L

Ih

GMa

Mb

c

E

ARAA

K�DhhI

1

rcsanc

p4

1h

Digestive and Liver Disease 45 (2013) 669– 676

Contents lists available at SciVerse ScienceDirect

Digestive and Liver Disease

j ourna l ho mepage: www.elsev ier .com/ lo cate /d ld

iver, Pancreas and Biliary Tract

n vitro differentiation into insulin-producing �-cells of stem cells isolated fromuman amniotic fluid and dental pulp

ianluca Carnevalea,∗,1 , Massimo Riccioa,1 , Alessandra Pisciottaa , Francesca Beretti a , Tullia Maraldia ,anuela Zavatti a, Gian Maria Cavallini a, Giovanni Battista La Salab, Adriano Ferrari c, Anto De Pola

Department of Surgery, Medicine, Dentistry and Morphological Sciences with Interest in Transplant, Oncology and Regenerative Medicine, University of Modena and Reggio Emilia,odena, ItalyDepartment of Medical and Surgical Sciences, IRCCS Arcispedale Santa Maria Nuova, University of Modena and Reggio Emilia, Reggio Emilia, ItalyDepartment of Biomedical, Metabolic and Neuroscience, Children Rehabilitation Special Unit, IRCCS Arcispedale Santa Maria Nuova, University of Modena and Reggio Emilia, Reggiomilia, Italy

a r t i c l e i n f o

rticle history:eceived 23 November 2012ccepted 6 February 2013vailable online 3 May 2013

eywords:-CellsiabetesAFSCsDPSCs

nsulin

a b s t r a c t

Aim: To investigate the ability of human amniotic fluid stem cells and human dental pulp stem cells todifferentiate into insulin-producing cells.Methods: Human amniotic fluid stem cells and human dental pulp stem cells were induced to differentiateinto pancreatic �-cells by a multistep protocol. Islet-like structures were assessed in differentiated humanamniotic fluid stem cells and human dental pulp stem cells after 21 days of culture by dithizone staining.Pancreatic and duodenal homebox-1, insulin and Glut-2 expression were detected by immunofluores-cence and confocal microscopy. Insulin secreted from differentiated cells was tested with SELDI-TOF MSand by enzyme-linked immunosorbent assay.Results: Human amniotic fluid stem cells and human dental pulp stem cells, after 7 days of differen-tiation started to form islet-like structures that became evident after 14 days of induction. SELDI-TOFMS analysis, revealed the presence of insulin in the media of differentiated cells at day 14, further con-firmed by enzyme-linked immunosorbent assay after 7, 14 and 21 days. Both stem cell types expressed,after differentiation, pancreatic and duodenal homebox-1, insulin and Glut-2 and were positively stainedby dithizone. Either the cytosol to nucleus translocation of pancreatic and duodenal homebox-1, either

the expression of insulin, are regulated by glucose concentration changes. Day 21 islet-like structuresderived from both human amniotic fluid stem cells and human dental pulp stem cell release insulin in aglucose-dependent manner.Conclusion: The present study demonstrates the ability of human amniotic fluid stem cells and humandental pulp stem cell to differentiate into insulin-producing cells, offering a non-pancreatic, low-invasivesource of cells for islet regeneration.

Gast

© 2013 Editrice

. Introduction

Diabetes is a chronic disease characterized by hyperglycemiaesulting from defects in insulin secretion or insulin action onells [1]. Although exogenous use of insulin remains the mostuccessful therapy, it fails to prevent hyperglycemia, resulting in

cute episodes of hypoglycemic and/or chronic complications likeephropathy, retinopathy, neuropathy, impotence and cardiovas-ular diseases [2,3]. For these reasons, over the past two decades

∗ Corresponding author at: Department of Surgery, Medicine, Dentistry and Mor-hological Sciences, University of Modena and Reggio Emilia, Via Del Pozzo 71,1125 Modena, Italy. Tel.: +39 059 2054828; fax: +39 059 2054859.

