Taxon-specific Models Research Articles

The rise of taxon-specific epitope predictors.

Computational predictors of immunogenic peptides, or epitopes, are traditionally built based on data from a broad range of pathogens without consideration for taxonomic information. While this approach may be reasonable if one aims to develop one-size-fits-all models, it may be counterproductive if the proteins for which the model is expected to generalize are known to come from a specific subset of phylogenetically related pathogens. There is mounting evidence that, for these cases, taxon-specific models can outperform generalist ones, even when trained with substantially smaller amounts of data. In this comment, we provide some perspective on the current state of taxon-specific modelling for the prediction of linear B-cell epitopes, and the challenges faced when building and deploying these predictors.

Organism-specific training improves performance of linear B-cell epitope prediction.

Motivation In silico identification of linear B-cell epitopes represents an important step in the development of diagnostic tests and vaccine candidates, by providing potential high-probability targets for experimental investigation. Current predictive tools were developed under a generalist approach, training models with heterogeneous datasets to develop predictors that can be deployed for a wide variety of pathogens. However, continuous advances in processing power and the increasing amount of epitope data for a broad range of pathogens indicate that training organism or taxon-specific models may become a feasible alternative, with unexplored potential gains in predictive performance.ResultsThis article shows how organism-specific training of epitope prediction models can yield substantial performance gains across several quality metrics when compared to models trained with heterogeneous and hybrid data, and with a variety of widely used predictors from the literature. These results suggest a promising alternative for the development of custom-tailored predictive models with high predictive power, which can be easily implemented and deployed for the investigation of specific pathogens.Availability and implementationThe data underlying this article, as well as the full reproducibility scripts, are available at https://github.com/fcampelo/OrgSpec-paper. The R package that implements the organism-specific pipeline functions is available at https://github.com/fcampelo/epitopes.Supplementary information Supplementary materials are available at Bioinformatics online.

Derivation of acute copper biotic ligand model-based predicted no-effect concentrations and acute-chronic ratio.

The copper biotic ligand model (BLM) can quantitatively describe the bioavailability depending on various environmental factors and has been used to derive the predicted no-effect concentrations (PNECs). The commonly employed acute BLM tool, HydroQual, which applies the biotic ligand constants of fathead minnow in the same model structure for all taxonomic groups, estimates lower acute copper toxicity values compared to the chronic copper PNECs of the European Union Risk Assessment Reports (EU-RAR), which are based on taxon-specific model structures and biotic ligand constants for vertebrates, invertebrates, and algae. In this study, the full-BLM approach was applied using an appropriate acute BLM for each taxonomic group to derive acute HC5s (fifth percentile value in the species sensitivity distribution [SSD]) and an acute-chronic ratio for copper. Two acute BLMs for vertebrates and invertebrates were used and validated against site waters using the new method to estimate the intrinsic sensitivity for each species across different environmental conditions. To derive acute copper full BLM-based HC5s in Korean freshwater, acute toxicity tests were performed with 10 indigenous species, which were used to build the acute BLM-based SSD at each site. The final estimated acute full-BLM HC5s were higher than the EU-RAR chronic PNECs within the BLM calibration range. Furthermore, a linear relationship was observed between the acute full-BLM HC5s and the EU-RAR chronic PNECs. This linear regression function was suggested as an acute to chronic transformation function that can be applied to calculate chronic PNEC values. In conclusion, if the chronic ecotoxicity database of indigenous aquatic organisms for copper is lacking, it may be more efficient to derive chronic PNECs using an acute-chronic ratio after deriving BLM-based acute copper SSDs for indigenous species within representative taxonomic groups. This study provides a scientific foundation for the derivation of water quality criteria for copper in freshwater.

Taxonomic resolution affects host-parasite association model performance.

Identifying the factors that structure host–parasite interactions is fundamental to understand the drivers of species distributions and to predict novel cross-species transmission events. More phylogenetically related host species tend to have more similar parasite associations, but parasite specificity may vary as a function of transmission mode, parasite taxonomy or life history. Accordingly, analyses that attempt to infer host−parasite associations using combined data on different parasite groups may perform quite differently relative to analyses on each parasite subset. In essence, are more data always better when predicting host−parasite associations, or does parasite taxonomic resolution matter? Here, we explore how taxonomic resolution affects predictive models of host−parasite associations using the London Natural History Museum's database of host–helminth interactions. Using boosted regression trees, we demonstrate that taxon-specific models (i.e. of Acanthocephalans, Nematodes and Platyhelminthes) consistently outperform full models in predicting mammal-helminth associations. At finer spatial resolutions, full and taxon-specific model performance does not vary, suggesting tradeoffs between phylogenetic and spatial scales of analysis. Although all models identify similar host and parasite covariates as important to such patterns, our results emphasize the importance of phylogenetic scale in the study of host–parasite interactions and suggest that using taxonomic subsets of data may improve predictions of parasite distributions and cross-species transmission. Predictive models of host–pathogen interactions should thus attempt to encompass the spatial resolution and phylogenetic scale desired for inference and prediction and potentially use model averaging or ensemble models to combine predictions from separately trained models.

