The strains C1 and C4 differ with respect to morphology and growth pattern.
Plant cells in suspension, exhibit morphological and physiological heterogeneity limiting their use for producing secondary metabolites (Dougall 1987). For Catharanthus roseus as well, suspension cell strains generated from single seedlings had shown unstable production of vinca alkaloids, indicative of metabolic heterogeneity (Deus-Neumann and Zenk 1984). As to design a strategy to study and/or improve the metabolic potential of these Catharanthus cell strains, it was imperative to first characterise them morphologically.
As prerequisite, the behaviour of the cultures had to be calibrated for stability. This was successful since growth indices (ratio of fresh weight at the end of the culture cycle over that of the initial inoculum) were constant over more than 20 cycles (Supplementary Figure S2a). However, the growth indices of C4 fluctuated more widely than those of C1. But even in C4, these fluctuations remained < 20% of the average level. Although both cell strains form aggregates in liquid culture, the aggregates of C4 were much larger (Fig. 2b) than those of C1 (Fig. 2a), both in terms of cell number and cell size. The aggregates of C1 exhibited a smooth and fine structure (Fig. 2c), whereas C4 was friable (Fig. 2d). These differences were also reflected on the level of cell size (Fig. 2e). The cells of strain C4 were significantly larger, both in length and in width, which on the level of estimated cell volume made up a factor of 1.6 times compared to strain C1 (Fig. 2f).
From the time course of cell density (Fig. 2g), the average doubling time could be inferred for both cell strains using an exponential growth model. Here, C4 duplicated slightly faster with 4.2 days as compared to C1 with 4.8 days. However, this proliferation was more persistent in C1 beyond 7 d (the usual time of sub-cultivation), while it was discontinued after day 7 in C4. As a result, C4 had increased in cell density by fourfold after 15 days, while density had increased sevenfold in the case of C1. The peak density for C4 was reached already on day 9, while for C1 the peak was reached only on day 13 (Fig. 2g). This time course of cell density was also mirrored on the level of fresh and dry weight, and Packed Cell Volume (Supplementary Figure S2). The fresh weight showed an exponential curve for C4 until day 8, followed by a plateau phase. However, in C1, fresh weight increased more slowly, at an almost constant rate, with only small increases observed between day 5 and day 6 and, again, between day 12 and day 13 (Supplementary Fig. S2b). The increase of fresh weight was strongly correlated with the increase in cell number. In contrast, the trend was reversed, if dry weight was considered (consistent with the finding that cells of C4 were larger (Fig. 2f). Here, the values of C1 were much higher than those of C4 over the entire measurement period (Supplementary Fig. S2c). If Packed Cell Volume as global readout was analysed, the increase over time was nearly linear for C1 (Supplementary Fig. S2d), while there was a pronounced sigmoidal time course for C4 with a lag phase prior to day 4, an exponential increase between days 4 and 7, and saturation from day 7. Thus, the two strains not only differ with respect to cellular morphology, but also with respect to their growth patterns.
Cells of strain C4 are more vacuolated and show mitochondrial fusions.
Alkaloid biosynthesis partitions to different sub-cellular compartments including cytoplasm, vacuole, tonoplast membrane and endoplasmic reticulum, are directly involved in alkaloid biosynthesis. The enzymes of the pathway are strictly compartmentalised, which requires pathway intermediates to be transported from one compartment to the next (Facchini 2001; Mahroug et al. 2007; Guirimand et al. 2011). Since the two cell strains, C1 and C4, differed in morphology, they might also differ with respect to their subcellular compartmentalisation. To get insight into their subcellular architecture, we transformed cells transiently with different fluorescent markers labelling actin filaments, Golgi vesicles, peroxisomes, and tonoplast, respectively (Fig. 3a-h). The most salient differences detected between the two cell strains concerned the actin cytoskeleton (Fig. 3a, e) and the tonoplast (Fig. 3d, h). While strain C1 displayed a finer meshwork of cortical actin, in C4 actin was forming prominent transvacuolar actin cables. Conversely, the tonoplast was more subdivided into smaller lacunae in C1, while in C4, the central vacuole was more prominent, filling most of the cell interior. In addition, peroxisomes appeared to be more abundant in C4 (Fig. 3g) as compared to C1 (Fig. 3c).