E-mail address: gianluca.carnevale@unimore.it (G. Carnevale).1 Contributed equally.

590-8658/$36.00 © 2013 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevierttp://dx.doi.org/10.1016/j.dld.2013.02.007

roenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.

the allogeneic transplantation of pancreatic islet cells has becomea subject of intense interest and activity as a potential care for dia-betes [4]. Unfortunately, islets transplantation faces a great numberof side effects. Particularly, the difficulty in obtaining sufficientlylarge numbers of purified islets due to shortage donor organs and tothe loss during the purification process [5], limits their therapeuticusage thus prompting a search for alternative sources of islet cells.The use of stem cells, a potential renewable source of pancreatic �-cells, is being currently investigated as an alternative therapeuticoption for the treatment of diabetes mellitus. Recently, attentionhas been aroused on embryonic [6], hematopoietic [7], pancre-atic islet [8], human biliary tree [9] and mesenchymal stem cells

[10] isolated from foetal and adult tissues, cultured and differenti-ated in vitro. Concerning mesenchymal stem cells, several reportshave focused great interest on human bone marrow-derived stemcells (hBMSCs), dental pulp stem cells (hDPSCs) and amniotic fluid

Ltd. All rights reserved.

6 nd Liv

saadsetspccsapi

mhtdetiir

2

2

dpNti

(bu

po

2

fi

wt

fl(Ames

2

sat

70 G. Carnevale et al. / Digestive a

tem cells (hAFSCs). Under appropriate experimental conditionsnd chemical stimuli, hBMSCs [11] and hDPSCs [12] can differenti-te into insulin-producing cells. On the contrary, Trovato et al. [13]emonstrated that under experimental conditions, used in theirtudy, cultured hAFSCs failed to differentiate into �-cells. Only Lit al. [14] demonstrated that hAFSCs can be differentiated into func-ional insulin-producing cells by knocking down neural restrictiveilencing factor. Despite their differentiation potential, the invasiverocedures required to isolate hBMSCs as well as the scarce per-entage of these cells in bone marrow can limit their use. On theontrary, hDPSCs show a large pool of donors and are easily acces-ible. Regarding hAFSCs, they can be obtained following humanmniocentesis backup and, as described by De Coppi et al. [15], areluripotent stem cells capable of giving rise to multiple lineages

ncluding representatives of all three embryonic germ layers.For these reasons the aim of the study is to develop a simple

ethod as well as to improve the differentiation of hDPSCs andAFSCs in pancreatic islet cells producing insulin and to compareheir ability along the differentiation pathways. Particularly, weemonstrated that hDPSCs and hAFSCs, under appropriate stimuli,xpress genes related to pancreatic �-cell development and func-ion, such as pancreatic and duodenal homebox-1 (PDX-1) andnsulin. Therefore, hDPSCs and hAFSCs are able to differentiate innsulin-producing cells. The two different stem cell lines could rep-esent a promising tool for the treatment of diabetes mellitus.

. Materials and methods

.1. Cell culture isolation and seeding

Human AFSCs were isolated and cultured as previouslyescribed [15,16]. Supernumerary amniocentesis samples wererovided by the Laboratorio di Genetica, Ospedale Santa Mariauova (Reggio Emilia, Italy). Back-up human amniocentesis cul-

ures were harvested by trypsinization and subjected to c-Kitmmunoselection by MACS® technology (Miltenyi Biotec).

Human DPSCs were isolated from third molar of adult subjects18 and 35 years of age) and were subjected to immunoselectiony magnetic cell sorting to obtain the CD34+/c-Kit+/STRO-1+ pop-lation as previously described by Riccio et al. [17].

All samples were collected with informed consent of theatients according to Italian law and guideline of ethical committeef Modena and Reggio Emilia University.

.2. In vitro multilineage differentiation

To induce osteogenic, adipogenic, myogenic and neurogenic dif-erentiation, hAFSCs and hDPSCs were cultured in the appropriatenduction media as previously described [18].

