2188 Diabetes Volume 66, August 2017

Temporal Transcriptomic and Proteomic Landscapes of Deteriorating Pancreatic Islets in Type 2 Diabetic Rats

Junjie Hou,1 Zonghong Li,1,2 Wen Zhong,1,3 Qiang Hao,1 Lei Lei,1 Linlin Wang,1,4 Dongyu Zhao,1 Pingyong Xu,1 Yifa Zhou,2 You Wang,1 and Tao Xu1,3,4

Diabetes 2017;66:2188–2200 | https://doi.org/10.2337/db16-1305

Progressive reduction in b-cell mass and function com- of which appear to play a primary role in b-cell function prise the core of the pathogenesis mechanism of type 2 rather than to affect insulin resistance, further highlighting diabetes. The process of deteriorating pancreatic islets, the importance of b-cells in the pathogenesis of T2D (5). in which a complex network of molecular events is in- T2D is a complex disease, and b-cell failure is likely caused volved, is not yet fully characterized. We used RNA sequenc- by altered expression of many genes and their products. ing and tandem mass tag–based quantitative proteomics Therefore, the use of system-oriented strategies is critical technology to measure the temporal mRNA and protein to investigate the complex changes that occur in b-cells or expression changes of pancreatic islets in Goto-Kakizaki pancreatic islets, which primarily comprise b-cells. Hence, (GK) rats from 4 to 24 weeks of age. Our omics data set large-scale and unbiased omics technologies, particularly outlines the dynamics of the molecular network during the microarray-based transcriptomics and mass spectrometry deterioration of GK islets as two stages: The early stage (MS)–based proteomics, have been used to analyze islets (4–6 weeks) is characterized by anaerobic glycolysis, in- isolated from various T2D animal models and human ca- flammation priming, and compensation for insulin synthe- b sis, and the late stage (8–24 weeks) is characterized by daver donors to elucidate the mechanisms underlying -cell inflammation amplification and compensation failure. Fur- failure (summarized in Supplementary Table 1). b-Cell fail- ISLET STUDIES ther time course analysis allowed us to reveal 5,551 dif- ure during diabetes progression is a gradual process that ferentially expressed genes, a large portion of which have undergoes various stages (6,7) wherein different molecule not been reported before. Our comprehensive and tem- events occur in chronological order. However, current stud- poral transcriptome and proteome data offer a valuable ies have focused primarily on single time points at relatively resource for the diabetes research community and for late stages of the disease, so mapping the order in which quantitative biology. these events occur and distinguishing causal molecular events (leading to diabetes) from those that occur as a consequence of glucolipotoxicity associated with diabetic Type 2 diabetes (T2D) is a major public health issue conditions are impossible. For this reason, prospective characterized by pancreatic islet b-cell failure in the pres- studies investigating the evolution of molecular events ence of insulin resistance. Accumulating evidence suggests in islet b-cells at various stages of T2D are of interest. that progressive deterioration of pancreatic b-cell function The study of b-cells in humans with T2D often has and gradual loss of b-cell mass could be the core events been hindered by the limited accessibility of human islets during T2D development, regardless of therapy status and by ethical considerations. In this context, appropriate (1–4). Genome-wide association and sequencing studies rodent models are essential for the identification of diabetic have identified multiple risk variants for T2D, the majority mechanisms (8). The Goto-Kakizaki (GK) rat, one of the

