MANF protects human pancreatic beta cells against stress-induced cell death

Aims/hypothesis There is a great need to identify factors that could protect pancreatic beta cells against apoptosis or stimulate their replication and thus prevent or reverse the development of diabetes. One potential candidate is mesencephalic astrocyte-derived neurotrophic factor (MANF), an endoplasmic reticulum (ER) stress inducible protein. Manf knockout mice used as a model of diabetes develop the condition because of increased apoptosis and reduced proliferation of beta cells, apparently related to ER stress. Given this novel association between MANF and beta cell death, we studied the potential of MANF to protect human beta cells against experimentally induced ER stress. Methods Primary human islets were challenged with proinflammatory cytokines, with or without MANF. Cell viability was analysed and global transcriptomic analysis performed. Results were further validated using the human beta cell line EndoC-βH1. Results There was increased expression and secretion of MANF in human beta cells in response to cytokines. Addition of recombinant human MANF reduced cytokine-induced cell death by 38% in human islets (p < 0.05). MANF knockdown in EndoC-βH1 cells led to increased ER stress after cytokine challenge. Mechanistic studies showed that the protective effect of MANF was associated with repression of the NF-κB signalling pathway and amelioration of ER stress. MANF also increased the proliferation of primary human beta cells twofold when TGF-β signalling was inhibited (p < 0.01). Conclusions/interpretation Our studies show that exogenous MANF protein can provide protection to human beta cells against death induced by inflammatory stress. The antiapoptotic and mitogenic properties of MANF make it a potential therapeutic agent for beta cell protection. Electronic supplementary material The online version of this article (10.1007/s00125-018-4687-y) contains peer-reviewed but unedited supplementary material, which is available to authorised users.

focal points of two experiments followed by counting total nucleus. Each image contained approximately 200-1000 cells.

Immunostaining and Immunoblotting of EndoC-βH1 cells
Immunohistochemistry was carried out as previously described [3]. In brief, the human pancreatic sections were deparaffinized and rehydrated using routine protocols. Sections were then treated with 1 mM EDTA buffer (pH 8) in a microwave oven to reveal the antigenic sites. Blocking was done using Ultra V Block (Thermo Scientific, Waltham MA, USA) for 10 min at RT to block non-specific binding sites followed by overnight incubation at 4°C with primary antibodies (ESM Table 2) diluted in PBS containing 0.1% Tween 20 (vol./vol). Nuclear staining was performed with DAPI (Vector Laboratories, Burlingame, CA, USA). Triple staining without nuclear staining were mounted with Vectashield (Vector Laboratories).
EndoC-βH1 cells were cultured on Matrigel and fibronectin coated 24-well TPP tissue culture plate or on glass coverslips and immunostained as described previously [4]. For primary antibodies see ESM Table 2. Hoechst 33342 was used to counterstain the nuclei. Images were acquired under EVOS fluorescence microscope (Life Technologies) or Zeiss AxioImager-3 microscope with Apotome and processed using Adobe Photoshop and ImageJ software.
Equal proteins (25 µg) were resolved by Any kD Mini-Protean -TGX gel (Bio-Rad), immunoblotted with corresponding antibodies as described in ESM Table 2 as described previously 4. Membranes were incubated with species-specific HRPlinked secondary antibodies (1:5000) and visualization was performed following ECL exposure with ChemiDoc XRS+ system and Image Lab Software (BioRad). A loading control was performed on the same blot for all western blot data. For MANF WB, cells were directly incubated with Laemmli buffer on ice for 30 min followed by 5 min heating at 95 °C then separated by SDS-PAGE. Densitometric analysis of bands from image were calculated using Image J (Media Cybernetics) software and intensities compared as phospho-p65 to actin; phospho-ERK and phospho-AKT to tubulin and MANF to GAPDH.

Quantitative RT-PCR
cDNA was synthesized using the random hexamer priming of the High Capacity cDNA Reverse Transcription kit according to the manufacturers recommendations (Applied Biosystems, Foster City, CA). The method for quantitative RT-PCR has been described previously [5]. Briefly, SYBR Green JumpStart Taq Ready Mix for quantitative PCR (Sigma-Aldrich) was used for the reactions with a Corbett Rotor-Gene 6000 (Qiagen, Hilden, Germany). The reactions were pipetted with a liquid handling system (Corbett CAS-1200, Qiagen). All reactions were performed in duplicates on at least three biological replicates. The median Ct values were used for 2 -ΔΔCt analysis. Cyclophilin G was used as an endogenous control. An exogenous positive control was used as a calibrator between all the real-time PCRs. Primer sequences for MANF, GRP78 (also known as HSPA5), CHOP (also known as DDIT3), sXBP1, ATF4, ATF6, ATF3, PreINS, INS, PDX1, MAFA, CyclophilinG (also known as PPIG), BCL10, Ki67, CDK1 and CDK4 are presented in ESM Table 3.