Genome-wide associations of human gut microbiome variation and implications for causal inference analyses.

Recent population-based1–4 and clinical studies5 have identified a range of factors associated with human gut microbiome variation. Murine quantitative trait loci6, human twin studies7 and microbiome genome-wide association studies (mGWAS)1,3,8–12 have provided evidence for genetic contributions to microbiome composition. Despite this, there is still poor overlap in genetic association across human studies. Using appropriate taxon-specific models along with support from independent cohorts, we show association between human host genotype and gut microbiome variation. We also suggest that interpretation of applied analyses using genetic associations is complicated by the likely overlap between genetic contributions and heritable components of host environment. Using fecal derived 16S rRNA gene sequences and host genotype data from the Flemish Gut Flora Project (FGFP, n=2223) and two German cohorts (FoCus, n=950, PopGen n=717), we identify genetic associations involving multiple microbial traits (MTs). Two of these associations achieved a study-level p-value threshold of 1.57×10−10; an association between Ruminococcus and rs150018970 near RAPGEF1 on chromosome 9, and between Coprococcus and rs561177583 within LINC01787 on chromosome 1. Exploratory analysis was undertaken using 11 other genome-wide associations with strong evidence for association (p-value < 2.5×10−08) and a previously reported signal of association between rs4988235 (MCM6/LCT) and Bifidobacterium. Across these 14 SNPs there was evidence of signal overlap with other GWAS including those for age at menarche and cardiometabolic traits. Mendelian randomization (MR) analysis was able to estimate associations between MTs and disease (including Bifidobacterium and body composition), however in the absence of clear microbiome driven effects, caution is needed in interpretation. Overall, this work marks a growing catalog of genetic associations which will provide insight into the contribution of host genotype to gut microbiome. Despite this, the uncertain origin of association signals will likely complicate future work looking to dissect function or use associations for causal inference analysis.

Time to branch out? Application of hierarchical survival models in plant phenology

The sensitivity of phenology to environmental drivers can vary across geography and species. As such, models developed to predict phenology are typically site- or taxon-specific. Generation of site- and taxon-specific models is limited by the intensive in-situ phenological monitoring effort required to generate sufficient data to parameterize each model. Where in-situ phenological observations exist, the data are often subject to analytical issues due to the limited duration of any individual monitoring program, spotty site- and species- level coverage, lack of standardized methodology, and infrequent or variable census intervals. Together, these characteristics constrain our ability to make phenological inferences outside of select sites and taxa where long-duration, intensive monitoring has occurred. In this study, we leveraged two national, standardized phenology datasets to develop a multi-species and multi-site state-space survival model of the onset of deciduous tree and shrub spring (leaf out) and fall (leaf-color) events across temperate ecoregions of the United States. We used data from two national-scale phenological databases, a 9-year, broadly distributed dataset from the USA National Phenology Network and a 4-year dataset from the National Ecological Observatory Network, to quantify regional and interspecific variation in sensitivity to environmental drivers for both spring and fall leaf phenophases. Spring leaf out was generally promoted by longer days, spring growing degree day accumulation, overwinter chilling, and was suppressed by frost events, whereas fall leaf color was promoted by shorter days and cold accumulation. The sensitivity to most environmental drivers tended to be more variable among species than among the regions as defined here (EPA ecoregions of North America, excluding desert and tropical areas). The results of this study lay the groundwork for incorporating the growing collection of phenological observations into a generalized framework for predicting the transition states for any species, in any location.

