Genomic and Chemical Markers of Contamination in Young of the Year Bluefish from Contrasting Estuaries

Abstract

To investigate the effects of habitat degradation on the genomic level, young of the year (YOY) bluefish, (Pomatomus saltatrix) from the highly urbanized Hudson Raritan Estuary (HRE) and the less impacted Mullica River-Great Bay (MRGB) estuary were analyzed for condition, level of PCBs (Polychlorobiphenyls), and pesticides, and compared in a preliminary transcriptomic study. For chemical analysis, whole juvenile fish were freeze-dried, extracted in DCM (Dichloromethane), extracts were cleaned by florisil/alumina/silica column chromatography and size-exclusion HPLC, and quantified by gas chromatography. For transcriptomic analysis, Suppression Subtractive Hybridization (SSH) libraries from the pooled livers of fish from the two sites were constructed to subtract common transcripts and recover cloned expressed sequence tag (EST) transcripts that were putatively more highly expressed in one or the other environment. Transcripts were Sanger-sequenced and identified by bioinformatic analysis. Relative quantitative PCR (polymerase chain reaction) was used to investigate a subset of the selected transcripts. MRGB fish were larger and in better condition than HRE fish. HRE fish had PCB concentrations that were, in sum, about seven times higher than MRGB fish and were more highly contaminated with pesticides including Dichlorodiphenyltrichloroethane (DDT) and its metabolites. Transcripts related to metabolic, transport, immune, detoxification, signaling, and environmental response and additional functions were associated with contrasting habitats and provide gene targets to investigate the hypothesis that the growth and condition of the HRE bluefish may be limited by their poor environmental quality.

1. Introduction

Anthropogenic impacts including chemical contamination can potentially limit the values and function of nursery habitats for reproduction and growth of estuarine-dependent fishes [1]. To investigate the genomic effects of habitat degradation, juveniles of the highly migratory bluefish, Pomatomus saltatrix, which has an estuarine-dependent juvenile stage, were compared in a preliminary transcriptomic investigation. Bluefish is a wide-ranging migratory species of commercial and sport fishery value [2]. Historically, fluctuations have been observed in bluefish populations and recent assessments (2025) indicate that bluefish stocks have been overfished in the past but are not experiencing overfishing at the current time [3]. With uncertainties in the variables surrounding recruitment from year to year in any fishery, it is particularly difficult to assess recruitment in bluefish given its complex migratory nature, little-understood spawning regime, and the attendant uncertainties of a pelagic larval stage followed by a shift to the near-shore or estuarine-dependent juvenile stage [4]. Another important variable in the dynamics of bluefish populations is that the east coast nursery areas for juveniles, the Mid-Atlantic bight (MAB), include some of the most anthropogenically altered and contaminated habitats in the species’ range [5,6].

Bluefish fry enter the estuaries along the coast after yolk development in the spring and later in the summer in two distinct cohorts due to a spring and summer spawn [7]. The first year of life is spent within the nursery estuary, feeding on progressively higher trophic-level prey and rapidly growing [2]. By the end of the first season, this ravenous predator has reached a length of ~200–250 mm and will join the offshore southerly migration [7,8]. Given the rapid growth (1.17–1.35 mm/d.) of the young of the year (YOY) [7] and their successive food choices (copepods, fry and juveniles of Atlantic silversides, bay anchovy, menhaden, clupeids, striped bass, sand shrimp, mysids, other fish and invertebrates) [2], the YOY bluefish has been shown to be a good integrator of the chemical condition of the estuary [9]. Although their first-year residence in the nursery estuaries is relatively short, significant bioaccumulation of polychlorinated biphenyls (PCBs), polybrominated diphenyl ethers (PBDEs), and organochlorine pesticides (OCPs) can occur in YOY bluefish [9,10,11]. Bluefish apparently continue to accumulate PCBs and other contaminants throughout their life [12,13].

To ensure optimal growth and development of fishes through their embryonic and juvenile stages, the underlying proper regulation of genes and the physiological processes they support is essential. Environmental degradation due to anthropogenic chemical contamination of persistent organic pollutants (POP) may contribute to poor growth and limited reproductive success due to disruption of physiological processes and mis-regulation of genes. To test this relationship, bluefish from two contrasting environments were compared to investigate the contaminant load of selected POPs and liver transcripts that might be associated with this load. Two sites with differing anthropogenic contamination were compared in this study. Site one was Newark Bay (Hoboken, NJ), which is a major east coast U.S. commercial shipping port within the Hudson Raritan Estuary (HRE) complex, with a well-documented history of industrial contamination [6]. In contrast, site two was the relatively less impacted Mullica River-Great Bay (MRGB) estuarine system primarily consisting of forested open space and suburban and rural uses, which is located within the Pinelands National Reserve, and the Jacques Cousteau National Estuarine Research Reserve (JCNERR), a part of the National Estuarine Research Reserve System administered by the National Oceanic and Atmospheric Administration (NOAA). To characterize the level of contamination at the contrasting field sites, PCBs and a suite of organochlorine pesticides including Dichlorodiphenyltrichloroethane (DDT) were determined as markers of habitat contamination in sampled fish. These are not a comprehensive inventory of potential site contamination in the HRE, which also includes Dioxins, PBDE, PFAS (Per- and Polyfluoroalkyl Substances), PAH (Polycyclic aromatic hydrocarbons), HC/VOC (hydrocarbon and volatile organic compounds), pharmaceutical and personal care products (PPCP), and heavy metal among others [5,14]. To preliminarily investigate transcriptomic differences, Suppression Subtractive Hybridization (SSH) cDNA libraries were used to enrich expressed sequence tag (EST) transcripts in bluefish livers that were present or absent in the contrasting habitats. SSH has been shown to be a useful tool to isolate potentially differentially expressed genes from the environment [15,16,17]. Isolation and identification of these putative environmentally responsive EST transcripts can also be an important step in developing bio-monitoring tools to relate genetic and physiological conditions to other indices, such as size and growth rate, which contribute to predictive models of year class recruitment.

2. Materials and Methods

2.1. Sites and Sampling

Bluefish collection occurred at two sites, at Newark Bay/Bayonne, NJ in the urbanized HRE (40°39′46.35″ N; 74° 7′59.56″ W) and Graveling Point, NJ in the less urbanized MRGB (39°32′0.85″ N; 74°23′27.14″ W), that were selected for contrasting levels of urbanized industrial anthropogenic contamination.

A total of 50 YOY bluefish were captured by hook and line or seine within one week from the two sites in August of 2004. Fish were immediately placed on dry ice after capture and stored for four months at −80 °C until processed. All fish were measured for total length, fork length and wet weight before further processing. Fulton’s condition index was calculated as K = 100,000 × weight (g)/standard length (mm)³ [18]. Twenty-two fish from each site (44 total) were randomly assigned to either chemical analysis (n = 11 per site × two) or genetic analysis (n = 11 per site × two). Chemical analyses were conducted on the whole individual fish per site, while livers were extracted and pooled by site (n = 2) for the genomic analysis to capture variability within the sites while providing enough material to perform contrasting site-specific transcript subtraction. All chemical and genomic analyses were completed within two years of initial processing. All experimental protocols were approved by the Stockton University Institutional Animal Care and Use Committee (IACUC) 20020521, 21 May 2002.

