Evidence of stress in β cells obtained with laser capture microdissection from pancreases of brain dead donors

ABSTRACT

Isolated islets used for transplantation are known to be stressed, which can result from the circumstances of death, in particular brain death, the preservation of the pancreas with its warm and cold ischemia, from the trauma of the isolation process, and the complex events that occur during tissue culture. The current study focused upon the events that occur before the islet isolation procedure. Pancreases were obtained from brain dead donors (n = 7) with mean age 50 (11) and normal pancreatic tissue obtained at surgery done for pancreatic neoplasms (n = 7), mean age 69 (9). Frozen sections were subjected to laser capture microdissection (LCM) to obtain β-cell rich islet tissue, from which extracted RNA was analyzed with microarrays. Gene expression of the 2 groups was evaluated with differential expression analysis for genes and pathways. Marked changes were found in pathways concerned with endoplasmic reticulum stress with its unfolded protein response (UPR), apoptotic pathways and components of inflammation. In addition, there were changes in genes important for islet cell identity. These findings advance our understanding of why islets are stressed before transplantation, which may lead to strategies to reduce this stress and lead to better clinical outcomes.

Introduction

The efficacy of islet transplantation continues to improve,¹ but concerns remain about changes found in islets isolated from deceased donors that make them vulnerable to injury and loss of β cells with the transplantation. It has long been known that kidneys and livers obtained from deceased donors undergo deleterious changes including inflammation that lead to less successful transplant outcomes.^2,3 Similar findings have been noted in isolated islets. For example, isolated human islets produce tissue factor (TF), which can trigger a thrombin reaction, called the instant blood-mediated inflammatory reaction (IBMIR), which is associated with cell loss when islets are transplanted into the portal vein.^4,5 There has been particular concern about the health of islets obtained from pancreases of brain-dead donors. Toyama et al.⁶ found that after brain injury in rats the isolated islets were infiltrated with macrophages and had increased gene expression of interleukin 1- β (IL1-β), interleukin 6 (IL6), Tumor necrosis factor-α (TNF-α), and monocyte chemoattractant protein-1 (MCP-1). Contreras et al. (7) performed similar experiments in rats and found that plasma measurements of TNF-α, IL1-β and IL6 were elevated and that gene expression of these same factors was elevated in pancreatic tissue. Moreover, there were increased numbers of apoptotic events in β cells as determined by Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling (TUNEL) staining. In another study, Saito et al.⁸ found that isolated islets from brain dead rats had increased gene expression of tissue factor (TF) and mono chemoattractant protein-1 (MCP-1) in isolated islets compared with islets obtained from control rats not subjected to brain injury. Isolated human islets have also been found to produce MCP-1, which could contribute to inflammation.⁹

To understand the threat to isolated islets, 4 points of variability must be considered: 1) brain injury and the stress of end of life care, 2) the combined detrimental effects of warm and cold ischemia of the harvested pancreas, 3) the trauma of the islet isolation process, and 4) the changes that occur when the isolated islets are cultured before transplantation.

To better understand the contribution of these stages, we sought to understand the changes in islets that occur before the isolation procedure, that is to determine the combined effects of end-of-life events including brain death and organ preservation. This question was examined by comparing gene expression in islets contained in deceased human donor pancreases with those in fresh pancreatic tissue obtained at surgery. A particular strength of this study is that LCM was used to obtain β cell rich tissue samples from frozen sections, which could then be analyzed for gene expression using Affymetrix gene arrays.

Methods and materials

Characteristics and preparation of pancreas specimens

Frozen blocks of 7 surgical pancreas specimens were provided by the Department of Pathology of Massachusetts General Hospital, and 7 cadaver pancreas were obtained from the New England Organ Bank. All of the samples were processed and evaluated at the Joslin Diabetes Center. The patients underwent pancreatic surgery for concerns about pancreatic cancer. We excluded specimens with evidence of pancreatitis, history of diabetes or glucose intolerance. The patient characteristics are shown in Table 1. All subjects were Caucasian; their mean age was 69.4 (8.7) years and body mass index (BMI) was 25.7 (4.3) kg/m². To select the 7 pancreas specimens from deceased donors, donors who were hyperglycemic before isolation of pancreas were excluded. One subject had a glucose of 248 mg/dl upon taken at admission, but the peak after this was recorded as 159 mg/dl. All donors were Caucasian, and their mean age was 50.4 (11.4) years and BMI was 25.6 (4.8) kg/m². Frozen pancreatic tissue was sectioned at 8 μm in a cryostat, mounted on uncoated glass slides at −20°C, and immediately stored at −80°C.

