LUND UNIVERSITY PO Box 117 221 00 Lund
Uptake and bioaccumulation of ionizable pharmaceuticals in aquatic organisms
Boström, Marja L.
2019
Document Version:
Publisher's PDF, also known as Version of record
Link to publication
Citation for published version (APA):
Boström, M. L. (2019). Uptake and bioaccumulation of ionizable pharmaceuticals in aquatic organisms. Media-Tryck, Lund University, Sweden.
Total number of authors: 1
General rights
Unless other specific re-use rights are stated the following general rights apply:
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal
Read more about Creative commons licenses: https://creativecommons.org/licenses/ Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
M A R JA L . B O ST R Ö M U pt ak e a nd b io ac cu m ula tio n o f i on iza ble p ha rm ac eu tic als i n a qu ati c o rg an ism s 20 19 MARJA L. BOSTRÖM
DEPARTMENT OF BIOLOGY | FACULTY OF SCIENCE | LUND UNIVERSITY
Uptake and bioaccumulation of ionizable
pharmaceuticals in aquatic organisms
Uptake and bioaccumulation of ionizable
pharmaceuticals in aquatic organisms
Uptake and bioaccumulation of ionizable
pharmaceuticals in aquatic organisms
Marja L. Boström
DOCTORAL DISSERTATION
by due permission of the Faculty of Science, Lund University, Sweden. To be defended in the Blue Hall, Ecology Building, Sölvegatan 37, Lund
on Friday 6th of December, 2019, at 9:00 a.m.
Faculty opponent
Prof. Katrine Borgå
Organization LUND UNIVERSITY
Document name: Doctoral Dissertation Department of Biology Date of issue: 12.11.2019
Author: Marja L. Boström Sponsoring organization
Titel: Uptake and bioaccumulation of ionizable pharmaceuticals in aquatic organisms Abstract
Pharmaceuticals are found at low concentrations (ng/L) in aquatic environments but bioaccumulation may result in aquatic organisms reaching internal effect levels (µg/L). Environmental hazard assessments include standardized bioaccumulation tests but contrary to the model substances around which the frameworks are built most pharmaceuticals are designed to mimic endogenic chemicals, ionizable, and less lipophilic. Hence, if using the same frameworks one may over- or underestimate hazard. I used the serotonin reuptake inhibitors (SSRIs) fluoxetine and sertraline, both weak bases, and the non-steroidal anti-inflammatory drugs (NSAIDs) ketoprofen, naproxen, diclofenac and ibuprofen, all four weak acids to evaluate possible over- or underestimation in hazard assessments. Also, to quantify the pharmaceuticals in organism tissue I developed a hollow fiber liquid phase microextraction (HF-LPME) method. The enrichment factor was high, 1900-3000 times, thus, the method is applicable for quantification at environmentally relevant concentrations.
Misestimation of predicted pharmaceutical bioaccumulation may be due to: pH-dependent uptake. Degree
of uncharged molecule uptake is greater than for ions and water pHs decreasing ionization will increase bioaccumulation and, thereby, also toxicity. Environmental pH typically ranges between 6 and 9 but hazard assessments are usually performed using toxicity data determined at one pH only. Using data from Daphnia magna toxicity testing at pH 7 and a pH distribution data set with over 4000 European running waters, I took a probabilistic modelling approach to study misestimations of hazard. European waters are often slightly basic and the model predicted underestimation by a median factor of 3 for the bases (90% of the results ranging from 1 to 6) and overestimation by a factor of 2 for acids (90% of the results ranging from 0.03 to 5). Because aquatic pH exhibited large variation both within and between countries, I advise the use of site-specific risk assessments for ionizable pharmaceuticals when making water management decisions. Organisms adapting to living in chronically polluted waters by reducing bioaccumulation. I compared fluoxetine bioaccumulation in a fish
population (Rutilus rutilus) residing in a by wastewater polluted environment to a population living upstream the polluted site. Bioaccumulation in fish from the polluted site was 10% lower than in fish upstream, and this still remained after exposing detoxified fish. This indicates adaptation and because it was not temporary, suggests alterations on a heritable genetic level. Consideration of the influence of pollution history on bioconcentration in hazard assessments could be called for, as identical experimental and environmental external exposure concentrations may result in different internal exposure. The standardized hazard assessment test species not being the ones bioaccumulating the most. Dietary transfer is an important route of uptake for the early
model substances and may result in trophic accumulation, but published data are inconclusive concerning such importance for pharmaceuticals. To study possible trophic transfer, I exposed two three-level aquatic food chains (leaf detritus, Acer platanoides; fed to Asellus aquaticus; in turn fed to Notonecta glauca or Pungitius pungitius) to the SSRIs. Bioaccumulation was 20-50% lower at higher trophic levels, indicating that dietary transfer is not of importance for internal concentrations. Organisms at low trophic levels had the highest internal concentrations, suggesting importance for their inclusion in hazard assessments.
My results conclude that to make informed water management decisions site specific conditions such as pH and history of pollution need to be considered if not to over- or underestimate hazard. Also, standardized bioaccumulation test species may not be the ones reaching the highest internal concentrations in the wild and hazard may, consequently, become underestimated.
Key words: Antidepressant, BAF, BCF, Bioaccumulation, Bioconcentration, Biomagnification, CYP, Extraction technique, HF-LPME, Ionizable pharmaceutical, Metabolism, pH, Risk assessment, Tolerance, Wastewater Supplementary bibliographical information Language: English
ISSN and key title ISBN 978-91-7895-302-8 (print) ISBN 978-91-7895-303-5 (pdf) Recipient’s notes Number of pages 135 Price
Security classification
I, the undersigned, being the copyright owner of the abstract of the above-mentioned dissertation, hereby grant to all reference sources permission to publish and disseminate the abstract of the above-mentioned dissertation.
Uptake and bioaccumulation of ionizable
pharmaceuticals in aquatic organisms
Cover and included illustrations by Marja L. Boström Copyright Marja L. Boström
Paper I, II, and IV are reprinted with permission from the publisher. Faculty of Science
Department of Biology
ISBN 978-91-7895-302-8 (print) ISBN 978-91-7895-303-5 (pdf)
Tryckt i Sverige av Media-Tryck, Lunds universitet Lund 2019
“Please relax,” said the voice pleasantly, like a stewardess in an airliner with only one wing and two engines one of which is on fire,
“you are perfectly safe.” – Douglas Adams, (1979) The Hitchhiker's Guide to the Galaxy
“Enthusiasm is followed by disappointment and even depression, and then by renewed enthusiasm.”
Table of contents
Popular science summary ...10
Populärvetenskaplig sammanfattning ...12 List of figures ...14 List of tables ...16 List of abbreviations ...17 List of papers ...18 Author contributions ...19
Scope and aim...20
Introduction ...24
Bioaccumulation of pharmaceuticals ...27
Ionization and pH-dependency ...27
Adapting to pollution ...30 Trophic transfer ...32 Extraction method ...36 Conclusion ...39 Future perspectives ...40 References ...42 Acknowledgement ...54 Paper I ...57 Paper II ...67 Paper III ...83 Paper IV ...107
Popular science summary
Pharmaceuticals that are excreted by consumers are incompletely eliminated in wastewater treatment plants, and the major source of pharmaceutical pollution in aquatic environments. Usually the environmental water concentrations are low, but aquatic organisms can accumulate some pharmaceuticals so that the internal concentration becomes higher than that in the surrounding water, denoted bioaccumulation. Internal concentrations may even reach levels known to cause effects like those seen in humans or other unexpected effects. In hazard assessments performed before market release bioaccumulation tests are included but the frameworks are built around chemicals similar to but not completely alike most pharmaceuticals. To evaluate possible over- or underestimation when performing hazard assessments using the current framework, bioaccumulation of painkillers (active substances: diclofenac, ibuprofen, naproxen and ketoprofen) and antidepressants (active substances: sertraline and fluoxetine) by aquatic organisms was studied in this thesis.
The pharmaceuticals addressed are ionizable, that is, the chemicals can exist in a neutral or a charged (positive or negative) form. The degree to which chemicals are present in each form is dependent on environmental pH. The uncharged form is more easily taken up by aquatic organisms and, consequently, a water pH favoring this form increases the risk of internal concentrations reaching effect levels. Environmental pH can vary due to time of the year, time of day and soil conditions for instance. When assessing the possible hazard of chemicals in the environment, toxicity and bioaccumulation tests are usually performed at only one pH which might not match the one in the environment where it is released. Results in this thesis show that if hazard assessments are done at a neutral pH 7, underestimation of hazard in European waters may occur for the basic antidepressants while the acidic painkillers may be overestimated.