2.2. Chemical Analysis

Organic chemical analyses for two classes of POP (PCBs and Pesticides) were undertaken at the NOAA James J Howard Marine Sciences Laboratory (Sandy Hook, NJ, USA), generally following Deshpande et al. [10,12]. Briefly, the whole individual fishes were partially thawed and cut into pieces. Samples were lyophilized until dry and then ground in a blender with anhydrous sodium sulfate. A blank and a certified reference material (CRM), CARP-2 [19], were included in the analysis. Method surrogate (MS) internal standards (4,4’-dibromooctafluorobiphenyl, Sigma Aldrich, St. Louis, MO, USA; Ronnel, Ultrascientific, North Kingstown, RI; and PCB 198, AccuStandard, New Haven, CT, USA) were added to the ground sample. Samples were extracted into 400 mL Dichloromethane, (DCM) (Pesticide Residue Analysis Grade, Fisher Scientific, Pittsburgh, PA, USA) for 24 h in a Soxhlet apparatus (Organomation, Berlin, MA, USA) over an 80 °C water bath and the extract concentrated to ca. 10 mL on removal. Extracted samples were passed over a florisil/aluminum oxide/silica gel column (10/10/20 g) to remove bulk polar internal interfering compounds. Column-cleaned, extracted samples were concentrated with a Zymark Turbovap, (Hopkinton, MA, USA) using nitrogen flow and then brought to exactly 5 mL volume with DCM.

Lipid concentration was estimated by dry weight of a subsample of the concentrated extracts. Liquid chromatography (LC) internal standards (1,2,3-Trichlorobenzene, Sigma-Aldrich, St. Louis, MO and PCB-192, AccuStandard, New Haven, CT, USA) were added to another subsample of the extracts, based on lipid concentration, and brought to 5 mL. Additional interfering substances were removed by semi-preparative HPLC (Hewlett Packard 1050, Palo Alto, CA, USA, λ-254 nm) on a size exclusion 600 × 21.2 mm: Phenogel 10, 100 Å, 10 μm column (Phenomonex, Torrance, CA, USA), preceded by a 50 × 7.8 mm: Phenogel 10 μm guard column. A calibration solution was made up of 4,4’-Dibromooctafluorobiphenyl, Biphenyl and Perylene. Appropriate fraction cuts were taken at calibrated intervals, exchanged into hexane, concentrated using the Zymark Turbovap and then brought to exactly 5 mL.

Samples were screened for contaminant levels using an Agilent 5890 (Santa Clara, CA, USA) Gas Chromatograph (GC) with an electron capture detector (ECD). Based on the screening levels, final dilutions or concentrations were established. Internal surrogate GC standards of 2,4,5-Tetrachloro-m-xylene (Sigma Aldrich, St. Louis, MO, USA), PCB-103, 3,3’,4,4’-Tetrabromobiphenyl, and Octachloronapthalene (AccuStandard, New Haven, CT, USA) were added to each extract. Final extracts with internal standards were analyzed on an Agilent 5890 GC with ECD after 1 µL injection with an Agilent 7673 auto-sampler. The sample was injected in the splitless injection mode into a fused-silica capillary guard column (10 m × 0.25 mm ID) concatenated into a fused-silica capillary column (DB-5 60 m × 0.25 mm ID, 0.25 µm film thickness). The GC column was heated from 50 °C to 155 °C at 5 °C/min, 155 °C to 215 °C at 1 °C/min, 210 °C to 315 °C at 4 °C/min and held at 315 °C for 16.75 min. Hydrogen served as carrier gas with nitrogen as the make-up gas. The carrier gas pressure was held for 1 min at 19.4 psi. It was then programmed to 23.2 psi at 0.17 psi/min, to 25.3 psi at 0.04 psi/min, and then to 29.3 psi for 0.15 psi/min where it was held for 17 min. The injector temperature was 280 °C and the detector temperature was 320 °C.

The primary calibration solution contained a mixture of the PCB Calibration Check Solution C-CCSEC (20 PCB congeners, AccuStandard, New Haven, CT, USA) and chlorinated pesticides in National Institute of Standards (NIST) Standard Reference Material (SRM) 2261 (15 chlorinated pesticides) [20]. The secondary calibration solution contained a mixture of eleven PCB congeners in NIST SRM 2274 [21], eight chlorinated pesticides in NIST SRM 2275 [22], and three individual PCB congeners 87, 132, and 201 (AccuStandard, New Haven, CT, USA). Target analyte peaks were quantified with the internal standard method using the HP 3365 Series II ChemStation Software (Version A.03.34).

Calibration solutions (primary and secondary) were run after every four samples. Eight coeluting compounds, PCB 77, PCB 101, PCB 110, PCB 169, p,p’-DDD, α-BHC, mirex, and dieldrin, were omitted from the interpretation of the chemical profiles. PCB nomenclature follows Ballschmiter and Zell [23]. Total PCB concentration was reported as the sum of concentration of 26 individual PCB congeners analyzed in this study. Aroclor-equivalent PCB concentrations were generated by summing the concentrations of PCB congeners 18, 28, 44, 52, 66, 105, 118, 128, 138, 153, 170, 180, 187, 195, 206, and 209; and then doubling this sum [24]. Total DDTs were calculated as the sum of concentrations of DDTs and metabolites: Dichlorodiphenyldichloroethane and Dichlorodiphenyldichloroethylene (DDD and DDE). Total chlordanes were calculated as the sum of concentrations of α-chlordane, γ-chlordane and τ-nonachlor. Concentrations of all analytes are expressed as ng/g wet weight. Method detection limits (MDL) for PCBs range from 1.1 to 3.1 ng/g (ppb), and pesticides range from 1.1 to 3.2 ng/g (ppb) [12].

The MS internal standards, CRM, and the method blank served as quality assurance checks. In all cases, the blank had no measurable level of analytes detected. Internal method standards were recovered from all samples. The MS internal standard recoveries were as follows: 4,4’-dibromooctafluorobiphenyl (34% + /− 9), Ronnel (14% +/− 19) and PCB 198 (106% +/− 30). Ronnel is a synthetic organic thiophosphate compound. Gehrt [25] reported that the low recoveries of ronnel could likely occur because of degradation. Low recoveries of DOB have also been reported in an avian egg study [26]. The LC internal standard recoveries were: 1,2,3-Trichlorobenzene (80% +/− 14) and PCB-192 (95% +/− 16). Recoveries of analytes from the CRM CARP-2 (Appendix A,

Table A1. CARP-2 certified reference material recovery.

Anylate	Certified Value Mean	Certified Value SD	Recovery Value	% Recovery	z-Score
PCB 18	27.3	4	11.16	40.88	−4.03
PCB 28	34	7.2	20.74	61.00	−1.84
PCB 44	86.6	25.9	56.44	65.17	−1.16
PCB 52	138	43	81.46	59.03	−1.32
PCB 118	148	33	101.84	68.81	−1.4
PCB 128	20.4	4.4	11.16	54.71	−2.1
PCB 153	105	22	58.08	55.31	−2.13
PCB 180	53.3	13	27.2	51.03	−2
PCB 194	10.9	3.1	8.14	74.68	0.187
PCB206	4.4	1.1	3.48	79.09	0.202

) ranged from 40 to 79% and are in line with previous recoveries based on these methods [12].

2.3. Statistical Analysis

A suite of univariate and multivariate statistical techniques was utilized to calculate differences in biometrics, chemical contaminants, and chemical assemblages between sampling locations. Significant differences in mean fish length, weight, condition, and PCB and pesticide concentrations between locations were tested with the two-tailed independent Student’s t-test following Levene’s test for equality of variance at an alpha of 0.05. Samples with unequal variance were evaluated under increased stringency of the t-test with reduced degrees of freedom. Univariate statistics were performed using IBM SPSS (package v25.0.0, Armonk, NY, USA).

To assess the relationship between PCB congeners and sample locations we employed an ordination technique known as principal component analysis (PCA). PCA reduces the number of original variables in a dataset into a smaller set of composite components by maximizing the variation among samples along their axes [27]. Thus, principal component 1 (PC1) explains the most variance in the dataset, principal component 2 (PC2), the second largest amount of variance, and so forth. The “loading” of a variable (in this case a PCB congener) describes how much it contributes to a given principal component. To perform the PCA the data were arranged in a variable-by-sample matrix, where the variables were the individual PCB congeners and normalized concentrations represented the value fields. The concentrations were normalized by dividing an individual congener concentration by the sum of the concentrations of all congeners within the sample. The PCA was conducted in R [28] and visualization was created using the “factoextra” package [29].

