*4.7. Proteomic Analyses of TiO2 Enriched Phosphopeptides*

Sample Preparation: Mouse liver tissue was homogenized in mass spectrometry grade water using a Potter Elvehjem Teflon on glass tissue homogenizer kept on ice. Homogenates were brought to 2% lithium dodecyl sulfate (LiDS) then heated to 95 ◦C for 5 minutes. Heat inactivated samples were exposed with 5 mM dithiothreitol (DTT) to reduce disulfide bonds, alkylated with 15 mM iodoacetamide (IAA) and then 5 mM additional DTT was added to quench the remaining IAA. Proteins were precipitated and LiDS removed by addition of 5 volumes of methanol. Methanol-washed pellets were resuspended in 0.5% deoxycholate (DOC) in 1X phosphate buffered saline plus 40 mM TEAB using sonication delivered by a QSonica cup-horn sonicator as necessary. Protein concentrations in the DOC solubilized samples were determined using a Bradford assay then 0.5 mg of protein was trypsinized using an overnight incubation with 2 μg Promega sequencing grade trypsin. Digested samples were centrifuged to remove particulates and the entire sample dried by speed-vac then resolubilized in 100 μL of 65% acetonitrile, 2% trifluoroacetic acid (TFA) and at 25% saturation with glutamic acid. Phosphopeptides were selected from the digests on an AssayMap Bravo (Agilent Technologies, Santa Clara, CA, USA) robot using TiO2 cartridges. Phosphopeptides were analyzed at the Wayne State Proteomics Core using LC-MS/MS on an Orbitrap Fusion MS system (Thermo Fisher Scientific, San Jose, CA, USA). Each sample was analyzed independently using reversed-phase chromatography on an Acclaim PepMap RSLC (Thermo Fisher Scientific, San Jose, CA, USA), 75 μm × 25 cm column (Dionex, Sunnyvale, CA, USA). Peptides were eluted from the column in a 2-h gradient from 5% to 30% acetonitrile and analyzed directly by MS/MS using the Orbitrap Fusion.

#### *4.8. Proteomic Data Analysis*

MS spectra were searched against the Uniprot mouse complete database downloaded on 14 July 2017 (16,884 entries) using MaxQuant v1.6.2.10. (Max Planck Institute, Munich, Germany) [66]. Phosphorylation at Serine, Threonine and Tyrosine residues was set as a variable modification and the default penalty for modified peptide identification was reduced so that a minimum score of 20 and a minimum delta score of 3 were required for modified peptides. Match between runs was enabled. All other parameters were left at their default values. All analyses except the kinase enrichment analysis used phosphorylation sites without regard to localization confidence. Kinase enrichment analysis used all sites localized with > 80% confidence by Maxquant. Subsequent analysis used R v3.4.3 (http://www.R-project.org/) (The R Foundation, Vienna, Austria).

Phosphorylation site abundances were normalized so that each sample had the same median. Differentially abundant sites were identified using a moderated t-test [67]. A q-value was calculated for each site to account for multiple testing [68]. Gene Ontology (GO) biological processes that were affected by exposure were identified using Platform for Integrative Analysis of Omics Data (PIANO) [40]. T-statistics from the moderated t-test were submitted to PIANO and pathway enrichment was determined using the t-statistic mean. Phosphoproteins with multiple sites were not summarized and each site was submitted to PIANO individually. PIANO uses a permutation test to calculate an FDR corrected *p*-value. To reduce redundancy due to multiple GO categories with similar membership, affected pathways were clustered by phosphoprotein membership similarity and only the pathway with the lowest *p*-value in each cluster was reported. Pathways clustering was done using dynamic tree cut [69]. Pathway overlap was calculated as: (number of sites common to both pathways/number of sites in the smaller pathway). Kinase analysis was carried out using the regular expression-type kinase motifs distributed with Perseus software [70]. Identified phosphorylation sites that matched kinase criteria were converted to PIANO "kinase sets" and each kinase was tested for enrichment by PIANO analysis as above. Fuzzy c-means clustering was used to identify dose-response patterns in the data [71]. Sites were considered members of a cluster if they had greater than 0.5 membership. The number of clusters was set to 6 based on inspection of a plot of the number of sites clustered vs. the number of clusters. All sites were submitted to clustering without regard to their statistical significance. Gene Ontology (GO) biological processes were tested for enrichment in dose-response clusters using Fisher's exact test for an increased frequency of pathway components in cluster members versus cluster non-members. As for the samples from mice exposed with 100 μg/kg of MC-LR, Reactome pathways were tested for enrichment in dose-response clusters using Fisher's exact test for an increased frequency of pathway components in cluster members versus cluster non-members.

