Pilocarpus

Allevato, Daniella M., Groppo, Milton, Kiyota, Eduardo, Mazzafera, Paulo & Nixon, Kevin C., 2019, Evolution of phytochemical diversity in Pilocarpus (Rutaceae), Phytochemistry 163, pp. 132-146 : 133-137

2.1. Phylogeny of Pilocarpus

All three methods (Parsimony, Bayesian, Maximum Likelihood) confirmed that Pilocarpus species are monophyletic, and the genus is made up of two major clades ( Fig. 2 ). These two major clades are geographically defined: Clade 1 ( P. microphyllus Stapf ex Wardlew. , P. trachyllophus Holmes , P. carajaensis Skorupa , P. jaborandi Holmes , P. racemosus Vahl , P. peruvianus (J.F. Macbr.) Kaastra ) present in the tropical northern region of Brazil as well as Central America, and Clade 2 ( P. grandiflorus Engl. , P. riedelianus Engl. , P. sulcatus Skorupa , P. spicatus A. St. -Hil., P. giganteus Engl. , and P. pauciflorus A. St. -Hil.) present in mid-coastal to southern regions of Brazil. In addition, these two clades are subtended by P. pennatifolius Lem. , which is determined to be the first branch of the Pilocarpus clade according to both Bayesian and Maximum Likelihood estimates. However, the Parsimony analysis resulted in the placement of both P. pennatifolius and P. peruvianus as the first branch. This uncertainty can be confirmed in both the Bayesian and Maximum Likelihood analyses, which portray less certainty (via posterior probabilities and bootstrap support) when placing P. peruvianus at the base of Clade 1 (Clade with P. microphyllus ) ( Fig. 2 ). There has only been one other study on the phylogeny of Pilocarpus , and it was estimated using Bayesian methods and with the following genes: trn G-S, 5.8S, ITS 1 and 2 ( Oliveira, 2008). In this study, P. peruvianus was placed at the base, and P. pennatifolius was nested within the monophyletic clade of Pilocarpus , as a sister clade of P. spicatus ( Oliveira, 2008) . For our comparative analyses we chose to use the branch lengths from our Bayesian phylogenetic tree, as we calibrated the tree using Ptelea fossils ( Fig. 2c ).

2.2. Diversity of coumarins and alkaloids across Pilocarpus has an inverse relationship

The diversity (total number) of compounds among the species in this clade for both alkaloids and coumarins was reconstructed on the phylogeny using Maximum Likelihood ( Fig. 3a total alkaloids, Fig. 3b total coumarins). The color-coded branches in Fig. 3 indicate low diversity (yellow) to high diversity (blue) of total compounds. In Fig. 3a , two major clades are observed: one with a higher diversity of alkaloids (Clade 1 = P. microphyllus , P. trachyllophus , P. carajaensis , P. jaborandi , and P. racemosus ) and the other with a lower diversity of alkaloids (Clade 2 = P. grandiflorus , P. riedelianus , P. sulcatus , P. spicatus , P. giganteus , and P. pauciflorus ).

Interestingly, the coumarin diversity results are reversed on the phylogeny ( Fig. 3b ). There are still two major clades: one with high coumarin diversity (Clade 2 = P. grandiflorus , P. riedelianus , P. sulcatus , P. spicatus , P. giganteus , and P. pauciflorus ) and the other with lower coumarin diversity (Clade 1 = P. microphyllus , P. trachyllophus , P. carajaensis , P. jaborandi , and P. racemosus ). However, we now see that Clade 1 (the clade with the greatest alkaloid diversity) has the lowest coumarin diversity, and Clade 2 (the clade with the lowest alkaloid diversity) has the greatest coumarin diversity. In addition, P. pennatifolius at the base contains very high numbers of coumarins and the more derived species have the opposite of escalation, a decline in numbers of coumarins, especially for Clade 1.