Organisms can adapt to a polluted environment by reducing bioaccumulation of chemicals, thereby reducing the risk of internal concentrations reaching effect levels. Adaptation by reducing pharmaceutical bioaccumulation may occur in waters receiving wastewater treatment effluent, but this is not well studied. In this thesis a fish population residing downstream of a wastewater treatment plant showed lower bioaccumulation of fluoxetine than a population living upstream, which suggests adaptation. The bioaccumulation was lower even after the downstream population was detoxified (kept in clean water for a week) before
11 fluoxetine exposure. The latter result suggests that the adaptation is on a genetic level, which would be heritable and not just developed temporarily by individuals.
Chemicals around which the hazard assessment framework is built reach higher concentrations in organisms that are higher up in the food chain, as exposed food adds to internal concentrations. However, it is unclear if this is also true for pharmaceuticals since published research is inconclusive. The results in this thesis show the contrary – in two aquatic food chains where all organisms were exposed at the same time and fed to the trophic level above, invertebrates low in the food chain had the highest internal concentrations. This suggests that food transfer does not add to bioaccumulation for the tested pharmaceuticals.
In summary, ionizable pharmaceutical hazard assessments may result in misestimation if: 1) pH in the environment differs from that used in bioaccumulation tests; 2) organisms in long-term exposed waters have adapted by reducing bioaccumulation; and 3) only organisms from high trophic levels are included. Hence, it is crucial that policymakers take this into account to make informed water management decisions.
Populärvetenskaplig sammanfattning
Främsta källan till läkemedel i miljön är att vi människor inte bryter ner det vi stoppar i oss och att avloppsreningsverk inte heller helt eliminerar resterna i reningsprocessen. Koncentrationerna i naturen är dock ofta låga men vattenlevande organismer kan ackumulera vissa läkemedel så koncentrationen i deras kroppar blir högre än i omgivningen, så kallad bioackumulation. Bioackumulation kan leda till att den invärtes koncentrationen blir så hög att det ger effekter liknande de hos människan eller andra helt oväntade effekter. Miljöriskanalyser, där bioackumu-lationstester inkluderas, genomförs innan läkemedel släpps ut på marknaden men ramverket är uppbyggt runt kemiska ämnen som till viss del, men inte helt, är lika de flesta läkemedlen. I denna avhandling har bioackumulering av värktabletter (verksamma substanser: diklofenak, ibuprofen, naproxen och ketoprofen) och antidepressiv medicin (verksamma substanser: sertralin och fluoxetin) hos vattenlevande organismer studerats för att utvärdera om användandet av nuvarande ramverk leder till att man riskerar att under- eller övervärdera risk.
Gemensamt för läkemedlen i avhandlingen är att de är joniserbara, det vill säga, de har en oladdad och en laddad form (positiv eller negativ). Hur mycket som finns av varje form varierar med pH i miljön. Oladdade ämnen tas lättare upp av vattenlevande organismer, så vid pH-värden i vattnet där denna form dominerar ökar upptaget och därmed risken att läkemedlet ger effekt. Naturligt varierar pH i vattenmiljöer med till exempel årstid, tid på dygnet och markförhållanden. Man använder oftast bara ett pH när man genomför tester som ligger till grund för riskbedömningar. Resultaten i denna avhandling visar att om riskbedömnings-testerna endast utförs vid neutralt pH 7 så riskerar man att i vattendrag i Europa undervärdera risken för de basiska antidepressiva medicinerna och övervärdera risken för de sura värktabletterna.
Organismer kan anpassa sig att leva i förorenade miljöer genom att minska bioackumuleringen av de ämnen som finns där, och därmed risken att invärtes koncentrationer når nivåer som ger effekt. Vattenlevande organismer som lever nedströms avloppsreningsverk skulle alltså potentiellt kunna anpassa sig genom att minska bioackumulering av läkemedelsrester. I avhandlingen visas att en mört-population nedströms ett avloppsreningsverk bioackumulerar fluoxetin till en lägre grad än en som lever uppströms. Om mörten som lever nedströms genomgick en avgiftning, det vill säga de hölls i rent vatten i en vecka, innan de blev exponerade så var bioackumulationen fortfarande lägre. Det senare skulle kunna tyda på att
13 anpassningen inte är individuell och tillfällig utan kan ligga på gennivå, vilket betyder att den är ärftlig.
De kemiska substanser som ramverket kring riskanalyser byggts upp runt når högre koncentrationer i djur högre upp i näringskedjan då exponerad föda adderar till bioackumulationen. Om detta också är sant för läkemedel är oklart. Någon slutsats kan inte dras från publicerade studier men resultaten i denna avhandling visar att när hela näringskedjor exponeras så sker ingen ökning uppåt till nästa födonivå. Snarare är koncentrationerna högre längre ner i näringskedjan.
Slutsatsen från denna avhandling är att riskbedömningar för joniserbara läkemedel kan bli missvisande om: 1) pH i miljön skiljer sig från det i laboratorie-tester, 2) organismer har minskat bioackumuleringen för att anpassa sig till ett liv i vatten där läkemedel kontinuerligt tillförs, 3) man bara gör bioackumulationstester på djur högt upp i näringskedjan. Det är av högsta vikt att detta tas i beaktning av beslutsfattare när de ska ta vattenförvaltningsbeslut.
List of figures
Figure 1. A. Major routes/factors associated with uptake and elimination of xenobiotics, and thus bioaccumulation. B. To quantify ionizable chemical concentrations (NSAIDs and SSRIs) in fish tissue an extraction method was developed in Paper I, to be used in Paper III and Paper IV. In Paper II, a
model was conducted to evaluate how spatial variability in pH may over- or underestimate environmental hazard in aquatic systems. The potential of adaptation of metabolic enzyme systems alongside reduced bioaccumulation was studied in organisms residing under historically long-term effluent pressure in
Paper III. In Paper IV the potential of trophic transfer of SSRIs was assessed
and empirically derived values from that study were also used to test a model predicting ionizable chemical bioaccumulation. ... 25 Figure 2. A visualization of how (A) fluoxetine, a weak base with pKa 10.1, and (B)
diclofenac, a weak acid with pKa 4.18, changes fraction (f, y-axis) of speciation
with increasing pH (x-axis). The solid line represents the uncharged speciation and the dashed line the charged speciation. The pH-dependent fractioning of weak base and weak acid pharmaceuticals is typically calculated using the acid dissociation constant, pKa, in a modified version of the Henderson-Hasselbalch
equation
𝑝𝑝𝑝𝑝 = 𝑝𝑝𝐾𝐾𝑎𝑎+ 𝑙𝑙𝑙𝑙𝑙𝑙10�∝∝𝐴𝐴−𝐴𝐴𝐴𝐴� (Eq 1)
where αA- and αAH are the ionic and uncharged fraction, respectively. ... 28
Figure 3. A simulated European river pH distribution (pH on x-axis, fraction on y-axis) using a data set from 4279 river sites spread over 21 countries. Hazard assessments made from toxicity tests performed at pH 7 only (indicated by the parallel lines) will potentially underestimate toxicity for both acids and bases, as not all rivers have neutral pH. As the pH distribution in Europe is more towards the basic side, toxicity of basic chemicals may more often be underestimated than acids. ... 29 Figure 4. Boxplot showing roach (Rutilus rutilus) fluoxetine logBCFs (y-axis, L/kg)
for the three different fluoxetine exposure treatments. Fish were exposed (7 d) either directly upon arrival from the upstream (A) and the downstream site (B), or after 7 d of detoxification of the fish from the downstream site (C). Different letters (a, b and c) in the boxplot denote significant differences among the exposures (Kruskal-Wallis test, n =6-9, followed by Mann-Whitney U tests). ... 32
15 Figure 5. Boxplots showing log transformed BCF (Eq 3) and BAF (Eq 8) values (L/kg) for (A) sertraline, and (B) fluoxetine, derived for the different species in
Paper IV. Trophic transfer was tested by using two three-level food-chains; Acer
platanoides (maple), fed to Asellus aquaticus (water hoglouse), in turn fed to Notonecta glauca (water boatman) or Pungitius pungitius (nine-spined
stickleback). * represents a statistically significant difference between groups,
p <0.05, ** = p <0.01 and *** = p <0.001 (Tukey test with 95% family-wise
confidence level). BAFs being lower than BCFs and BAFs decreased with increasing trophic level (x-axis, left to right) interprets as dietary exposure not adding to internal concentration for either SSRI. ... 35 Figure 6. Ion trapping is used to extract ionizable chemicals (here a weak acid) in
the hollow fiber liquid phase microextraction technique (HF-LPME). By soaking the hollow fiber (dashed structure) in organic solvent it allows uncharged organic chemicals, HA, to pass (yellow, straight arrow) but hindering ions, A- (yellow, curved arrow). The pH in the donor solution, i.e. the sample, is shifting the dissociation equilibrium towards the uncharged speciation. In the acceptor phase, i.e. liquid to be analyzed, the pH shifts to favor the ionic speciation. This results in enrichment of the chemical in the acceptor solution. ... 37 Figure 7. A hollow fiber prepared for extracting. The copper wire is used to weigh
the fiber down for total submergence. Fiber size: wall thickness 50 μm; pore size 0.1 μm; inner diameter 280 μm; and length 20 cm (Membrana GmbH, Wuppertal, Germany). Image source: Henrik Engström. ... 37
List of tables
Table 1. Chemical information of the four NSAIDs and the two SSRIs studied. MW = molecular weight, pKa = acid dissociation constant, logKow = logarithm of the
octanol/water partition coefficient, PBT = hazard score.
