Structure-Dependent Toxicokinetics of Selected Pyrrolizidine Alkaloids In Vitro [original]

Citation: Buchmueller, J.; Kaltner, F.;

Gottschalk, C.; Maares, M.;

Braeuning, A.; Hessel-Pras, S.

Structure-Dependent Toxicokinetics

of Selected Pyrrolizidine Alkaloids In

Vitro. Int. J. Mol. Sci. 2022,23, 9214.

https://doi.org/10.3390/

ijms23169214

Academic Editor: Alessandro

Attanzio

Received: 4 July 2022

Accepted: 13 August 2022

Published: 16 August 2022

Publisher’s Note: MDPI stays neutral

with regard to jurisdictional claims in

published maps and institutional affil-

iations.

Licensee MDPI, Basel, Switzerland.

This article is an open access article

distributed under the terms and

conditions of the Creative Commons

Attribution (CC BY) license (https://

creativecommons.org/licenses/by/

4.0/).

International Journal of

Molecular Sciences

Article

Structure-Dependent Toxicokinetics of Selected Pyrrolizidine

Alkaloids In Vitro

Julia Buchmueller 1, Florian Kaltner 2,3 , Christoph Gottschalk 1,2, Maria Maares 1,4 , Albert Braeuning 1

and Stefanie Hessel-Pras 1,*

1German Federal Institute for Risk Assessment, Max-Dohrn-Str. 8-10, 10589 Berlin, Germany

2Chair of Food Safety, Veterinary Faculty, Ludwig-Maximilians-Universität München, Schoenleutnerstr. 8,

85764 Oberschleissheim, Germany

3Institute of Food Chemistry and Food Biotechnology, Justus Liebig University of Giessen,

35392 Giessen, Germany

4Institute of Food Chemistry and Toxicology, Technische Universität Berlin, Straße des 17. Juni 135,

10623 Berlin, Germany

*Correspondence: stefanie.hessel-pras@bfr.bund.de; Tel.: +49-30-18412-25203

Abstract:

Phytochemicals like pyrrolizidine alkaloids (PAs) can affect the health of humans and

animals. PAs can occur for example in tea, honey or herbs. Some PAs are known to be cytotoxic,

genotoxic, and carcinogenic. Upon intake of high amounts, hepatotoxic and pneumotoxic effects were

observed in humans. This study aims to elucidate different toxicokinetic parameters like the uptake

of PAs and their metabolism with

in vitro

models. We examined the transport rates of differently

structured PAs (monoester, open-chained diester, cyclic diester) over a model of the intestinal barrier.

After passing the intestinal barrier, PAs reach the liver, where they are metabolized into partially

instable electrophilic metabolites interacting with nucleophilic centers. We investigated this process by

the usage of human liver, intestinal, and lung microsomal preparations for incubation with different

PAs. These results are completed with the detection of apoptosis as indicator for bioactivation

of the PAs. Our results show a structure-dependent passage of PAs over the intestinal barrier.

PAs are structure-dependently metabolized by liver microsomes and, to a smaller extent, by lung

microsomes. The detection of apoptosis of A549 cells treated with lasiocarpine and monocrotaline

following bioactivation by human liver or lung microsomes underlines this result. Conclusively, our

results help to shape the picture of PA toxicokinetics which could further improve the knowledge of

molecular processes leading to observed effects of PAs in vivo.

Keywords: pyrrolizidine alkaloids; metabolism; structure-dependency; uptake

1. Introduction

Pyrrolizidine alkaloids (PAs) are a large group of phytochemicals with more than

660 different chemical structures identified, synthesized by a wide variety of plants. PA-

producing plants are, for example, Boraginaceae, Asteraceae or Fabaceae [

]. Humans are

exposed to PAs via the plants themselves or via contaminated food. Prominent exposure

sources are tea, leafy vegetables, herbs, plant-based food supplements or honey [3].