2.4. RNA Extraction of Liver Tissue

Genomic analyses were performed on pooled liver samples of the sub-population at each site as the SSH subtraction method was used to compare the transcriptome of one habitat versus another. As such, this study consists of one biological replicate of the transcriptomic and quantitative PCR analyses and results must be considered as preliminary. Pooling individuals can incorporate a larger range of variations in the samples while reducing costs but can lead to bias in the libraries, particularly of certain genes; therefore, alternate methods of gene expression such as quantitative polymerase chain reaction (qPCR) can be used as a check on expression [30]. Fish for genetic analysis, 11 from each site, were partially thawed and the liver dissected out on ice. A 0.25 g subsample of each liver was taken for pooled RNA extraction and stored on dry ice. Liver tissue was homogenized on ice using a Brinkman Polytron (Kinematics, AG, Lucerne, Switzerland) and extracted using the guanidine thiocyanate method with an RNAgents total RNA Isolation System (Promega, Madison, WI, USA) per standard protocol. Total RNA for each site was resuspended in nuclease-free water and quantified by Ultraviolet (UV) spectroscopy at A260 using an extinction coefficient of 40 ug/mL/A260. RNA integrity was ascertained using a BIO-Rad (Hercules, CA, USA) Experion Bioanalyzer with Experion RNA Analysis kits (chips and standards). Messenger RNA (mRNA) was purified by a streptavidin affinity magnetic separation from the total RNA using a PolyATract mRNA Isolation System (Promega, Madison, WI, USA). Isolated mRNA, eluted in nuclease-free water, was concentrated by precipitation in ethanol, resuspended in nuclease-free water and quantified by UV spectroscopy. A final precipitation, quantification by UV spectrophotometry and adjustment to concentration of 0.5 µg/µL for library construction was necessary. Total RNA and mRNA were extracted, stored at −80 °C and used within four months of fish capture. RNA for qPCR was used within two years.

2.5. Suppression Subtractive Hybridization (SSH) Libraries

SSH libraries, forward- and reverse-selected, were constructed using Clontech PCR-Select^tm cDNA (copy DNA) library construction kits (BD Biosciences, Palo Alto, CA, USA) following the manufacturer’s protocol [31]. An amount of 2 μg of mRNA was used for each first-strand synthesis. Following second-strand synthesis, the cDNA was digested with RsaI and one aliquot was ligated to Adaptor 1, another to Adaptor 2R, and the remaining used as drivers in the hybridization reaction. The forward library was selected by hybridizing adaptor-ligated cDNA from the more highly anthropogenically impacted HRE fish as the tester in the presence of cDNA from the less impacted MRGB control as the driver. The forward reaction was designed to yield clones that were more highly expressed or up-regulated in the chemically impacted fish. The reverse library used the adaptor-ligated cDNA from the control (MRGB fish) as the tester and the impacted (HRE fish) cDNA as the driver. This reverse library was designed to recover clones that are under-expressed or down-regulated in the impacted fish versus the controls. The driver cDNA was added in each case to remove common cDNAs by hybrid selection, leaving the respective unique tester cDNAs for recovery and cloning. The differentially expressed cDNAs were amplified by PCR with Advantage^R 2 polymerase mix (BD Biosciences, Palo Alto, CA, USA). Primary and secondary amplification of the reciprocal cDNA subtractions produced two SSH libraries (up- and down-regulated with anthropogenic impact) that were enriched for putative differentially expressed cDNAs. The libraries were cloned using a pGEM Easy T/A cloning vectors kit with JM108 competent E. coli (Promega, Madison, WI, USA). Transformants were plated on Luria–Bertani (LB) agar with ampicillin (100 μg/mL), 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) (0.5 mM) and Isopropyl β-D-1-thiogalactopyranoside (IPTG) (80 μg/mL), and white colonies were selected for overnight growth in LB broth with ampicillin (50 μg/mL). DNA plasmids were prepared with a Wizards Miniprep kit (Promega, Madison, WI, USA). Plasmid DNA was quantified by UV spectrophotometry at A260 using an extinction coefficient of 50 μg/mL/A260.

2.6. Sanger Sequencing

Plasmid double-stranded DNA template (50–100 fmoles or ~120 ng) was preheated at 94 C for 1 min and cycle sequenced in a CEQ^TMDCTS Quick Start PCR dye-termination reaction (Beckman Coulter, Fullerton, CA, USA) with 10 pmoles of T-7 or Sp6 sequencing primer (Promega, Madison, WI, USA). Sequence reactions were purified by ethanol precipitation in the presence of glycogen (0.8 μg/μL), 0.24 M sodium acetate (pH 5.2), and 8 mM of Na₂-EDTA, washed in 70% ethanol, and dried and resuspended in deionized formamide as per the DCTS kit protocol (Beckman, Fullerton, CA, USA). DNA sequencing was performed on a Beckman Coulter CEQ 8000 using the CEQ LFR-1, long fast read, protocol (Beckman, Fullerton, CA, USA). One hundred and sixty clones were prepped and sequenced. Sequence data were analyzed with Invitrogen Vector NTI (Thermo-Fisher, Waltham, MA, USA) and trimmed to remove vector sequence. Putative transcript identifications were made using BLAST-X web resources [32] of the National Center for Biotechnology Information (NCBI) at the National Institutes of Health (NIH).

2.7. Gene Expression by Quantitative Polymerase Chain Reaction (qPCR)

To further investigate gene expressions, a relative reverse transcription (RT) qPCR method was used to compare the expression of selected transcripts from each of the contrasting sites [33]. Primers for PCR were designed from sequences of seven clones from the up-regulated contamination library using Vector NTi (Thermo-Fisher, Waltham MA, USA). Primers were synthesized by IDT (Coralville, IA) and tested by PCR of the respective plasmid clones to ensure specificity and production of a single product (dissociation curve analysis). Total RNA (5 μg/reaction), as extracted from each sample set, was reverse-transcribed with an Invitrogen Superscript II first strand synthesis kit (Thermo-Fisher, Carlsbad, CA, USA) using the kit’s polyT primer to produce cDNA first strand template for qPCR. A total of five RT reactions for each sample site, HRE and MRGB, were prepared and combined within site to provide adequate template for qPCR comparison. For each gene primer set, serial dilutions (1:10×, 1:100×, 1:1000×) of the template expected to have a higher target concentration were used to generate a standard dilution curve [33]. Standard dilutions and 1× samples (~240 ng) were cycled in triplicate (technical replicates) in a 25 μL reaction with a Stratagene Brilliant SYBR^R Green qPCR master mix (Agilent, La Jolla, CA, USA) with 0.3 μM of each gene-specific primer in the presence of fluorescent, DNA binding dye SYBR Green and the reference dye ROX. An ABI 7500 Real Time PCR System (Thermo-Fisher, Waltham, MA, USA) was used to cycle the reaction using 1 cycle of 15 min. at 95 °C followed by 40 cycles of 94 °C for 40 s., 55 °C for 40 s. and 72 °C for 1 min. A no-template blank was included in each experiment. Triplicate fluorescence measurements were collected during the 72 °C step. A dissociation curve was generated for each amplification run with 15 s at 95 °C and 15 s at 60 °C followed by a 20 min ramp from 60 °C to 95 °C. Fluorescence was analyzed with the ABI 7500 Sequence Detection Software, SDS v1.2, (Thermo-Fisher, Waltham, MA, USA) as delta Rn vs. cycle, normalized to the reference dye (ROX), to generate crossing temperature Ct with a threshold value of 0.009. The mean Ct value of triplicate reference samples was plotted against the log₁₀ of microliter template dilution (0, −1, −2, −3) to produce a straight line. The equation derived from the standard curve data was used to convert the sample Ct values to the exponent (base 10) of the equivalent microliter template volume of the tested sample, and solving the exponential equation gives the equivalent microliter template volume of the tested sample. The fold induction or inhibition was calculated by dividing the 1× volume of sample (1 μL) by the equivalent microliter template volume [33]. No endogenous control for the bluefish was available at the time, so these fold inductions are not normalized (for potential variation in the efficiency of the RT reactions) and can only be interpreted as relative differences between the sites for the purpose of supporting the subtractive library method.