1National Laboratory of Biomacromolecules, CAS Center for Excellence in Bio- Received 18 November 2016 and accepted 17 May 2017. macromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing, This article contains Supplementary Data online at http://diabetes China .diabetesjournals.org/lookup/suppl/doi:10.2337/db16-1305/-/DC1. 2School of Life Sciences, Northeast Normal University, Changchun, China J.H., Z.L., and W.Z. contributed equally to this work. 3College of Life Science and Technology, HuaZhong University of Science and Technology, Wuhan, China © 2017 by the American Diabetes Association. Readers may use this article as 4College of Life Sciences, University of Chinese Academy of Sciences, Beijing, long as the work is properly cited, the use is educational and not for profit, and the China work is not altered. More information is available at http://www.diabetesjournals .org/content/license. Corresponding author: You Wang, [email protected], and Tao Xu, xutao@ ibp.ac.cn. diabetes.diabetesjournals.org Hou and Associates 2189 best-characterized animal models of spontaneous T2D (9), instructions from the manufacture (Thermo Fisher Scien- shares many characteristics with human patients with di- tific). TMT-labeled peptide mixtures were equally pooled, abetes (10). Similar to human T2D, the core cause under- separated by offline high pH reversed-phase chromatogra- lying hyperglycemia in GK rats is b-cell failure (11,12). phy, and repeatedly analyzed using a nano–liquid chro- In the current study, to understand the process of matography-tandem MS technique (Supplementary Fig. deteriorating pancreatic islets at the molecular level, we 1A). The raw MS data were processed with Proteome Dis- used RNA sequencing (RNA-seq) and tandem mass tag coverer 1.4 software. The peptide confidence value was set (TMT)–based quantitative proteomics technology to gener- as 0.01. At the protein level, a precursor intensity fraction ate integrated transcriptomic and proteomic profiles of of 50% was selected as an optimal trade-off value for both pancreatic islets in GK rats after the establishment of hyper- identification and quantification (Supplementary Fig. 1B). A glycemia (from 4 to 24 weeks). Subsequent bioinformatics pseudocount representing relative protein abundance was analysis in a time course fashion revealed the chronological calculated by using the TMT ratio and the normalized spec- order of T2D-related molecular events during the deterio- tral abundance factor (15). ration of pancreatic islets. This large quantitative data set Bioinformatics Analysis represents a valuable resource that provides a comprehen- Both mRNA raw counts and protein pseudocounts were sive picture of the mechanisms responsible for islet dysfunc- normalized by using the remove unwanted variation approach tion and will allow us to identify potential interventions to (Supplementary Fig. 1C) (16). Differentially expressed (DE) prevent b-cell failure and deterioration in human T2D. genes were assessed by ANOVA with a false discovery rate , RESEARCH DESIGN AND METHODS of 0.01. The Database for Annotation, Visualization and Integrated Discovery (DAVID) Web service API Perl Client Brief descriptions of key experimental procedures are (17,18) was used to perform gene ontology (GO) functional provided below. More details are provided in the Supple- enrichment analysis (with a false discovery rate of ,0.05). mentary Data. The k-means clustering algorithm was used to classify dy- Animals namic gene expression patterns. Kyoto Encyclopedia of Founders of GK/Jcl diabetic rats were purchased from RIKEN Genes and Genomes (KEGG) signaling pathway enrichment BioResource Center (Ibaraki, Japan). All GK/Jcl diabetic rats analysis was carried out by using the Generally Applicable and Wistar (WST) rats were maintained under specific Gene Set Enrichment (GAGE) package in R software (with pathogen-free conditions and were used between 4 and an adjusted P , 0.05) (19). 24 weeks of age in accordance with the animal experimental Data Resources guidelines set forth by the Institutional Animal Care and For this work, 92.4 gigabyte sequencing data were gener- Use Committee of the Institute of Biophysics, Chinese ated. All RNA-seq data were deposited in the National Academy of Sciences. Center for Biotechnology Information Gene Expression Preparation of Pancreatic Islets From GK and WST Rats Omnibus under accession number GSE 81811. The MS data Pancreatic islets from male GK and age-matched control were deposited in the ProteomeXchange Consortium through fi WST rats were isolated through collagenase digestion. After the PRIDE (Proteomics Identi cations) (20) partner reposi- fi separation on a Ficoll density gradient, the islets were tory with the data set identi er PXD004709. handpicked in Hanks’ buffer under a dissection microscope. RESULTS N-Acetyl-L-Cysteine Treatment Experiments Transcriptomic and Proteomic Profiles of Rat SixteenmaleGKratswereusedinthefollowingexperi- Pancreatic Islets Over Time ments. Littermates of GK rats were randomly divided into To investigate the global molecular dynamics of T2D islets, N-acetyl-L-cysteine (NAC) and control groups. Four-week- we analyzed the transcriptomes and proteomes of pancre- old rats were orally administered NAC 200 mg/kg of body atic islets isolated from male GK rats and age- and sex- weight (616-91-1; Sigma) or drinking water by gavage once matched control WST rats at five consecutive time points a day for 12 weeks. Random blood glucose assay and glucose (weeks 4, 6, 8, 16, and 24) (Fig. 1A). Transcriptomes and tolerance, insulin tolerance, and glucose-stimulated insulin proteomes of islets were measured by using the MAPS- secretion (GSIS) tests were performed as described in the based RNA-seq technique and TMT labeling–based proteo- Supplementary Data. mic method, respectively. Combined analysis of all samples yielded the identification of 15,101 mRNAs and 8,362 pro- RNA-Seq Analysis teins, of which 7,395 overlapped (Fig. 1B). Furthermore, RNA-seq was performed by using a multiplex analysis of 13,866 mRNAs (minimal counts of 10 detected in at least polyA-linked sequences (MAPS) approach as previously three samples) and 5,631 proteins (identified in at least two described (13). biological replicates by peptides of precursor intensity TMT-Based Proteomics Analysis fraction #50%) were considered a quantifiable data set, Proteins extracted from isolated islets were digested of which 5,015 overlapped (information for all identified and labeled by 6-plex TMT reagents (14) according to the genes is provided in Supplementary Table 2). We estimated 2190 Deteriorating Pancreatic Islets in T2D Diabetes Volume 66, August 2017

Figure 1—Transcriptomic and proteomic analysis of pancreatic islets in diabetic GK rats over time. A: Experimental workflow. Pancreatic islets isolated from GK and age-matched control WST rats of five different ages (4, 6, 8, 16, and 24 weeks) were subjected to MAPS-based RNA-seq and TMT labeling–based proteomics analysis. After performing data quality control and normalization, differentially expressed mRNAs and proteins were analyzed by ANOVA followed by integrated bioinformatics analysis and biological validation. B: Venn diagram of identifiable and quantifiable mRNAs and proteins in this study. C and D: Unsupervised hierarchical clustering and PC analysis of all quantifiable mRNAs and proteins from GK and WST islets indicated the reproducibility of biological replicates; however, islet gene expression in both GK and WST rats was highly variable between time points at both the mRNA and the protein level.

relative protein abundance, which was noted as the protein separated from other GK islets and instead clustered with pseudocount in this study, on the basis of TMT ratio and WST islets, suggesting minimal changes in protein expres- normalized spectral abundance factor. Compared with TMT sion at the early stages of T2D, even when there are signif- ratios, protein pseudocounts resulted in the generation of icant changes at the mRNA level. more-reasonable clusters that matched the experiments We next performed a principal component (PC) analysis (Supplementary Fig. 2A). Protein pseudocounts positively to investigate transcriptome and proteome dynamics over correlated with mRNA counts, with a mean Spearman cor- time (Fig. 1D). PC2 primarily reflected age-related develop- relation coefficient of 0.37 (Supplementary Fig. 2B), which mental changes, whereas PC1 represented the differences is similar to reported values in other biological systems between normal WST and diabetic GK rats. Similar to the (21,22). These results demonstrate that our method of above results obtained from unsupervised hierarchical clus- estimating protein pseudocounts was reasonable and tering, PC1 highlighted two notable diabetic stages in GK unbiased. islets. At the mRNA level, islets at 4, 6, and 8 weeks clus- We first carried out an unsupervised hierarchical clus- tered as one stage, and islets at 16 and 24 weeks repre- tering analysis of the transcriptomes and proteomes from sented another stage. However, at the protein level, islets at GK and WST islets (Fig. 1C). At the mRNA level, WST islets 4 and 6 weeks clustered as one stage, and the remaining clustered together and clearly separated from GK islets as time points clustered as a separate stage. GK islets progres- expected, representing different pathophysiological states sively develop into disorganized structures exhibiting pro- of islets in control rats versus diseased rats. Overall, the nounced fibrosis separating strands of endocrine cells (23). transcriptomes of GK and WST rats demonstrated appro- Of note, these changes were not present or rare in islets priate clustering at different time points, representing the at 4–6 weeks but became prominent at older ages (8– developmental stages of islets. Of note, GK islets at 4, 6, 24 weeks) (Supplementary Fig. 2C), correlating well with and 8 weeks formed one branch that was distinct from the our proteome-defined two stages (Fig. 1C and D). Taken branch formed at 16 and 24 weeks, likely reflecting two together, global profiling at the mRNA and protein level different diabetic stages of islets in GK rats. Upon proteome roughly characterizes the deterioration of islets in GK rats clustering examination, GK islets at 4 and 6 weeks were from 4 to 24 weeks of age into two stages: an early stage at diabetes.diabetesjournals.org Hou and Associates 2191