Table 1.

Surgical samples and cadaveric samples.

Donor	Age	Sex	BMI	BG	Diagnosis / Cause of Death
1	54	M	34	121	Intracerebral Hemorrhage
2	38	F	22	117	Motor Vehicle Accident & CA
3	57	M	33	133	Head trauma & CA
4	44	M	27	248	Asphyxia
5	63	M	29	135	Intracerebral Hemorrhage
6	62	F	22	136	Subarachnoid Hemorrhage
7	35	F	23	137	Intracerebral Hemorrhage
9*	81	M	23	92	Adenocarcinoma
8*	60	M	24	85	Adenocarcinoma
10*	67	F	26	NA	Cystadenoma
11*	72	M	35	92	Adenocarcinoma
12*	72	M	22	81	Tubular Adenoma
13*	57	F	24	90	Neoplasm
14*	77	F	26	91	Neoplasm

Starred samples are the surgical specimens. Blood glucose (BG mg/dL) Diagnosis/Cause of Death: Cardiac Arrest (CA), BG in Cadavers are the initial recording.

Laser capture microdissection (LCM)

Prior to LCM, frozen pancreatic sections were dehydrated in 70% ethanol for 30 seconds, diethylpyrocarbonate (DEPC)-treated water for 30 seconds, 70% ethanol for 30 seconds, 100% ethanol twice for 1 min, and xylene for 4 minutes. Slides were immediately air-dried. LCM was performed using PixCell II Laser Capture Microdissection System (Arcturus Engineering, Mountain View, CA). The dissections were performed by melting thermoplastic films mounted on transparent LCM caps (Arcturus) targeting selected populations of cells with intrinsic fluorescence.¹⁰ To obtain optimal size of the laser pulse, the system parameters were set as follows: laser power 35 mW; pulse duration 3.0 msec; and spot size 7.5 μm. Each LCM session was completed within 30 minutes to avoid RNA degradation. The cells attached to the thermoplastic transfer film were incubated with 10 μl of a guanidine thiocyanate and polyethylene glycol octylphenol ether based buffer for 30 minutes at 42°C. Each section typically had 3–10 islets, and each islet typically contained 2–4 clumps of intrinsic fluorescence representing β-cells. On average, 6–7 sections were used to obtain 800 pulses, which were needed to obtain sufficient RNA for arrays.

RNA extraction, amplification and labeling

Total RNA was extracted from the laser-captured cells using PicoPure RNA Isolation Kit (Arcturus). DNase treatment was performed for 15 minutes using the RNase-free DNase Set (Qiagen, Valencia, CA) to reduce risk of DNA interference. The extracted RNA was amplified by T7-based linear amplification using T7-oligo-dT-primers. Two rounds of amplification were performed using RiboAmp HS RNA Amplification Kits (Arcturus). The quantity and quality of amplified RNA were determined by measuring absorbance at 260 nm and 280 nm using NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE). Biotinylated complementary RNA (cRNA) was generated from cDNA by in vitro transcription reaction using the BioArray High Yield RNA Transcript Labeling Kit (Enzo Diagnostics, Farmingdale, NY). RNA products were then purified using the MiraCol Purification Columns (Arcturus).

Microarray profiling

For comparison of gene expression profiles, single RNA preparations were run on the GeneChip® Human X3P Array (Affymetrix, Santa Clara, CA), containing 61,000 probe sets representing 47,000 transcripts. Ten μg of biotinylated RNA were fragmented to nucleotide stretches of 30–200 nucleotides and hybridized to the chips by the Joslin Diabetes Center DRC Genomics Core.