All PAs share a common structure of a necine base esterified with one or two necic

acids [

]. It is generally accepted that 1,2-unsaturated PAs can exert toxic effects to humans

and animals [

]. Several cases of intoxications after uptake of high doses of PAs were

observed. Symptoms like ascites, liver fibrosis, and liver cirrhosis were described as well

as hepatomegaly, and the hepatic sinusoidal obstruction syndrome (HSOS) [

–

]. PAs ex-

hibit cytotoxic, genotoxic, and carcinogenic properties [

]

. To

exert toxicity, PAs require

bioactivation comprising enzymatic oxidation to reactive metabolites. These intermediates

can spontaneously be dehydrated, resulting in reactive pyrrolic esters that interact with

nucleophilic counterparts like proteins or DNA [

]. This reaction happens mainly in

Int. J. Mol. Sci. 2022,23, 9214. https://doi.org/10.3390/ijms23169214 https://www.mdpi.com/journal/ijms

Int. J. Mol. Sci. 2022,23, 9214 2 of 16

the liver, which explains why most PA-caused effects are detected here. Nevertheless,

the lung can also be affected by PAs

. It

is well accepted that some PAs induce pulmonary

arterial hypertension (PAH) [

]. However, PA-mediated toxicity is a consequence of an

interplay of many more factors than metabolism

. Yang

et al. (2001) investigated the oral

absorption of PAs in rats and showed that PAs are passively absorbed [

]. However, for the

PAs senecionine N-oxide, retrorsine N-oxide and lycopsamine N-oxide, an additional efflux

mechanism was detected. A similar effect was investigated by

Hessel et al. (2014)

. They ob-

served that echimidine and heliotrine are transported back into the intestinal lumen by an

active P-glycoprotein (ABCB1)-mediated efflux mechanism [

]

. After

passing the intestinal

epithelium, the PAs are transported to the liver. PAs are actively taken up by human hepatic

HepaRG cells. Thereby, diester PAs demonstrated higher uptake rates than the examined

monoester PAs [

]

. Moreover

, the knockdown of two liver influx transporters (SLC10A1

and SLC22A1) reduced the uptake of retrorsine into human HepaRG cells compared to

control cells without transporter knockdown indicating a considerable influence of liver

transporters in PA-mediated toxicity [

]

. The

role of the uptake transporter solute carrier

(Slc)22a1 (also known as Oct1) in transporting retrorsine into rat hepatocytes and SLC22A1-

overexpressing Madin–Darby canine kidney cells (MDCK)-cells was also examined by

Tu et al. (2014) The results indicate that the transporter plays an important role in trans-

porting retrorsine [

]

. Monocrotalin

e was also demonstrated to be transported by Slc22a1

into MDCK cells [

]. In the hepatocytes, PAs are structure-dependently bioactivated by

cytochrome P450 (CYP) monooxygenases [20–22].

Taken together, PA-mediated toxicity depends on a variety of factors comprising oral

bioavailability, different influx and efflux transport processes in the intestine and liver,

structure-activity relationships in bioactivation, different distribution processes within the

body, and maybe some more yet unexplored factors.

In the present study, we aimed to elucidate the efficiency of the passage over a model

for the intestinal barrier with structurally different PAs (Figure 1). To include organ-specific

metabolism into this study, we analysed the metabolism rate of the intestine, liver, and

lung using human microsomal fractions in order to associate possible (reactive) metabolite

formation with toxic cellular effects.

Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 2 of 17

described as well as hepatomegaly, and the hepatic sinusoidal obstruction syndrome

(HSOS) [6–8]. PAs exhibit cytotoxic, genotoxic, and carcinogenic properties [1,4,9]. To

exert toxicity, PAs require bioactivation comprising enzymatic oxidation to reactive

metabolites. These intermediates can spontaneously be dehydrated, resulting in reactive

pyrrolic esters that interact with nucleophilic counterparts like proteins or DNA [10,11].

This reaction happens mainly in the liver, which explains why most PA-caused effects are

detected here. Nevertheless, the lung can also be affected by PAs. It is well accepted that

some PAs induce pulmonary arterial hypertension (PAH) [12,13]. However, PA-mediated

toxicity is a consequence of an interplay of many more factors than metabolism. Yang et

al. (2001) investigated the oral absorption of PAs in rats and showed that PAs are passively

absorbed [14]. However, for the PAs senecionine N-oxide, retrorsine N-oxide and

lycopsamine N-oxide, an additional efflux mechanism was detected. A similar effect was

investigated by Hessel et al. (2014). They observed that echimidine and heliotrine are