4–6weeksandalatestageat8–24 weeks, with a turning with enriched functions by using EnrichmentMap software point at ;8weeks. (Supplementary Table 4) (25). On the basis of the time- course expression patterns of clustered genes, we classified Analysis of DE mRNAs and Proteins the biological events into the following categories: DE genes between GK and WST rats at each time point were first assessed by one-way ANOVA (P , 0.05) (Supple- 1. Constant up (m2, p8) and down (p2, m3, m10) genes, mentary Table 2). The results revealed a remarkable increase which were either up- or downregulated at all the time of DE mRNAs over time, whereas only a mild increase was points. The constant up genes were highly enriched for found in the number of DE proteins, suggesting a more such functions as cell redox homeostasis, translation elon- dynamic regulation of gene expression at the mRNA level gation, cytoskeleton organization, antiapoptosis, and so than at the protein level during the development of GK forth. In contrast, the constant down genes were mainly diabetes (Supplementary Fig. 3A). When we compared the associated with mitochondrion, metabolism, lysosome, fold changes in mRNA and protein levels, the Pearson cor- protein transport, and so forth. relation value (0.11) was relatively poor at 4 weeks but increased 2. Up early genes (p1, p6, m11) that were upregulated at to 0.23–0.28 at later time points (Supplementary Fig. 3B). 4–8 weeks, which include those participating in the From our time-resolved expression data set, we analyzed glucose metabolism and innate immune response. In the temporal significance of gene expression changes addition, many proteasome proteins were dramatically by using two-way ANOVA, which considered weeks upregulated at 4 weeks. (five different time points) and rats (GK vs. WST) as the 3. Up late genes (p9, m8, p7, m1, m12) that were upregu- two statistical factors. In total, we identified 5,551 DE lated at the late stage of 8–24 weeks. These genes are genes, including 3,910 mRNAs and 2,387 proteins (Supple- highly enriched for cell adhesion and cytoskeleton orga- mentary Table 2), of which 746 were identified at both the nization at the protein level, probably associating with mRNA and the protein level. The correlations among these the development of islets fibrosis, whereas at the 746 DE genes varied from anticorrelation to full accordance, mRNA level, apoptosis was significantly upregulated with an average Pearson coefficient of 0.39 (Fig. 2A and at 24 weeks. Supplementary Table 3), which is larger than the correlation 4. Down early genes (p3, m6) that were downregulated at coefficients at individual time points. DAVID functional 4–8 weeks. The most noticeable feature is the downregu- clustering analysis indicated carbon metabolism and ribo- lation of oxidative phosphorylation (OXPHOS) and tri- some enrichment among genes with concordant mRNA and carboxylic acid (TCA) cycle at the protein level, probably protein levels (Fig. 2B), potentially representing the set of suggesting the insufficient energy supply and oxidative genes exhibiting stable mRNA and protein expression (24). stress in GK islets at the early stage (26–28). At the In contrast, no significant functional clustering was identi- mRNA level, the genes associated with cell cycle and fied for negatively correlated DE genes. nuclear lumen were highly enriched, likely contributing For the confirmation of expression data, we randomly to the loss of b-cells in GK islets. selected six DE genes to measure their mRNA expression by 5. Down late genes (p5, m9, p4, p7, m5, m4) that were quantitative RT-PCR (qRT-PCR) in independent GK/WST downregulated at 8–16 weeks. The GO functions of in- islet samples. The expression patterns of these genes sulin secretion, lysosome, and secretory granule were measured by RNA-seq and qRT-PCR were similar (Supple- representatively enriched. mentary Fig. 3C). Moreover, by comparing our GK/WST data set with the published data set of islets from individ- Such a time course clustering analysis from genes to uals with and without T2D (Supplementary Table 1) (10), biological functions suggests that the progression of GK we found that on average, 68.9% of DE genes were consis- islet deterioration was stage based and dynamically regu- tent between GK rats and humans with diabetes (Supple- lated. Furthermore, because genes exhibiting similar ex- mentary Fig. 3D), indicating high relevance of the current pression patterns generally share functional relationships, study in GK rat to human islets in T2D. clustering analyses also allow the prediction of genes that share similar temporal expression patterns, with previously Dynamic mRNA and Protein Expression Patterns Over Time in GK Islets validated diabetes-related genes as potential new candidates The primary purpose for the generation of the time course for further investigation. data set in this study is to reveal the temporal properties of Pathway Dynamics in GK Islets During Diabetes biological pathways relevant to the development of GK Progression diabetes at the system level. Therefore, we performed time To gain a deeper understanding of temporal pathway course pattern analysis for all DE genes by using the sequences during the deterioration of GK islets, we k-means clustering method and successfully identified performed GAGE analysis with the quantitative transcrip- 12 mRNA expression patterns (m1–m12) and 9 protein tomic and proteomic data sets and identified 161 KEGG expression patterns (p1–p9) (Fig. 2C). To explore the pathways significantly enriched for at least one time point biological functions of these expression patterns, we carried (Benjamini-Hochberg–adjusted P , 0.05) (Supplementary out a DAVID analysis and organized the identified networks Table 5). Consistent with the above gene expression 2192 Deteriorating Pancreatic Islets in T2D Diabetes Volume 66, August 2017