17
List of abbreviations
BAF Bioaccumulation factor BCF Bioconcentration factor BMF Biomagnification factor CYP Cytochrome P450 (protein)
cyp Cytochrome P450 (gene)
DDT Dichlorodiphenyltrichloroethane
HF-LPME Hollow fiber liquid phase microextraction
Kow 1-octanol/water partition coefficient
NSAID Non-steroidal anti-inflammatory drug PCB Polychlorinated biphenyl
pKa Acid dissociation constant
SPE Solid phase extraction SPME Solid phase microextraction
SSRI Specific serotonin reuptake inhibitor TMF Trophic magnification factor
WWTP Wastewater treatment plant
List of papers
This thesis is based on the following papers, referred to by their Roman numerals: I Boström M.L., Huang C., Engström H., Larsson E., Berglund O., Jönsson
J.Å. (2014). “A specific, highly enriching and “green” method for hollow fiber liquid phase microextraction of ionizable pharmaceuticals from fish tissue”. Analytical Methods 6(15): 6031-6037
II Boström M.L. & Berglund O. (2015). “Influence of pH-dependent aquatic toxicity of ionizable pharmaceuticals on risk assessments over environmental pH ranges”. Water Research 72: 154-161
III Boström M.L., Hultin C.L., Hansson M.C., Berglund O. “Identification of a
cyp3a gene in wild roach (Rutilus rutilus): gene expression and its
relationship to long-term chemical exposure history and short-term fluoxetine exposure”. (Manuscript)
IV Boström M.L., Ugge G., Jönsson J.Å., Berglund O. (2017). “Bioaccumulation and trophodynamics of the antidepressants sertraline and fluoxetine in laboratory three-level aquatic food chains”. Environmental
19
Author contributions
I MLB, JÅJ and EL designed the method development. CH and HE carried out the laboratory work, and together with MLB analyzed the data. MLB wrote the manuscript with contributions from OB, JÅJ, EL, CH and HE.
II OB developed the original idea, and together with MLB designed the study. MLB carried out the laboratory work, and together with OB analyzed the data. MLB wrote the manuscript with support from OB.
III MLB and CLH developed the original idea. MLB and CLH designed the study with contributions from OB and MH. MLB carried out the field work, and together with CLH the laboratory work. MLB analyzed the data, and together with CLH wrote the manuscript with support from OB and contribution from MH.
IV OB and MLB developed the original idea and MLB designed the study. MLB and GU carried out the field and laboratory work, and MLB analyzed the data with contribution from OB. MLB wrote the manuscript with support from OB and contributions from GU and JÅJ.
List of Authors: Marja L. Boström (MLB), Jan Åke Jönsson (JÅJ), Estelle Larsson (EL), Chuixiu Huang (CH), Henrik Engström (HE), Olof Berglund (OB), Cecilia L. Hultin (CLH), Maria Hansson (MH), and Gustaf Ugge (GU).
Scope and aim
The scope of this thesis is the uptake and bioaccumulation of ionizable pharmaceuticals in aquatic systems. The overall aim was to further the under-standing of processes affecting bioaccumulation of ionizable pharmaceuticals to inform environmental risk assessments and water management. To fulfill the aim four objectives were set:
1. Develop an efficient extraction method to analyze and quantify different ionizable pharmaceuticals in organism tissue, in order to assess bioaccumu-lation in the further studies in the thesis.
2. Assess pH-dependent uptake and toxicity under environmentally relevant pH gradients.
3. Test the influence of pollution history on bioaccumulation, i.e. the possibility of gained tolerance in exposed populations in the wild.
4. Test food chain bioaccumulation and potential biomagnification of pharmaceuticals in different aquatic food chains.
The pharmaceuticals used in the papers composing this thesis are two specific serotonin reuptake inhibitors (SSRIs, Box 1) (Paper I, II, III, IV): fluoxetine and
sertraline, both weak bases (Table 1), and four non-steroidal anti-inflammatory drugs (NSAIDs, Box 1) (Paper I and II): ketoprofen, naproxen, diclofenac and
ibuprofen, all weak acids (Table 1). These pharmaceuticals were selected as they 1) are used in large volumes (Box 1); 2) are not classified as bioaccumulative (bioconcentration factor, BCF (Box 2, Eq B3), >2000 L/kg ([KEMI] 2015)) in hazard assessments based on standardized tests (Table 1), but have been found to reach such levels in aquatic organisms in non-standardized tests (Brown et al. 2007, Du et al. 2015, Grabicova et al. 2017); 3) have been detected at concentrations close to, or in some cases above, concentrations affecting aquatic organisms in surface waters impacted by wastewater (Hong et al. 2007, Lajeunesse et al. 2011, Nentwig 2007).
21
Table 1. Chemical information of the four NSAIDs and the two SSRIs studied. MW = molecular weight, pKa =
DFLGdissociation constant, logKow = logarithm of the octanol/water partition coefficient, PBT = hazard score.
aScifinder database (© 2013 American Chemical Society), calculated values
bData retrieved from Stockholms läns landsting ([SLL] 2018). International agreements have been made to identify,
ban, or rigorously regulate anthropogenic chemicals with potential to harm the environment ([UNEP] 2008). Regulatory agencies screen for hazardous chemicals by evaluating the criteria persistence (P), bioaccumulation (B), and toxicity (T) from standardized tests ([EC] 2013, [ECA] 2017, [UNEP] 2004, [USEPA] 1976). Each criteria is scored from 0 to 3 (low-high) and the final range used in hazard assessments is thus from 0 to 9 (low-high).
Type of
pharmaceutical MW pKaa logKOWa PBT
b Fluoxetine SSRI 309 10.1 3.9 6 (P 3, B 0, T 3) Sertraline SSRI 306 9.47 5.1 6 (P 3, B 0, T 3) Ketoprofen NSAID 254 4.23 2.9 4 (P 3, B 0, T 1) Naproxen NSAID 230 4.84 2.9 4 (P 3, B 0, T 1) Diclofenac NSAID 296 4.18 4.5 4 (P 3, B 0, T 1) Ibuprofen NSAID 206 4.41 3.5 1 (P 0, B 0, T 1) Box 1
Fluoxetine and sertraline (Table 1) are specific serotonin reuptake inhibitors (SSRIs), raising serotonin levels in neuron synapses by blocking the “drainage” of this neuro-transmitter from the synaptic cleft. Serotonin is involved in regulating mood and emotional behavior and SSRIs are used to treat for instance depression and anxiety (Finley 1994). SSRIs were purchased 300 times per 1 000 inhabitants in 2018 in Sweden ([Socialstyrelsen] 2018). Ketoprofen, naproxen, diclofenac and ibuprofen (Table 1) are non-steroidal anti-inflammatory drugs (NSAIDs), painkillers also having anti-anti-inflammatory properties. Their main effect is inhibition of the enzyme cyclo-oxygenase (COX), which is involved in inflammatory processes in the body (Vane and Botting 1998). Painkillers excluding opioids were purchased 700 times per 1 000 inhabitants in 2018 in Sweden ([Socialstyrelsen] 2018).