transported back into the intestinal lumen by an active P-glycoprotein (ABCB1)-mediated

efflux mechanism [15]. After passing the intestinal epithelium, the PAs are transported to

the liver. PAs are actively taken up by human hepatic HepaRG cells. Thereby, diester PAs

demonstrated higher uptake rates than the examined monoester PAs [16]. Moreover, the

knockdown of two liver influx transporters (SLC10A1 and SLC22A1) reduced the uptake

of retrorsine into human HepaRG cells compared to control cells without transporter

knockdown indicating a considerable influence of liver transporters in PA-mediated

toxicity [17]. The role of the uptake transporter solute carrier (Slc)22a1 (also known as

Oct1) in transporting retrorsine into rat hepatocytes and SLC22A1-overexpressing

Madin–Darby canine kidney cells (MDCK)-cells was also examined by Tu et al. (2014) The

results indicate that the transporter plays an important role in transporting retrorsine [18].

Monocrotaline was also demonstrated to be transported by Slc22a1 into MDCK cells [19].

In the hepatocytes, PAs are structure-dependently bioactivated by cytochrome P450

(CYP) monooxygenases [20–22].

Taken together, PA-mediated toxicity depends on a variety of factors comprising oral

bioavailability, different influx and efflux transport processes in the intestine and liver,

structure-activity relationships in bioactivation, different distribution processes within

the body, and maybe some more yet unexplored factors.

In the present study, we aimed to elucidate the efficiency of the passage over a model

for the intestinal barrier with structurally different PAs (Figure 1). To include organ-

specific metabolism into this study, we analysed the metabolism rate of the intestine, liver,

and lung using human microsomal fractions in order to associate possible (reactive)

metabolite formation with toxic cellular effects.

Figure 1.

Structures of the PAs used in this study. The PAs were chosen to represent the different struc-

tures: monoester (intermedine and heliotrine), open-chained diester (echimidine and lasiocarpine)

and cyclic diester (monocrotaline, senkirkine, senecionine, and retrorsine).

Int. J. Mol. Sci. 2022,23, 9214 3 of 16

2. Results

2.1. Intestinal Transfer Rate of PAs Is Structure-Dependent

The intestinal passage influences the oral bioavailability and thus the amount of

a substance at the target structure

. The

passage of heliotrine, echimidine, senecionine,

and senkirkine over the intestinal Caco-2 cell monolayer has already been published

Hessel et al. (2014) [15]. However

, to get a broader overview concerning structure-

dependency of the passage, we here used the additional PAs intermedine, retrorsine,

lasiocarpine, and monocrotaline to complete the results with further PA-representatives

from each structure type (monoester, open-chained diester and diester). As in the afore-

mentioned study, we used the differentiated human Caco-2 cell model mimicking the small

intestine, to investigate which PAs pass the intestinal barrier and to which extent.

The results of the transport assay (Figure 2) illustrated a structure-dependent pas-

sage of PAs. The monoester PAs (intermedine and heliotrine) together with echimidine

showed a low recovery rate (12.3%, 32.3%, and 13.1%) in the basolateral compartment

when administered apically. The open-chained diester PA lasiocarpine revealed a high

passage rate comparable with the rates of the cyclic diester PAs (monocrotaline, senkirkine,

senecionine, and retrorsine). The investigated PAs revealed no toxic effects in the concen-

tration used. This was verified with cytotoxicity testing, as well as with measurements

of the transepithelial electrical resistance (TEER) as indicator for monolayer integrity

(Supplementary Material Figure S1).

Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 3 of 17

Figure 1. Structures of the PAs used in this study. The PAs were chosen to represent the different

structures: monoester (intermedine and heliotrine), open-chained diester (echimidine and

lasiocarpine) and cyclic diester (monocrotaline, senkirkine, senecionine, and retrorsine).

2. Results

2.1. Intestinal Transfer Rate of PAs Is Structure-Dependent

The intestinal passage influences the oral bioavailability and thus the amount of a

substance at the target structure. The passage of heliotrine, echimidine, senecionine, and

senkirkine over the intestinal Caco-2 cell monolayer has already been published by Hessel

et al. (2014) [15]. However, to get a broader overview concerning structure-dependency

of the passage, we here used the additional PAs intermedine, retrorsine, lasiocarpine, and

monocrotaline to complete the results with further PA-representatives from each

structure type (monoester, open-chained diester and diester). As in the aforementioned

study, we used the differentiated human Caco-2 cell model mimicking the small intestine,

to investigate which PAs pass the intestinal barrier and to which extent.