Determination of β-cell enrichment

To clarify the degree of β–cell enrichment in β–cell samples isolated by LCM, insulin, glucagon, and somatostatin transcripts were quantified by real-time PCR. We measured the expression of these genes in samples obtained by dissected clumps of auto fluorescent cells (β–cells) from within islets compared with samples obtained from randomly dissected islet cells. The LCM avoided acinar tissue. Three pancreases were used for this comparison. Total RNA was extracted as described previously,¹ and RNA quantity and purity were evaluated by NanoDrop ND-1000 Spectrophotometer. cDNA templates were prepared from 35.5 ng of total RNA with random hexamers using TaqMan Reverse Transcription Reagents (Applied Biosystems, Foster City, CA). The housekeeping gene ribosomal protein S18 (RPS18) transcript was selected as reference. ΔCT values of glucagon or somatostatin in the β–cell samples were used to calculate the differences between glucagon or somatostatin and insulin gene expression in β–cell or islet samples using the ΔΔCT method. The fold changes obtained were used to calculate ratios of expression of glucagon or somatostatin to insulin in β–cell enriched and randomly selected islet cells.

Data analysis

All array data were normalized by Robust Multi-Array Average (RMA).¹¹ Control probes (whose label begins with “AFFX”) were removed. To visualize whether the data significantly clustered into cadaver and donor groups, we plotted the samples according to their first 2 principal components from principal component analysis (PCA), but visual inspection was inconclusive (Fig. 1). To test the clustering formally, we calculated a clustering score (using the Calinski-Harabasz Index¹²), and compared this score to those scores calculated from 10,000 permutations of the group labels. This analysis showed that the arrangement of the 2 groups is unlikely to be random (Fig. 2). We moved forward with the analysis by identifying differentially expressed genes between these 2 groups using the limma package¹³ accounting for empirical array quality weights.¹⁴ Then, we looked for genes in particular that were related to stress and inflammation. In addition, we ran a pathway analysis against the Broad Institute's Molecular Signature Database's (MSigDB)¹⁵ Hallmark pathways, GO terms gene sets, C2 curated gene sets, and C2: Biological Processes independently using the sigPathway package.¹⁶ Analyses were conducted in the R/Bioconductor software (www.r-project.org) by the Joslin DRC Bioinformatics Core.

Figure 1.

Principal component analysis did not show differential clustering.

Figure 2.

Calinski-Harabasz index describes the ratio of between-cluster variance to within-cluster variance. calinski-Harabasz clustering index between our data and 10,000 permutations of the data showed that although these groups are close, the arrangement of the groups is far from random. Calinski-Harabasz Index describes the ratio of between-cluster variance to within cluster variance.

Data presentation

Data are presented as mean and standard error of the mean [mean(SEM)]. In table 3–5, upregulation is designated with a plus symbol and means upregulated in cadaver pancreases as compared with surgical pancreases.

Table 3.

Genes associated with er stress, UPR, inflammation, and apoptosis.