Figure 2—Temporal gene expression patterns during GK diabetes progression. A: Pearson correlation analysis of the temporal mRNA and protein expression of 746 overlapping DE genes. In total, 76.8% of genes were positively correlated, of which 41.3% were significantly correlated (adjusted P < 0.01). The mean Pearson correlation coefficient was 0.388. B: DAVID analysis of positively correlated DE genes revealed two enriched functional annotations primarily associated with carbon metabolism and ribosomes. No functional annotation was enriched for negatively correlated DE genes. C: Time course dynamic expression clustering analysis of DE genes. First, fold changes in DE genes were transformed into z scores. Next, the k-means clustering method was used to classify the genes into 12 mRNA and 9 protein patterns, displayed as a Circos figure. Functional enrichment analyses of the genes within each pattern were carried out by using EnrichmentMap software. The enriched GO functional groups are selectively highlighted with transparent pink (upregulated) and blue (downregulated) circles. ER, endoplasmic reticulum; ETC, electron transport chain; FA, fatty acid; IKK, IkB kinase; MAPKKK, mitogen-activated protein kinase kinase kinase; w, week. diabetes.diabetesjournals.org Hou and Associates 2193 clustering analysis results, these KEGG pathways also roughly comprised down- and upregulation-dominated tem- poral classes (Fig. 3). For example, insulin secretion, and SNARE interactions in vesicular transport were downregu- lated at the late stage with the same temporal pattern (representative genes illustrated in Supplementary Fig. 4). These two pathways were partially downregulated at 4 and 6 weeks with no significant change in statistics and became significantly downregulated after 8 weeks, clearly demon- strating the dynamic features of various pathways involved in insulin secretion during the progression of T2D. We also noticed that the pathway of glycolysis/gluconeogenesis was gradually upregulated since 6 weeks, probably indicating that in GK islets, the anaerobic metabolism is the dominant approach for supplying the energy because of the defect in OXPHOS and TCA cycle. In addition, we found that signaling pathways associated with inflammation were upregulated at the mRNA level, including the nucleotide oligomerization domain (NOD)– like receptor, tumor necrosis factor (TNF), and nuclear factor-kB(NF-kB) signaling pathways. This finding is con- sistent with previous reports that islet inflammation plays an important role in the pathogenesis of GK diabetes (26,29) and provides more comprehensive details of path- way dynamics at both the mRNA and the protein level. Mitochondrial Signatures in GK Islets We identified 311 DE genes as mitochondria related by using GO terms for cellular components that contain the keywords mitochondrion or mitochondrial and divided them into four groups on the basis of the unsupervised hierarchical clustering analysis of their temporal profiles (Supplementary Table 6). Further manual annotation revealed the details regarding mitochondrial dysfunction during the progression of T2D (Fig. 4). We found that OXPHOS complexes, mitochondrial ribosome proteins, trans- locase outer/inner membrane complex (TOM/TIM), and some metabolite transporters were downregulated early at the protein level, and conversely, most of corresponding mRNAs were upregulated at 4 and 6 weeks, indicating tran- scription compensation at the early stage of T2D. However, this compensation ability was eventually lost at the later stage during b-cell deterioration. In addition, several pro- Figure 3—Heat map of KEGG pathway enrichment analysis. Normal- teins responsible for protein assembly and quality control, ized counts/pseudocounts of the DE genes were subjected to GAGE mitochondrial biogenesis, and mitochondrial DNA tran- analysis by using the Bioconductor package gage. Pathways with adjusted P values (Benjamini-Hochberg procedure) of <0.05 are in- scription were downregulated, such as Hspd1, Grpel1, dicated by asterisks. The Stat.mean values represent the averaged Lonp1, Letm1 and Pmpcb, Tfam, Mtfr1l, Mfn2, the transfer magnitude and direction of fold changes at the gene set level corre- RNA ligases (Rars2, Nars2, Dars2, and Vars2), and 12 mito- sponding to the color-coded upregulated (red) and downregulated chondrial RNAs (Supplementary Fig. 5A). Taken together, (blue) changes. KEGG pathway maps were used to perform classifi- cations. Akt, protein kinase B; ECM, extracellular matrix; HIF-1, hyp- mitochondrial dysfunction was considered one of earliest oxia-inducible factor 1; Jak, Janus kinase; MAPK, mitogen-activated pathogenic events in islets of GK rats. kinase; PI3K, phosphatidylinositol 3-kinase; TGF, transforming growth factor; tRNA, transfer RNA. Overview of Metabolism in GK Islets To gain a deeper understanding about how the metabolism in GK islets changed with the development of diabetes, we mapped our quantitative omics data to KEGG metabolic (26,27) and the downregulation of the TCA cycle, OXPHOS, pathways (Fig. 5). The most notable change of metabolism and fatty acid metabolism, which suggests that the primary in GK islets was the upregulation of glycolysis metabolism metabolism of GK islets switches from aerobic metabolism to 2194 Deteriorating Pancreatic Islets in T2D Diabetes Volume 66, August 2017

Figure 4—Mitochondrial signatures in GK islets. On the basis of the locations or biological functions of GO annotations, the mitochondria- related DE genes were mapped to the outer and inner membrane, membrane transporter, OXPHOS complex, ribosomal proteins, protein quality control, transcription, translation, biogenesis, and antioxidant. The heat maps for mRNA and protein expression at the five time points are color- coded according to the log2 fold-changes for GK vs. WST. anaerobic metabolism (the so-called Warburg-like effect). Fur- data show that several genes functionally associated with thermore, we found that besides FAD-dependent glycerol- neogenesis are downregulated in GK islets, including Pdx1, 3-phosphate dehydrogenase (Gpd2) reported previously Nkx2-2, Nkx6-1, Mafa,andFev, at the mRNA and/or protein (30), Got1, Got2, Mdh1, and Mdh2 were also downregu- level (Fig. 6). Moreover, several genes specificfora-cells and lated, representing a comprehensive picture of defective pancreatic polypeptide cells (i.e., Arx, Isl1, Pax6, Pou3f4, malate-aspartate shuttle and glycerol phosphate shuttle in Ppy) were downregulated at the early stage. Certain genes GK islets (Supplementary Fig. 5B). Such a defect could cause required for the differentiation of pancreatic progenitors an increase of the cytoplasmic NADH/NAD+ ratio, enhance into endocrine progenitors (i.e., Foxa1, Gata6, Sox9, Onecut1) fi the formation of lactate from pyruvate, and further disrupt were signi cantly upregulated at the late stage of GK the link between cytosolic glycolysis and mitochondrial diabetes. metabolism. Low levels of b-cell proliferation in GK rats constitute In addition, some genes participating in amino acid another factor contributing to decreased b-cell mass metabolism were also downregulated, partially contributing (11,26,36). In the current data set, 38 genes among cluster to a defect in GSIS (31–33). However, several enzymes in- m6, including Aurkb, Ccna2,andKifc1, were associated with volved in glutathione metabolism were upregulated at both cell cycle and nuclear lumen and exhibited similar ex- the mRNA and the protein level in GK islets, including Gclc, pression patterns at the early stage (Supplementary Fig. 6A). These genes gradually decreased over time in WST rats, Ggct, Ggt1, Gsta3, and Gsto1, which may reflect an adaptive reflecting an aging-dependent reduction in proliferation mechanism to combat increased reactive oxygen species over time (38). In contrast, all these genes were dramati- (ROS) in GK islets (34,35). cally downregulated at 4 weeks in GK islets and then pro- Reduced Neogenesis and b-Cell Proliferation gressively decreased to even lower levels at 24 weeks Insufficient insulin secretion is caused by either impaired (Supplementary Fig. 6B). Consistently, Ki67-positive GSIS or reduced b-cell mass. b-Cell mass is regulated by a b-cells were significantly decreased in GK islets at 4, 6, balance among b-cell neogenesis, proliferation, and apopto- and 8 weeks (Supplementary Fig. 6C), suggesting reduced sis. Consistent with previous studies (26,36), the current b-cell proliferation (11,12,26). Conversely, the apoptosis diabetes.diabetesjournals.org Hou and Associates 2195