#### *4.9. Statistical Analysis*

Statistical analysis of all non-proteomic data was done using GraphPad PRISM 7 software (San Diego, CA, USA) and comparison within groups was done using Unpaired Student's t-test and Analysis of Variance (ANOVA) with Dunnette's multiple comparisons test. All data are presented as mean ± standard error of the mean (SEM) and a *p*-value of < 0.05 was considered to be statistically significant.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6651/11/9/486/s1, Figure S1: Effect of MC-LR on survival and gross liver morphology, Figure S2: Effect of MC-LR exposure on liver injury enzymes in NAFLD mice, Figure S3: Phosphorylation sites affected by MC-LR exposure, Figure S4: Fuzzy c-means clusters of phosphorylation site abundance versus microcystin dose. 6 clusters were generated, Table S1: Effect of MC-LR exposure on tissue weights in Leprdb/J mice, Table S2: Effect of MC-LR exposure on tissue weights in C57Bl/6J mice, Table S3: Effect of MC-LR exposure on blood chemistry in Leprdb/J mice, Table S4: Effect of MC-LR exposure on blood chemistry in normal C57Bl/6J (WT) mice, Table S5: Hematoxylin & Eosin (H&E), Periodic Acid-Schiff (PAS), and Composite Liver Injury Scores in Leprdb/J (db) mice, Table S6: Genetic analysis of hepatotoxicity in liver tissues, Table S7: Genetic analysis of oxidative stress response in liver tissues, Table S8: Identification of the clusters of pathways affected by 50 μg/kg MC-LR versus control using Reactome

*Toxins* **2019**, *11*, 486

database, Table S9: GO Biological Process enrichment analysis, Table S10: REACTOME enrichment analysis, Table S11: Phosphorylation sites that were affected by microcystin exposure at 50 or 100 μg/kg.

**Author Contributions:** Conceptualization, A.L., D.J.K. and S.T.H.; data curation, A.L., R.C.S., J.D.B., P.M.S., N.J.C., N.K.S., F.K.K., S.Z., A.L.K., P.D., C.J.M., J.A.W., E.C., D.P., D.B.-R., D.I., B.L., N.M., A.F.G., D.M., S.T.H., D.J.K.; formal analysis, A.L., J.D.B., N.M., D.J.K. and S.T.H.; funding acquisition, S.T.H. and D.J.K.; investigation, A.L., P.M.S., N.J.C., N.K.S., D.P., D.B.R., D.I., B.L., N.M., D.J.K. and S.T.H.; methodology, A.L., D.J.K. and S.T.H.; project administration, D.M., S.T.H. and D.J.K.; resources, A.L., R.C.S., J.D.B., P.M.S., N.J.C., N.K.S., F.K.K., S.Z., A.L.K., P.D., C.J.M., J.A.W., E.C., D.P., D.B.R., D.I., B.L., N.M., A.F.G., D.M., S.T.H., D.J.K.; software, A.L., D.J.K. and S.T.H.; supervision, S.T.H. and D.J.K.; validation, P.M.S., N.J.C., J.A.W., D.I., B.L., N.M., A.F.G., D.M., D.J.K. and S.T.H.; visualization, A.L., J.D.B., P.M.S., N.J.C., D.P., D.B.R., D.I., B.L., N.M., S.T.H. and D.J.K.; writing—original draft preparation, A.L.; writing—review and editing, A.L., P.M.S., N.J.C., R.C.S., J.D.B., N.M., S.T.H., D.J.K.

**Funding:** This research was funded by Harmful Algal Bloom Research Initiative grants from the Ohio Department of Higher Education, David and Helen Boone Foundation Research Fund, University of Toledo Women and Philanthropy Genetic Analysis Instrumentation Center, The University of Toledo Medical Research Society, the Center for Urban Responses to Environmental Stressors (CURES) NIH Grant # P30 ES020957 and the Wayne State University Proteomics Core that is supported through NIH grants P30 ES020957, P30 CA 022453 and S10 OD010700.

**Acknowledgments:** Some of these data were presented in abstract form at the 2018 and 2019 Midwest Clinical and Translational Research Meeting in Chicago, IL. The authors are very grateful for the technical assistance provided by Pamela Brewster, Adam Spegele, Aaron Tipton and Dalal Mahmoud as well as instrumentation support provided by the Air Force Office of Scientific Research (DURIP 14RT0605) for the acquisition of the Orbitrap Fusion instrument.

**Conflicts of Interest:** The authors declare no conflict of interest.