Use of taxon-specific models of phytoplankton assemblage composition and biomass for detecting impact

Phytoplankton is a functionally important taxonomic group commonly used in the assessment of lakes. Using data from minimally disturbed lakes (n = 151) and relatively easily obtained environmental variables, we modeled the probability of taxon occurrence and taxon-specific biomass (mg L−1) of 183 phytoplankton taxa using variables characterizing spatial context, ecosystem size and land cover. Data from lakes putatively affected by one or more pressures (n = 162) were used to evaluate responses to anthropogenic stress. Longitude and latitude were the two best predictors of phytoplankton biomass, followed by the percent of catchment area classified as coniferous forest area and altitude. Model accuracy was highest for relatively predominant taxa, and taxa with relatively wide distributions did not improve model accuracy. Assemblage similarity (1-BC index) and the phytoplankton trophic index (PTI) calculated using predicted and observed phytoplankton assemblages were used to evaluate model performance. Significant changes in phytoplankton assemblages were associated with agricultural and forestry pressures. Encouragingly the PTI index responded as predicted to agricultural land use but not forestry, implying that PTI was a robust indicator of agricultural effects and that forestry effects on assemblage composition were not related to increased nutrients. Combined, our findings showed that phytoplankton assemblages can be reliably predicted using a few, often readily available, environmental variables, and that predictions of assemblage composition and taxon-specific biomass and biological metrics based on phytoplankton taxon-specific biomass are reliable indicators of anthropogenic pressures, supporting their use in lake management.

Predicting mayfly recovery in acid mine-impaired streams using logistic regression models of in-stream habitat and water chemistry.

Mayflies (Order Ephemeroptera) require high quality water and habitat in streams to thrive, so their appearance after restoration is an indicator of ecological recovery. To better understand the importance of restoring in-stream habitat versus water chemistry for macroinvertebrate communities, we developed taxon-specific models of occurrence for five mayfly genera (Caenis, Isonychia, Stenonema, Stenacron, and Baetis) inhabiting streams in the Appalachian Mountains, USA. Presence/absence records from past decades were used to develop single and multiple logistic predictive models based on catchment characteristics (drainage area, gradient), in-stream habitat variables (e.g., substrate, channel morphology, pool and riffle quality), and water chemistry. Model performance was evaluated using (a) classification rates and Hosmer-Lemeshow values for test sets of data withheld from the original model-building dataset and (b) a field comparison of predicted versus observed mayfly occurrences at 53 sites in acid mine drainage-impaired watersheds in 2012. The classification accuracies of final models for Caenis, Stenacron, and Baetis ranged from 50 to 75%. In-stream habitat features were not significant predictor variables for these three taxa, only water chemistry. Models for Isonychia and Stenonema had higher classification rates (81%) and included both habitat and chemical variables. However, actual occurrences of Isonychia and Stenonema at study sites in 2012 were low, consistent with the calculated probability of occurrence (Po) < 0.60. Caenis occurred at test sites 35% of the time when the model predicted a Po > 0.40. Stenacron showed the greatest consistency of actual versus predicted occurrences, occurring at 56% of sites when the Po (based on pH and conductivity) was > 0.50 and only at 1 site when Po < 0.5. The results demonstrate how predictive models of individual indicator taxa could be valuable for evaluating the relative impacts of restoring physical habitat versus water chemistry during stream remediation.

Prediction of Pesticide Toxicity in Midwest Streams.

The occurrence of pesticide mixtures is common in stream waters of the United States, and the impact of multiple compounds on aquatic organisms is not well understood. Watershed Regressions for Pesticides (WARP) models were developed to predict Pesticide Toxicity Index (PTI) values in unmonitored streams in the Midwest and are referred to as WARP-PTI models. The PTI is a tool for assessing the relative toxicity of pesticide mixtures to fish, benthic invertebrates, and cladocera in stream water. One hundred stream sites in the Midwest were sampled weekly in May through August 2013, and the highest calculated PTI for each site was used as the WARP-PTI model response variable. Watershed characteristics that represent pesticide sources and transport were used as the WARP-PTI model explanatory variables. Three WARP-PTI models-fish, benthic invertebrates, and cladocera-were developed that include watershed characteristics describing toxicity-weighted agricultural use intensity, land use, agricultural management practices, soil properties, precipitation, and hydrologic properties. The models explained between 41 and 48% of the variability in the measured PTI values. WARP-PTI model evaluation with independent data showed reasonable performance with no clear bias. The models were applied to streams in the Midwest to demonstrate extrapolation for a regional assessment to indicate vulnerable streams and to guide more intensive monitoring.