Gene Title	Gene Name	p-value	Up/ Down	Function	INF	ER/ UPR	APO
CD9	tetraspanin-29	3.92E-04	4.27	regulates and prevents inflammation	+		+
HSP90B1	heat shock protein 90 Da	1.13E-04	4.25	molecular chaperone, PI3K-AKT-mTOR survival signaling		+
HSPA5	heat shock protein family A (Hsp70) member 5	1.56E-04	3	folding proteins, monitors protein transport, ER stress		+
CEBPD	CCAAT/enhancer binding protein (C/EBP), delta	1.00E-04	2.47	inflammation, growth arrest, apoptosis	+		+
RHOB	Ras homolog	1.06E-03	2.38	promotes intraluminal vesicle delivery	+	+
ARNTL	aryl hydrocarbon receptor nuclear translocator-like	1.90E-04	2.36	heterodimer with CLOCK	+	?
ZFP36	ZFP36 ring finger protein	7.29E-03	2.2	regulated by NF κ B, regulates TNF	+
CARD16	caspase recruitment domain family, member 16	8.97E-03	2.11	inflammation response and NF κ B activator, cell death, caspase 1	+		+
NAIP	NLR family, apoptosis inhibitory protein	0.013	2.1	apoptosis inhibition			+
KDELR3	endoplasmic reticulum protein retention receptor 3	3.92E-03	2.09	retains proteins in ER		+
CHAC1	ChaC glutathione specific gamma-glutamylcyclotransferase 1	2.78E-04	2.06	downstream of ATF4, ATF3 and CHOP, proapoptotic		+	+
DnaJC3	heat shock protein family (Hsp40) member C3	0.015	2.02	stress inducible protein		+
IL17RB	interleukin 17 receptor B	2.81E-04	1.99	ER induced apoptosis	+		+
SDF2L1	stromal cell derived factor 2 like 1	1.41E-03	1.96	ER stress-induced gene		+
SLC33A1	ER membrane transporter for acetyl-CoA	3.99E-03	1.93	regulates autophagy		+
DDIT4	DNA damage inducible transcript 4	1.06E-03	1.92	modulates apoptosis, mTOR related ∼ p53			+
ISG20	interferon stimulated exonuclease gene 20kDa	3.62E-03	1.88	degrades single stranded DNA and RNA, apoptotic agent			+
ETS2	vets avian erythroblastosis	3.48E-03	1.88	regulates apoptosis			+
IFNGR1	interferon gamma receptor 1	0.020	1.86	inflammation	+		+
F3	coagulation factor III (thromboplastin, tissue factor)	0.032	1.79	inflammation and cellular stress	+	+
ATF3	activating transcription factor 3 –cellular stress response	0.040	1.74	cellular stress response		+
DnaJB9	DnaJ heat shock protein family (Hsp40) member B9	6.00E-03	1.73	induced by ER stress, ER chaperone, related to NF κ B		+
HYOU 1	hypoxia upregulated 1 -HSP 70 family	1.29E-03	1.72	UPR pathway		+	+
NFKBIZ	nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, zeta	0.020	1.7	inflammation	+
PDIA6	protein disulfide isomerase family A member 6	3.86E-04	1.7	formation, reduction and isomerization of disulfide bonds		+
GADD45G	growth arrest and DNA-damage-inducible, gamma	8.95E-03	1.68	can activate JNK pathways			+
EIF2AK3	eukaryotic translation initiation factor 2-α kinase 3	7.96E-03	1.68	induced by ER stress		+
DERL 2	derlin 2	2.70E-04	1.66	detects misfolded proteins, ERAD complex		+
KDELR2	endoplasmic reticulum protein retention receptor 2	3.54E-04	1.65	retains proteins in ER		+