GLC β-Ala

s

i

s

y

l

o

c

y

l

FAFA FAFA GlycolysisG BiosynthesisBiosynthesis β-oxidation-oxidation PYR LA Ser Gly AminoAmino AAcidcid Ala ACCOA Val Arg Asp Leu Ile

TCATCA Lys Met

Glu Cys Gln OXPHOSOXPHOS GSH GABA

46816244681624 Week Acaca Aacs Acsl1 mRNA Protein Abat Acsl3 Mccc2 Gck Acsl5 Mcee Pfkm Fasn Oxct1 Gpi Mcat Auh Aldoa Oxsm Pccb Aldob Acaa1a Mccc1 Gapdh Acaa1b Gad2 Eno1 Acaa2 Gls Pklr Acads Glud1 Ldha Acadsb Gclc Aco2 Acat2 Ggct Cs Acox1 Ggt1 Dhtkd1 Acox3 Gsta3 Dlat Cpt1a Gsto1 Fh Cpt2 Idh2 Echs1 Category Idh3a Eci2 Idh3B Gcdh Glycolysis Mdh2 Hadh TCA Cycle Ogdhl Hadha Pdhb Fatty acid biosynthesis Sdha Sdhb Log2 (GK/WST) Fatty acid degradation Sdhd 2 0 -2 Amino acid metabolism Suclg1 Suclg2 Glutathione metabolism

Figure 5—Metabolic overview of GK islets. DE enzymes were mapped on the KEGG global metabolism map. Gold lines represent differentially expressed metabolic enzymes. Metabolic pathways are selectively highlighted with pink and green lines representing upregulation and down- regulation, respectively. The heat maps for mRNA and protein expression at five time points are displayed and grouped as glycolysis, TCA cycle, fatty acid biosynthesis, fatty acid degradation, amino acid metabolism, and glutathione metabolism. ACOOA, acetyl CoA; Ala, alanine; Arg, arginine; Asp, aspartic acid; Cys, cysteine; FA, fatty acid; GABA, g-aminobutyric acid; GLC, glucose; Gln, glutamine; Glu, glutamic acid; GSH, glutathione; Gly, glysine; Ile, isoleucine; LA, lactate; Leu, leucine; Lys, lysine; Met, methionine; PYR, pyruvate; Ser, serine; Val, valine.

pathway was only elevated at 24 weeks at the mRNA level, proinflammatory cytokines revealed two distinct stages: implying that apoptosis was not responsible for b-cell loss early priming and late amplification. During the priming during the early phase of the disease. stage, interleukin-1b (IL-1b) and IL-6 were only elevated to low levels between 4 and 8 weeks followed by a rapid Two-Stage Inflammation in GK Islets increase to much higher levels (84-fold increase in GK at Chronic inflammation in GK islets has been demonstrated 24 weeks for IL-6) during the late amplification stage (be- andconsideredasapathophysiological contributor in tween 16 and 24 weeks) (Fig. 7). Because immune cell in- T2D (29,39). In the current study, temporal expression of filtration was hardly detectable in GK islets before 8 weeks 2196 Deteriorating Pancreatic Islets in T2D Diabetes Volume 66, August 2017

Figure 6—b-Cell neogenesis defects in GK rats. The illustration depicts an overview of pancreatic endocrine cell development, which was reconstructed by adapting a figure from a reference article (37) with minor modifications. The master transcription factors within each type of cell are listed; upregulation and downregulation are presented in red and green, respectively. Gene expression line chart data are mean 6 SEM of repeated experiments (n = 3 [except for WST week 4 mRNA data where n = 2]). *P < 0.01, **P < 0.001, ***P < 0.0001 by adjusted ANOVA. ns, no significant difference between GK and WST; PP, pancreatic polypeptide.

(39), the early priming stage is likely induced by metabolic The NLRP3 (NLR family, pyrin domain containing 3) dysfunction in GK islets. The identity of specificsensors inflammasome activates both NF-kB (to induce pro-IL-1b that are triggered to produce this priming of inflammation production) and caspase-1 (to process pro-IL-1b into its is not fully understood. mature active form). ASC (apoptosis-associated speck-like diabetes.diabetesjournals.org Hou and Associates 2197

