. The total reaction volume of 10 ml was composed of 100 mM of tropinone, 200 μM of NADPH, 100 mM buffer of potassium phosphate, and 5 μg of the recombinant protein. After incubating for 2 h at 30°C, the reaction products were extracted as described by Kushwaha et al. (2013) and they were analyzed by GC-MS (QP2010 Ultra, Shimadzu, Kyoto, Japan) according to published methods (Kushwaha et al., 2013). Briefly, the GC-MS was equipped with an Rtx-5 (5% diphenyl) dimethylpolysiloxane capillary column (30 m, 0.25 mm ID, 0.25 μm df). Helium was used as the carrier gas, with a constant flow rate set at 1 ml min^-1. The injection volume was 1 μl with a split ratio of 20:1. The inlet temperature was set at 250°C. The temperature of the flame ionization detector was set at 300°C. The GC program went as follows: 65°C as the initial temperature, up to 150°C (10°C/min), isotherm for 2 min, then up to 240°C (10°C/min), and finally, at 240°C for 5 min.

The coding sequence of SlTRI was inserted into pBI121 within the BamHI and SacI. Then, the constructs harboring SlTRI were introduced into the Agrobacterium C58C1 (pRiA4) that was used for genetic transformation of S. lurida. The primers used for constructing the plant expression vectors are listed in Table 1. Hairy roots were induced from the bacteria-free leaves of S. lurida, as we did previously (Qin et al., 2014). The root lines without bacterial contamination were cultivated in a 100-ml liquid MS medium in 250 ml flasks (110 rpm) on an orbital shaker at 25°C in the dark. The SlTRI-overexpressing root cultures and the control root cultures were harvested after 40 days of cultivation. The control root cultures came from three independent hairy root lines not transformed by SlTRI. There were three biological replicates in these experiments. The kanamycin-resistant gene, as well as two rooting genes, rolB (root loci B) and rolC (root loci C), were detected as done in our other work (Qin et al., 2014; Xia et al., 2016). To confirm the authentic transgenic root cultures, the specific fragment of the transgene (SlTRI) fused with the 35S promoter was amplified, and the primers for this are listed in Table 1.

The experimental plant materials, including the leaves, stems, roots, transgenic root cultures, and control root cultures, were thoroughly dried at 40°C and then ground into a fine powder for use in the extraction of alkaloids (Xia et al., 2016) and their HPLC analysis (Wang et al., 2011). This alkaloid analysis was performed on the LC-20A system of HPLC with a photo-diode array detector (Shimadzu, Kyoto, Japan), while a CTP-ODS column (Phenomenex, Torrance, CA, United States) was used at 40°C with a 226-nm detecting wavelength. A sample of 20 μl was injected each time. The authentic samples included the hyoscyamine and scopolamine, as purchased from Sigma–Aldrich (St. Louise, MO, CA, United States).

The cDNA of SlTRI was 1148 bp in length, encoding a 29.4 kDa polypeptide that contained 273 amino acids (Figure 2). The BLASTP analysis showed that SlTRI belonged to the family of SDRs, the NADPH-dependent short chain dehydrogenases/reductases (Dräger, 2006). At the sequence level, SlTRI was highly similar to the tropine-forming reductases from D. stramonium (Portsteffen et al., 1994) and H. niger (Dräger et al., 1988). The SDR-specific motifs were found in SlTRI (Figure 3). Serving as the NADPH-binding site, the conserved TGXXXGXG motif (Oppermann et al., 2003) was present at the 27–34 position in SlTRI. The signature motif of SDRs was found at the 105–108 position in SlTRI. SlTRI contained the catalytic tetrad motif characterized by N-S-Y-K, in which the tyrosine (Y) residue is considered essential for the catalytic activity of TRs (Nakajima et al., 1999; Qiang et al., 2016). The bioinformatics analysis suggested that SlTRI might be one of the TRs, and the follow-up phylogenetic analysis gave clear clues to SlTRI’s classification. According to the phylogenetic analysis, the TRs of Solanaceae were definitively classified into two groups. One group (Group 1) was the tropine-forming reductases (TRI), in which the functionally identified TRI proteins from D. stramonium (Portsteffen et al., 1994) and H. niger (Nakajima et al., 1993a) were included; the other one (Group 2) was the pseudotropine-forming reductases (TRII), including the TRII proteins functionally characterized from D. stramonium (Portsteffen et al., 1994) and H. niger (Nakajima et al., 1993a). SlTRI was grouped into the TRI clade according to the phylogenetic analysis we performed (Figure 4). Interestingly, the TR from Cochlearia officinalis (CoTR), which belongs to Cruciferae plant family (Brock et al., 2008), formed an independent cluster (Group 3) that was separated from either the TRI or TRII group. In fact, CoTR was a non-specific TR that catalyzed tropinone, to yield both tropine and pseudotropine (Brock et al., 2008). SlTRI was deposited in GenBank with the accession No. MF578231.

