Bioavailability of Polycyclic Aromatic Hydrocarbons and their Potential Application in Eco-risk Assessment and Source Apportionment in Urban River Sediment

Traditional risk assessment and source apportionment of sediments based on bulk polycyclic aromatic hydrocarbons (PAHs) can introduce biases due to unknown aging effects in various sediments. We used a mild solvent (hydroxypropyl-β-cyclodextrin) to extract the bioavailable fraction of PAHs (a-PAHs) from sediment samples collected in Pearl River, southern China. We investigated the potential application of this technique for ecological risk assessments and source apportionment. We found that the distribution of PAHs was associated with human activities and that the a-PAHs accounted for a wide range (4.7%–21.2%) of total-PAHs (t-PAHs), and high risk sites were associated with lower t-PAHs but higher a-PAHs. The correlation between a-PAHs and the sediment toxicity assessed using tubificid worms (r = −0.654, P = 0.021) was greater than that from t-PAH-based risk assessment (r = −0.230, P = 0.472). Moreover, the insignificant correlation between a-PAH content and mPEC-Q of low molecular weight PAHs implied the potiential bias of t-PAH-based risk assessment. The source apportionment from mild extracted fractions was consistent across different indicators and was in accordance with typical pollution sources. Our results suggested that mild extraction-based approaches reduce the potential error from aging effects because the mild extracted PAHs provide a more direct indicator of bioavailability and fresher fractions in sediments.

Scientific RepoRts | 6:23134 | DOI: 10.1038/srep23134 the bioavailability declines as PAHs persist, or age, within a heterogeneous sediment matrix 17 . Therefore, assessment based on bioavailability is considered to be a valuable tool in risk-based approach for remediation or management of contaminated sites 18,19 .
Numerous approaches have been developed for assessing the risk of PAHs. Among these, the effects-based, sediment quality guidelines (SQGs) have been implemented worldwide for more than 20 years 14,20,21 , and are extensively applied in predicting sediment quality based on the toxicity of living organisms 22 . However, doubts have been cast on the applicability of total dose-based approaches, including the SQGs, in different sediment matrixes because the bioaccessibility of PAHs depends on the sediment properties and the aging effects 18 . Hence, biases might occur when the total dose-based approach is used in different sediment matrixes, and some governments have tried to build their local SQGs 23,24 . A variety of studies have been conducted on the relationship between bioaccumulation and extraction of non-ionic organic compounds, such as predicting the bioaccumulation using C18-/octadecyl-modified silica [25][26][27] , mild solvents 26,28 , and mixed-solvents 26,28,29 . Biomimetic extraction technologies were recently developed to predict the bioaccessibility of PAHs in sediment environments 30 . Studies have suggested that the fractions extracted from some mild solvents (e.g., n-butanol, methanol, Tenax, and hydroxypropyl-β -cyclodextrin) were equal to the effect dose of bioaccumulation and biodegradation [30][31][32][33] . Therefore, mild extraction of PAHs may be a suitable new approach for risk assessment of bioavailable PAHs. On the other hand, benthic organisms have been universally used in biomonitoring assays to reflect the organic pollution in aquatic sediments 34,35 . Tubificid (such as Tubifex sp. and Limnodrilus sp.) are the oligochaetes use in heavily contaminated sediment, because they frequently dominate the macrobiotic community in freshwater habitats and generally tolerant to organic pollutants [36][37][38][39] . These worms have developed antioxidant defense mechanisms to prevent cellular damage from reactive oxygen species when exposed to organic pollutants 40 . Glutathione S-transferases (GSTs) are a superfamily of multifunctional enzymes involved in the antioxidant defenses (phase II metabolism), whose activities have been widely used as a biomarker for predicting the toxicity level of organic pollution, including PAHs [40][41][42][43] . Therefore, the relationship between GST activity and mild-extracted PAHs might provide insight into the potential use of mild-extracted PAHs in risk assessment.