Figure 7—ROS and inflammation contribute to the pathogenesis of islet dysfunction in GK rats. The illustration depicts ROS signaling flux as a core hub linking metabolic dysfunction with islet inflammation and fibrosis. The time course gene expression data for several key genes involved in the generation of ROS, antioxidants, inflammation, and fibrosis are displayed as a line chart. Gene expression line chart data are mean 6 SEM of repeated experiments (n = 3 [except for WST week 4 mRNA data where n =2]).*P < 0.01, **P < 0.001, ***P < 0.0001 by adjusted ANOVA. ECM, extracellular matrix; ER, endoplasmic reticulum; HIF-1a, hypoxia-inducible factor 1a; Mito, mitochondrion; ns, no significant difference between GK and WST. 2198 Deteriorating Pancreatic Islets in T2D Diabetes Volume 66, August 2017 protein containing a CARD, also called PYCARD), which methods used here. More importantly, the construction of interacts with NLRP3 during inflammasome assembly, time course–based gene expression and protein profiles was upregulated in GK islets, indicating possible activation allowed us to identify the chronological order of biological of the NLRP3 inflammasome during the initiation of sterile events contributing to the pathogenesis of T2D. inflammation. The NLRP3 inflammasome is also activated The data suggest two early events that likely contribute by TXNIP (thioredoxin-interacting protein), a key node link- to T2D in GK islets: a reduction in b-cell mass and a shift in ing glucotoxicity and endoplasmic reticulum stress to metabolism. Many transcription factors required for the NLRP3 inflammasome activation (40). We observed that specification of endocrine cells were downregulated early TXNIP was gradually upregulated in GK islets after 6 weeks. in GK islets, whereas those required for trunk and exocrine cells were not altered at 4 weeks but increased at later time Invasive ROS Contributing to the Deterioration points (Fig. 6). Indeed, GK rats from the Paris colony ex- of Islets in GK hibit a significant reduction in b-cell mass at the fetal stage Oxidative stress is a pathogenic factor caused by chronic that precedes the onset of hyperglycemia at ;4weeksafter hyperglycemia in GK pancreatic islets (41). However, sour- birth (11), similar to our GK colony. Besides neogenesis, fi ces of ROS in T2D have not been clearly de ned (42). The defective proliferation also causes a reduction in b-cell current data suggest multiple sources of ROS accumulation mass, as has been suggested for GK rats from the Paris in GK islets at the early stage, including dysfunctional mi- colony (11,26). In our GK rats, many genes required for tochondria, nitric oxide synthase (Nos2/iNOS), NADPH ox- the cell cycle and proliferation were significantly downregu- idase (Nox4), cyclo-oxygenase (Ptgs1/Cox1 and Ptgs2/Cox2), lated at the early stage (4–6 weeks) (Supplementary Fig. 6C and cytochrome P450 mono-oxygenase (Cyp2s1, Cyp7b1, and D), suggesting an early defect in proliferation. and Cyp4f5) (Fig. 7). We also observed increased expression Another notable feature of the current data is the obser- of several antioxidants, such as Sod1, Prdx4,andGpx2 (Fig. vation of an early shift in metabolism. As early as 4 weeks, 7), in GK islets, consistent with previous reports (27,35,43). the primary metabolism in the islets of GK rats switched To validate the involvement of ROS in the pathogen- from aerobic metabolism to anaerobic metabolism (Warburg- esis of T2D, we treated 4-week-old GK rats with an like effect). Although metabolism switch has been proposed antioxidant, NAC, for 10 weeks. NAC treatment signif- in previous research (34), our time course–based quantita- icantly reduced random blood glucose of GK-NAC rats tive data allowed us to detect this metabolic shift as an early (n = 8, NAC 200 mg/kg) compared with that of age-matched event of T2D in GK rats and permitted us to gain a deeper – GK-control rats (n = 8, sham) (Supplementary Fig. 7A D), insight into the mechanism underlying this shift. Insuffi- and GK-NAC rats exhibited greater tolerance to high glu- cient reduction in b-cell mass alone may not necessarily fi cose in the oral glucose tolerance test. No signi cant cause T2D because autopsy studies of patients with T2D differences in insulin sensitivity between GK-NAC and have revealed an ;50% decrease in b-cell mass compared GK-control rats were observed. The islets in GK-NAC with BMI-matched control patients (44,45). Furthermore, a groups had higher GSIS than those in GK-control groups reduction in b-cell mass of ;50% is required for dogs and (Supplementary Fig. 7E). Furthermore, we measured the rats to develop diabetes (46). This metabolic shift is likely mRNA expressions in islets between GK-NAC and GK-control caused by mitochondrial dysfunction. Many proteins asso- by using RNA-seq. The results show that the insulin secretion ciated with the OXPHOS system, metabolite transporters, fl pathway was upregulated, whereas ROS and in ammation- and the TCA cycle were downregulated starting at 4 weeks related genes (Tnf, Ggt1, Pycard, Cxcl1, etc.) and signaling (Fig. 3). Thus, the mitochondrial limitation of glucose oxi- pathways (NOD-like receptor signaling pathway, TNF signal- dation in GK islets occurred during the early stage. To com- ing pathway, metabolism of xenobiotics by cytochrome P450, pensate for this energy deficiency, GK islets improved the etc.) were downregulated in GK islets after NAC treatment rate of glycolysis and upregulated the expression of Ldha, (Supplementary Fig. 7F). Thus, the neutralization of in- which converted pyruvate to lactate and further disrupted creased ROS by NAC ameliorates its impact on GSIS and the link between cytosolic glycolysis and mitochondrial me- fl in ammation and protects GK b-cell function. tabolism (Fig. 4). Anaerobic glucose metabolism with NADH accumulation in the b-cell of mitochondrial diabetes caused DISCUSSION by ethidium bromide treatment can impair the transcription of We carried out a large-scale analysis of gene and protein mitochondria DNA, halt the TCA cycle, and affect GSIS (47). dynamics in pancreatic islets of GK rats at various stages Therefore, this Warburg-like metabolic shift may contribute of T2D. Combined transcriptome and proteome analysis to the early impairment of GSIS in GK b-cells because GSIS revealed sufficient depth of coverage and quantitative requires the production of both a triggering signal (ATP) accuracy to generate functional portraits of healthy and and amplifying signals (i.e., cAMP, short-chain acyl-CoA com- diseased pancreatic islets with unprecedented detail. Many pounds, NADPH) produced during aerobic metabolism (48). DE genes and proteins identified previously were confirmed Despite the early reduction in b-cell mass, insulin and in this study, such as those associated with OXPHOS, the enzymes required for proinsulin processing were com- mitochondrial function, metabolism, insulin secretion, oxi- parableorevenhigheratboththemRNAandtheprotein dative stress, and inflammation, thus validating the analytic level at 4–6 weeks, suggesting compensation during the diabetes.diabetesjournals.org Hou and Associates 2199 early phase in response to hyperglycemia. Absolute