The tissue profiles of the three genes involved in TA biosynthesis (SlPMT, SlTRI, and SlH6H) were analyzed in S. lurida (Figure 5A). These genes showed a near-specific expression in the roots. SlPMT, the first committed-enzyme, had a very high expression level in the roots that was 164-fold that seen in the stems. SlPMT was not expressed in leaves, however. SlH6H, the last-committed enzyme, was exclusively expressed in the roots. SlTRI was preferentially expressed in roots, and its expression in leaves and stems were barely detected. Since the tissue profile of SlTRI was similar to that of SlPMT and SlH6H, it is not unreasonable to expect SlTRI’s involvement in the tropine-derived alkaloid biosynthesis. The tissue distribution patterns of hyoscyamine and scopolamine were analyzed by HPLC (Figure 5B). Both hyoscyamine and scopolamine were detected in the roots, stems, and leaves of S. lurida, a species that features the hyoscyamine-rich chemotype much like A. belladonna does (Xia et al., 2016). In each plant organ, hyoscyamine was the main alkaloid occurring with the highest content. In the roots and stems, the hyoscyamine content doubled the scopolamine content; in the leaves, the hyoscyamine content (3.13 mg g^-1 DW) was more than twice the scopolamine content (1.44 mg g^-1 DW). Additionally, the hyoscyamine content peaked in the leaves, suggesting that they were the main storage organs for alkaloids in S. lurida.

Enzymatic assays were performed to identify the functions of SlTRI. The recombinant His-tagged SlTRI was purified (Figure 6) from the engineered E. coli. On the SDS-PAGE gel, the molecular weight of this recombinant SlTRI was c. 29 kDa, consistent with its calculated molecular weight. The purified recombinant protein was used for the enzymatic assay. When tropinone was fed to the recombinant SlTRI, SlTRI catalyzed the reduction of tropinone to form tropine, as confirmed by GC-MS. As Figure 7 shows, compared to the spectrum of standards (Figures 7A,B), the retention time for the tropine produced by SlTRI (7.741 min) was consistent with that of authentic tropine (7.736 min) (Figure 7C). The chemical structure was characterized by the MS spectrometric data, together with the Kovats retention indices (RI), quasi-molecular ions ([M]⁺), and characteristic ions with a relative intensity (Figure 8). The two compounds, including tropinone and tropine, were, respectively, identified through their comparison with published data (Zayed and Wink, 2004) and the NIST databases. Further, we also analyzed the enzymatic kinetics of the SlTRI. The K_m, V_max, K_cat, and K_cat/K_m values of SlTRI were, respectively, 40.90 ± 7.059 mM, 32.43 ± 3.175 nkat mg^-1, 0.98 ± 0.096 s^-1, and 23.96 s^-1 M^-1 at pH 6.4 (Table 2). The K_m value of SlTRI was much higher than that of DsTRI, which includes the purified TRI (1.3 mM) from D. stramonium (also assayed at pH 6.4) (Portsteffen et al., 1994) and its recombinant TRI (4.18 mM at pH 6.4) (Qiang et al., 2016). The K_m value of the recombinant SlTRI was 9.78-fold that of the recombinant DsTRI, suggesting the lower affinity of SlTRI for tropinone. By contrast, both the V_max and K_cat values of SlTRI were significantly lower than those of DsTRI, thus indicating that SlTRI has the lower catalytic activity than does DsTRI.

Expression of the rooting genes (i.e., rolB and rolC) was detected in the hairy roots and the positive control C58C1 (pRiA4), but it was not detected in the wild-type roots of S. lurida. The fragments of rolB and rolC were, respectively, 423 bp (Figure 9A) and 626 bp (Figure 9B). Transgenic hairy root cultures (i.e., overexpressing SlTRI) were confirmed by PCR amplification of the 502-bp NPTII gene fragment (Figure 9C). The 914-bp PCR products, including the fragments from the 35S promoter and SlTRI, were specifically amplified from the positive control and transgenic root cultures (Figure 9D). Compared with the control, the expression level of SlTRI in all the SlTRI-overexpressing root cultures was dramatically higher (9–287-fold greater; Figure 10B). At the same time, the expression of SlPMT (Figure 10A) and SlH6H (Figure 10C) were also detected in the root cultures. In most cases, each of the two genes (SlPMT and SlH6H) showed no significant different gene expression between the transgenic root cultures and the control. This suggested that the enzymatic step defined by TRI was substantially improved at the transcriptional level in the transgenic root cultures. The hyoscyamine content in SlTRI-overexpressing root lines ranged from 0.892 DW to 1.492 mg g^-1 DW (Figure 11A), with an averaged 1.163 mg g^-1 DW, whereas the average content of hyoscyamine in the control was 0.515 mg g^-1 DW (Figure 11A). In the SlTRI-overexpressing root cultures, the hyoscyamine contents were 1.7- to 2.9-fold that found in the control, with a difference that was significant. On average, the detected scopolamine content was 0.482 mg g^-1 DW in the SlTRI-overexpressing root culture lines (range: 0.268–0.632 mg g^-1 DW), for which most had significantly higher scopolamine contents than did the control (0.199 mg g^-1 DW) (Figure 11B). Moreover, the total amount of both hyoscyamine and scopolamine was notably higher in the SlTRI-overexpressing root lines (range: 1.238–2.068 mg g^-1 DW) than that in control (0.714 mg g^-1 DW) (Figure 11C). These results suggested that overexpressing SlTRI significantly promoted tropane alkaloid production in the hairy root cultures of S. lurida.