Box 2
Guidelines to test a chemical’s potential to bioaccumulate in aquatic organisms are based on quantifications of rates of uptake and elimination (e.g. [OECD] 2012). The rate of uptake of a chemical from water with a certain concentration, cW (g/L), through respiratory organs
and skin, k1 (L water/kg organism/day), together with the rate of elimination, i.e. sum of
rates of elimination over respiratory organs and skin, k2 (1/day); egestion, kE (kg food/kg
organism/day); metabolism, kM (1/day); and growth kG (1/day), determines organism
chemical concentration, cB (g/kg organism) (Fig 1A) (parameter labeling adopted from
Gobas and Morrison 2000). The rate constants originate from a general kinetic first order, two-compartment model (Gobas and Morrison 2000)
d𝑐𝑐𝐵𝐵
d𝑡𝑡 = 𝑘𝑘1𝑐𝑐𝑊𝑊− (𝑘𝑘2+ 𝑘𝑘𝐸𝐸+ 𝑘𝑘𝑀𝑀+ 𝑘𝑘𝐺𝐺)𝑐𝑐𝐵𝐵 (Eq B1)
When a steady state is reached between the rates of uptake and elimination and cB is higher
than cW, it is defined as bioconcentration
𝑐𝑐𝐵𝐵=𝑘𝑘2+𝑘𝑘𝑘𝑘𝐸𝐸1+𝑘𝑘𝑐𝑐𝑊𝑊𝑀𝑀+𝑘𝑘𝐺𝐺> 1 (Eq B2)
The foremost metric used in hazard assessments is the laboratory derived bioconcentration factor, BCFs, defined when cB and cW have reached equilibrium
𝐵𝐵𝐵𝐵𝐵𝐵 =𝑐𝑐𝐵𝐵
𝑐𝑐𝑊𝑊 (Eq B3)
or before reaching equilibrium measuring rate of uptake during exposure and rate of elimination during depuration
𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑘𝑘1
𝑘𝑘2+𝑘𝑘𝐸𝐸+𝑘𝑘𝑀𝑀+𝑘𝑘𝐺𝐺 (Eq B4)
In field and in mesocosm studies including consumption of exposed food, there will be a dietary contribution to cB (Fig 1A). Disregarding the contribution from water the
two-compartment, first order kinetic model yields
d𝑐𝑐𝐵𝐵
d𝑡𝑡 = 𝑘𝑘𝐷𝐷𝑐𝑐𝐷𝐷− (𝑘𝑘2+ 𝑘𝑘𝐸𝐸+ 𝑘𝑘𝑀𝑀+ 𝑘𝑘𝐺𝐺)𝑐𝑐𝐵𝐵 (Eq B5)
were cD is the concentration in the food and kD the dietary uptake rate. The dietary
contribution is hence
𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑐𝑐𝑙𝑙𝑐𝑐𝑑𝑑𝑑𝑑𝑑𝑑𝑐𝑐𝑐𝑐𝑑𝑑𝑑𝑑𝑙𝑙𝑐𝑐 = 𝑘𝑘𝐷𝐷𝑐𝑐𝐷𝐷
𝑘𝑘2+𝑘𝑘𝐸𝐸+𝑘𝑘𝑀𝑀+𝑘𝑘𝐺𝐺 (Eq B6) Adding the dietary contribution to Eq B2 defines bioaccumulation
23 When food is included in mesocosms or if data derives from field studies a bioaccumulation factor, BAF, also often used in hazard assessments is at equilibrium defined (Gobas and Morrison 2000)
𝐵𝐵𝐵𝐵𝐵𝐵 =𝑐𝑐𝐵𝐵
𝑐𝑐𝑊𝑊 (Eq B8)
or before equilibrium (Gobas and Morrison 2000) 𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑘𝑘1+𝑘𝑘𝐷𝐷
𝑘𝑘2+𝑘𝑘𝐸𝐸+𝑘𝑘𝑀𝑀+𝑘𝑘𝐺𝐺 (Eq B9)
Also used in hazard assessments is the metric for trophic bioaccumulation, the biomagnification factor, BMF, derived from feeding studies. The BMF is determined on species level by calculating the ratio between a predator’s body concentration, cB, and the
prey’s body concentration, cD
𝐵𝐵𝐵𝐵𝐵𝐵 =𝑐𝑐𝐵𝐵
𝑐𝑐𝐷𝐷 (Eq B10)
An extension of BMF is the trophic magnification factor, TMF, a field quantified metric including all organisms sampled in a food web, introduced by Broman et al. (1992), and further developed by Fisk et al. (2001). All trophic interactions are taken into account and TMF can be seen as an average BMF throughout the food web. The TMF is a rate of (potential) increase in concentration with trophic position given by the slope, m, derived by linear regression of the logarithmically transformed cB and relative trophic position among
all species sampled (Borgå et al. 2012, Conder et al. 2012). The latter is determined by stable nitrogen isotope ratio [δ15N], measured using tissue N15/N14, typically increasing with
trophic level.
𝑇𝑇𝐵𝐵𝐵𝐵 = 10𝑚𝑚 (Eq B11)
So far, the application of TMFs in hazard assessments is still evaluated (Perceval et al. 2017).
Introduction
Worldwide sales of human pharmaceuticals have been increasing yearly ([OECD.Stat] 2018), and further growth is predicted in the coming years (data collected from 2010 to 2017 and predicted until 2024 by [EvaluatePharma®] 2018).
Certain active ingredients of pharmaceuticals have been identified as environmental contaminants since they end up in aquatic environments as excreted residues incompletely eliminated in wastewater treatment plants (WWTPs) (Bendz et al. 2005, Gracia-Lor et al. 2012, Larsson et al. 2009, Martin et al. 2012, Rivera-Utrilla et al. 2013, Ying et al. 2009). They are now frequently detected in aquatic environments worldwide, both in the water and in aquatic organisms (e.g. Bai et al. 2014, Bradley et al. 2017, Du et al. 2015, Grabic et al. 2012, Grabicova et al. 2017, Liu et al. 2015, Petrović et al. 2014, Wang and Gardinali 2012). Pharmaceuticals are produced to affect biochemical and/or physiological processes in target organisms and the target sites of action are often genetically conserved (Gunnarsson et al. 2008). Consequently, their presence in the environment is of concern as they may affect non-target organisms. The physiology (Bossus et al. 2014, Hoeger et al. 2005, Hong et al. 2007, Mehinto et al. 2010, Schultz et al. 2011, Schwarz et al. 2017, Wang et al. 2016) and/or behavior (Bossus et al. 2014, Brodin et al. 2013, Conners et al. 2009, Hedgespeth et al. 2014, Luna et al. 2013, Schwarz et al. 2017) of both invertebrate and vertebrate aquatic organisms have been shown to be affected at environmental concentrations.
Because of the conservation of pharmaceutical target sites the read-across hypothesis propose that human therapeutic concentrations also can be assumed to affect fish in a similar manner (Huggett et al. 2005). Water concentrations of pharmaceuticals in wastewater treatment recipients are often lower (e.g. Gros et al. 2010, Kolpin et al. 2002, Metcalfe et al. 2010) than therapeutic levels (Schulz et al. 2012), but since several pharmaceuticals have the potential to bioaccumulate in aquatic organisms (e.g. Du et al. 2016, Grabicova et al. 2015, Meredith-Williams et al. 2012, Metcalfe et al. 2010, Moreno-González et al. 2016, Nakamura et al. 2008, Rodrigues et al. 2015), internal concentrations may still reach human therapeutic levels and have proven to reach effect levels (reveiwed in Corcoran et al. 2010, Cuklev et al. 2011) in non-target organisms.