Phylogenetic signal, a statistical measure assessing phylogenetic dependence, was calculated with Blomberg's K and we found that it was not significant for alkaloid diversity or coumarin diversity (p-value> 0.05). This implies that evolutionary history was not the main contributor for the diversity of compounds, and instead there are other factors affecting the diversity of compounds. One thing to keep in mind is that these two clades are regionally distinct; Clade 1 is mostly found in northern Brazil, the Amazon, and Central America, whereas Clade 2 is mostly found in the southern and eastern regions of Brazil. This distinction is important because as these species dispersed and developed into these different regions they were exposed to a variety of environmental pressures. Evolutionary conservation of naturally occurring compounds comes with a metabolic cost (i.e. carbon), and this cost is a tradeoff that can lead to decreases in other processes such as plant growth ( Nitao and Zangerl, 1987). Therefore, in new environments directional selection of adaptive evolution could lead to a reduction of compounds that were needed in an old environment, as they have a high cost for no reward, and instead lead to an increase in production of beneficial compounds for the new environment, as they improve fitness and survival ( Olson-Manning et al., 2012). One interesting example of possible convergence is the similarity of coumarin and alkaloid diversity in P. sulcatus and P. trachyllophus , two sympatric species growing in the Caatinga of Brazil. It is also important to note that P. pennatifolius , at the base of the Pilocarpus clade, has a higher diversity for both alkaloids and coumarins.

2.3. Reconstruction of coumarin biosynthetic diversity across Pilocarpus

Specialized metabolites can also be considered non-independent traits, since they can be part of the same biosynthetic pathway or network. Although the pathway of imidazole alkaloids is unknown, the relationship between coumarin compounds and pathway enzymes is mostly known or approximated ( Bourgaud et al., 2006). Therefore, we were able to calculate the Sørensen distance between coumarin compounds for each species, taking into consideration the proportion of shared enzymes to account for shared biosynthetic origins, in order to determine whether there was a large correlation between coumarin compounds ( Sørensen, 1948; Junker et al., 2017). Fig. 4 depicts the reconstruction of biosynthetic diversity using the mean of Sørensen distances for each pairwise compound comparison. Clade 2 mostly depicts larger Sørensen distances or few shared enzymes in the pathways for compounds present in Clade 2 species. The majority of species in Clade 1 have small Sørensen distances; therefore, they have more shared enzymes and possibly a greater correlation between compounds present. Overall the majority of species in the Pilocarpus clade have larger mean Sørensen distance values, representing fewer shared biosynthesis enzymes and a lower correlation of coumarin chemical traits ( Sørensen, 1948). A larger Sørensen distance is important because the independence of traits is an important consideration for comparative methods. In addition, the mean Sørensen distance values for these species are approximately in the same range as the values found in previous studies for the correlation of chemical traits in plant species ( Sørensen, 1948; Junker et al., 2017). When we calculated the phylogenetic signal of the mean Sørensen distance we found that it was not significant (λ = 7.58 −0.05 and p-value> 0.05), thus there is a more random distribution of biosynthetic diversity across the genus.

2.4. Reconstruction of pilocarpine reveals a species with the greatest concentration

The discrete (stochastic character mapping of presence/absence) and continuous (Maximum Likelihood analysis of quantitative variation) reconstructions of the important medicinal compound pilocarpine ( A 4) are visualized in Fig. 5 . Fig. 5a depicts the presence/absence of pilocarpine, establishing its presence in all but two species: P. spicatus and P. giganteus . The pilocarpine trait is lost at the orange “/” on the phylogeny and later regained at the black “/“, this is further confirmed by the marginal probabilities at each node. Next, in our continuous trait reconstructions we determined that P. sulcatus had the greatest concentration of pilocarpine, followed by P. trachyllophus and then P. grandiflorus (Star symbol indicates P. sulcatus in Fig. 5b ). It is intriguing that the majority of other species in the same clade as P. sulcatus (Clade 2) have some of the lowest values for concentrations of pilocarpine, including the absence of pilocarpine itself.

As we see in the figures, pilocarpine concentration is not conserved on the phylogeny and appears to be a divergent trait. In this case, there must be other factors affecting pilocarpine expression. A comparison of the distributions of species in Clade 2 confirmed that the distribution of P. sulcatus is distinct. Pilocarpus sulcatus is the only species in Clade 2 that has a limited distribution in the Caatinga vegetative areas of southern Bahia and northern Minas Gerais, and it can be found occurring alongside P. trachyllophus from Clade 1 ( Skorupa, 1996). Although P. sulcatus is one of the smallest shrubs in the genus (height = 1–2 m), its growth habit is unique with a low density of branches and simple leaf clusters present at long internodes. In addition, P. sulcatus differs from other species in Pilocarpus as it has very conspicuous veins, unique perforated pollen, reddish leaf trichomes, and few pellucid dots. This is the first instance that this species has been assessed for imidazole alkaloids, and this study found that P. sulcatus has the largest concentration of pilocarpine compared to the extractions of the 11 other species in this study.