Macrobenthic–mud relations strengthen the foundation for benthic index development: A case study from shallow, temperate New Zealand estuaries

Globally the input of sediment to coastal systems, in particular to estuaries, is predicted to increase due to anthropogenic activities. Sediment mud content is a powerful driver of ecologically important, macrobenthic taxa in estuarine intertidal flats. Accordingly, forecasting species responses to increased sedimentation is fundamental for effective ecosystem management, particularly for productive, geologically young, and sand-dominated estuaries that characterise many countries, including New Zealand (NZ). Modelling studies have highlighted the non-linear, highly variable responses of taxa to mud concentration. However, existing taxon-specific models have not adequately accounted for the full mud gradient, the influence of potentially confounding variables (e.g. organic enrichment, heavy metal concentrations), or regional differences in species responses. Furthermore, such models are often based on qualitative expert consensus of the membership of taxa in ecological groups that characterise their sensitivity to mud content. In this study, data from 25 unmodified to highly disturbed, shallow NZ estuaries, were used to develop an ecologically relevant model to relate the responses of 39 taxa to sediment mud content for use in the intertidal flats of shallow, temperate estuaries. Preliminary analyses indicated that sediment mud content was the dominant driver of macroinvertebrate community composition among sites, total organic carbon was of secondary importance and heavy metals did not explain significant variation in composition. Regression analysis revealed a significant linear relationship between sediment mud and total organic carbon content, which permitted subsequent analyses to be based on mud content alone. Generalised additive models were used to develop taxon-specific models that, according to k-fold cross validation, accurately predicted both probability of presence (up to 79% deviance explained) and maximum density (up to 96% deviance explained) along the sediment mud gradient (0.1–92.3%). Estimates of “optimal mud range” and “distribution mud range” were quantitatively-derived for each taxon and used to categorise taxa into one of five ecological groups (identical to those used in existing biotic indices), based on their individual sensitivities to increasing mud content. By removing expert consensus from the grouping process, the classification methods established herein provide strong support for the use of quantitative indices for the assessment and management of estuarine condition in response to increasing sediment mud content. The findings indicate that NZ estuarine sediments (2–25% mud) support a more diverse and abundant macroinvertebrate assemblage and exhibit low organic enrichment (<1% total organic carbon) compared to estuaries with >25% mud content.

Modeling patterns of coral bleaching at a remote Central Pacific atoll

A mild bleaching event (9.2% prevalence) at Palmyra Atoll occurred in response to the 2009 ENSO, when mean water temperature reached 29.8–30.1°C. Prevalence among both abundant and sparse taxa varied with no clear pattern in susceptibility relating to coral morphology. Seven taxon-specific models showed that turbidity exacerbated while prior exposure to higher background temperatures alleviated bleaching, with these predictors explaining an average 16.3% and 11.5% variation in prevalence patterns, respectively. Positive associations occurred between bleaching prevalence and both immediate temperature during the bleaching event (average 8.4% variation explained) and increased sand cover (average 3.7%). Despite these associations, mean unexplained variation in prevalence equalled 59%. Lower bleaching prevalence in areas experiencing higher background temperatures suggests acclimation to temperature stress among several coral genera, while WWII modifications may still be impacting the reefs via shoreline sediment re-distribution and increased turbidity, exacerbating coral bleaching susceptibility during periods of high temperature stress.

Comparing environmental influences on coral bleaching across and within species using clustered binomial regression

Differential susceptibility among reef-building coral species can lead to community shifts and loss of diversity as a result of temperature-induced mass bleaching events. We evaluate environmental influences on coral colony bleaching over an 8-year period in the Florida Keys, USA. Clustered binomial regression is used to develop models incorporating taxon-specific responses to the environment in order to identify conditions and species for which bleaching is likely to be severe. By building three separate models incorporating environment, community composition, and taxon-specific responses to environment, we show observed prevalence of bleaching reflects an interaction between community composition and local environmental conditions. Environmental variables, including elevated sea temperature, solar radiation, and reef depth, explained 90% and 78% of variability in colony bleaching across space and time, respectively. The effects of environmental variables were only partially explained (33% of variability) by corresponding differences in community composition. Taxon-specific models indicated individual coral species responded differently to local environmental conditions and had different sensitivities to temperature-induced bleaching. For many coral species, but not all, bleaching was exacerbated by high solar radiation. A 25% reduction in the probability of bleaching in shallow locations for one species may reflect an ability to acclimatize to local conditions. Overall, model results indicate predictions of coral bleaching require knowledge of not just the environmental conditions or community composition, but the responses of individual species to the environment. Model development provides a useful tool for coral reef management by quantifying the influence of the local environment on individual species bleaching sensitivities, identifying susceptible species, and predicting the likelihood of mass bleaching events with changing environmental conditions.