KDELR2	KDEL (Lys-Asp-Glu-Leu) endoplasmic reticulum protein retention receptor 2	3.54E-04	1.65	ER associated gene		+
DNAJB11	DnaJ (Hsp40) homolog, subfamily B, member 11	0.015	1.62	ER chaperone complex subunit		+
HILPDA	hypoxia inducible lipid droplet-associated	3.22E-03	1.59	stress response gene		+	+
IL13RA1	interleukin 13 receptor	4.63E-04	1.59	linked to IL4 ∼ related to JAK/STAT signaling	+		+
NFKBIA	nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor, α	0.011	1.58	Inflammation	+
LMAN1	lectin-mannose binding	2.02E-04	1.57	cycles between ER and Golgi		+
HIF3A	hypoxia inducible factor 3, α subunit	4.94E-03	1.56	mitochondrion response to hypoxia and stress		+	+
SERP1	stress-associated endoplasmic reticulum protein 1	0.017	1.55	UPR and ER and Golgi transport		+
IL6ST	interleukin 6 signal transducer	2.48E-03	1.55	transducer shared by many cytokines regulates apoptosis			+
SLC9A6	solute carrier family 9 member A6	0.018	1.54	mediator of ER stress-induced apoptosis		+	+
TNFRSF1A	tumor necrosis factor receptor superfamily member 1A	7.93E-03	1.51	regulates inflammation, mediate apoptosis	+		+
MCL1	myeloid cell leukemia 1	2.16E-03	1.51	anti-apoptotic gene – member of Bcl−2 family			+
PDCD4	programmed cell death 4 (neoplastic transformation inhibitor)	0.021	1.49	apoptosis and translation inhibitor			+
CASP6	caspase 6, apoptosis-related cysteine peptidase	0.019	1.45	regulates ER stress and apoptosis			+
VIMP	VCP interacting membrane protein	2.36E-03	1.45	central to ER stress and ERAD	+	+
SRPRB	signal recognition particle receptor	2.76E-04	1.44	ER response to apoptosis		+	+
BBC3	BCL2 binding component 3	5.92E-03	1.43	essential mediator of apoptosis with BCL−2		+	+
CALR	calreticulin, major calcium binding protein in the ER	5.51E-03	1.43	major Calcium binding protein in the ER		+
CALR	calreticulin	5.51E-03	1.43	ER associated gene		+
LDHA	lactate dehydrogenase A	0.048	1.42	LDHA is not expressed in normal β cells		+
DERL1	derlin 1	9.73E-03	1.4	detects misfolded proteins, ERAD complex		+
TNIP2	TNFAIP3 interacting protein 2	4.13E-03	1.38	inhibitor NF κ B and involved in apoptosis	+		+
COPB2			coatomer protein complex subunit β 2	2.32E-03	1.38	Important for Golgi budding		+
ERGIC1	endoplasmic reticulum-Golgi intermediate compartment (ERGIC) 1	1.03E-03	1.35	transportation between ER and Golgi		+
DAD1	defender against cell death	3.19E-03	1.27	anti-apoptotic			+
MAPK8IP3	mitogen-activated protein kinase 8 interacting protein 3	9.63E-03	−1.23	JNK related protein			+
WFS1			Wolfram Syndrome 1 (Wolframin)	0.015	−1.29	link to apoptosis			+
GGA2	golgi associated, gamma adaptin ear containing, ARF binding protein 2	2.64E-03	−1.49	regulates Golgi's traffic and lysosome		+
MX1	myxovirus (influenza virus) resistance 1, interferon-inducible protein p78 (mouse)	2.54E-04	−1.62	anti-cell-death and ER stress		+	+
IGF1R			insulin-like growth factor 1 receptor	1.74E-06	−1.88	anti-apoptotic enhancing the cell survival			+