3. Results

3.1. Bluefish Sites and Condition

Juvenile bluefish (YOY) from the highly urbanized HRE were typically smaller in length and weight than fish from the less impacted MRGB, and mean weights for MRGB fish were significantly higher than HRE fish, as were mean fork lengths (

Table 1. Mean fork length and wet weight for YOY bluefish from the Hudson Raritan Estuary (HRE) and from the Mullica River-Great Bay (MRGB) Estuary. Probability (p) for t-test, * significant at α = 0.05.

		Fork Length (mm)		Weight (g)
Site	N	Mean	S.D.	Mean	S.D.
HRE	22	155.18	17.19	45.67	17.61
MRGB	22	176.08	11.02	69.85	13.92
p			<0.001 *		<0.001 *

). Fish from both sites showed a strong linear relationship between length and weight (Figure 1). Fulton’s condition index was significantly different between sites, with a wider distribution of fish in HRE exhibiting less optimal condition (Figure 2).

3.2. Polychlorobiphenyl (PCB) Analysis

Twenty-six PCB congeners were detected in all samples of bluefish. The mean concentration for each of the PCB congeners separated was significantly higher (t-test, n = 11, α = 0.05) at HRE than MRGB (Figure 3). The sums of PCB concentrations (sum total and sum of Aroclors) at the HRE site were approximately seven times those seen at MRGB (

Table 2. Polychlorobiphenyls (PCBs ng/g) in YOY bluefish from Hudson Raritan Estuary (HRE) and from the Mullica River-Great Bay (MRGB) Estuary. Σ = sum, p = probability of significant difference (*) in means by t-test, a < 0.05).

Location	HRE		MRGB
No. of Samples	11		11		Means/t-test
Analyte	mean	S.D.	mean	S.D.	p
Sum (Σ) PCBs	356.23	114.85	53.44	14.731	<0.001 *
Sum (Σ) Aroclors	712.46	229.70	106.886	29.462	<0.001 *

). While individual congeners and concentrations varied by site, they represent a clear chemical fingerprint of the respective estuaries (Figure 3). Principal component analysis of the PCB congener data revealed two distinct sample groups (Figure 4). The first principal component (PC1) accounted for almost 90% of the variance between the samples.

Samples from MRGB were found on the positive side of the axis while samples from HRE grouped on the negative side of the axis. The second principal component (PC2) accounted for an additional 4.5% of the variance within the dataset. Five PCB congeners had PCA loadings that contributed greater than 0.3 to PC1: PCB 153 (0.48), PCB 138 (0.36), PCB 187 (0.34), PCB 52 (−0.39), and PCB 49 (−0.31) (See Appendix A,

Table A2. PCA loading scores for the PCB congeners included in the analysis. PCB congeners with loading scores larger than 0.3 for PC1 are in bold and are displayed in the PCA biplot (Figure 4).

Congener	PC1	PC2
PCB 52	−0.391243187	0.046030553
PCB 49	−0.306668276	0.082704940
PCB 44	−0.263111572	0.132184922
PCB 95	−0.249304212	−0.025604428
PCB 66	−0.215865398	−0.063642875
PCB 28	−0.200490635	0.302478694
PCB 18	−0.085965593	0.234496099
PCB 31	−0.074297373	0.214990640
PCB 156	−0.018221960	−0.023559163
PCB 105	−0.017929630	−0.202887928
PCB 151	0.008978884	−0.134698813
PCB 209	0.019003407	0.028315064
PCB 206	0.025528966	0.017271277
PCB 195	0.031234064	0.024457992
PCB 194	0.032328440	−0.027942848
PCB 126	0.035287536	−0.299178745
PCB 170	0.042712994	−0.148183076
PCB 180	0.045287309	−0.626611108
PCB 149	0.052592261	0.061598326
PCB 128	0.062649181	0.003196016
PCB 183	0.070039534	−0.050735404
PCB 99	0.076831258	0.060819762
PCB 118	0.140269441	−0.172217258
PCB 187	0.338456138	0.375925319
PCB 138	0.362698627	0.023100243
PCB 153	0.479199806	0.167691821

for all loading scores). Since individual fish were randomly assigned to organic analysis and to genetic analysis groups per site, these results provide confidence that the fish analyzed for gene expression were similarly contaminated with the levels and chemical species observed.

3.3. Pesticide Analysis

Eighteen organochlorine pesticides were resolved in each of the bluefish samples. Overall, pesticide concentrations were higher in the HRE fish than in the MRGB fish (

Table 3. Organochlorine pesticides (ng/g) in YOY bluefish from the Hudson Raritan Estuary (HRE) and from the Mullica River-Great Bay Estuary (MRGB). Samples (n) = 11 per site. Σ = sum. p = probability of significant difference (*) in means by t-test, α < 0.05; ^ value is below method detection limits (MDL).

Location	HRE		MRGB
Analyte	Mean	S.D.	Mean	S.D.	p
HCB	1.721	0.774	0.192 ^	0.083	<0.001 *
β-BHC	0.264	0.131	0.032 ^	0.014	<0.001 *
Lindane	0.164 ^	0.25	0.09 ^	0.131	0.398
Heptachlor	0.511 ^	0.222	0.045 ^	0.021	<0.001 *
Aldrin	0.067 ^	0.069	0.03 ^	0.079	0.259
Heptachlor Epoxide	2.581	1.18	0.182 ^	0.06	<0.001 *
Oxychlordane	2.271	1.183	0.239 ^	0.096	<0.001 *
γ-chlordane	6.712	2.092	0.457 ^	0.208	<0.001 *
α-chlordane	14.113	4.415	1.195 ^	0.525	<0.001 *
τ-nonachlor	14.576	5.528	2.337	1.122	<0.001 *
Endrin	8.459	3.185	0.642 ^	0.365	<0.001 *
Endosulfan II	0.073 ^	0.242	0.204 ^	0.344	0.311
Endosulfan Sulfate	5.113	1.369	0.958 ^	0.294	<0.001 *
o,p,p’-DDE	25.34	10.307	0.552 ^	0.592	<0.001 *
p,p’-DDE	116.252	41.681	21.765	6.641	<0.001 *
o,p’-DDD	9.101	3.429	1.28	0.311	<0.001 *
o,p’-DDT	6.219	1.744	1.223 ^	0.404	<0.001 *
p,p’-DDT	4.053	1.819	0.722 ^	0.708	<0.001 *
Sum (Σ) DDTs	160.965	53.220	25.542	7.894	<0.001 *
Sum (Σ) chlordanes	35.401	11.238	3.989	1.771	<0.001 *

). In particular, the persistent organochlorine pesticide DDT (dichlorodiphenyltrichloroethane) and its breakdown products DDE (Dichlorodiphenyldichloroethylene) and DDD (dichlorodiphenyldichloroethane) were six times the total concentration (sum of DDTs) in HRE compared to MRGB. Other organochlorines, HCB (hexachlorobenzene), β-BHC (β-Hexachlorocyclohexane), heptachlor, heptachlor epoxide, oxychlordane, γ-chlordane, α-chlordane, τ-nonachlor, endrin, and endosulfan sulfate, were significantly elevated in HRE fish over MRGB fish, while lindane, aldrin, and Endosulfan II were not significantly different between sites. Total chlordane concentrations (sum of α-chlordane, γ-chlordane and τ-nonachlor) were approximately nine times higher in HRE fish than MRGB fish. These chemical analyses support the level of anthropogenic impact within each of the contrasting survey areas but are by no means the only contaminants in the environment that could potentially be affecting bluefish growth and development.