MANF Assay
For the quantitation of secreted MANF protein, the conditioned medium samples from human islets and EndoC-βH1 cells were centrifuged at 5000 rpm for 5 min, and the supernatants were analyzed on an in-house sandwich ELISA specific for human MANF [6]. The dynamic range of the assay was 62.5 to 2000 pg/ml and the sensitivity was 45 pg/ml. Intra-and inter-assay coefficients of variation were 8.1% and 5.5%, respectively. For the measurement of cellular MANF cells were lysed with non-SDS lysis buffer TETG solution contains 20 mM Tris pH 8.0; 0.1% Triton X-100; 1% Glycerol; 137 mM NaCl; 2 mM EGTA and anti-protease tablet (Roche) for 30 min on ice. The lysate was next centrifuge at 5000 rpm for 5 min and store at -20 °C until MANF ELISA and BCA assay for total protein measurement. Total cellular MANF is presented as ng /ml /1.5 x 10 5 seeded cells after value corrected for the total protein content.
Apoptotic cells were quantified with BD Annexin-V: FITC Apoptosis Detection Kit (#556547, BD Biosciences, San Jose, CA, USA) following manufacturer's instructions. Cells were analyzed by flow cytometry, using BD Accuri C6 (BD Biosciences) and at least 10,000 events were collected per sample. Data was analyzed using BD Accuri C6 analysis software (version 1).

RNA Sequencing using STRT protocol
For RNA sample collection, the islets were cultured in suspension in non-adherent plates in Ham's F10 medium supplemented with 0.5% BSA, penicillin (100 IU/ml) and streptomycin (100 µg/ml). Total RNA from human islets was isolated using the RNeasy mini kit (Qiagen) according to the manufacturer's instructions. The samples used for transcriptome analysis were evaluated by Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA). The RNA integrity number (RIN) values for all samples was >8.0.
High-quality total-RNA (10 ng) was taken from the islets of six organ donors from the conditions described above (control, MANF, four cytokines and four cytokines with MANF) and processed according to highly multiplexed Single-cell Tagged Reverse Transcription (STRT) RNA sequencing method [7,8]. Instead of processing as single cells, the method was used for generating barcoded cDNA libraries from low quantity (10ng) RNA samples. Poly-dT primers were used to enrich polyA+ mRNAs for sequencing from total RNA samples. High-quality 10 ng total-RNA was extracted from the islets of six organ donors from four conditions: control, MANF, four cytokines and four cytokines with MANF. The samples were processed with Singlecell Tagged Reverse Transcription (STRT) RNA sequencing protocol [7,8] with minor modifications. Briefly, one microliter of total-RNA (10 ng/µl) samples were added to forty-eight barcoded plate where four microliters of cell capture buffer, contained 0.1% Triton X-100, 800 nM T30-VN-oligo, 2 mM deoxy-nucleotide (dNTP) mix, and 2 mM template-switching oligonucleotide (TSO) without magnesium chloride, was previously prepared. ERCC spike-in Mix A was diluted 1,000 times before adding 1 µl to 48-plex reverse transcriptase master mix. After cDNA synthesis, all 48 cDNAs were pooled into one 2-mL tube using 10% PEG-6000 and 0.9 mol NaCl (final concentration). The purified cDNA was first amplified using 14 cycles of PCR and later an additional 10 cycles to introduce a complete set of adapters for Illumina single read sequencing. Ready library was size-selected using sequential AMPure XP bead selection protocol where 0.73 and 0.223 bead/PCR product ratios were used. Ready library was analyzed on three lanes of Illumina HiSeq2000 instrument.

RNA sequencing data processing and gene expression analysis
Data processing of the sequenced RNA libraries was performed using the STRTprep version v3dev pipeline (https://github.com/shka/STRTprep; commit 6389622).
Briefly, the reads were de-multiplexed into individual samples using the samplespecific barcodes. Redundant reads were excluded according to unique molecular identifiers (UMIs), then mapped to the human genome assembly hg19/GRCh37 with RefSeq annotations [9] using Bowtie v. 1.1.0 [10] and Tophat v.2.0.12 [11]. For quality control, samples with low mapped read counts (< 200,000 reads/sample), high redundancy (>10), shallow spike-in counts (< 700 reads/sample), low spike-in map rate (< 90%), and low map rate to transcript start sites (< 70%) were excluded from subsequent analyses. The read counts were normalized to relative amounts compared to total spike-in counts. Differential expression analysis was performed using SAMstrt [12]. In addition to differential expression significance between control and patient samples, only transcripts with more biological variation than the background technical noise was considered as significant. Variation caused by technical noise was estimated from technical replicates using a generalized linear model with a gamma distribution, as described in [8].
For identification of genes and pathways specific to MANF rescue effects, a different pipeline was used. Mapped RNA reads were obtained from the pipeline as described in the previous section. Genes with low counts in more than three samples were removed before the identification of differentially expressed genes using the  transfected with siNT or siMANF for 72h, further exposed to cytokine cocktail II or left untreated for 24h and BCL10 expression was determined. qRT-PCR data were normalized to Cyclophilin G and plotted as fold over siNT-control cells.