The results of the transport assay (Figure 2) illustrated a structure-dependent passage

of PAs. The monoester PAs (intermedine and heliotrine) together with echimidine showed

a low recovery rate (12.3%, 32.3%, and 13.1%) in the basolateral compartment when

administered apically. The open-chained diester PA lasiocarpine revealed a high passage

rate comparable with the rates of the cyclic diester PAs (monocrotaline, senkirkine,

senecionine, and retrorsine). The investigated PAs revealed no toxic effects in the

concentration used. This was verified with cytotoxicity testing, as well as with

measurements of the transepithelial electrical resistance (TEER) as indicator for

monolayer integrity (supplementary material Figure S1).

Figure 2. Transfer rates of structurally different PAs over the differentiated Caco-2 cell monolayer.

Caco-2 cells were seeded and differentiated in Transwell inserts and incubated with 0.25 µM of the

specific PAs from the apical compartment of the Transwell chamber. At the indicated time points,

the basolateral PA amounts were determined with LC-MS/MS. Mean values ± SD of the PA

concentrations in the basolateral compartment in % in comparison to the applied PA concentration

are shown from three individual experiments.

Figure 2.

Transfer rates of structurally different PAs over the differentiated Caco-2 cell monolayer.

Caco-2 cells were seeded and differentiated in Transwell inserts and incubated with 0.25

M of

the specific PAs from the apical compartment of the Transwell chamber. At the indicated time

points, the basolateral PA amounts were determined with LC-MS/MS. Mean values

SD of the PA

concentrations in the basolateral compartment in % in comparison to the applied PA concentration

are shown from three individual experiments.

2.2. Human Intestinal Microsomes Do Not Significantly Degrade PAs

In the human intestinal tract, the expression of several xenobiotic metabolizing en-

zymes like CYP3A4, CYP2C9, SULT1A, and UGT1A and others were demonstrated. Caco-2

cells show less phase I enzyme activities than human intestinal tissue

. Therefore

, we ex-

amined the metabolism of PAs with human intestinal microsomes incubated with each

PA. The remaining concentration in comparison to the initial concentration at t = 0 h was

determined with LC-MS/MS.

Int. J. Mol. Sci. 2022,23, 9214 4 of 16

The results in Figure 3illustrated that the remaining PA concentrations after incubation

with human intestinal microsomes compared to t = 0 h were not significantly reduced.

However, some tendencies were observable, as e.g., lasiocarpine and senkirkine showed

slight decreases of the concentration with considerable standard deviations.

Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 4 of 17

2.2. Human Intestinal Microsomes Do Not Significantly Degrade PAs

In the human intestinal tract, the expression of several xenobiotic metabolizing en-

zymes like CYP3A4, CYP2C9, SULT1A, and UGT1A and others were demonstrated.

Caco-2 cells show less phase I enzyme activities than human intestinal tissue. Therefore,

we examined the metabolism of PAs with human intestinal microsomes incubated with

each PA. The remaining concentration in comparison to the initial concentration at t = 0 h

was determined with LC-MS/MS.

The results in Figure 3 illustrated that the remaining PA concentrations after incuba-

tion with human intestinal microsomes compared to t = 0 h were not significantly reduced.

However, some tendencies were observable, as e.g., lasiocarpine and senkirkine showed

slight decreases of the concentration with considerable standard deviations.

Figure 3. Reduction of the concentration of PAs after incubation with human intestinal microsomes.

10 µM PAs were incubated with microsomes at 37 °C for up to 4.5 h. At the indicated time points,

the remaining PA concentration in comparison to t = 0 h was determined. For each incubation three

individual experiments were performed. Mean values ± SD are shown. Statistical differences to t =

0 h were determined with One-Way ANOVA followed by Dunnett’s post hoc test.