3.4. Suppression Subtraction Hybridization (SSH) Libraries

The SSH up-regulated library yielded 139 clones, of which 92 unique sequences were identified by Blast search of the DNA sequences from the library (

Table 4. Identification of up-regulated genes with contamination bluefish (BFu-) expressed sequence tags (ESTs) from suppressive subtraction hybridization (SSH) library. Homologs and species names were identified using BlastX (NCBI) search. The Genbank accession number of closest homolog and the expected value (E value) for each alignment.

EST	Accession	Gene	Homolog	Homolog Accession	E Value
Bfu-			Immune system
21	DY223235	C3	complement C3	XP_008315469.1	3.00 × 10−51
158	DY223311	serping1	C1 (complement) inhibitor	AFO64913.1	8.00 × 10−86
30	DY223239	cbln1	precerebellin-1-like-C1q	XP_033503736.1	3.00 × 10−68
124	DY223292	flnb	tapasin-TAP associated glycoprotein	XP_031139445.1	8.00 × 10−29
147	DY223303	mfap4	microfibril associated glycoprotein 4 like	XP_018548939.1	8.00 × 10−31
			Environmental response
39	DY223245	wap65-1	warm temperature acclimation protein 65-1	CCA29189.1	2.00 × 10−41
70	DY223261	isp2	type-2 ice-structuring protein	XP_032362960.1	4.00 × 10−23
69	DY223260	afp	antifreeze polypeptide precursor	AAA49617.1	3.00 × 10−19
			Metabolism
10	DY223228	txn	thioredoxin	XP_005807280.1	5.00 × 10−19
19	DY223233	cyp1a1	cytochrome P450 1A	AEX63356.1	2.00 × 10−36
32	DY223241	cyp2n	cytochrome P450 CYP2N	ABO21079.1	2.00 × 10−5
87	DY223273	cyb561	cytochrome b ascorbate	XP_028434070.1	4.00 × 10−40
108	DY223283	cyp2a10	cytochrome P450 2A10	XP_030264280.1	2.00 × 10−98
137	DY223298	cyp1a1	cytochrome P450 1A1	XP_029282397.1	7.00 × 10−115
48	DY223251	selenot1a	selenoprotein T1a thioredoxin reductase like	XP_022063960.1	2.00 × 10−64
6	DY223225	naca	nascent polypeptide-assoc. complex subunit α	ADX97194.1	3.00 × 10−63
27	DY223238	pdia4	protein disulfide-isomerase A4	XP_018548628.1	1.00 × 10−33
37	DY223243	minpp1	multiple inositol polyphosphate phosphatase 1	XP_029306261.1	1.00 × 10−36
45	DY223249	pla1a	phospholipase A1 member A	KAE8299747.1	1.00 × 10−36
46	DY223250	lysc	c-type lysozyme	ADZ44620.1	4.00 × 10−37
55	DY223255	gcdh	glutaryl-CoA dehydrogenase	XP_026204710.1	4.00 × 10−27
74	DY223263	aldh	aldehyde dehydrogenase	TNN41590.1	9.00 × 10−63
77	DY223266	smase	sphingomyelin phosphodiesterase	XP_030293881.1	8.00 × 10−48
81	DY223269	smci	carboxypeptidase inhibitor SmCI	XP_020493765.1	9.00 × 10−46
83	DY223270	rpn1	ribophorin1	XP_014043427.1	1.00 × 10−54
84	DY223271	hoga	4-hydroxy-2-oxoglutarate aldolase	XP_020442478.1	3.00 × 10−39
85	DY223272	gapdh	glyceraldehyde-3-phosphate dehydrogenase	XP_028984894.1	7.00 × 10−24
92	DY223277	pdi	protein disulfide-isomerase	TWW59456.1	9.00 × 10−11
93	DY223278	pdi	protein disulfide isomerase precursor (PDI) gene	JX891477.1	7.00 × 10−96
119	DY223290	cth	cystathionine gamma-lyase	XP_022593808.1	3.00 × 10−94
121	DY223291	spp24	secreted phosphoprotein 24	XP_018516447.1	2.00 × 10−38
132	DY223295	adh1	alcohol dehydrogenase 1	XP_018553249.1	7.00 × 10−53
150	DY223305	nmt3	phosphoethanolamine N-methyltransferase 3	XP_023263083.1	3.00 × 10−111
7	DY223226	bhmt	betaine-homocysteine S-methyltransferase	XP_028852308.1	1.00 × 10−11
8	DY223227	limk2	LIM domain kinase 2	XP_024660306.1	2.30 × 10−1
17	DY223231	eno3	beta-enolase	XP_012715925.2	1.00 × 10−35
20	DY223234	shmt	serine hydroxymethyltransferase, cytosolic	XP_026201746.1	7.00 × 10−121
52	DY223252	bhmt1	betaine-homocysteine S-methyltransferase 1	XP_022048002.1	4.00 × 10−53
54	DY223254	gstt1	glutathione S-transferase theta-1	XP_018548814.1	1.00 × 10−40
71	DY223262	dhpr	dihydropteridine reductase	XP_030001857.1	2.00 × 10−110
112	DY223286	gsta	glutathione S-transferase	AAQ91198.1	7.00 × 10−56
152	DY223307	ubp1	beta-ureidopropionase	XP_006797654.1	2.00 × 10−47
154	DY223308	marc1	mitochondrial amidoxime-reducing component 1	XP_022606172.1	3.00 × 10−57
			Signal proteins
4	DY223223	srprb	signal recognition particle receptor subunit beta	XP_010766055.1	1.00 × 10−66
18	DY223232	sec11a	signal peptidase complex catalytic subunit	XP_023153367.1	2.00 × 10−24
44	DY223248	eif2b4	translation initiation factor eIF-2B subunit delta	XP_029966559.1	1.00 × 10−35
90	DY223275	agt	angiotensinogen	XP_022619392.1	6.00 × 10−41
78	DY223267	agt	angiotensinogen	XP_023147340.1	5.00 × 10−117
113	DY223287	dnajc11	dnaJ homolog subfamily C member 11	XP_029363207.1	2.00 × 10−81
136	DY223297	eef1a1	elongation factor 1-alpha	CAX33861.1	2.00 × 10−108
141	DY223300	f1p1	pre-mRNA 3’-end-processing factor FIP1	XP_026216478.1	2.00 × 10−7
144	DY223301	eif3c	eukaryotic translation initiation factor 3 subunit C	XP_018925779.1	1.00 × 10−53
			Transport and Tissue proteins
5	DY223224	cfl2	cofilin-2	XP_015248032.1	1.00 × 10−37
151	DY223306	apoa1	apolipoprotein A-I	XP_032365910.1	3.00 × 10−47
130	DY223294	plg	plasminogen	XP_023285692.1	1.00 × 10−105
109	DY223284	hpx	hemopexin	ASV48204.1	5.00 × 10−63
104	DY223282	gtpb1	GTP-binding nuclear	AEM37693.1	1.00 × 10−23
70	DY223261	habp2	hyaluronan binding protein 2	XP_029925861.1	3.00 × 10−43
42	DY223246	apob	apolipoprotein B-100	TKS75535.1	1.00 × 10−54
1	DY223221	vtn	vitronectin	XP_031136743.1	8.00 × 10−80
2	DY223222	apoa1	apolipoprotein A1	TNN34591.1	4.00 × 10−13
11	DY223229	ftl	ferritin	XP_008291313.1	9.00 × 10−77
22	DY223236	hp	haptoglobin	QBF53745.1	1.00 × 10−21
24	DY223237	vtn	vitronectin	XP_031136743.1	2.00 × 10−78
31	DY223240	fgb	fibrinogen beta chain	XP_023152926.1	2.00 × 10−22
33	DY223242	hba1	hemoglobin subunit alpha	XP_029383058.1	6.00 × 10−62
38	DY223244	kng1	kininogen-1 isoform X1	XP_023284563.1	2.00 × 10−37
57	DY223256	tf	transferrin	CAC19468.1	2.00 × 10−51
13	DY223230	tfa	transferrin-a	XP_008283972.1	6.00 × 10−73
53	DY223253	proc	vitamin K-dependent Protein C	KAE8290483.1	1.00 × 10−52
68	DY223259	fetub	fetuin-b	XP_054646644.1	3.00 × 10−27
76	DY223265	hepl1	ferroxidase	XP_071316308.1	2.00 × 10−61
89	DY223274	krt18	keratin, type I cytoskeletal 18	XP_020473591.1	7.00 × 10−61
91	DY223276	bsg	basigin isoform X1	XP_022615278.1	2.00 × 10−54
94	DY223279	tubb	beta tubulin	AAL24510.1	9.00 × 10−74
139	DY223299	tf	serotransferrin-like	XP_023254217.1	3.00 × 10−67
146	DY223302	serpina1	alpha-1-antitrypsin homolog	XP_028460829.1	5.00 × 10−59
160	DY223312	a2m	alpha-2-macroglobulin	AAR06589.1	6.00 × 10−69
80	DY223268	actb	actin	XP_022594016.1	4.00 × 10−60
			Ribosomal proteins
155	DY223309	rps7	S7 (40S)	AWP20434.1	1.00 × 10−75
110	DY223285	rps3	S3 (40S)	XP_003448999.4	1.00 × 10−112
75	DY223264	rps6	S6 (40S)	AAT01908.1	6.00 × 10−24
59	DY223258	rps2	S2 (40S)	XP_004558416.1	7.00 × 10−58
129	DY223293	mrps28	28S	XR_004329248.1	1.00 × 10−125