Bioaccumulation occurs when the rate of uptake of chemicals exceeds the rate of elimination (reviewed in Arnot and Gobas 2006) (Box 2). Chemicals are taken up by aquatic organisms directly from water, for aquatic unicellular organisms
25 this takes place over the cell membrane, and for primary producers via exposed outer cell layers (Gobas et al. 1991). Aquatic animals take up chemicals in the water through the skin and respiratory systems, the latter of which is often the main uptake route (Bruggeman et al. 1984) (Fig 1A). Since lower trophic levels already have taken up the chemical, consumers will also be exposed through diet (Bruggeman et al. 1984) (Fig 1A). Elimination is dependent on metabolism, storage, excretion, and growth dilution (Fig 1A). Metabolic elimination refers to enzymatic biotrans-formation, storage to translocation into organism tissues, and excretion to egestion and elimination through skin and respiratory systems (reviewed in Arnot and Gobas 2006). Growth dilution is due to a lowering of body concentration as a result of an increase in tissue volume (Gobas and Morrison 2000).
Figure 1. A. Major routes/factors associated with uptake and elimination of xenobiotics, and thus
bioaccumulation. B. To quantify ionizable chemical concentrations (NSAIDs and SSRIs) in fish tissue an extraction method was developed in Paper I, to be used in Paper III and Paper IV. In Paper II, a model was
conducted to evaluate how spatial variability in pH may over- or underestimate environmental hazard in aquatic systems. The potential of adaptation of metabolic enzyme systems alongside reduced bioaccumu-lation was studied in organisms residing under historically long-term effluent pressure in Paper III. In Paper IV the potential of trophic transfer of SSRIs was assessed and empirically derived values from that study were
also used to test a model predicting ionizable chemical bioaccumulation.
Paper IV Trophic transfer
Paper II
pH-dependent uptake and elimination Metabolic adaptationPaper III
Paper I Tissue extraction Dietary uptake
Uptake and elimination through gills and skin
Egestion Metabolic elimination Growth dilution Storage A B 25
When pesticides started to be commonly used in the 1960s hydrophobic persistent organic chemicals were found to bioaccumulate and also that bioaccumulation increased with trophic level (denoted biomagnification) (Robinson et al. 1967, Woodwell et al. 1967). Dichlorodiphenyltrichloroethane (DDT, an insecticide) and polychlorinated biphenyls (PCBs, used for example as flame retardants and plasticizers) became model chemicals in the early studies (Branson et al. 1975, Bruggeman et al. 1984, Oliver and Niimi 1988, Thomann and Connolly 1984), and have laid the foundation to how bioaccumulation assessments are performed today. Bioaccumulation processes are governed by the properties of a chemical (Erickson et al. 2008), and organism traits such as respiration strategy, locomotion, and behavior (Borgå et al. 2004, Rubach et al. 2010), as well as physiology, e.g. lipid content, metabolic enzyme systems, and size (Arnot and Gobas 2006, Borgå et al. 2004). In this thesis I will address the question: Is bioaccumulation of ionizable pharmaceuticals governed by the same processes and parameters as the legacy model chemicals, and may use of the classical framework for bioaccumulation and hazard assessments lead to under- or overestimation? Specifically, I evaluate:
• pH-dependent toxicity and how environmental pH variability affects
hazard assessments of ionizable pharmaceuticals, i.e. if toxicity metrics measured under laboratory conditions at a fixed pH are appropriate to assess toxicity in natural waters with differing pH.
• if organisms, specifically fish, are able to adapt to life-long exposure to effluents from wastewater treatment plants by developing tolerance
through reduced bioaccumulation when continuously exposed.
• the potential of trophic transfer of pharmaceuticals, i.e. the importance of
dietary contribution to bioaccumulation at different trophic levels including fish and invertebrates
.
27
Bioaccumulation of pharmaceuticals
The three bioaccumulation processes addressed in this thesis are presented in sections, starting with pH-dependent uptake, next, adaptations to pollution, and last, trophic transfer.
Ionization and pH-dependency
Pharmaceuticals are commonly designed as organic chemicals to facilitate uptake (Sugano et al. 2010) and mimic biologically active endogenous molecules (Krogsgaard-Larsen and Bunch 2009). The early model chemicals of bioaccumu-lation are also organic but while the model chemicals are uncharged, almost 80% of pharmaceuticals are ionizable (Manallack 2008). Ionizable pharmaceuticals are classified as acids, bases or ampholytes, and the different classes will have shifts in their dissociation equilibrium at different parts of the pH scale (Fig 2). The chemical species have different chemical properties, which will affect their bioavailability, defined as “that which is freely available to cross an organism’s cellular membrane from the medium the organism inhabits at any given time” (Semple et al. 2004). As uncharged molecules pass biomembranes to a higher degree than ions (Erickson et al. 2008), uptake becomes pH-dependent (Stehly and Hayton 1990). A water pH which increases ionization thus reduces bioaccumulation (Paper II, Armitage et al.
2017 and references therein, Hlina et al. 2017, Karlsson et al. 2017) and, accordingly, toxicity (Fent and Looser 1995, Kobayashi and Kishino 1980, Liu et al. 2016, Nakamura et al. 2008, Neuwöhner and Escher 2011, Rendal et al. 2011, Valenti et al. 2009).
Environmental risk assessments performed before market release do not take into account bioaccumulation and toxicity being pH-dependent ([EMEA] 2006). For instance, guidelines for testing the pharmaceutical toxicity on aquatic organisms, such as the OECD test “No. 202: Daphnia sp. acute immobilization test” ([OECD] 2004), accept variation in pH by three units (within the typical environmental range of 6 to 9), but only one specific pH is used when performing the test. Environmental pH is governed by spatial (geological) and temporal (diel or seasonal fluctuations) pH variations (Valenti et al. 2009, Valenti et al. 2011) so the experimentally
determined toxicity, consequently, might not match the environmental one depen-ding on the chosen experimental pH (Karlsson et al. 2017). In Paper II (Fig 1B),
the OECD test No. 202 modified to keep pH at 6, 7.5 and 9, respectively, was performed to test for relative aquatic toxicity of SSRIs and NSAIDs. A model was then developed to evaluate how spatial variability in pH in European waters may result in under- or overestimation of negative impacts on organism health. The European rivers are generally slightly basic (7.8 ±0.5, mean ±SD) and, conse-quently, the model simulation showed that underestimation of toxicity is more likely for bases than for acids (Fig 3). The median underestimation for bases was by a factor of 3, with 90% of the model results ranging from factor 1 to 6. On the other hand, model simulation showed that weak acid toxicity may be overestimated by a factor of 2. Predicted median toxicity was 0.5 with the results ranging from 0.03 to 5. The results from Paper II show the importance of accompanying reported
toxicity data with the pH at which the study was performed, to allow informed water management decisions.
Figur 2. A visualization of how (A) fluoxetine, a weak base with pKa 10.1, and (B) diclofenac, a weak acid with
pKa 4.18, changes fraction (f, y-axis) of speciation with increasing pH (x-axis). The solid line represents the
uncharged speciation and the dashed line the charged speciation. The pH-dependent fractioning of weak base and weak acid pharmaceuticals is typically calculated using the acid dissociation constant, pKa, in a
modified version of the Henderson-Hasselbalch equation 𝑝𝑝𝑝𝑝 = 𝑝𝑝𝐾𝐾𝑎𝑎+ 𝑙𝑙𝑙𝑙𝑙𝑙10�∝∝𝐴𝐴−
𝐴𝐴𝐴𝐴� (Eq 1)
where α- and α are the ionic and uncharged fraction, respectively.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 (f) pH Cation Uncharged speciation A 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 (f) pH Anion Uncharged speciation B
29 Uncharged organic chemicals partition into organisms’ storage lipids and lipophilicity is an important parameter when predicting bioaccumulation (Arnot and Gobas 2006, Isnard and Lambert 1988, Mackay 1982, Meylan et al. 1999, Neely et al. 1974, Oliver and Niimi 1988, Veith et al. 1979). Lipophilicity is usually quantified by measuring the partition coefficient, Kow, between octanol and water,
defined
𝐾𝐾𝑂𝑂𝑊𝑊=𝑐𝑐𝑐𝑐𝑊𝑊𝑂𝑂 (Eq 2)
where cO and cW are the chemical equilibrium concentrations in octanol and water,
respectively. The affinity of organic chemicals to octanol is regarded as an acceptable surrogate for organism lipid phase partitioning (Leo et al. 1971). For uncharged organic chemicals a positive linear relationship between logKow < 6
(Bintein et al. 1993, Gobas et al. 1988) and the bioconcentration factor (BCF, Box 2, Eq B3) has been proposed (Arnot and Gobas 2006, Isnard and Lambert 1988, Mackay 1982, Meylan et al. 1999, Neely et al. 1974, Oliver and Niimi 1988, Veith et al. 1979). Interactions between storage lipids and charged compounds are not favorable (Avdeef et al. 1998) and ionizable pharmaceuticals do not show the same strong relationship with logKow (Paper IV and Haddad et al. 2018,
Meredith-Williams et al. 2012).