2.3. Human Liver Microsomes Reduce the Concentration of PAs in a Structure-Dependent

Manner

Reduction of the concentration of structurally different PAs with human liver micro-

somes were investigated to get more information about possible structure-activity rela-

tionships. Therefore, human liver microsomes were incubated with each PA, and the re-

maining PA concentrations in comparison to t = 0 h were determined with LC-MS/MS

(Figure 4).

The results illustrated different reduction of the concentrations for the specific PAs.

The remaining PA concentration after 4.5 h of incubation with human liver microsomes

varied between 37.4% ± 10.8 for lasiocarpine and 87.3% ± 5.1 for monocrotaline. The con-

centration reduction seemed to be influenced by structural properties. The diester PAs

echimidine, lasiocarpine, senkirkine, and senecionine demonstrated high degradation

rates. Whereas the monoester PAs intermedine, and heliotrine revealed no significant re-

duction of the initial concentration. However, no significant reduction of the concentra-

tion was observed for the cyclic diester PAs monocrotaline and retrorsine.

Figure 3.

Reduction of the concentration of PAs after incubation with human intestinal microsomes.

M PAs were incubated with microsomes at 37

◦

C for up to 4.5 h. At the indicated time points,

the remaining PA concentration in comparison to t = 0 h was determined. For each incubation three

individual experiments were performed. Mean values

SD are shown. Statistical differences to

t = 0 h were determined with One-Way ANOVA followed by Dunnett’s post hoc test.

2.3. Human Liver Microsomes Reduce the Concentration of PAs in a Structure-Dependent Manner

Reduction of the concentration of structurally different PAs with human liver mi-

crosomes were investigated to get more information about possible structure-activity

relationships. Therefore, human liver microsomes were incubated with each PA, and the

remaining PA concentrations in comparison to t = 0 h were determined with LC-MS/MS

(Figure 4).

The results illustrated different reduction of the concentrations for the specific PAs.

The remaining PA concentration after 4.5 h of incubation with human liver microsomes

varied between 37.4%

10.8 for lasiocarpine and 87.3%

5.1 for monocrotaline. The

concentration reduction seemed to be influenced by structural properties. The diester PAs

echimidine, lasiocarpine, senkirkine, and senecionine demonstrated high degradation rates.

Whereas the monoester PAs intermedine, and heliotrine revealed no significant reduction

of the initial concentration. However, no significant reduction of the concentration was

observed for the cyclic diester PAs monocrotaline and retrorsine.

Int. J. Mol. Sci. 2022,23, 9214 5 of 16

Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 5 of 17

Figure 4. Reduction of PA concentrations after incubation with human liver microsomes. 10 µM

PAs were incubated with liver microsomes at 37 °C for up to 4.5 h. At the indicated time points, the

remaining PA concentration in comparison to t = 0 h was determined. For each incubation three

independent experiments were performed. Mean values ± SD are shown. Statistical significance was

determined with One-Way ANOVA followed by Dunnett’s post hoc test and is indicated as * p <

0.05, ** p < 0.01.

2.4. Human Lung Microsomes Reduce the Concentration of PAs in a Structure-Dependent

Manner

As secondary target organ, the lung is exposed to PAs after they passed the intestinal

tract and the liver. In the lung as well as in the intestine CYP enzymes are expressed

[23,24]. For this reason, reduction of the concentration of PAs with human lung micro-

somes was determined in order to investigate whether PAs are possibly metabolized also

by extrahepatic tissues containing CYP enzymes.

The incubation of PAs with human lung microsomes revealed differences in the re-

duction of concentration (Figure 5). The monoester PAs intermedine and heliotrine re-

vealed only slight reduction of the initial concentrations (89.9% ± 2.3, 71.9% ± 11.4)

whereas the incubation with the diester PAs lasiocarpine, senkirkine, senecionine, mono-

crotaline, and echimidine demonstrated a more pronounced reduction of the PA amount

in comparison to t = 0 h (55.8% ± 24.9, 56.2% ± 12.4, 70.6% ± 20.9, 75.9% ± 3.6, 67.8% ± 19.9).

Even though, the reduction of senecionine is not statistically significant due to the high

standard deviations.

Figure 4.