). The SSH down-regulated library yielded 138 clones, and 23 unique sequences were identified (

Table 5. Identification of the down-regulated genes with contamination bluefish (BFd-) expressed sequence tags (ESTs) from suppressive subtraction hybridization (SSH) library. Homologs and species name were identified using BlastX, GenBank search. The accession number of closest homologs, length of amino acid query, percent identity (%ID), percent positives (%Pos), and the expected value (E value) for each alignment.

EST	Accession	Gene	Homolog	Homolog Accession	E Value
Bfd-			Immune system
37	DY223202	lect2	leucocyte cell derived chemotaxin-2	KAF3703988.1	1.00 × 10−50
			Environmental response
113	DY223215	isp ls-12	type-4 ice-structuring protein LS-12	XP_022603083.1	6.00 × 10−36
			Metabolism
6	DY223199	eef1a1	elongation factor 1-alpha	KAE8282309.1	2.00 × 10−154
71	DY223207	mt-nd1	NADH dehydrogenase	ACR19960.1	3.00 × 10−47
150	DY223219	mt-atp6	ATP synthase F0 subunit 6	YP_008593748.1	2.00 × 10−44
86	DY223209	mt-co1	cytochrome c oxidase subunit I	YP_008593745.1	5.00 × 10−51
129	DY223217	azin1	predicted antizyme inhibitor 1 (azin1) like	XM_035671848.1	6.00 × 10−18
			Signal proteins
4	DY223198	serpinf1	pigment epithelium-derived factor	XP_026213308.1	3.00 × 10−109
77	DY223208	mid1ip1	mid1-interacting protein 1-B	XP_029302452.1	2.00 × 10−18
44	DY223204	if2m	translation initiation factor IF-2-like	XP_037553016.1	3.00 × 10−30
			Transport and Tissue proteins
46	DY223206	hba1	hemoglobin subunit alpha	XP_008301375.1	7.00 × 10−82
39	DY223203	fgg	putative fibrinogen gamma chain isoform 3	AWP05360.1	2.00 × 10−29
92	DY223210	col1a1	collagen alpha-1(I) chain	RXN15115.1	8.00 × 10−64
118	DY223216	hbb	hemoglobin subunit beta2	APS87122.1	3.00 × 10−95
			Ribosomal proteins
20	DY223200	rpl15	L15 (60S)	XP_008293876.1	2.00 × 10−65
33	DY223201	rps6	S6 (40S)	XP_018551167.1	1.00 × 10−73
45	DY223205	rps4	S4 (40S)	ACM41869.1	5.00 × 10−82
153	DY223220	rps9	S9 (40S)	AER42648.1	3.00 × 10−3
1	DY223197	rpl3	L3 r	AEM37694.1	2.00 × 10−80

). The 115 unique up- and down-regulated sequences have been archived at NCBI under GenBank accession #s DY223197-DY223312. It is not known if the down-regulated library construction was less efficient or indicates that there were in fact less down-regulated genes than up-regulated genes. For both, there appears to be a mix of gene transcripts that are putatively differentially expressed across functional systems of the fish, including the immune system, environmental response, metabolism, signaling, transport and tissue proteins and ribosomal proteins.

3.5. Quantitative Polymerase Chain Reaction (qPCR)

Seven putative up-regulated genes were selected for qPCR analysis. Test genes were selected as relevant targets for hepatic genes from a variety of metabolic systems. Gene-specific forward and reverse primers were designed and tested (

Table 6. Gene-specific primers for quantitative polymerase chain reaction (qPCR) of bluefish up-regulated (bfu) clones isolated by SSH. Sequences are 5′-3′.

Homolog	Bfu Clone	5′ Forward Primer	5′ Reverse Primer
signal recognition particle receptor subunit beta	4	CTC ATT CAG CAG CAG CTG G	ACT CCA CCT TCA TGG GCA
ferritin	11	ATG AGT GGG GCA GTG GTG TC	GCA GGT TGG TCA CCC AGT CT
serotransferrin-like	13	CTA CGC CGT TGC CGT GGT G	GCT CAC CGC CTC CTC CAC TG
cytochrome P450 1A	19	GCT GGA GGA ACA CAT CTG C	CCA GCT CTT GGT CGT TGT G
complement C3	21	GTG GGA CTG GTG GCA GTT G	GGA CTC AAA TAA CAG CCC AGC
cerebellin-1-like	30	GCC CGA GTG GAG AAA CTG G	GTG AAG ACG CCT GTG TCG G
warm temperature acclimation protein	39	GGT GAC CAT CTG TTC AAG GGC	GTG GTC ATG ATC GGA GGG GTT GTC

). Each primer set was specific and produced a single product when tested with the respective plasmid DNA clone. The standard curve method of relative quantification was employed [33] with the HRE, or high-anthropogenic-impact samples, diluted (1×, 0.1×, 0.01×, 0.001×) for the standard curve and compared to MRGB controls or less impacted samples as calibrators (1x). Each of the seven gene-specific primer pairs showed a higher expression level in the high-anthropogenic-impact HRE fish versus the MRGB (Figure 5). Cytochrome P-450 1A was the most highly differentially expressed of the seven clones at nearly 72-fold relative increase. Acute phase response (APR)-related transcripts, complement component C-3, ferritin, transferrin and precerebellin all were 4 to 10 times relative induction of expression with more highly anthropogenically impacted fish versus less impacted controls. Warm temperature acclimation protein (Wap65-1) is increased 21 times and signals recognition particle receptor subunit B is increased 13 times with higher anthropogenic impact relative to less impacted control fish. Each of these gene clones, isolated from the up-regulated with contamination ibrary, supports the directional effectiveness of the SSH differential selection but, like the SSH library, are representative of only one biological replicate, so no statistical confidence can be attached to the magnitude.