ER stress and unfolded protein response (UPR) gene expression increased in β cells of cadaver pancreases compared those from surgical pancreases. These are designated as plus and minus symbols for fold change. These upregulated genes indicate major changes of genes involved with ER mechanisms, including ER Stress and the unfolded protein response (UPR). Also, a variety of gene concerned with apoptosis (APO) and cell viability are markedly upregulated in the β cells of cadaver pancreases. Genes associated with Inflammation upregulated in β cells of cadaver pancreases. Furthermore, genes that are associated with inflammation (INF) and stress response were observed to be affected in the cadaver group. A group of these genes is related to BCL2 and NFκB inflammatory molecules.

Table 4.

Islet cell identity genes.

Gene Title	Gene.Name	p-value	FC	Function
Differentially Expressed Genes
SLCO5A1	solute carrier organic anion transporter family, member 5A1	1.69E-05	2.98	differentiation
CHGA	chromogranin A (parathyroid secretory protein 1)	2.86E-03	−1.23	secretory cell marker
SYT13	synaptotagmin XIII	1.04E-04	−1.29	exocytosis
ABCC8	ATP-binding cassette, sub-family C (CFTR/MRP), member 8	0.011	−1.31	sulfonylurea – SUR1 receptor
STXBP1	syntaxin binding protein 1	1.62E-03	−1.33	exocytosis
PPDPF	pancreatic progenitor cell differentiation and proliferation factor	0.024	−1.33	cell differentiation and exocrine regeneration
SCAMP	secretory carrier membrane protein 1	4.96E-03	−1.38	secretory carrier membrane proteins
ISL1	ISL LIM homeobox 1	0.045	−1.39	differentiation
PAX6	paired box 6	0.013	−1.46	differentiation
NKX6–1	NK6 homeobox 1	1.77E-03	−1.51	β cell maturation
MAFB	v-maf avian musculoaponeurotic fibrosarcoma oncogene homolog B	4.18E-04	−1.52	β and α cell identity
KCNG3	potassium voltage-gated channel modifier subfamily G member 3	3.11E-03	−1.82	cell viability, nuclear migration speed, ER
ARX	aristaless related homeobox	1.17E-03	−2.02	α-cell marker
GRIN3A	glutamate receptor, ionotropic, N-methyl-D-aspartate 3A	3.90E-06	−3.76	related to ion channels
SLC6A4	solute carrier family 6 (neurotransmitter transporter), member 4, serotonin	3.38E-03	−3.92	insulin secretion
Non-differentially Expressed Genes
NKX2–2	NK2 homeobox 2	0.149	1.33	β cell development
SLC2A2	solute carrier family 2 (facilitated glucose transporter), member 2	0.625	1.11	glucose transporter 2 (GLUT2)
AKT3	v-akt murine thymoma viral oncogene homolog 3	0.805	1.1	key signaling pathway
NEUROD1	neuronal differentiation 1	0.816	−1.02	β cell identity
FOXA2	forkhead box A2	0.729	−1.03	important β cell transcription factor
SST	somatostatin	0.84	−1.03	Inhibits insulin and glucagon secretion
INS	insulin	0.515	−1.05	β cell product
GCG	glucagon	0.324	−1.06	stimulates gluconeogenesis and glycogenolysis
PCSK2	proprotein convertase subtilisin/kexin type 2	0.504	−1.07	proinsulin cleavage
NEUROG3	neurogenin 3	0.219	−1.08	Islet cell development
IAPP	islet amyloid polypeptide	0.408	−1.11	β cell secretory product
G6PC2	glucose-6-phosphatase, catalytic, 2	0.551	−1.13	important for neogenesis
GCK	glucokinase (hexokinase 4)	0.053	−1.34	glucose metabolism control

The expression of some of the key transcription factors for β cells are downregulated, but no changes were seen for insulin, glucagon or somatostatin expression.

Table 5.

Noteworthy genes.

Gene Title	Gene Name	p-value	Up/Down	Function
REG1B	regenerating islet-derived 1 β	2.78E-03	8.19	regeneration and β cell proliferation
REG1A	regenerating islet-derived 1 α	0.01	4.63	regeneration and β cell proliferation
SYT16	synaptotagmin XVI	4.86E-06	3.44	exocytosis
HYOU1	hypoxia upregulated 1	1.29E-03	1.72	related to heat shock 70 family, hypoxia-related stress and apoptosis
TM9SF3	transmembrane 9 superfamily member 3	5.68E-06	1.66	related to interferon
GRHPR	glyoxylate reductase/hydroxypyruvate reductase	1.43E-03	−1.41	metabolism
MRPL10	mitochondrial ribosomal protein L10	9.63E-05	−1.57	protein synthesis
NDRG1	N-myc downstream regulated 1	6.47E-05	−1.92	stress responses
STAT4	signal transducer and activator of transcription 4	1.15E-06	−2.21	important element for β cell dysfunction Induced by Inflammatory cytokines
ACSM1	acyl-CoA synthetase medium-chain family member 1	1.14E-05	−2.61	fatty acid synthesis
CCNB1	cyclin B1	6.41E-06	−2.90	cell cycle, cell division, growth arrest if downregulated

Results

Quantity and quality of RNA

RNA extraction and amplification from samples obtained by LCM yielded 16.9 (3.9) μg of aRNA with A260:A280 ratios ranging from 2.44 to 2.58. The A260:A280 ratios of the aRNA were within the acceptable limits and not different between surgical and deceased donor pancreas specimens. After amplification and labeling, the integrity of the RNA was evaluated by 3′/M’ ratio of the housekeeping gene (glyceraldehyde-3-phosphate dehydrogenase, GAPDH). The mean 3′/M’ ratio of GAPDH measured from surgical and cadaver pancreas specimens was 8.4 (1.1) and 7.8 (1.5), which were within the acceptable ranges.