insulin Duality of Interest. No potential conflicts of interest relevant to this article insufficiency only occurred during the late stage of T2D. were reported. Consistently, we also observed compensation for OXPHOS Author Contributions. J.H. performed the proteomic experiments and MS fi complexes and mitochondrial machinery at the mRNA level data analysis. J.H. and W.Z. prepared the gures. J.H., W.Z., and D.Z. carried out the bioinformatics analyses. J.H., Y.W., and T.X. wrote the manuscript with help from the at 4–8 weeks, despite reduced protein levels. Thus, the pro- other authors. Z.L. performed the RNA-seq experiments and immunohistochemistry gression of T2D in GK rats from 4 to 24 weeks can be divided – imaging. Q.H. carried out rat breeding. L.L. and L.W. performed the qRT-PCR and into two stages: compensation (4 6 weeks) and compensation GSIS experiments. Y.W. carried out the islet preparation and animal experiments. – failure (8 24 weeks). Further data mining to examine tem- Y.W. and T.X. conceived the project. All authors read the manuscript and discussed poral expression patterns will help to elucidate the mecha- the interpretation of results. T.X. is the guarantor of this work and, as such, had full nisms underlying compensation and decompensation. access to all the data in the study and takes responsibility for the integrity of the data Mitochondrial dysfunction and the Warburg effect gen- and the accuracy of the data analysis. erate greater ROS invasion, which in turn induces chronic low-grade inflammation (49). Of note, proinflammatory References cytokines also exhibit two distinct stages (Fig. 7): a priming 1. Saisho Y. b-Cell dysfunction: its critical role in prevention and management of stage at 4–6weeksandanamplification stage after 8 weeks. type 2 diabetes. World J Diabetes 2015;6:109–124 Given that no inflammatory cell infiltration was observed in 2. Turner RC. The U.K. Prospective Diabetes Study. A review. Diabetes Care 1998; – GK islets before 8 weeks (39), the priming stage of inflam- 21(Suppl. 3):C35 C38 mation was likely induced by intracellular signals generated 3. Cnop M, Welsh N, Jonas JC, Jorns A, Lenzen S, Eizirik DL. Mechanisms of pancreatic beta-cell death in type 1 and type 2 diabetes: many differences, few by metabolic stress, such as the presence of increased ROS – fl similarities. Diabetes 2005;54(Suppl. 2):S97 S107 or free fatty acids. However, the second stage of in amma- 4. Del Prato S, Bianchi C, Marchetti P. b-Cell function and anti-diabetic phar- fi tion ampli cation may be induced by a complex combina- macotherapy. Diabetes Metab Res Rev 2007;23:518–527 tion of intracellular and extracellular (i.e., macrophage 5. Brun T, Li N, Jourdain AA, et al. Diabetogenic milieus induce specificchanges infiltration) inducers. The current data provide clues to in mitochondrial transcriptome and differentiation of human pancreatic islets. Hum unravel the mechanism underlying the initiation and am- Mol Genet 2015;24:5270–5284 plification of sterile inflammation; understanding this mech- 6. Fonseca VA. Defining and characterizing the progression of type 2 diabetes. anism is necessary to developing novel anti-inflammation Diabetes Care 2009;32(Suppl. 2):S151–S156 therapies to treat T2D. Islet inflammation is undoubtedly 7. Weir GC, Bonner-Weir S. Five stages of evolving beta-cell dysfunction during – an early event during T2D pathogenesis, but it is not likely a progression to diabetes. Diabetes 2004;53(Suppl. 3):S16 S21 8. King AJ. The use of animal models in diabetes research. Br J Pharmacol 2012; causal event because IL-1b and TXNIP were not significantly 166:877–894 expressed at 4 weeks, although they gradually increased 9. Goto YKM, Masaki N. Spontaneous diabetes produced by selective breeding of during later stages of the disease. normal Wistar rats. Proc Jpn Acad 1975;51:5 In summary, the data reveal two stages during the 10. Akash MS, Rehman K, Chen S. Goto-Kakizaki rats: its suitability as non-obese progression of T2D in GK islets. The early stage (4–6weeks)is diabetic animal model for spontaneous type 2 diabetes mellitus. Curr Diabetes Rev characterized by anaerobic glycolysis, inflammation priming, 2013;9:387–396 and compensation for insulin synthesis, whereas the late stage 11. Movassat J, Saulnier C, Serradas P, Portha B. Impaired development of pan- (8–24 weeks) is characterized by inflammation amplification creatic beta-cell mass is a primary event during the progression to diabetes in the GK and compensation failure (Supplementary Fig. 8). We did rat. Diabetologia 1997;40:916–925 not observe significant apoptosis during the early stage. 12. Plachot C, Movassat J, Portha B. Impaired beta-cell regeneration after partial The apoptosis pathway was only significantly elevated at pancreatectomy in the adult Goto-Kakizaki rat, a spontaneous model of type II di- abetes. Histochem Cell Biol 2001;116:131–139 24 weeks at the mRNA level. The time course transcriptome 13. Fox-Walsh K, Davis-Turak J, Zhou Y, Li H, Fu XD. A multiplex RNA-seq strategy and proteome data sets for GK rat islets depict a compre- to profile poly(A+) RNA: application to analysis of transcription response and 39 end hensive landscape of dynamic changes in gene expression at formation. Genomics 2011;98:266–271 various stages of diabetes, representing a valuable resource 14. Dayon L, Hainard A, Licker V, et al. Relative quantification of proteins in human for the research community to further explore the molecu- cerebrospinal fluids by MS/MS using 6-plex isobaric tags. Anal Chem 2008;80: lar etiology and progression of diabetes. In-depth explora- 2921–2931 tion of this resource will aid in the discovery of potential 15. Zhang Y, Wen Z, Washburn MP, Florens L. Refinements to label free proteome diagnostic and therapeutic targets for human T2D. quantitation: how to deal with peptides shared by multiple proteins. Anal Chem 2010; 82:2272–2281 16. Risso D, Ngai J, Speed TP, Dudoit S. Normalization of RNA-seq data using Acknowledgments. The authors thank the staff of the Institute of Biophysics factor analysis of control genes or samples. Nat Biotechnol 2014;32:896–902 Core Facilities, in particular, Yan Teng for technical support with confocal imaging, 17. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of Dr. Jifeng Wang for MS operation, and Zhen Fan and Xiaowei Chen for RNA-seq large gene lists using DAVID bioinformatics resources. Nat Protoc 2009;4:44–57 design and data collection. 18. Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths Funding. This work was supported by grants from the National Key Basic toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res Research Project of China (2014CB910503 and 2015CB910303), the National Key 2009;37:1–13 Research and Development Program of China (2016YFC0903301), the Strategic 19. Luo W, Friedman MS, Shedden K, Hankenson KD, Woolf PJ. GAGE: generally Priority Research Program of the Chinese Academy of Sciences (XDA12030101), and applicable gene set enrichment for pathway analysis. BMC Bioinformatics 2009; the National Science Foundation of China (31421002, 31400703, and 31400658). 10:161 2200 Deteriorating Pancreatic Islets in T2D Diabetes Volume 66, August 2017