Identifying the possible sources and contributions for PAHs in sediment has been proposed in environmental management worldwide [44][45][46] . Several useful methods have been developed to identify the possible PAH sources in sediments 47 , such as ratios of different PAHs and receptor models. Generally, these methods assume that the compositions of source emissions are constant over the period of ambient and source sampling. However, in reality, the PAHs do not arrive at the receptor during the same period and hence exhibit different bioavailabilities because of variable aging. The fraction with higher bioavailability (less aged) is likely to be of greater interest to environmental managers.
The aim of this study is to reveal the potential of mild extracted PAHs for risk assessment and source apportionment of urban river sediments for future use by regulators. We therefore compare the risk assessments based on total dose with those from the mild extracted fraction of PAHs, and apply the mild extracted fraction to PAHs source apportionment in the urban river sediments.

Materials and Methods
Sampling. Fifteen sample sites were selected in the Guangzhou Section of the Pearl River, China (Fig. 1).
Four of these (F1-F4) were located within the front channel, which is a heavily engineered waterway used for sightseeing and water-bus traffic. Eight sites (R1-R8) were located within the rear channel, which serves as freight upstream and industrial buildings are located along the banks. Sites U1-U2 and D were located at the up and down confluences of the front and rear channels, respectively. The surroundings of the sample sites are described in Supplementary Table S1. hydroxypropyl-β -cyclodextrin extractable fraction, respectively. Sites F1-F4 and R1-R8 were located within the front and rear channels, respectively. Sites U1-U2 and D were located at the up and down confluences of the front and rear channels, respectively. The map was created and edited using ArcGIS (version 9.3, ESRI, USA, http://www.esri.com/software/arcgis/) and Origin (OriginLab, Northampton, MA).
Grab samples of surface sediments (triplicate) were collected, freeze-dried (Freezone 4.5, Labconco, USA) and sieved through a 2 mm sieve to remove large debris. To analyze the total content of PAHs, 5 g of sediment was then ground through a 0.15 mm sieve and Soxhlet-extracted with dichloromethane for 24 h. Activated copper was added for desulfurization. The extracts were concentrated and solvent-exchanged to hexane. Each hexane extract was subject to a silica gel-alumina (2:1) glass column (30 cm) for cleanup and fractionation. The column was eluted with 15 mL of hexane to remove the aliphatic hydrocarbons, and the second fraction containing PAHs was eluted with 70 mL of dichloromethane-hexane mixture (3:7). The PAH fraction was concentrated to 2 mL under the gentle N 2 stream. To analyze the mild extracted fraction (accessible/available fraction, a-PAHs), 1.5 g of sediment were extracted with 30 mL of hydroxypropyl-β -cyclodextrin (HPCD, 50 mM) for 12 h under 150 rpm horizontal vibration. After centrifuging, the HPCD extracts underwent liquid-liquid extraction with hexane. The cleanup and fractionation steps were as described above, and then the extracts were concentrated to 0.2 mL. A known quantity of the recovery standard, m-diphenylbenzene, was added to the sample prior to instrumental analysis.
Total organic carbon content (TOC) was determined using the potassium dichromate dilution heat colorimetric method. Grain sizes of sediments were determined by Mastersizer (Malvern 3000, UK).
Bioassay with tubificid worms. The tubificid worms, Limnodrilus hoffmeisteri (Oligochaeta Tubificidae) were maintained and acclimatized under laboratory conditions. They were washed with distilled water prior to analysis. For analysis, the worms were kept in a baker with sediment and river water for 14 days in the dark. A glass bead was used as a control. At the end of the exposure period, the survival of the L. hoffmeisteri was evaluated, and the worms were washed and immediately deep frozen. The tissues were homogenized at 4 °C using 0.5 g wet weight L. hoffmeisteri per 1 ml phosphate buffer PBS (pH 7.2). The homogenate was centrifuged at 4 °C at 2,500 × g, and the supernatant was used to analyze the GST activities. The GST activity in the L. hoffmeisteri extract was evaluated as the formation of a conjugate between glutathione and 1-chloro-2,4-dinitrobenzene at 340 nm in 50 mM potassium phosphate buffer, pH of 6.5 with 1 mM EDTA 48 . The enzyme activity is reported as the number of micromoles of conjugate formed per minute per mg protein. Protein concentrations were estimated using the Coomassie brilliant blue method. Unfortunately, due to an accident, the experiment failed in samples from sites F1 and R1.