β cell enrichment by LCM

LCM samples were obtained from selected β cell rich autofluorescent areas of islets or from generalized non-selected areas LCM of islet tissue. Real time PCR was performed on unamplified RNA to avoid artifacts from amplification. The ratio of GLUCAGON/INSULIN expression in β cell selected tissue to GLUCAGON/INSULIN expression in the generalized non-selected islet tissue was 0.238 (0.132). The ratio of SOMATOSTATIN/INSULIN expression in β cell selected tissue to SOMATOSTATIN/INSULIN expression in the generalized non-selected islet tissue was 0.584 (0.277). This means that the selected LCM samples had 74% less GLUCAGON expression than randomly isolated tissue and 42% less SOMATOSTATIN expression. This result is mostly in agreement with our earlier report finding about an 80% reduction in GLUCAGON expression and a 60% reduction in SOMATOSTATIN expression.¹⁰

Pathways for differential expression

In Table 2, the 9 most differentially expressed pathways are listed. The most notable changes are those associated with the endoplasmic reticulum, apoptosis, inflammation, and β cell identity. The most differentially expressed pathway was Hallmark: Glycolysis, which identifies a variety of changes that include genes related to chaperones and apoptosis.

Table 2.

GSEA pathway analysis.

Group	Noteworthy Pathways	Z-score	FDR
Cell Function	Hallmark: Glycolysis	4.05	0.003
UPR	Hallmark: Unfolded Protein Response	3.88	0.003
Stress	Hallmark: mTORC1 Signaling	3.62	0.004
Cell Identity	Hallmark: Pancreas β Cells	−3.58	0.004
Stress	Hallmark: TNFA Signaling via NF κ B	3.09	0.007
ER	GO: Endoplasmic Reticulum	3.84	0.009
Apoptosis	KEGG: Apoptosis	3.1	0.011
Hypoxia	Hallmark: Hypoxia	2.75	0.013
Apoptosis	Reactome: Intrinsic Pathway for Apoptosis	2.88	0.016

Z-score represents “Nek* Stat” from sigPathway package, which is standardized enrichment score. Positive Z score means upregulated in cadaver pancreases and conversely negative means downregulated in cadaver pancreases. For instance, β Cell identity is downregulated in cadavers whereas unfolded protein response (UPR) is upregulated. False Discovery Rate (FDR).

Expression changes related to ER stress, UPR, inflammation, and apoptosis

As shown in Table 3, there are impressive increases in the number of upregulated genes in the β cells of deceased donor pancreases that are related to ER function and ER stress with its related Unfolded Protein Response (UPR).¹⁷ There were notable increases in chaperone genes including HSP90B1, HSPA5, DNAJC3, DNAJB9, and DNAJB11. There were also many upregulated genes related to apoptosis and inflammation although it is important to point out the complexity of these changes in that the genes related to apoptosis included a mixture of changes that were both pro- and anti-apoptotic. Likewise activation of NFκB can lead to changes that are pro-apoptotic or protective.¹⁸ For example, some proapoptotic genes that were notably upregulated include CHAC1,¹⁹ CASP6,²⁰ BBC3²¹ and CEBPD.²² With regard to NFκB signaling, IL17RB²³ is thought to be an activator and TNIP2²⁴ is inhibitory. In spite of so many changes, islets isolated from cadaver pancreases usually function very well when used for clinical transplants. It is of interest that so many of the genes in these categories are upregulated while a much smaller number were downregulated.

Islet cell identity genes

As shown in Table 4, there were several surprising changes. There was notable downregulation in the deceased donor pancreases of MAF B²⁵ and ARX,²⁶ which are important for α cell identity. SYT13²⁷ and STXBP1,²⁸ which are important components of the β cell exocytotic machinery, were both downregulated. Also downregulated were the important β cell genes NKX6.1, a key transcription factor²⁹; ABCC8, the sulfonylurea receptor³⁰; and PAX6.³¹ On the other hand, most of the key genes for β cell identity including INS, IAPP, GCK, PCSK2, NKX2.2, and NEUROD1 were not differentially expressed.