Since alkaloid biosynthesis is an energy-intensive process, mitochondria were studied in more detail. In both cell strains, mitochondria displayed different forms, ranging from punctate to filamentous, mesh-like structures (Fig. 4a). This variation of mitochondrial shape was then followed over a period of 15 days for both cell strains (Fig. 4b). Here, the punctate pattern was found to be dominant in C1, while in C4, the frequency of punctate mitochondria was significantly lower (Fig. 4b). There was an undulating rise and fall in the punctate pattern in both cell strains. Punctate mitochondria became more frequent until day 7, followed by a decline until day 10, whereupon their incidence grew again. In summary, while the overall pattern was parallel in both Catharanthus cell strains, the amplitude for the frequency of punctate mitochondria was higher in C1 as compared to C4. In contrast to mitochondrial shape, total mitochondrial coverage behaved in a similar way for both strains, being higher in the beginning of the culture cycle (Fig. 4c) and then decreasing slowly. Thus, the two strains do not differ so much in mitochondrial number, but rather in the partitioning between different mitochondrial states. While mitochondria in strain C1 are mainly small, they tend to expand and fuse in strain C4.
Sugar consumption rate is higher in strain C1.
Disaccharides in the culture medium are an important source of energy to support cell growth, and a source of carbon for generating the carbon-rich alkaloids as well. In a range between 15 and 60 g/l, viability was close to 100% (Supplementary Fig. S3a). When the sucrose concentration exceeded 60 g/l, viability dropped progressively. At 100 g/l sucrose, cell viability in C4 was about 75 % over the 6-day growth period. Sugar consumption rates were also quantified for both cell strains by following the sugar content in the medium over time using a Brix refractometer (Supplementary Fig. S3b). For strain C1, sugar concentration showed a sharp decline after just 3 days. Already on day 6, the sugar content had decreased to a barely detectable level. In contrast, the sugar content in strain C4 started to decline from day 4, which then continued less rapidly compared to C1 until day 8, after which it stabilised at a value around 0.5 %. Thus, the rate of sugar consumption is higher in C1 as compared to C4.
Strain C1 shows a strong expression of peroxidase 1.
Expression of several key biosynthesis genes for the various biosynthetic branches (seco-iridoid pathway, shikimate pathway, vindoline pathway and vinblastine-vincristine pathway) involved in the synthesis of vinca alkaloids were analysed using semi-quantitative reverse transcriptase PCR (Fig. 5). We observed that geraniol 10-hydroxylase (g10h), as key enzyme of the seco-iridoid pathway, was expressed strongly from between days 2 and 6 in C1, while in C4 it remained high even at day 7. Anthranilate synthase (as), as initial step of the indole pathway, was strongly and constitutively expressed in C1, while its expression was weaker in C4 and remained transient (days 3–5). Tryptophan decarboxylase (tdc), which converts tryptophan into tryptamine, was constitutively expressed in both the cell strains with a stronger expression in C4. Strictosidine synthase (str) and strictosidine β-D-glucosidase (sgd), which are involved in preparing the precursors for the TIA biosynthesis pathway, were constitutively expressed in C4, albeit to lower levels. In C1, sgd was stable at a higher level, while str increased significantly during the second half of the culture cycle. The expression of genes within the vindoline biosynthesis pathway exhibited even more profound differences between the two cell strains: Tabersonine 16-hydroxylase 1 (t16h1) showed expression in both cell strains, however, with an upregulation in C1. The expression of tabersonine 16-hydroxylase 2 (t16h2) was weak in cell line C1, with a stronger expression at day 3, while a very weak expression could be detected in C4 at days 2, 3 and 7. For the genes involved in the final steps of the vindoline biosynthesis pathway, namely, deacetoxyvindoline 4-hydoxylase (d4h) and deacetylvindoline 4-O-acetyltransferase (dat), no expression could be detected in C1. In C4, a very weak signal was found for d4h on days 3, 5, 6 and 7, and a comparably weak signal for dat on days 3, 4 and 7. The peroxidase gene (prx1) which is proposed to be involved in the final step of the production of the bisindole alkaloid vinblastine from anhydrovinblastine, was strongly upregulated in strain C1 throughout the growth period but was much lower in strain C4. This difference in the amplitude of prx1 transcripts was the most significant among the tested genes.
Auxin depletion promotes the accumulation of catharanthine in C4.