Risk assessment with sediment quality guidelines. The consensus-based SQG approaches were used in this study. The consensus-based probable effect concentration (PEC) value represents a concentration above which adverse effects to benthic organisms are likely. The incidence of sediment toxicity could be evaluated with the ranges of mean PEC quotient (mPEC-Q) 49,50 . As established in the SQGs, the PECs are concentrations of individual chemicals above which adverse effects in sediments are expected to frequently occur. Nine PAHs (naphthalene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzo[a]pyrene) were evaluated using the SQGs, according to MacDonald et al. 22 . The threshold values of consensus-based PECs 22,50 are listed in Supplementary Table S2.
For each chemical in each sample in the database, a PEC quotient (PEC-Q) was calculated by dividing the concentration of that chemical by the PEC for that chemical 50 : where C i is the concentration (ng g −1 ) of the chemical i, and PEC i is the threshold effect concentration (ng g −1 ) for the chemical i listed in Supplementary Table S2. In particular, the total PAHs (denoted as Σ 9 PAHs) served as an individual chemical when predicting the total effect of PAHs. For each sample, a mPEC-Q was calculated by dividing the sum of individual quotient for each chemical by the number of PECs evaluated:

Results and Discussion
Distribution of PAHs and their bioavailability in Pearl River sediment. The Pearl River supports the development of Guangzhou city. Its main channel and waterways are responsible for shipping, flood discharge, and receiving pollution. Figure 1 shows that high total PAHs (Σ t-PAHs) were observed in the sediments that were subjected to frequent human activities, such as the water traffic artery (F1-F3), ship building and repairing industries (F4, R1-R5) and areas with riverside human communities (R1-R4, R8). As a result of their high hydrophobicity and persistence, PAHs entering the aquatic ecosystem tend to rapidly adsorb onto suspended particles and settle to the sediment where they become accumulated 54 . Similar to other studies 55,56 , we found the highly hydrophobic PAHs such as 4-ring and 5-ring PAHs dominated in sediments (31%-49% and 13%-36%, respectively). However, the concentrations of PAHs did not decrease along the flow direction. The sites subject to frequent human activities (F1-F4, R1-R5) had a concentration of Σ t-PAHs around two times higher than sites located in the upstream (U1) and suburban areas (R6 and R7, which is close to orchard land use) (Fig. 1). This confirms that the geographic distribution of PAHs in urban river sediment is associated with hotspots of urban activities. Although high concentrations of total PAHs have been detected in many river sediments, the presence of PAHs in sediments does not always signify toxicity to the ecosystem 18 . Recent studies suggest that the bioavailable fraction of PAHs is able to account for the eco-toxic effects 57,58 and mild extraction with hydroxypropyl-β -cyclodextrin was demonstrated to be relevant to bioavailability [59][60][61] . In this study, we extracted the bioavailable PAHs (a-PAHs) from Pearl River sediments with hydroxypropyl-β -cyclodextrin. We found that the dominant a-PAHs were 3-ring and 4-ring PAHs (Fig. 2), which accounted for 4.7%-21.2% of t-PAHs ( Supplementary Fig. S1). These percentages were similar to the results from other research 30 . However, the high variability in the percentage of a-PAHs implies that the toxicity of PAHs was not only dependent on t-PAH doses, but might be also controlled by environmental factors. Previous research has suggested that the bioavailability of PAHs in soils and sediments were controlled by the octanol-water partition coefficient (K OW ) of individual PAHs, and the organic compounds content and grain size of sediments 18,32,62,63 . The ratio of a-PAHs to t-PAHs decreased as the lgK OW increased (Supplementary Fig. S2). However, significant correlations were observed only between the LMW-PAHs and TOC (r = − 0.660, P < 0.01) and between HMW-PAHs and fine particles (r = 0.730, P < 0.05) but not for the HMW-PAHs with TOC and LMW-PAHs with fine particles (Supplementary Fig. S3). The reasons might be that sediment organic matter may vary in properties and consist of aggregates of material that may affect the fate of adsorbed PAHs 18 . Furthermore, the PAH distribution relative to organic carbon content might differ between particle-size fractions 64 . The complex interrelation of these physiochemical properties makes it difficult to directly correlate the PAHs concentration with bioavailability even through normalizing to the physiochemical properties (organic carbon) which was conducted in some SQGs 14 . Therefore, sites with relatively low t-PAHs but high a-PAHs (e.g., U1, R6, and R7) could be even more toxic than the heavily polluted sites with high t-PAHs concentration (e.g., F2 and R2). To illustrate this, we compare the a-PAH-based and t-PAH-based eco-risk assessments in the following section.

Comparison of a-PAHs to traditional eco-risk assessment and bioassay. Recent studies by
Harmsen et al. 65 and Duan et al. 18 suggest that the assessment of PAHs pollution remediation should be based on the bioavailability, which will lead to a realistic appraisal of the potential risks from exposure to contaminants. We compare a-PAH content and consensus-based SQGs in this study. As illustrated in Fig. 3, the correlation between Σ 9 a-PAHs (hereafter the symbol Σ 9 represents the sum over the 9 selected PAHs) and PEC-Q of Σ 9 PAHs was significantly high (r = 0.730, P < 0.01, Fig. 3a), whereas the correlation between Σ 9 a-PAHs and mPEC-Q of each individual PAHs was comparatively lower (r = 0.541, P = 0.048, Fig. 3b). To gain further insights, LMW-PAHs and HMW-PAHs were analyzed separately. As illustrated in Fig. 3c, there is positive correlation between a-PAH content and mPEC-Q of HMW-PAHs but no significant correlation is found between a-PAH content and mPEC-Q of LMW-PAHs. This could explain the discrepancy between a-PAH and t-PAH based approach in this study where LMW-PAHs contributed to 16% to 38% of the total PAHs in sediment samples from different sites (see Fig. 2). The predicted results by SQGs did not correlate well with the bioavailability, possibly due to the regional differences in geochemistry of sediments 14 and aging effects that might lead to different fate of LMWand HMW-PAHs 18 (Supplementary Fig. S3).   We next compare the results of sediment toxicity test with L. hoffmeisteri with those from the a-PAHs and the mPEC-Q. We selected GST activity as the toxic biomarker because GST plays a crucial role in cellular protection against oxidative stress and toxic compounds such as PAHs 41,42 . We found that the GST level of worms in some cases was lower than that in controls (4.69 ± 0.14 mg mL −1 ) (Fig. 4). The depleted GST activities have also been recorded in mussels in the sites polluted with high PAH concentrations 41 . As Akcha et al. 41 suggested, the competition between endogenous substrates and those produced by PAH biotransformation could be responsible for an inhibition of GST. Our results demonstrate a negative correlation between a-PAHs and GST activities in L. hoffmeisteri (r = − 0.654, P = 0.021), suggesting that PAHs inhibited the GST activity (either LMW or HMW PAHs, Fig. 4a). However, no significant correlation was observed between GST and t-PAHs (r = − 0.252, P = 0.430) or t-PAHs-based assessments (including the correlations of either mPEC-Q or PEC-Q of the Σ 9 PAHs, Supplementary Fig. S4). The lack of correlation between t-PAHs and GST activities suggested that the t-PAHs-based method established in northern America might be unable to represent the bioavailability of PAHs in the sediment matrix of this study. Though the eco-toxic tests are generally respond to the status of all contaminants, as for the assessing the GST inhibition by PAHs, however, the a-PAHs has better prediction ability. In all, this suggested the potential applicability of the mild-extraction approach to pollution assessment.