The predictability of extinction: biological and external correlates of decline in mammals

Extinction risk varies among species, and comparative analyses can help clarify the causes of this variation. Here we present a phylogenetic comparative analysis of species-level extinction risk across nearly the whole of the class Mammalia. Our aims were to examine systematically the degree to which general predictors of extinction risk can be identified, and to investigate the relative importance of different types of predictors (life history, ecological, human impact and environmental) in determining extinction risk. A single global model explained 27.3% of variation in mammal extinction risk, but explanatory power was lower for region-specific models (median R2=0.248) and usually higher for taxon-specific models (median R2=0.383). Geographical range size, human population density and latitude were the most consistently significant predictors of extinction risk, but otherwise there was little evidence for general, prescriptive indicators of high extinction risk across mammals. Our results therefore support the view that comparative models of relatively narrow taxonomic scope are likely to be the most precise.

Time Constraints Limit Group Sizes and Distribution in Red and Black-and-White Colobus

Researchers have shown that, in frugivorous primates, a major constraint on group size is intra group feeding competition. The relationship is less obvious in folivorous primates. We investigated whether colobine group sizes are constrained by time limitations as a result of their low energy diet and ruminant-like digestive system. We used climate as an easy to obtain proxy for the productivity of a habitat. Using the relationships between climate, group size, and time budget components for populations of Colobus and Piliocolobus at different research sites, we created 2 taxon-specific models. In both genera, feeding time increased with group size or biomass. The models for Colobus and Piliocolobus correctly predicted the presence or absence of the genera at, respectively, 86% of 148 and 84% of 156 African primate sites. Median predicted group sizes where the respective genera were present are 19 for Colobus and 53 for Piliocolobus. We show that the differences between the 2 genera are due mainly to intrinsic differences in the way each taxon’s digestive physiology interacts with climatic variables to influence resting time requirements. The models may help us explore their responses to climatic change in both the past and the future.

Evaluation of SeaWiFS chlorophyll algorithms in the Southwestern Atlantic and Southern Oceans

Bio-optical measurements of spectral upwelling radiance and surface chlorophyll- a concentration have been conducted during 15 cruises between 1995 and 2004. The bio-optical data were divided into two sub-sets: the Southwestern Atlantic Ocean (SwAO), comprising a variety of biogeochemical provinces, from the oligotrophic waters in the South Atlantic gyre to the coastal waters influenced by La Plata River and Patos Lagoon discharge, and the Southern Ocean (SO) data set, comprising sampling stations south of the mean position of the Polar Front, with most stations being located in the vicinity of the Antarctic Peninsula. We derived regional chlorophyll algorithms for both regions and comparisons were made with the NASA's OC4v4 (operational algorithm) and OC2v4. For the Southwestern Atlantic region, the NASA OC4v4 algorithm presented a reasonable performance ( r 2=0.87, rmse-L=0.475, N=136) as compared to the revised algorithm for SwAO data ( r 2=0.89, rmse-L=0.426, N=136). A few stations under strong river plume influence were not considered in the analyses. These were detected by a higher reflectance at 670 nm, at low in situ chlorophyll concentration (<2 mg m −3). These results show that empirical algorithms applied to in-situ radiance data have a limited ability to extract accurate chlorophyll estimates below a 30% uncertainty level. For Southern Ocean stations, a 2-band linear-type model was generated ( r 2=0.64, rmse-L=0.347, N=77), which significantly improved the bias (6.4%) as compared to NASA's OC4v4 algorithm (bias=−21.7%). An evaluation of some published high-latitude algorithms on our data set has shown a better performance by taxon-specific models, even from distant regions. A validation experiment of the normalized spectral water-leaving radiances and chlorophyll- a SeaWiFS products was also conducted using the FURG-SwAO/SO data set, through a match-up exercise. Despite the relatively low number of pairs of radiometric measurements, SeaWiFS estimations compare well with in situ data (0.77< r 2<0.98, N=21), although the satellite estimate show a marked bias (−35.6%) in the blue band nL w (412). Regarding the chlorophyll- a concentration, an overall agreement was observed ( r 2=0.77, rmse-L=0.66, N=28), with a mean absolute percentage difference of 66%, which is above the goal generally accepted of 35% for satellite ocean color chlorophyll estimates. For the studied Southern Ocean area (mainly the Bransfield Strait), NASA's OC4v4 algorithm systematically underestimates chlorophyll above 0.2 mg m −3, as previously demonstrated by other researchers.