Figure 3. A simulated European river pH distribution (pH on x-axis, fraction on y-axis) using a data set from
4279 river sites spread over 21 countries. Hazard assessments made from toxicity tests performed at pH 7 only (indicated by the parallel lines) will potentially underestimate toxicity for both acids and bases, as not all rivers have neutral pH. As the pH distribution in Europe is more towards the basic side, toxicity of basic chemicals may more often be underestimated than acids.
Models predicting bioaccumulation (reported as a bioaccumulation factor, BAF, Box 2, Eq B8) for the early model uncharged organic chemicals (Arnot and Gobas 2004, Mackay 1982, Thomann 1989) are being revised for a better fit for ionizable pharmaceuticals. The pH-dependent chemical speciation is introduced by adding parameters like the acid partitioning constant pKa (Fig 2), and water and
5,0% 90,0% 5,0% 3,8% 90,5% 5,7% 6,88 8,35 3 4 5 6 7 8 9 10 pH 0,0 0,2 0,4 0,6 0,8 1,0 1,2 Probability density Input ExtValueMin 3 4 5 6 7 8 9 10 European river pH Potentially underestimating
risk for NSAIDs Potentially underestimating risk for SSRIs
(f)
internal pH (Armitage et al. 2013, Neuwöhner and Escher 2011). In Paper IV (Fig
1B) empirical data for sertraline and fluoxetine were entered into such a model (Armitage et al. 2013) to test its applicability, with a poor fit as a result. The only input value in the model explaining the poor fit of the output was the difference in logKow between the SSRIs. However, as only two chemicals were used to test the
model one needs to be careful in the interpretation. Nevertheless, the outcome does suggest the need for further evaluation of the use of the Kow in ionizable
pharma-ceutical bioaccumulation models.
Adapting to pollution
When continuously exposed to pollutants over a long period of time, aquatic animals can adapt to their environment by developing tolerance to these chemicals (e.g. Greytak et al. 2010, Nacci et al. 1999, Nacci et al. 2010, Reid et al. 2016). Metabolism (biotransformation) of pollutants affects the elimination rate and is an important factor controlling internal concentrations (Arnot et al. 2008a, Burkhard 2003, de Wolf et al. 1993, Koenig et al. 2012, Nichols et al. 2007) (Fig 1A). Adaptation due to alterations in pollutant metabolizing enzyme systems possibly affects elimination rates and, consequently, bioaccumulation (Brown Jr 1992, Koenig et al. 2012). Intraspecies variation in bioaccumulation has been shown among populations living in environments with dissimilar pharmaceutical pressure (Rodrigues et al. 2015). Still, bioaccumulation data in hazard assessments are derived from laboratory bred standard species or from organisms collected at sites with no or low pollution ([ECA] 2017). Knowledge of the potential of adaptation to long-term pharmaceutical exposure should accompany these laboratory derived data to make informed water management decisions, for instance downstream WWTPs. Tolerance can be achieved through physiological acclimation, which is an uninheritable, temporary, and reversible response to toxicant exposure (Weis and Weis 1989), or through heritable genetic adaptation, which will remain after removal of the toxicant (Van Veld and Nacci 2008, Weis and Weis 1989). Physiological acclimation altering bioaccumulation can occur as alterations in expression levels of biotransformation enzyme systems (Huang et al. 2016, Pujolar et al. 2012, Reid et al. 2016), or as changes in protein activity levels (Huang et al. 2016). Genetic adaptation would occur through genomic alterations in biotrans-forming enzyme systems or through genes controlling these systems (Reid et al. 2016). Both physiological changes in protein activity (e.g. Fisher et al. 2006, Liu et al. 2015, Wang et al. 2013) and genetic changes in biotransforming enzyme systems (e.g. Nacci et al. 2010, Reid et al. 2016) have been found in organisms in polluted environments.
31 Important in human pharmaceutical metabolism (Guengerich 2008) and present in all animals (Nelson et al. 2013) is a superfamily of biotransforming enzymes named cytochrome P450 (gene denominated cyp; and protein, CYP). Differences in CYP activity have been related to dissimilarities in bioaccumulation among species (Koenig et al. 2012). Thus, in environments with chronical pharmaceutical pressure, CYP mediated biotransformation may be specifically important to focus on when considering the possibility of adaptation through alterations in bioaccumulation levels. Pharmaceuticals indeed alter cyp expression and CYP activity in aquatic animals (e.g. Beijer et al. 2013, Burkina et al. 2015, Kim et al. 2018, Wang et al. 2016) and exposing fish to WWTP effluent (that contains a mixture of pollutants, particularly pharmaceuticals (Buser et al. 1999, Gros et al. 2010, Meador et al. 2016)) results in altered CYP gene expression (Gräns et al. 2015, Gunnarsson et al. 2009, Huang et al. 2016, Lister et al. 2009, Pujolar et al. 2012, Vidal-Dorsch et al. 2013). Tolerance to organic chemicals due to changes in cyp expression has been studied in fish residing in chronically polluted industrial areas (Reid et al. 2016), and in environments burdened by PCBs (Pujolar et al. 2012) but, to my knowledge, not in waters chronically exposed to WWTP effluent.
In Paper III (Fig 1B), indications of adaptation were assessed by comparing
level of bioaccumulation as well as cyp3a gene expression (the main cytochrome P450 subfamily biotransforming pharmaceuticals (Bugel et al. 2014, Gräns et al. 2015, Zanger et al. 2008)) after fluoxetine exposure between fish from two sites. The fish were either collected from (1) water with historically long-term WWTP effluent pressure, or (2) water from an upstream site with no known source of pollution. The gene expression was also compared between fish directly caught at the two sites. Results showed that fish residing downstream the WWTP had 10% lower fluoxetine logBCF after a 7-d exposure than fish collected upstream (Fig 4), and the lower logBCF still remained after fish had been detoxified for 7 days (Fig 4). It is thus possible that a genetic adaptation in enzyme systems involved in fluoxetine biotransformation has occurred. No difference was found when comparing gene expression levels between fish directly caught at the two sites, neither before nor after fluoxetine exposure. This suggests that the adaptation is not due to alteration in this enzyme system. However, the results should be viewed in the light of the low replication (n =3-9). Still, BCF being altered suggests the possibility of adaptation to chronic WWTP effluent exposure and may need to be considered when extrapolating laboratory data to field conditions. If confirmed, the result points out the importance of site-specific hazard assessment if to make informed water management decisions.
Figure 4. Boxplot showing roach (Rutilus rutilus) fluoxetine logBCFs (y-axis, L/kg) for the three different
fluoxetine exposure treatments. Fish were exposed (7 d) either directly upon arrival from the upstream (A) and the downstream site (B), or after 7 d of detoxification of the fish from the downstream site (C). Different letters (a, b and c) in the boxplot denote significant differences among the exposures (Kruskal-Wallis test, n =6-9, followed by Mann-Whitney U tests).
Trophic transfer
Increased internal concentration by dietary uptake (Fig 1A) occurs when the rate of chemical absorption from food exceeds that of the rate of uptake from water passing the respiratory organs (reviewed in Mackey et al. 2013, Walters et al. 2016). Dietary transfer has been identified as an important route of uptake of the early model chemicals (Bruggeman et al. 1984, Thomann 1981, Thomann 1989). Dietary absorption is in some cases also faster than elimination processes (Gobas and Morrison 2000, Thomann 1981), and if this occurs throughout a food chain (i.e. biomagnification) top predators are at higher risk of reaching internal effect concentrations (Thomann 1981). If dietary uptake is excluded or if it is not the major uptake route, species from lower trophic levels may reach higher internal concentrations than those at higher trophic levels (e.g. Meredith-Williams et al. 2012). Studies addressing trophic transfer of pharmaceuticals are just recently being published, and results are inconclusive (Du et al. 2014, Du et al. 2016, Grabicova et al. 2017, Heynen et al. 2016, Ruhi et al. 2016, Xie et al. 2015). The importance of dietary uptake needs to be resolved to facilitate predictions of species at risk of reaching high internal concentrations. Also, identifying species at risk of high
Fish from:
A) reference lake, 7d exposed B) WWTP pond, 7d exposed
C) WWTP pond, 7d detoxified + 7d exposed
a b c 2.4 0.8 0 A lo gBCF B C 1.6
33 bioaccumulation levels is crucial to know which test organisms to use in hazard assessments ([ECA] 2017).