20. Vizcaino JA, Csordas A, del-Toro N, et al. 2016 update of the PRIDE database 36.ChaveyA,BailbeD,MaulnyL,RenardJP,MovassatJ,PorthaB.A and its related tools. Nucleic Acids Res 2016;44:D447–D456 euglycaemic/non-diabetic perinatal environment does not alleviate early beta cell 21. Zhang B, Wang J, Wang X, et al. Proteogenomic characterization of human maldevelopment and type 2 diabetes risk in the GK/Par rat model. Diabetologia colon and rectal cancer. Nature 2014;513:382–387 2013;56:194–203 22. Gry M, Rimini R, Stromberg S, et al. Correlations between RNA and protein 37. Khadra A, Schnell S. Development, growth and maintenance of beta-cell mass: expression profiles in 23 human cell lines. BMC Genomics 2009;10:365 models are also part of the story. Mol Aspects Med 2015;42:78–90 23. Guenifi A, Abdel-Halim SM, Hoog A, Falkmer S, Ostenson CG. Preserved beta- 38. Wang P, Fiaschi-Taesch NM, Vasavada RC, Scott DK, Garcia-Ocana A, Stewart cell density in the endocrine pancreas of young, spontaneously diabetic Goto- AF. Diabetes mellitus–advances and challenges in human beta-cell proliferation. Kakizaki (GK) rats. Pancreas 1995;10:148–153 Nat Rev Endocrinol 2015;11:201–212 24. Schwanhausser B, Busse D, Li N, et al. Global quantification of mammalian 39. Calderari S, Irminger JC, Giroix MH, et al. Regenerating 1 and 3b gene ex- gene expression control. Nature 2011;473:337–342 pression in the pancreas of type 2 diabetic Goto-Kakizaki (GK) rats. PLoS One 25. Merico D, Isserlin R, Stueker O, Emili A, Bader GD. Enrichment map: a network- 2014;9:e90045 based method for gene-set enrichment visualization and interpretation. PLoS One 40. Devi TS, Lee I, Huttemann M, Kumar A, Nantwi KD, Singh LP. TXNIP links 2010;5:e13984 innate host defense mechanisms to oxidative stress and inflammation in retinal 26. Portha B, Giroix MH, Tourrel-Cuzin C, Le-Stunff H, Movassat J. The GK rat: a Muller glia under chronic hyperglycemia: implications for diabetic retinopathy. Exp prototype for the study of non-overweight type 2 diabetes. Methods Mol Biol 2012; Diabetes Res 2012;2012:438238 933:125–159 41. Ihara Y, Toyokuni S, Uchida K, et al. Hyperglycemia causes oxidative stress in 27. Portha B, Lacraz G, Chavey A, et al. Islet structure and function in the GK rat. In pancreatic beta-cells of GK rats, a model of type 2 diabetes. Diabetes 1999;48:927– Islets of Langerhans. Islam MS, Ed. Dordrecht, Springer Netherlands, 2015, p. 743– 932 765 42. Cantley J, Biden TJ. Sweet and sour beta-cells: ROS and Hif1alpha induce 28. Giroix MH, Saulnier C, Portha B. Decreased pancreatic islet response to Warburg-like lactate production during type 2 diabetes. Diabetes 2013;62:1823– L-leucine in the spontaneously diabetic GK rat: enzymatic, metabolic and secretory 1825 data. Diabetologia 1999;42:965–977 43. Ehses JA, Lacraz G, Giroix MH, et al. IL-1 antagonism reduces hyperglycemia 29. Homo-Delarche F, Calderari S, Irminger JC, et al. Islet inflammation and fibrosis and tissue inflammation in the type 2 diabetic GK rat. Proc Natl Acad Sci U S A in a spontaneous model of type 2 diabetes, the GK rat. Diabetes 2006;55:1625–1633 2009;106:13998–14003 30. Ostenson CG, Abdel-Halim SM, Rasschaert J, et al. Deficient activity of FAD-linked 44. Yoon KH, Ko SH, Cho JH, et al. Selective beta-cell loss and alpha-cell expansion glycerophosphate dehydrogenase in islets of GK rats. Diabetologia 1993;36:722–726 in patients with type 2 diabetes mellitus in Korea. J Clin Endocrinol Metab 2003;88: 31. Gheni G, Ogura M, Iwasaki M, et al. Glutamate acts as a key signal linking 2300–2308 glucose metabolism to incretin/cAMP action to amplify insulin secretion. Cell Reports 45. Butler AE, Janson J, Bonner-Weir S, Ritzel R, Rizza RA, Butler PC. Beta-cell 2014;9:661–673 deficit and increased beta-cell apoptosis in humans with type 2 diabetes. Diabetes 32. Sener A, Malaisse WJ. The stimulus-secretion coupling of amino acid-induced 2003;52:102–110 insulin release. Insulinotropic action of L-alanine. Biochim Biophys Acta 2002;1573: 46. Matveyenko AV, Veldhuis JD, Butler PC. Mechanisms of impaired fasting glu- 100–104 cose and glucose intolerance induced by an approximate 50% pancreatectomy. 33. Yang J, Dolinger M, Ritaccio G, et al. Leucine stimulates insulin secretion via Diabetes 2006;55:2347–2356 down-regulation of surface expression of adrenergic alpha2A receptor through the 47. Noda M, Yamashita S, Takahashi N, et al. Switch to anaerobic glucose me- mTOR (mammalian target of rapamycin) pathway: implication in new-onset diabetes tabolism with NADH accumulation in the beta-cell model of mitochondrial diabetes. in renal transplantation. J Biol Chem 2012;287:24795–24806 Characteristics of betaHC9 cells deficient in mitochondrial DNA transcription. J Biol 34. Sasaki M, Fujimoto S, Sato Y, et al. Reduction of reactive oxygen species Chem 2002;277:41817–41826 ameliorates metabolism-secretion coupling in islets of diabetic GK rats by sup- 48. Prentki M, Matschinsky FM, Madiraju SR. Metabolic signaling in fuel-induced pressing lactate overproduction. Diabetes 2013;62:1996–2003 insulin secretion. Cell Metab 2013;18:162–185 35. Lacraz G, Figeac F, Movassat J, et al. Diabetic beta-cells can achieve self- 49. Talero E, Garcia-Maurino S, Avila-Roman J, Rodriguez-Luna A, Alcaide A, protection against oxidative stress through an adaptive up-regulation of their anti- Motilva V. Bioactive compounds isolated from microalgae in chronic inflammation oxidant defenses. PLoS One 2009;4:e6500 and cancer. Mar Drugs 2015;13:6152–6209