Both Catharanthus cell strains, C1 and C4, were analysed, qualitatively as well as quantitatively, for the levels of the alkaloids they accumulated. During the growth phase, both strains were cultivated in the growth medium in presence of the artificial auxin 2,4-D to support biomass accumulation. As a more sensitive analytical approach, LC-MS technology was employed to qualitatively assess the profile of alkaloids. Catharanthine and tabersonine could be detected in both cell strains (Fig. 6a). Vindoline and the other high-value alkaloids vinblastine and vincristine could not be detected in any of the strains during the growth phase. These growth phase samples were then also subjected to quantitative analysis using a less sensitive HPLC-DAD platform. In C1 cells, neither catharanthine, nor tabersonine could be quantifiably detected. However, in C4 cells, atleast catharanthine was detected in small amounts (Fig. 6b). Intracellular catharanthine accumulated significantly over time, however, the amount of catharanthine recovered from the filtered medium, was much lower. These findings were congruent with our LC-MS results, indicating that the alkaloids accumulated, however only to very low abundance during growth phase, irrespective of the cell strain.
In the next step, the alkaloid profile was determined in response to auxin depletion, established by cultivation medium without 2,4-D. Since auxin is essential for cell division, thus, both strains were in stationary phase which should promote the synthesis of secondary metabolites. During pilot studies, both, catharanthine and tabersonine, could be detected in both cell strains, with catharanthine content increasing over time. Therefore, cells were sampled early (at day 4), in the middle (at day 7), and late (at day 10) during this stationary phase caused by auxin depletion. Once again, catharanthine was the only alkaloid that could be quantitatively detected in both strains (Fig. 6c). In addition, strain C4 accumulated substantially more catharanthine over time as compared to the growth phase. While the levels were just under 0.2 mg/g in the auxin-supplemented medium on day 6, they had increased around 20-fold to ~ 4 mg/g on the day 10 of the stationary phase.
Catharanthine accumulation can be elicited in strain C4 by exogenous jasmonates.
Since auxin depletion was successful in promoting alkaloid accumulation, we further studied the effect of chemical elicitation. No matter, whether 2,4-D was present (growth phase), or omitted (stationary phase), viability for both cell strains was maintained at a very high-level during growth phase level (Supplementary Fig. S4a, b). During the stationary phase, the viability of C4 decreased slightly to 93 % on day 6, whereas for C1 there was no change. The additional closing of the filter caps of the flasks which disrupted oxygen access to the cells, had a slight negative effect on the viability of C1, whereas it showed a comparatively more pronounced effect on the viability of C4. In case of stationary cells, the 0.04% EtOH (control solvent for phytohormones), significantly reduced the viability of C1 from 97 % on day 1 to 89 % on day 6, while C4 displayed less sensitivity to EtOH (dropping only from 94 % to 90 % over the same period). The effect of 0.04% and 0.08% MetOH (the solvent controls for the precursor feeding experiment) on C4 is also lower than the effect of EtOH, even at higher concentrations. In case of MeJA treatment (which included combination effects of EtOH, closed caps and MeJA), C1 showed a reduction in viability from 96 % to 80 %, whereas C4 remained quite stable with a decrease from 95 % to 90 %. In summary, the viability of C4 was at a slightly lower level for untreated samples, but remained more stable than C1, when subjected to various treatments indicating better cellular homeostasis for C4 cells. To test, whether sugar would be limiting, we measured catharanthine accumulation over higher sucrose concentrations (Suppl. Fig. S5a). In fact, catharanthine accumulation could be stimulated in both lines, albeit to a different extent and only up to 60 g/l of sucrose. For higher concentrations, the abundance of catharanthine dropped again.
Since alkaloid accumulation, in the biological context, represents a strategy against herbivory, a stress that is conveyed by activation of jasmonate signalling, we used jasmonic acid (JA) as elicitor. This stimulated a clear accumulation of catharanthine in strain C4, but not in strain C1 (Fig. 7a). The effect in C4 was most prominent at day 4 with a factor of 4.5 over the catharanthine levels in the control without JA (Fig. 6c). However, this surplus decreased with time, because catharanthine also accumulated over time in the auxin-depleted, but non-elicited condition (Fig. 6c). Still, at day 7, the addition of JA had more than doubled the catharanthine content compared to the non-elicited control. Even on day 10, an increase of around 50% over the control was seen. The highest abundance of catharanthine, 7.7 mg/g dry weight, was found on day 7 of JA-elicited, stationary C4 cells. For C1, only very low values of catharanthine were detected, which further decreased steadily along the course of the JA treatment.