In order to verify adipogenic differentiation cells were stainedith oil red O solution and counterstained with Harris haema-

oxylin for 1 min.Myogenic differentiation was verified by double immuno-

uorescence staining, using an anti-human nuclei antibody (Ab)anti-hNu; Millipore, Billerica, MA, USA) and an anti-Myosinb (Sigma–Aldrich), in order to verify the formation of hybridyotubes between hDPSCs or hAFSCs and murine C2C12 cells. The

xpression of �3-Tubulin was analyzed by immunofluorescencetaining as marker of neurogenic differentiation.

.3. In vitro ˇ-cells differentiation of hAFSCs and hDPSCs

Differentiation of hAFSCs and hDPSCs was carried out in threetages. Briefly, cells were seeded at the density of 10,000 cell/cm2

nd cultured until reaching 80% of confluence. On day 1 cul-ure medium was replaced with Pre-induction medium [L-DMEM

er Disease 45 (2013) 669– 676

(Euroclone), supplemented with 5% FBS, 5 �M trans-retinoic acid(RA) and 0.5 mM �-mercaptoethanol (BME)] for two days. Onthe third day, the pre-induction medium was replaced with theinduction medium [H-DMEM (Euroclone), serum-free, plus 10 mMnicotinamide (NA), 5 �M RA, 0.5 mM BME]. On the fifth day themedium was replaced with maintaining-1 medium (H-DMEM sup-plemented with 5% FBS, 10 mM NA, 10 �M zinc sulphate and10 �M selenium). The maintaining-1 medium was replaced at day7 with maintaining-2 medium (L-DMEM supplemented with 5%FBS, 10 mM NA, 10 �M zinc sulphate and 10 �M selenium). Themaintaining-1 and the maintaining-2 media were alternated every2 days until day 21. A time table of treatments is shown in Fig. 1.Cells cultured in medium without inducers were used as controls.All chemicals were purchased from Sigma–Aldrich unless other-wise indicated.

2.4. Medium protein profile by SELDI-TOF MS

Proteomic analyses to evaluate the presence of insulin in dif-ferentiating and control medium were performed using SELDI-TOFMS (Bio-Rad). Particularly after 14 days of differentiation, the pro-teomic profiles, between maintaining-1 medium of hAFSCs, hDPSCsand the medium from control cells were compared to the proteomicprofiles obtained from maintaining-1 medium supplemented with21 pg of recombinant human insulin.

Chip preparation and sample application were performedaccording to the manufacturer’s instructions using the Biomek3000 Laboratory Automation Workstation (Beckman Coulter,Fullerton, CA, USA). Briefly, 3 �l of each medium was loaded on thespot of gold ProteinChip arrays and incubated for 15 min at roomtemperature.

After air drying, 1 �l of matrix (saturated sinapinic acid solutionin 50% acetonitrile in water containing 1% trifluoroacetic acid) wasapplied to each spot twice. Spot adsorbed protein patterns wereanalyzed on a ProteinChip Reader (Series 4000, Bio-Rad Laborato-ries, Hercules, CA, USA) and spectra detection was performed withProteinChip Data Manager software (Bio-Rad Laboratories Inc., Her-cules, CA, USA).

The obtained spectra were baseline subtracted, normalized fortotal ion current (TIC) in the range of mass-to-charge ratios (m/z)between 4000 and 10,000 Da and finally mass aligned.