Acoustical study of the spatial distribution of plankton on Georges Bank and the relationship between volume backscattering strength and the taxonomic composition of the plankton

High frequency (420 kHz) sound was used to study the volume backscattering from plankton and micronekton over Georges Bank as part of a study designed to determine the correlation length scales of plankton spatial patterns in relation to physical structure and to intercompare different kinds of sampling and remote-sensing instrumentation. Two physically distinct areas were studied: a well-mixed area in a shallow portion of the Bank and a stratified area on the deeper southern flank of the Bank. A submersible echo sounder with a down-looking transducer was mounted in a towed V-fin. Volume backscattering data were collected from near the sea surface to the bottom (40–80 m).Vertical and horizontal volume backscattering structure in the stratified region differed from that in the well-mixed area in both mean and variance, providing evidence that physical forcing of the pattern varied significantly between the two areas. Internal waves appeared to modulate the depth of dense mid-depth volume scattering layers in the stratified sites. In the mixed area, there was little horizontal layering or coarse-scale horizontal structure. However, fine-scale vertical lineations were evident with horizontal length scales on the order of the depth of the water column. One hypothesis to explain these vertical lineations in the well-mixed areas involves the development of secondary vertical circulation cells associated with the tidal flows over a rough bottom. Although volume backscattering at the stratified sites was 4–7 times higher than at the mixed site, there was no significant difference in MOCNESS (Multiple Opening/Closing Net and Environmental Sensing System) collected biovolumes between these locations. The difference in volume backscattering was due to differences in both the acoustic scattering properties of zooplankton taxa and the taxonomic composition of the plankton between the sites. Correlations between taxon abundance and volume scattering were positive and significant only for pteropods and euphausiid larvae. The abundances of copepods, chaetognaths, fish larvae, and amphipods were not significantly correlated with volume scattering. When taxon-specific model predictions of acoustic backscattering cross-section, developed by Stanton et al. (ICES Journal of Marine Science, 51(1994) 505–512), were used with field collected individual size and abundance data to predict measured volume backscattering data, good agreement was found between observed and predicted volume backscattering strengths.

Cytochrome P4501A induction and inhibition by 3,3′,4,4′-tetrachlorobiphenyl in an Ah receptor-containing fish hepatoma cell line (PLHC-1)

The induction of cytochrome P4501A1 (CYP1A1) in rat hepatoma cells has been used by some investigators to determine ‘dioxin equivalents’ in environmental samples, including extracts of fish tissues. However, the relative potency of inducing compounds may vary between species, suggesting the need for taxon-specific model systems. In this paper we present an initial characterization of CYP1A induction in one such system, a teleost liver cell line (PLHC-1) derived from a hepatocellular carcinoma of Poeciliopsis lucida (Hightower, L.E. and Renfro, J.L., 1988. J. Exp. Zool. 248, 290). Specific binding of the photoaffinity ligand 2-azido-3-[ 125I]iodo-7,8-dibromodibenzo- p-dioxin([ 125I]N 3Br 2DD) to proteins in PLHC-1 cytosol indicated the presence of the Ah receptor, which is known to control CYP1A induction in mammals. 3,3′,4,4′-Tetrachlorobiphenyl (TCB) induced a microsomal protein in PLHC-1 cells that was recognized by monoclonal antibody (MAb) 1-12-3 to scup CYP1A1 (P450E) on immunoblots. Immunohistochemical staining of whole cells with MAb 1-12-3 showed specific recognition of CYP1A induced by TCB. No staining was seen in untreated or vehicle-treated cells. There was an excellent quantitative correlation between amounts of CYP1A protein detected immunohistochemically and in immunoblots of cell homogenates. In a dose response experiment, maximal induction of ethoxyresorufin O-deethylase (EROD) activity occurred at 0.1 μM TCB; at higher concentrations (1 and 10 μM), EROD activity was reduced as compared to the activity at 0.1 μM TCB. In contrast, immunoreactive CYP1A protein increased with increasing TCB concentration up to 10 μM. The loss of EROD activity at high concentrations of TCB did not result from changes in cell number or viability. The apparent inhibition or inactivation of CYP1A catalytic activity by the higher concentrations of halogenated biphenyls has been seen, but not generally recognized, both in vivo and in cultured cells from diverse vertebrate species. PLHC-1 cells may be a good model system for studying Ah receptor-mediated regulation of gene expression, for determining the fish-specific toxic or inducing potency of halogenated aromatic hydrocarbon congeners, and for investigating the mechanism of CYP1A inhibition or inactivation by environmental contaminants such as TCB.