4. Discussion

Degraded habitat, as evidenced by the elevated legacy contaminant levels in fish from the HRE are likely impacting the overall fitness of YOY bluefish using that habitat as a first-year nursery. Juvenile bluefish (YOY) from the HRE were significantly smaller and lighter, and had a lower Fulton’s condition index and a higher body burden of total PCBs and pesticides than the same spawning cohort found in the MRGB. An earlier laboratory study found that bluefish fed a diet of common prey (menhaden, mummichog, killifish) from contaminated habitats in the HRE for four months showed bioaccumulation of contaminants (PCBs and pesticides) and displayed significantly reduced feeding behaviors, swimming activity, and growth compared to controls fed with prey from uncontaminated sites in the MRGB [34]. Disruption of behaviors such as feeding, predator avoidance, and reproduction with increased contamination has been documented in a number of taxa [35,36] and while certain taxa may ameliorate or avoid effects of contamination through behavior, feeding modifications, and accelerated depuration, the higher trophic levels such as predatory fishes with more advanced neuroendocrine systems may be more likely affected in growth and ultimately survivorship [37].

As confirmed in this study, HRE environments have been extensively documented to be highly contaminated with PCBs and pesticides, as well as heavy metals and PAH [38]. Many of these substances have been shown to bio-magnify up the food web [37], and there have been significant fishery restrictions in the HRE system due to concern over risks to human health. In addition to total PCBs, the total congener-normalized fingerprint of PCBs was distinctive at each site, as confirmed by principal component analysis in this study and as seen by Smalling et al. [11] in Sandy Hook Bay (HRE) and Great Bay (MRGB). Chemical fingerprints are a well-established method of relating specific contamination of sites with fish body burden such as with PAH [39]. As previously noted by Deshpande et al. [9], specific PCB congener fingerprints indicate site-specific fidelity of YOY bluefish to specific estuaries where they feed and rapidly grow through their initial year(s) before undertaking adult migrations throughout the range.

Exposure to persistent organochlorinated pesticides, including many of those surveyed in this study, have also been shown to affect growth and reproductive success. These pesticides accumulate in tissues with high lipid content and can potentially affect many cellular processes, including disruption of hormones, enzymes, growth factors and neurotransmitters, and induce many metabolic pathways leading to dysregulation, contributing to the development of tumor growth [40,41]. Chronic exposure to DDT and its metabolites has been shown to disrupt estrogen activity in reproductive tissue, leading to decreased reproductive success, induce microsomal liver detoxification genes for cytochrome P450s, disrupt lipid and sugar metabolism, and promote neoplasia/carcinogenesis in some tissues [42,43,44]. Exposure of YOY bluefish to these contaminants may have significant effects on the success of year-class growth and recruitment into the fishery.

Investigation of gene expression in the liver of bluefish was undertaken in this study as the liver serves an important role in vertebrates as a site of chemical detoxification as well as a site of synthesis and degradation for many important cell constituents, including carbohydrates, fats, proteins and hormones. These and other key processes of the liver have been shown to be affected by chemical contamination [17,45]. Liver hepatocytes synthesize plasma proteins for transport, as well as oxidize triglycerides, synthesize amino acids, deaminate and transaminate amino acids, and remove toxic ammonia from protein metabolism and bilirubin from hemoglobin breakdown. The Phase I oxidation systems of the liver, including the cytochrome P450 superfamily (CYP), are induced by a wide range of alkyl and aromatic hydrocarbons, including co-planar PCBs (such as PCB 126) and DDT, either through the Aryl hydrocarbon receptor-2 (AHR2) or other receptors [46], and along with Phase II enzymes (glutathione transferase, etc.), which have served as enzymatic biomarkers of chemical contamination. As part of the detoxification role of the liver, Phase-I enzymes oxidize chemical substrates to increase their polarity. In some cases, this increases the toxicity of the chemical and may lead to increased damage of the liver [47].

In this study, metabolic protein transcripts observed that were putatively up-regulated with anthropogenic impact included the enzyme systems associated with Phase I and Phase II liver detoxification pathways. This finding was expected as Phase I involves oxidative enzyme metabolism of xenobiotics including PCBs, PAH, aromatic amines, heterocyclic amines, pesticides, herbicides, and most compounds used for drugs [48]. Although these other xenobiotic categories were not quantified in this study, the HRE is undoubtedly contaminated with many of these compounds as well as heavy metals [5]. Phase I transcripts for cyp1, cyp1A1, cyp2N, cyp2A10 and mitochondrial amidoxime-reducing component-1 (marc1) were observed in those putatively up-regulated with the anthropogenic impact library in this study. Phase II metabolism involves enzymatic conjugation to the reactive site of phase I metabolites, generally resulting in a more water-soluble compound that can be further metabolized or eliminated. Phase II-related, putative up-regulatedtranscripts with anthropogenic impact observed in this study include glutathione-S-transferases (gst) and thioredoxin (trx). Up-regulation of these phases I and II transcripts in the HRE are consistent with the increased impact of anthropogenic contaminants. Other important liver detoxification enzymes seen to be up-regulated were alcohol dehydrogenase (adh) and aldehyde dehydrogenase (aldh). Additional metabolic-related transcripts observed to be putatively up-regulated in the more anthropogenically impacted environment may be involved in pathways related to the energy metabolism of lipids, carbohydrates and proteins involved in the condition of the fish related to feeding depression documented with contamination [34]. Potentially related to fish condition as well, genes observed to be down-regulated in the more impacted environment included metabolic genes involved with general oxidative phosphorylation/energy production such as ATP synthase F0 subunit 6 (mt-atp6), NADH dehydrogenase (mt-nd1) and cytochrome c oxidase subunit 1 (mt-co1). In a transcriptome study of the Atlantic Killifish (Fundulus heteroclitus) populations from several superfund and reference sites, Oleksiak [49] found that Newark Bay, NJ in the HRE had more differentially expressed genes than other superfund sites, and that a large number of these were related to metabolism and “suggests that pollution may have a significant effect on energy metabolism in these fish” [49].

The liver is also the source of many circulating immune-related proteins. Immune-related proteins with up-regulated transcripts with anthropogenic impact that were observed included complement component-3 (C3), precerebellin-1-like C1q (cbln1), and microfibril associated glycoprotein (mfap4), which are associated with the innate immune complement system and acute phase response (APR) to inflammation and infection. Cbln1 has been shown to be part of the APR in rainbow trout challenged with bacterial infection [50]. Tapasin associated glycoprotein (flnb) was also observed as up-regulated and is part of the antibody-acquired immune system acting to mediate antigen processing and display [51]. A down-regulated immune-related transcript for leucocyte cell-derived chemotaxin-2 (lect2) was observed. Lect2 is related to cytokine-induced immune activity, including phagocytosis and bacterial activity of white blood cells [52].

Environmental response protein transcripts observed to be up-regulated in the liver with anthropogenic impact were warm temperature acclimation protein 65-1 (wap65-1), type 2-ice structuring protein (isp2) and antifreeze polypeptide precursor (afp). Wap65-1 is a hemopexin-like protein that may be involved in scavenging heme and protection under warm conditions or bacterial infection [53,54]. The function of expression of isp2 and afp, both of which contribute to low temperature survival in the blood of fish, is not known to be related to contamination but may play some protective role. A down-regulated, with contamination, transcript for type-4-ice structuring protein (isp-ls-12) was also observed.

Ribosomal proteins (Rp) levels have generally been considered to be relatively constant but recent work has shown that differential expression of Rps is correlated to the proliferative state of cells and may indicate underlying pathologies linked to nucleolar stress responses, such as P53 activation, and cancer risk [55,56]. Five up-regulated rps transcripts, S2, S3, S6, S7, and 28S, and five down-regulated rps, L3, L15, S4, S6 and S9, were observed.