2.5. Immunofluorescence and confocal microscopy

Undifferentiated or differentiated hAFSCs and hDPSCs werefixed for 20 min in 4% ice-cold paraformaldehyde in phosphate-buffered saline (PBS), permeabilized with 0.1% Triton X-100 in PBSfor 5 min and then processed as previously described [19]. The fol-lowing primary Abs were used: rabbit anti-c-Kit, mouse anti-CD34,mouse IgM anti-STRO-1, rabbit anti-Glut-2, Santa Cruz Biotechnol-ogy; goat anti-PDX-1, rabbit anti-insulin, rabbit anti-�3-Tubulin,AbCam; rabbit anti-myosin, Sigma–Aldrich; mouse anti-humannuclei, Millipore; diluted 1:100. Secondary Abs (goat anti-mouseAlexa405, goat anti-rabbit Alexa488, bovine anti-goat Alexa488and goat anti-rabbit Alexa546, Molecular Probe; goat anti-mouseIgM Cy3, Jackson) were diluted 1:200. Nuclei were stained by1 �g/mL 4′,6-diamidino-2-phenylindole (DAPI) in PBS. Negativecontrols consisted of samples not incubated with the primary Ab.The multi-labelling immunofluorescence experiments were car-ried out avoiding cross-reactions between primary and secondary

Abs.

Confocal imaging was performed by a Nikon A1 confocal laserscanning microscope as previously described [20]. The confocalserial sections were processed with ImageJ software to obtain

G. Carnevale et al. / Digestive and Liver Disease 45 (2013) 669– 676 671

F o gene�

tf

2

hod

igncu

tiiaD

2

de4S

A1amclta0o

2

4wm

ig. 1. Schematic representation of the multistep differentiation protocol used t-mercaptoethanol; NA: nicotinamide.

hree-dimensional projections and image rendering was per-ormed by Adobe Photoshop Software.

.6. Determination of insulin release and dithizone staining

Insulin concentrations were measured in the culture media ofAFSCs and hDPSCs at 0, 7, 14 and 21 days of differentiation. More-ver, after 21 days of differentiation the total release of insulin wasetected following the stimulation with 25.0 mM glucose.

Briefly, cells were gently washed with PBS once and then werencubated with L-DMEM and H-DMEM containing 5.5 or 25.0 mMlucose respectively, at 37 ◦C for 3 h. After 3 h of incubation, super-atants from wells were harvested and the levels of insulin fromells exposed to low or high glucose were measured. Media fromndifferentiated cells were used as control.

The quantitative measurement of human insulin concentra-ion was determined using the Insulin Human enzyme-linkedmmunosorbent assay (ELISA) kit (AbCam Cambridge, UK) accord-ng to the manufacturer’s instructions. Islet-like structures weressessed in hAFSCs and hDPSCs after 21 days of culture followingithizone staining [21].

.7. Western blotting

Whole cell lysates were obtained from undifferentiated andifferentiated hAFSCs and hDPSCs at 7, 14 and 21 days of differ-ntiation, and were processed as described by Pisciotta et al. [18].0 �g of total proteins from each sample were separated by 12%DS-PAGE and then transferred to PVDF membranes.

Blots were incubated with primary Abs (goat anti-PDX-1,bCam; rabbit anti-Glut-2, rabbit anti-PARP, Santa Cruz) diluted:1000, revealed by HRP-conjugated secondary Abs (anti-goat;nti-rabbit; Pierce Antibodies, Thermo Scientific) diluted 1:300. Allembranes were revealed by using Enhanced Chemio Lumines-

ence (Amersham). Anti-actin Ab was used as control of proteinoading in timing experiments. Densitometry was performed onhree independent experiments by NIS software (Nikon). An equalrea was selected inside each band and the mean of grey level (in a–256 scale) was calculated. Data were then normalized to valuesf background and of control actin band.

.8. Data analysis

Values are reported as the mean ± SD obtained by groups of–5 samples each. Differences between two experimental samplesere analyzed by paired, Student’s t test. Three or more experi-ental samples were analyzed by ANOVA followed by Dunnett’s

rate islet-like structure from hAFSCs and hDPSCs. RA: trans-retinoic acid; BME:

test (GraphPad Prism Software version 5 Inc., San Diego, CA, USA).In any case significance was set at p < 0.05.