Two ways to determine the potential of trophic transfer and biomagnification are (1) measuring field-concentrations and calculating BAFs or trophic magnify-cation factors (TMFs, Box 2, Eq B11), and (2) performing feeding trials and comparing BAFs among organisms or calculate biomagnification factors (BMFs, Box 2, Eq B10). Field studies include the complexity of dietary transfer within a food web with all food resources included. Spatial and temporal differences can also be studied (Du et al. 2016, Haddad et al. 2018), and potential effects by pollutant mixtures are included (Backhaus and Karlsson 2014). Several field BAFs for pharmaceuticals have been reported (e.g. Grabicova et al. 2015, Grabicova et al. 2017, Metcalfe et al. 2010, Moreno-González et al. 2016, Ruhi et al. 2016), but whether or not diet contributes to internal concentrations of pharmaceuticals is inconclusive. Calculated TMFs for ionizable pharmaceuticals have so far been close to or below 1 (Du et al. 2014, Haddad et al. 2018), suggesting no pharmaceutical biomagnification.
Laboratory set-ups in simple constructed food chains can also be used to study trophic transfer (Paper IV and Heynen et al. 2016), although field-derived
data is considered more conclusive than laboratory-derived data from short food chains (Gobas et al. 2009, Weisbrod et al. 2009). However, as environmental concentrations are generally close to or below detection limits in field studies (Du et al. 2014, Grabicova et al. 2017, Ruhi et al. 2016, Xie et al. 2015), laboratory set-ups of constructed food chains at quantifiable concentrations (Heynen et al. 2016) are a useful additional approach. To assess trophic transfer of sertraline and fluoxetine in Paper IV (Fig 1B), two three-level aquatic food chains were
constructed in microcosms. To the best of my knowledge, only Heynen et al. (2016) have before presented laboratory derived results for pharmaceuticals from constructed food chains. As with the field-derived results, the two studies are contradictory, with Heynen et al. (2016) suggesting addition to internal concentration by dietary uptake (10% - 40%) contrary to the results in Paper IV,
where logBAFs even became lower with trophic level (20% - 50%) and logBAFs were not greater than logBCFs (Fig 5). However, the use of different pharmaceuti-cals in the two studies, the ampholyt oxazepam by Heynen et al. (2016) and the bases sertraline and fluoxetine in Paper IV, with dissimilar chemical properties,
may explain this difference. Worth noticing though, is that an increase in sertraline concentration has been found in fish during periods when feeding activity has been higher (Grabicova et al. 2017), suggesting dietary contribution to internal concentration also for this chemical. In summary, there is no strong indication of trophic transfer or biomagnification of pharmaceuticals from neither field nor laboratory derived data, but results are ambiguous, encouraging further studies.
If trophic transfer is low, biotransformation becomes a strong driver of interspecies differences among aquatic animals, as uptake and elimination rates by
passive diffusion tends to be more similar (Arnot et al. 2008a, Brown Jr 1992, Burkhard 2003, de Wolf et al. 1993, Koenig et al. 2012, Nichols et al. 2007). In the absence of metabolic elimination, uncharged organic chemicals with logKow >5 have
biomagnification potential (Burkhard et al. 2012, Gobas et al. 2009) which is less likely if biotransformation rates are high (van der Oost et al. 2003). Interestingly, trophic dilution has been reported for pharmaceuticals with logKow close to or above
5 (amitriptyline and sertraline) (Paper IV, Haddad et al. 2018). In fact,
pharmaceutical BCFs can differ more than 100,000-fold among species (literature search on fluoxetine in Paper IV), sometimes with internal concentrations higher
in animals at low trophic levels (Paper IV, Du et al. 2014, Haddad et al. 2018,
Meredith-Williams et al. 2012). Enzyme systems biotransforming pharmaceuticals generally have reduced activity in aquatic invertebrates compared to fish (Koenig et al. 2012, Livingstone 1998). Thus, this could explain higher internal concentrations at lower trophic levels, and in turn, leave biomagnification unlikely (reviewed in Borgå et al. 2004). In the light of this, focus in hazard assessments may need to be on organisms at lower trophic levels having low activity in pharmaceutical biotransformation systems.
35
Figure 5. Boxplots showing log transformed BCF (Eq 3) and BAF (Eq 8) values (L/kg) for (A) sertraline, and
(B) fluoxetine, derived for the different species in Paper IV. Trophic transfer was tested by using two
three-level food-chains; Acer platanoides (maple), fed to Asellus aquaticus (water hoglouse), in turn fed to Notonecta glauca (water boatman) or Pungitius pungitius (nine-spined stickleback). * represents a statistically significant difference between groups, p <0.05, ** = p <0.01 and *** = p <0.001 (Tukey test with 95% family-wise confidence level). BAFs being lower than BCFs and BAFs decreased with increasing trophic level (x-axis, left to right) interprets as dietary exposure not adding to internal concentration for either SSRI.
Extraction method
A novel extraction method was developed to perform the chemical analyses in
Paper III and Paper IV. For the purpose of bioaccumulation studies and
monitoring at environmentally relevant concentrations, as well as evaluating WWTP treatment techniques, simple and inexpensive analytical methods for measuring low chemical concentrations in environmental field and biological samples are needed. An ion trapping process over a selective membrane can be used as an extraction technique for enriching ionizable chemicals from various samples (Jönsson 2012). In Paper I (Fig 1B), analyte ion trapping was used in hollow fiber
liquid phase microextraction (HF-LPME). The HF-LPME methods have been developed for ionizable chemicals and used on matrices such as water, sewage sludge and biological material (e.g. cecal content) (Ghaffarzadegan et al. 2014, Larsson et al. 2009, Sagristà et al. 2012, Sagristà et al. 2010). In Paper I it was
adapted for extraction of NSAIDs and SSRIs in fish tissue to be able to quantify concentrations and calculate BCFs and BAFs in Paper III and Paper IV.
Ion trapping is a process where uncharged ionizable chemicals cross a semipermeable membrane into a medium favoring the ionic speciation, which have lower (or even negligible) membrane permeability, thereby enriching the chemical in the new medium (de Duve et al. 1974, Neuwöhner and Escher 2011, Zarfl et al. 2008) (Fig 6). Acidic chemicals will be trapped if the pH is higher in the second medium, and the opposite is true for bases. This enrichment process increasing internal concentration can be used to extract ionizable chemicals from organisms before final analysis (Liu et al. 2016, Menck et al. 2013).
The HF-LPME sample preparation technique is based on creating a miniature scale liquid-liquid extraction device using hollow polypropylene fibers (Fig 7). When used for extracting ionizable chemicals, the fiber walls form a semipermeable membrane, and a pH gradient between the donor solution outside and the acceptor solution inside the fiber creates an ion trap within the fiber (Fig 6). The technique can enrich ionizable trace compounds from the donor solution (sample) several thousandfold (Jönsson 2012) (the extraction efficiency in Paper I was 1900-3000)
and simultaneously cleans up specifically target ionizable chemicals thereby minimizing co-extraction (Jönsson 2012). The method can in principle be used for any sample that can be dissolved or mixed in water and the acceptor solution can be directly injected on an analytical instrument (for example gas or liquid chromatography, and mass spectrometry) without further cleanup.
37
Figure 6. Ion trapping is used to extract ionizable chemicals (here a weak acid) in the hollow fiber liquid phase
microextraction technique (HF-LPME). By soaking the hollow fiber (dashed structure) in organic solvent it allows uncharged organic chemicals, HA, to pass (yellow, straight arrow) but hindering ions, A- (yellow, curved
arrow). The pH in the donor solution, i.e. the sample, is shifting the dissociation equilibrium towards the uncharged speciation. In the acceptor phase, i.e. liquid to be analyzed, the pH shifts to favor the ionic speciation. This results in enrichment of the chemical in the acceptor solution.