Reduction of PA concentrations after incubation with human liver microsomes. 10

M PAs

were incubated with liver microsomes at 37

◦

C for up to 4.5 h. At the indicated time points, the remaining

PA concentration in comparison to t = 0 h was determined. For each incubation three independent

experiments were performed. Mean values

SD are shown. Statistical significance was determined

with One-Way ANOVA followed by Dunnett’s post hoc test and is indicated as * p< 0.05, ** p< 0.01.

2.4. Human Lung Microsomes Reduce the Concentration of PAs in a Structure-Dependent Manner

As secondary target organ, the lung is exposed to PAs after they passed the intestinal

tract and the liver. In the lung as well as in the intestine CYP enzymes are expressed [

For this reason, reduction of the concentration of PAs with human lung microsomes

was determined in order to investigate whether PAs are possibly metabolized also by

extrahepatic tissues containing CYP enzymes.

The incubation of PAs with human lung microsomesrevealed differences in the reduction

of concentration (Figure 5). The monoester PAs intermedine and heliotrine revealed only slight

reduction of the initial concentrations (89.9%

2.3, 71.9%

11.4) whereas the incubation

with the diester PAs lasiocarpine, senkirkine, senecionine, monocrotaline, and echimidine

demonstrated a more pronounced reduction of the PA amount in comparison to t = 0 h

(

55.8% ±24.9

, 56.2%

12.4, 70.6%

20.9, 75.9%

3.6, 67.8%

19.9). Even though, the

reduction of senecionine is not statistically significant due to the high standard deviations.

Int. J. Mol. Sci. 2022, 23, x FOR PEER REVIEW 6 of 17

Figure 5. Reduction of concentration of PAs after incubation with human lung microsomes. 10 µM

PAs were incubated with human lung microsomes at 37 °C for up to 4.5 h. At the indicated time

points, the remaining PA concentration compared to t = 0 h was determined. For each incubation

three independent experiments were performed. Mean values ± SD are shown. Statistical signifi-

cance was determined with One-Way ANOVA followed by Dunnett’s post hoc test and is indicated

as * p < 0.05, ** p < 0.01.

2.5. Human Liver and Lung Microsomes Can Bioactivate Lasiocarpine and Monocrotaline to

Induce Apoptosis in A549 Cells

Based on the result that liver and lung microsomes do efficiently metabolize lasiocar-

pine, we examined if both organs are able to bioactivate the PAs. Thereby we used apop-

tosis as marker for cellular damage in the human lung cell line A549. Lasiocarpine was

chosen as one of the most hepatotoxic PA [25–27]. Moreover, we used monocrotaline as a

well-characterized PA candidate able to induce lung damage [13,28,29]. The PAs were

incubated with human liver or lung microsomes and applied to A549 cells, which were

subsequently investigated regarding apoptosis.

The results illustrated that the number of cells indicating apoptotic and necrotic prop-

erties increased when incubated with liver microsomes and PAs (Figure 6A). Lasiocarpine

in combination with liver microsomes induced a stronger effect than monocrotaline incu-

bated with liver microsomes. 150 µM lasiocarpine incubated with liver microsomes in-

duced an increase of apoptotic cells from 1.5% ± 0.8 in the control treatment to 30.3% ±

11.8. The number of apoptotic cells increased further with 250 µM lasiocarpine to 36.7% ±

11.2. Similarly, the number of necrotic cells was increased upon the incubation with 150

µM and 250 µM lasiocarpine with liver microsomes from 1.9% ± 0.9 for cells in the control

treatment to 18.8% ± 11.8 and 21.7% ± 11.2, respectively. Lung microsomes together with

150 µM lasiocarpine induced an increase of apoptotic cells to 5.6% ± 3.6 and 2.3% ± 0.7 of

necrotic cells. This effect was comparable to the effect of 250 µM lasiocarpine with lung

microsomes (apoptotic cells: 4.2% ± 1.3 necrotic cells: 3.5% ± 1.3). The induction of Caspase

3 activity also revealed the ability of lasiocarpine to induce apoptosis when incubated with

liver and lung microsomes. However, the differences in the effect strength between liver

and lung microsomes are less pronounced compared to the results of the FACS measure-

ments. The induction of Caspase 3 activity is comparable for the incubations with liver

and lung microsomes. This effect can be explained with the different endpoints measured.

Figure 5. Cont.

Int. J. Mol. Sci. 2022,23, 9214 6 of 16