Other noteworthy changes

As shown in Table 5, there were a variety of other notable changes. REG1A and REG1B have marked increases in expression in the β cells of the deceased donor pancreases. The roles of Reg family of proteins continue to be puzzling. They have been associated with regeneration in several tissues including β cells³² and a knockout of Reg1 led to reduced β cell mass.³³ The expression of the Reg genes can undergo marked changes, as was found in the present study and in an earlier study when β cells gene expression from subjects with T2D were studied after LCM.³⁴

Discussion

The use of LCM is a unique strength of this study, which makes it possible to obtain RNA from β cell enriched tissue of human pancreatic frozen sections. Most other studies measure gene expression from isolated islets, which can be expected have different patterns of gene expression. First, the trauma of islet isolation and time spent in tissue culture must lead to changes in β cell gene expression. Second, other cell types, notably duct and acinar cells, inevitably contaminate islet cell preparations. In particular, using electron microscopy for cell identification of islet preparations used for clinical islets transplants from 37 pancreases, we found that on average only 36% of cells were β cells.³⁵

In the present study there were important changes in gene expression in β cells dissected from deceased donor pancreases compared with those dissected from fresh surgically obtained pancreatic tissue. These changes resulted from a combination of events occurring during the end-of-life events that occur before death is pronounced, including brain death, fluctuations of blood pressure, administration of drugs, etc. Then another set of variables occurs with warm ischemia as the pancreas is procured followed by cold ischemia during the period of organ preservation, which last for periods of hours. It was not possible to determine the effects of brain death alone because of the many other variables that were involved. Moreover, we could not determine potential effects of hyperglycemia. Even though we chose deceased donors without a recorded history of hyperglycemia, there could easily have been periods of hyperglycemia as so often happens in intensive care situations. This issue is important because hyperglycemia, even modest elevations of glucose concentrations, can induce changes in gene expression through a process often called glucotoxicity.³⁶

Several pathways were very significantly differentially expressed, as shown in Table 2, including those related to glycolysis, the unfolded protein response (UPR), hypoxia, H-MTORC1, H-TNFA-NFkB, and the ER membrane. The most striking finding was how many gene related to the endoplasmic reticulum were differentially expressed. Many of these were genes were associated with UPR, including the heat shock proteins that serve a chaperone function, DERL1 and 2, which detect misfolded proteins, PDIA6, which is important for formation, reduction and isomerization of disulfide bonds, and ATF3, which is an important mediator of UPR.¹⁷ The UPR is mostly a protective response that is activated when the ER's capacity for handing the production of new proteins is exceeded, which is characterized by increased production of chaperones, which should enhance folding capacity and slowing down peptide synthesis by inhibiting both transcription and translation.³⁷ As ER stress becomes more intense the protective effects of the UPR can be overwhelmed leading to apoptosis, with several proapoptotic gees including the so-called “executioner” gene CHOP and JNK being turned on,³⁸ but in this case neither was differentially expressed. It is not clear why the UPR was activated. There was no reason to think insulin synthesis and secretion would be turned on by severe illness and organ preservation.

Pathways concerned with apoptosis were also differentially expressed, but it is not clear what these changes mean. Perhaps they indicate that the β cells are vulnerable to apoptosis. We know there is a low incidence of apoptosis as determined by TUNEL staining of β cells in paraffin sections of pancreases obtained at autopsy, and that the incidence is a little higher³⁹ in type 2 diabetes. There are also data indicating that the incidence of apoptosis is low in pancreases obtained from organ banks,⁴⁰ although it is possible that the cold preservation temperature inhibited the active process of apoptosis.

In conclusion, the present findings show that β cells obtained from deceased donor pancreases are stressed even before the trauma of islet isolation and culture. Finding ways to improve organ retrieval and preservation could lead to healthier β cells following the isolation process. Improvements at many points during the process including better pre-death care, organ procurement, organ preservation, islet isolation, and islet culture should lead to better clinical outcomes.

Data access

All the microarray data has been submitted to GEO under GSE89120 accession number.

Abbreviations

ER

endoplasmic reticulum

FDR

false discovery rate

IBMIR

instant blood-mediated inflammatory reaction

LCM

laser capture microdissection

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

UPR

unfolded protein response

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Grace Daher of the Advanced Genomics/Genetics Core, Chris Cahill of the Advanced Microscopy Core, and the Bioinformatics Core of the Joslin Diabetes Research Center [DRC].

Funding

This study was supported by grants from the NIH (R01 DK093909 [SBW], NCRR ICR U4Z 16606 (GCW), NCRR ICR U42 RR0023244–01(GCW), P30 DK036836 Joslin Diabetes Research Center (DRC), the Diabetes Research and Wellness Foundation and an important group of private donors. No conflicts of interest relevant to this article are reported.