2.5. Reconstruction of chemical traits across Pilocarpus

In Fig. 6 , the discrete character reconstructions for three specific compounds are visualized. In Fig. 6a coumarin (C25) was detected in all of Clade 2 with one subsequent loss on the P. spicatus branch. This is in contrast to Clade 1, which appears to lose this trait at the base of Clade 1, though it is regained by P. trachyllophus . As C25 is the precursor to a variety of coumarins and furanocoumarins in the genus, it is possible that its disappearance in Clade 1 could be due to its role in forming other specialized coumarins further down the pathway. If there were selection for downstream products, the precursor (C25) would be utilized to form other intermediates and compounds along the path, leading to a concentration that is too low to be detected. Since these two major clades are also geographically restricted, it is not possible to determine if these differences are due to random mutation, environmental specialization, or phylogenetic relationships. Next in Fig. 6b for citropten (C27), there appears to be two losses and two gains in each clade. Finally, in Fig. 6c the reconstruction of the coumarin osthol (C29) appears to have a somewhat random distribution across the genus, with multiple gains and losses. Although some compounds were not detected in certain species, it is possible that these species could still produce the compounds. This lack of detection could be due to a variety of reasons including low compound concentrations, developmental mechanisms, and environmental plasticity. All of the continuous chemical coumarin traits were reconstructed on the phylogeny and are visualized in Supplemental Fig. 1 .

Interestingly it appears that C25, C27, and C29 have a similar distribution in Clade 1, being absent in all but one species, P. trachyllophus . One possibility is that P. trachyllophus needed to adapt to the precipitation and temperature extremes, through the expression of more coumarins for defense ( Zobel and Brown, 1995). This could be tested with a common garden experiment, verifying if P. trachyllophus continued to produce these compounds in a tropical humid environment. The other species in Clade 1 could also be tested to evaluate whether translocation to a drier environment could lead to expression of C25, C27, and C29. On the other hand, the absence of C25, C27, and C 29 in Clade 1 species could be a trade-off. All the other species in Clade 1 can be found in tropical humid regions, facing a greater diversity of pathogens and herbivores. These biological pressures could lead to an increase of other beneficial defense compounds, which could be accompanied by a decrease of the C25, C27, and C29 coumarins.

Figures

Fig. 2. Phylogeny of Pilocarpus estimated using three methods. (a) Parsimony in tnt, (b) Maximum Likelihood in RaxML, and (c) Bayesian in BEAST. The red star in (c) refers to the node where two fossils of Ptelea, one from the mid-Eocene and one from mid-Miocene, were used to date the Bayesian tree in BEAST. The calibrated phylogeny suggests a Miocene diversification of Pilocarpus (in yellow). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3. Chemical diversity (total number of compounds) reconstructed across the phylogeny of Pilocarpus is reversed when comparing (a) alkaloid diversity with (b) coumarin diversity. The trait value gradient bar depicts the number of compounds present for that taxon ranging from a low (yellow) to high (blue) number of compounds. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4. Reconstruction of the biosynthetic diversity or similarity of coumarin biosynthetic enzymes in Pilocarpus. The trait value gradient represents the mean of the Sørensen distances for pairwise compounds present in each species, and this exact value is labeled to the right of each taxon name. A small Sørensen distance (yellow) signifies small differences in enzymes, therefore a greater amount of shared enzymes in biosynthesis. A greater Sørensen distance (blue) signifies few or no shared enzymes in the biosynthesis. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5. Pilocarpine trait reconstructions. (a) Stochastic character mapping (SCM) of presence (black)/absence (orange) for pilocarpine. Orange “/” marks loss of the trait and black “/” the trait is gained. (b) Maximum Likelihood analysis of quantitative variation in pilocarpine, with the star symbol referring to the species with the greatest pilocarpine concentration. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6. Ancestral chemical trait reconstructions. Stochastic character mapping (SCM) of discrete presence (black)/absence (orange) for each trait; C25 (one gain and one loss), C27 (two gains and two losses), C29 multiple gains and losses which appears random. (a) C25 = Coumarin (b) C27 = Citropten (c) C29 = Osthol.. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 1. Pilocarpus species collection localities. Dots on the map represent collection sites. The colors of the dots correspond to the species on the phylogeny to the right, with some species collected at multiple localities. *This phylogeny was estimated using Bayesian methods described in this paper. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)