Since the access of JA to the relatively compact cell clusters might be limiting, we tested for the volatile derivative MeJA, sampling on days 1, 3 and 6 after the treatment, and using the more sensitive LC-MS analysis. Now, in addition to catharanthine, also tabersonine could be detected. Interestingly, minute traces of vindoline were observed in some of the samples (Fig. 7c). However, catharanthine remained the only tangible alkaloid, accumulating to levels that could be quantified, also in these MeJA treated cell strains (Fig. 7b). While the content in C1 increased only to minute levels, C4 was more responsive. Here, from day 3, we observed a significant increase of catharanthine in the C4 strain to ~ 10–11 mg/g dry weight on day 6, which was much higher than the values seen for JA elicitation (Fig. 7a). Vinblastine and vincristine were not detected in any of the strains, irrespectively of, whether JA or MeJA was used for elicitation.
To get more insight into these strain-differences in the accumulation of alkaloids, we examined the expression of the vindoline pathway genes under these conditions (Fig. 7d). In C1, neither t16h1 nor t16h2 were expressed, despite MeJA elicitation (which did cause a slight expression of dat at day 6). In contrast, in C4, there was a clear induction of t16h1 and t16h2 as early as 1 d after elicitation, and from day 3, also a significant increase in steady-state transcript levels of dat in C4. Thus, several genes required for the conversion of tabersonine into vindoline, became activated in strain C4.
Feeding alkaloid pathway precursors to strain C4 leads to vincristine.
The above results gave clear indications that MeJA elicitation was able to elicit catharanthine production in C4 cells (but not in C1) catharanthine production (Fig. 7b). Moreover, some of the transcripts needed for the conversion of tabersonine into vindoline were induced as well (Fig. 7d). Still, only trace amounts of vindoline became detectable. Thus, there seems to be a bottleneck, here. We wondered, whether the vindoline precursor, tabersonine, might be limiting, and whether we would be able to remove this bottleneck, by feeding tabersonine (1.2 µM). We asked further whether feeding of the downstream product of this limiting metabolic branch, vindoline (0.8 µM), might lead to the accumulation of vinblastine or vincristine. Since these compounds emerge from the fusion of the catharanthine and the vindoline moieties (Fig. 1), we conducted these feeding experiment also in a variant, where catharanthine (1.2 µM) was added as well and allowed accumulation over a period of 6 days (Fig. 8a). To ensure that metabolic competence was fully unfolded, the cells were again elicited by MeJA (Fig. 8b).
In fact, in absence of MeJA, none of the tested precursors had any discernible effect on catharanthine biosynthesis (Fig. 8a). In combination with MeJA, 1.2 µM tabersonine stimulated the accumulation of catharanthine, but only for small amounts of tabersonine (Fig. 8b). When the concentration of tabersonine was doubled or tripled, the catharanthine levels decreased; for 3.6 µM of tabersonine, they dropped even below the level seen without tabersonine, indicative of channelling the common precursors (stemmadenine) towards the tabersonine-vindoline branch of the pathway, if tabersonine levels were high. In contrast, feeding of vindoline (0.8 µM), the downstream product of the tabersonine branch, was not effective in changing catharanthine accumulation. When we probed with LC-MS, we found for the combination treatment of MeJA and vindoline, in addition to tabersonine and catharanthine, small amounts of vincristine (Fig. 8c). The identity was verified by the presence of the qualifier fragment ions, 765.51 and 807.59, which were confirmed by the standard compound (Fig. 8d). Encouraged by this finding, we then tried to promote vincristine accumulation by a combination of MeJA and vindoline (3 different concentrations − 0.88, 1.6, 2.4 µM) on C4 in several replications. Unfortunately, although the qualifier fragment ions (765.51 and 807.59) for vincristine were found in these treated samples for the peak of a compound at an appropriate retention time, they were superimposed by other masses, which suggested extremely low amounts, or an extremely unstable production (data not shown). Vinblastine was not detected in any of the elicited treatments. However, we observed an unknown peak at a retention time of 17.9 min with m/z 811.60 (specific for vinblastine). However, the expected retention time of vinblastine was quite different (10.2 min). Thus, the identity of this compound remains unresolved.