As discussed earlier, the TOC, the percentage of clay, and even other sediment conditions may influence the aging of PAHs and hence affect their bioavailability. The assessment method based on the total dose of PAHs is unable to account for the various sediment properties in different environments. Therefore, we suggest that a-PAHs represent a more suitable and accurate index to determine eco-toxicity.
The application of mild extraction in source apportionment. In addition to eco-risk assessment, aging effects should also be considered in source apportionment of PAHs. Most PAH source apportionment in soil or sediment is based on t-PAHs and assumes that the PAHs were mass conservative from the sources to acceptors (i.e., there was no change in source profile between the source and receptor) 51,52 . However, PAHs age to different extents after reaching the sediment matrix, where degradation rates are affected by various environmental factors. All of such processes affect the accuracy of the source apportionment. Here, we define the a-PAHs as the fresher (labile or less aged) fraction and the residual PAHs (r-PAHs, the difference between t-PAHs and a-PAHs) as the aged fraction. The source apportionment results for both t-PAHs and r-PAHs differ with types of molecular ratio being used to diagnose (Fig. 5). For example, source based on FLT/(FLT + PYR) suggested that t-PAHs and r-PAHs in sites were primarily determined by biomass combustion, whereas apportionment based on BaA/(BaA + CHR) or IPY/(IPY + BPE) suggested petroleum or petroleum combustion to be the primary source of t-PAHs and r-PAHs. The complex aging and degradation processes led to different fates of PAHs, which made it difficult to appoint the sources of r-PAHs and hence decreased the accuracy of the source apportionment based on the t-PAHs content. These inconsistent source apportionment results imply that further study should be conducted on the accumulation of PAHs in sediments in order to identify the behavior and fate of PAHs. In contrast, the source apportionments based on a-PAHs produced consistent predicting sources under the two pairs of molecular ratios (Fig. 5). This suggests that the a-PAHs obtained through mild extraction, which represents the near-term (or fresher) pollution, might be more meaningful for source apportionments than t-PAHs.
The t-PAHs data were decomposed into a-PAHs and r-PAHs, and then each of the three (t-PAHs, a-PAHs, and r-PAHs) was input to the CMB. Using t-PAHs or r-PAHs as input gave similar source contribution profiles; however, those using a-PAHs as input were substantially different (Fig. 6). This implies that the source apportionments of t-PAHs were mainly dependent on the aged fraction, namely the r-PAHs, which is not surprising as in this study r-PAHs accounted for 79% to 96% of t-PAHs in sediment samples. However, the predicted sources of a-PAHs were diesel engines (DE) and coke ovens (CO) (Fig. 6), which were more indicative of the typical PAH pollution sources in urban rivers (dominated by shipping activities and riverside discharges). These results suggest that source apportionment based on a-PAHs provides more accurate prediction than those based on t-PAHs. Furthermore, the a-PAHs represent the least aged PAHs with the highest toxicity potential, and should therefore be of special concern with respect to the current status of contamination.

Conclusion
In the present work, mild extracted fraction of PAHs showed avantage of predicting the toxicity of PAHs and exhibited higher consistency and rationality in source apportionment than t-PAHs-based approaches. Therefore, for environmental management, we suggest that the mild extracted fraction of PAHs, which was demonstrated to be the bioavailable and un-aged fraction of PAHs, should be taken into account to reduce the error from aging effects. This work suggests that the implementation of mild extracted fraction in risk assessments and source apportionments would be beneficial, though further investigation is still required.