Transport and tissue protein transcripts that were up-regulated in the liver in the more anthropogenically impacted environment included a number related to iron and heme metabolism and transport such as ferritin (ft1), transferrin (tfr1), hemoglobin A (hba), haptoglobin (hpt), hephaestin-like protein (hpl1) and hemopexin (hpx). Also present in the up-regulated library were the glycoproteins, vitronectin (vtnc), apolipoprotein-A1 (apoa1), and alpha-2-macroglobulin (a2mg); the serine proteases, alpha-1-antitrypsin (serpina1), and fetuin-b (fetub); and coagulation factors, plasminogen (plg), fibrinogen (fgb), kininogen (kng1), and vitamin K-dependent Protein C (proc). Chronic inflammation of the liver and APR to infection have been shown to affect iron homeostasis and a large number of APR-related liver proteins [57]. Transport/tissue proteins in the down-regulated library included fibrinogen gamma chain (fgg), hemoglobin subunit beta2 (hbb), and hemoglobin subunit alpha (hba) which is also identified in the up-regulated library but is a different unique sequence and may be an alternative splice variant.

Signal proteins observed in the liver to be up-regulated with anthropogenic impact libraries included angiotensinogen (agt), which is produced in the liver in response to low blood pressure. When agt is activated by the enzymes Renin and Angiotensin converting enzyme (ACE), the product, Angiotensin, causes constriction of smooth muscle and increased blood pressure. Potential mis-regulation of angiotensinogen could contribute to stress-related hypertension, cardiac issues and issues with ionic and osmotic balance [58]. Signal recognition particle receptor subunit beta (srpb) is a ribonucleoprotein GTPase that binds signal recognition particle (SRP) to target nascent secretory particles to the Endoplasmic reticulum, and signal peptidase complex catalytic subunit (sec11A) removes the signal peptide upon translocation into the lumen of the ER. Up-regulation of these proteins with anthropogenic impact may signal increased secretory protein metabolism related to inflammation. Other putative up-regulated-with-anthropogenic-impact transcripts involved protein synthesis, processing and targeting, including translation initiation factor (eif2b4), elongation factor-1-alpha (eef1a1), eukaryotic translation initiation factor-3 (eif3c), pre-mRNA 3’-end processing factor (f1p1), and dnaJ homolog (dnajc11). Putative down-regulated transcripts involved in signaling included pigment epithelium-derived factor (serpinf1), mid-1-interacting protein 1-B (mid1ip1) and translation initiation factor IF-2 like (if2m). Serpinf1 is a serine protease inhibitor involved in inhibiting angiogenesis through inhibition of VEGF (Vascular epidermal growth factor) and is related to wound healing as well as to the onset of age-related macular degeneration.

Experimental exposure to individual and mixed POPs such as PCBs has been shown to differentially affect gene expression in a number of fish taxa. Agrawal et al. [59] exposed zebrafish and medaka to PCB 126 (a co-planar PCB) [59]. One of the nine candidate biomarkers of PCB 126 exposure observed was cyp1a, which was also isolated in bluefish as an EST in this study. Exposure of Atlantic cod (Gadus morhua) to PCB 153 (a non-co-planar PCB) had effects on transcripts for lipid metabolism, cell cycle, tissue remodeling and wound repair, immune response, stress response, apoptosis and various signaling pathways of the liver transcriptome by microarray [60]. Breese et al. also showed significant effects on the transcriptome with single and PCB mixtures and found that PCB metabolites (produced in part by liver detoxification pathways), in many cases, had a larger effect on developing zebrafish than parent compounds [61]. In a transcriptomic study of hornyhead turbot (Pleuronichthys verticalis) dosed with PCBs and PBDE, transcripts related to immune response, lipid metabolism, xenobiotics and endocrine disruption were observed relating to each of the chemical species [62]. To approach realistic contaminant-loading in the wild, Vogs et al. [63] modeled Baltic Sea salmon exposure utilizing chemical profiles from salmon serum (including nine organohalogen mixtures of PCB, PFAS and DDT metabolites) in a zebrafish embryo RNA sequencing transcriptomic study. Thirteen transcripts isolated in this study correspond to homologs co-expressed under the various treatments in the Vogs study [63], including: cyp1a1, gsto1, gstp1, hpx, aldh1, cyp2, apob, dnajb1, gstp1, serpina1, lect2, pdia, and tubb4a.

In general, interruption of metabolic function by interference of xenobiotics can lead to effects on growth and development that can limit the success of larval fishes [64]. Reid and Whitehead [65] reviewed the applications of functional genomics as biosensors in the assessment of biological responses to marine pollution in fishes, concluding that modern genomics-based tools such as transcriptomics and proteomics offer high-throughput and information-rich methods for the characterization of molecular mechanisms by which pollutants exert toxic effects and for the estimation of pollutant impacts in the short and long term. Genomics data can also offer insight into evolutionary adaptations in polluted environments that allows the potential for population survival. For example, Wirgin et al. [66] reported that the resistance of Microgadus tomcod (tomcod) populations in the HRE to high PCB levels was related to variation in the AHR2, pointing to rapid evolutionary change due to selection at a single locus in contaminated environments. There is also extensive evidence for evolutionary adaptation in the estuarine resident Fundulus heteroclitus, which is linked with repeated exposure, high nucleotide diversity, functional divergence of gene expression and in the case of PCB and PAH exposure with changes in the sensitivity of the AHR2 signaling pathways [67].

This functional genomic study observed preliminary effects on gene expression, as determined by SSH enrichment, in the liver of bluefish from contrasting habitats with increased anthropogenic impact. SSH is useful for enriching transcripts from non-model organisms where no reference genome is available but lacks the statistical power of RNA-sequencing. Availability of these EST transcripts can also contribute to the development of environmental RNA-Sequencing and Next Gen sequencing technologies [68] for interrogating the bluefish transcriptome in further studies. Recent studies of fish populations related to pollutants [59] indicate the increased power of these RNA sequencing methods to interrogate the transcriptome over older comparative methodology such as SSH. These EST transcripts are related to important pathways of energy, transport, detoxification, immune status, stress and environmental response and signaling. While up- or down-regulation of the expression of some of these genes may be adaptive responses to habitat quality, it is likely that many may involve negative consequences to normal development, feeding, growth and survival mechanisms necessary to ensure successful recruitment into the reproducing adult population. While long-term trends in the HRE indicate that legacy contamination inputs have decreased over the past 50 years [38,69], the refractory nature and persistence of these chemicals as well as emerging classes of pollutants such as PBDE, PFAS, PPCP and microplastics may continue to exert their effects at very low concentrations by affecting physiological and developmental processes at the genomic level.

5. Conclusions

This study observed a differential effect on growth and condition and isolated a preliminary group of expressed sequence tags (EST) transcripts related to habitat quality in the liver of YOY bluefish from contrasting habitats with increased anthropogenic impact. Habitat degradation, documented in part with elevated contaminant levels measured in the HRE fish, is likely impacting the overall fitness of YOY bluefish using that habitat as a first-year nursery. Juvenile bluefish from the HRE were significantly smaller and lighter, and had a poorer condition index and a higher body burden of total PCBs and pesticides than the same spawning cohort found in the MRGB. A suite of EST transcripts for liver genes related to detoxification, metabolism, transport, signaling, immune and environmental response were putatively elevated, or depressed, in the HRE compared to the MRGB. These preliminary transcript observations can be used to investigate the hypothesis that habitat contamination is contributing to observed depression of growth and development of juvenile bluefish. As such, the putative genomic biomarkers observed may contribute to further study leading to an understanding of the cumulative effects of sub-lethal exposure to a suite of anthropogenic contaminants found in industrialized estuarine systems. This study also contributes an important time point, the early 2000s, which documents transcripts and levels of many critical POPs in bluefish that have subsequently decreased over time in the HRE due to a system-wide contaminant reduction program (CARP) that has prioritized removal of contaminated sediments through dredging and superfund site designations and cleanups [69].