Figure 7. A hollow fiber prepared for extracting. The copper wire is used to weigh the fiber down for total
submergence. Fiber size: wall thickness 50 μm; pore size 0.1 μm; inner diameter 280 μm; and length 20 cm (Membrana GmbH, Wuppertal, Germany). Image source: Henrik Engström
Organic solution in pores Acceptor solution Donor solution 𝑝𝑝++ 𝐵𝐵−→ 𝑝𝑝𝐵𝐵 𝑝𝑝𝐵𝐵 → 𝑝𝑝++ 𝐵𝐵− 37
Other techniques commonly used for fish tissue extraction are solid phase extraction (SPE) (e.g. Brooks et al. 2005, Brown et al. 2007, Brozinski et al. 2011, Cuklev et al. 2011) and solid phase microextraction (SPME) (Ouyang et al. 2011, Togunde et al. 2012). These techniques are typically more suitable for hydrophobic compounds and not for ionizable molecules, and in the case of SPME generally limited to volatile compounds (Jönsson 2012). Some of the general HF-LPMEs benefits over SPE are the specificity, the one-step cleanup, the high enrichment and the small amounts of organic solvents used (Jönsson 2012). The acceptor in the hollow fiber technique can usually be directly injected into the analytical instrument, thereby reducing labor time compared to other extraction techniques. HF-LPME is useful for both extensive environmental data collection and studying pharmacokinetics, as it is highly enriching, labor time efficient (cleanup and extraction are done in one single step) and solvent use is kept to a minimum (performed at the μL scale).
39
Conclusion
The main conclusions in this thesis are:
• The HF-LPME was highly effective for extracting ionizable pharmaceuti-cals from organism tissues and therefore applicable for bioaccumulation studies and monitoring at environmentally relevant concentrations.
• Due to pH-dependent toxicity and natural variation in environmental pH,
hazard assessments of ionizable pharmaceuticals based on results from
laboratory toxicity tests performed at a fixed pH may over- or underestimate hazard depending on the pKa of the substance and the pH of the recipient.
• The investigated fish species (roach; Rutilus rutilus) had lower fluoxetine BCF when living under life-long exposure of wastewater treatment effluent, suggesting induced tolerance to pharmaceuticals when continuously
exposed to effluents.
• Trophic transfer of the SSRIs in this thesis did not add to internal
concentrations. Organisms at lower trophic levels (such as invertebrates) had the highest BAF, suggesting the importance of including them in hazard and risk assessments.
Future perspectives
As discussed above, trophic transfer and biomagnification are uncertain for pharmaceuticals. Perhaps it is important to start focusing on identifying species most likely to reach high internal concentrations irrespective of trophic level. The ability of organisms to biotransform pharmaceuticals seems to be important for differences in bioaccumulation among species (Arnot et al. 2008a, Brown Jr 1992, Burkhard 2003, de Wolf et al. 1993, Hutchinson et al. 2014, Koenig et al. 2012, Nichols et al. 2007). Determining pharmaceutical metabolism for several species is a way to find those at high risk of reaching internal effect concentrations. Hutchinson et al. (2014) have put attention to the importance of comparative metabolism and species sensitivity to pharmaceuticals and the matter does need further consideration.
Biotransformation rate is frequently included in models predicting BAF (Armitage et al. 2013, Arnot and Gobas 2004) and TMF (Kim et al. 2016, Walters et al. 2016). Compilation of metabolic rate data of organic chemicals (Arnot et al. 2008b), and collections of publications reporting fish pharmacokinetics in databases (Reimschuessel 2012, Reimschuessel et al. 2005) have been created to be used as resources for hazard assessments. However, fish often being at higher trophic levels might not be the most important organisms to make these compilations for as the highest internal concentrations of pharmaceuticals often are found in invertebrates (Paper IV, Du et al. 2014, Haddad et al. 2018, Meredith-Williams et al. 2012). It
may be time to start deriving and gather bioaccumulation rate data also for invertebrates.
Organisms at lower trophic levels such as crustaceans and molluscs are especially sensitive to SSRIs (Du et al. 2014, Grabicova et al. 2015, Meredith-Williams et al. 2012) and at the same time they demonstrate high BCF and BAF values (reviewed in Fong and Ford 2014). If I may hypothesize, the sensitivity to SSRIs in combination with high bioaccumulation could affect organism composition in environments with chronic pharmaceutical pressure through bottom-up effects. It would be interesting to test this hypothesis by, for instance, performing long-term monitoring downstream a newly introduced WWTP, or a WWTP where new purification techniques are introduced.
HF-LPME has been evaluated for use in predictions of pH-dependent toxicity (Liu et al. 2016). Using a logistic model, Liu et al. (2016) showed a relationship between change in HF-LPME acceptor phase concentration with donor phase pHs,
41 and changes in sulfadiazine (an ionizable chemical used as an antibiotic) EC50
values (50% of effect concentration) from standardized D. magna acute toxicity tests ([OECD] 2004) adjusted to corresponding pHs. If validated, this model could reduce animal testing as pH-dependent toxicity could be calculated rather than empirically tested at multiple pHs. The model could also be used to calculate site specific toxicity which is needed in local water management decisions (Paper II).
It would be intriguing to follow up on using HF-LPME for this purpose focusing on basic pharmaceuticals as their toxicity more likely than acids is underestimated (Paper II).
References
Databases, guidelines, policies, rapports, and websites
[EC], Environment and Climate Change Canada (2013). "Toxic Substances Management Policy". Retrieved Sept 11, 2017, from
https://www.ec.gc.ca/toxiques-toxics/default.asp?lang=En&n=2A55771E-1.
[ECA], European Chemicals Agency (2017). "Guidance on information requirements and chemical safety assessment. Chapter R.11: PBT/vPvB assessment. Version 3.0". Retrieved Sept 11, 2017, from
https://echa.europa.eu/documents/10162/13643/information_requirements_part_c _en.pdf.
[EMEA], European Medicines Agency (2006). "Guideline on the environmental risk assessment of medicinal products for human use". Doc. Ref.
EMEA/CHMP/SWP/4447/00 Corr 2. Retrieved Jun 15, 2016, from
https://www.ema.europa.eu/en/environmental-risk-assessment-medicinal-products-human-use.
[EvaluatePharma®] (2018). "World preview 2018, outlook to 2024". London, Evaluate Ltd.
[KEMI], KEMI Swedish Chemical Agency (2015). "Criteria for assessing whether the substance is a phase-out substance - PBT/vPvB". Retrieved Jan 19, 2018, from
https://www.kemi.se/en/prio-start/criteria/the-criteria-in-detail/pbtvpvb. [OECD.Stat], OECD Statisics (2018). "Health/Pharmaceutical Market/Pharmaceutical
consumption". Retrieved Aug 09, 2018, from https://stats.oecd.org/.
[OECD] (2004). "Test No. 202: Daphnia sp. Acute Immobilisation Test, OECD Guidelines for the Testing of Chemicals, Section 2". Paris, OECD Publishing.
[OECD] (2012). "Test No. 305: Bioaccumulation in Fish: Aqueous and Dietary Exposure, OECD Guidelines for the Testing of Chemicals, Section 3". Paris, OECD Publishing.
[SLL], Stockholms läns landsting (2018). "Miljö och läkemedel". Retrieved Sept 18, 2018, from http://www.janusinfo.se/Beslutsstod/Miljo-och-lakemedel.
[Socialstyrelsen], Socialstyrelsen (2018). "Statistical database, pharmaceuticals". Retrieved Jan 6, 2018, from
http://www.socialstyrelsen.se/statistik/statistikdatabas/lakemedel. [UNEP], United Nations Environment Program (2004). "Stockholm convention on
persistent organic pollutants, Appendix II, Article 8, Annex D". Retrieved Sept 11, 2017, from http://www.pops.int/Portals/0/Repository/conf/UNEP-POPS-CONF-4-AppendixII.5206ab9e-ca67-42a7-afee-9d90720553c8.pdf#Article. [UNEP], United Nations Environment Program (2008). "Overview of The Stockholm Convention on Persistent Organic Pollutants". Retrieved Sept 11, 2017, from