3. Results

3.1. In vitro multilineage differentiation

In order to verify the phenotype of isolated hAFSCs and hDPSCssurface immune-labelling was performed by using c-Kit, CD34, andSTRO-1 specific Abs. Isolated hAFSCs showed a clear expression ofc-Kit confirming that the correct population was obtained. In thesame manner hDPSCs showed positivity to c-Kit, CD34, and STRO-1Abs, by co-expressing these antigens on their surface (Fig. 2A).

Multi-differentiation experiments were also performed for boththe cell types to confirm multipotent properties of isolated stemcells.

After 3 weeks of culture with osteogenic medium, both hAFSCsand hDPSCs showed nodular structures and the presence of a denseextracellular matrix. Particularly, extracellular matrix showed apositivity to the Alizarin Red staining (Fig. 2B).

The adipogenic potential of hAFSCs and hDPSCs was assessedby treating cells with adipogenic medium for 21 days. Particularly,lipid vacuoles were noticeable under light microscopy as early as10 days after adipogenic induction and visualized by staining withOil Red O after 21 days (Fig. 2B).

The ability of hAFSCs and hDPSCs to differentiate towards myo-genic lineage was verified by co-culturing the two human stem celltypes with C2C12 mouse myoblasts. After 14 days of co-culturemyotubes formation was observed in both co-cultures. Particu-larly, myotubes appear multi-nucleated indicating that cell fusionoccurs. Double staining with anti-hNu and anti-myosin Abs indi-cates that mature hybrid myotubes were formed between mouseC2C12 and hAFSCs or hDPSCs. Myotubes not labelled by anti-hNuAb and therefore formed only by C2C12 cells were also present(Fig. 2B).

After 10 days of addition of neurogenic medium, most of hAFSCsand hDPSCs showed a typical neuronal appearance. Furthermore,cells with neuron-like morphology showed a positive stainingagainst the �3-Tubulin neuronal marker (Fig. 2B).

3.2. ˇ-Cell differentiation

hAFSCs and hDPSCs were cultured in differentiation medium

for 21 days according to the multistep protocol described in Fig. 1.The induced cells were examined daily for morphological changesunder light microscope. Particularly, undifferentiated hAFSCs andhDPSCs showed a typical fibroblast-like morphology. Conversely,

672 G. Carnevale et al. / Digestive and Liv

Fig. 2. Surface phenotype and multi-differentiation ability of hAFSCs and hDPSCs. A:Immunofluorescence staining of hAFSCs by anti-c-Kit Ab (green) and triple immuno-fluorescence of hDPSCs by anti-CD34 (blue), anti-c-Kit (green) and anti-Stro-1 (red)Abs. B: In the first line is shown Alizarin staining of hAFSCs and hDPSCs osteogenicdifferentiation. The second line shows oil red/haematoxylin staining of hAFSCs andhDPSCs differentiated for two weeks towards adipogenic lineage. The third linerepresents immunofluorescence images of hAFSCs/C2C12 and hDPSCs/C2C12 co-culture stained by DAPI, anti-hNu (green) and anti-myosin (red) Abs. Images, infh

asna(cpamas

ourth line represent anti-�3-tubulin immunofluorescence labelling on hAFSCs andDPSCs differentiated towards neurogenic lineage. Bar 10 �m.

fter 7 days of differentiation the cells started to form islet-liketructures that became evident after 14 days of induction. Theumber and size of these islet-like structures markedly increasednd their boundary became clearer on 21st day of differentiationFig. 3A). In order to evaluate whether morphological changes wereoupled with the release of insulin in the culture medium, theroteomic profiles of media from undifferentiating and differenti-

ting hAFSCs and hDPSCs were compared. Spectrum obtained fromedium supplemented with recombinant human insulin was used

s positive control. Particularly, after 14 days of differentiation thepectra obtained from hAFSCs and hDPSCs culture media showed

er Disease 45 (2013) 669– 676

the peaks with mass-to-charge ratio (m/z) 5800 Da that matchedwith the detection peak obtained in the medium supplementedwith insulin (Fig. 3B). Moreover, as shown in Fig. 3B, the spec-tra analysis of undifferentiated hAFSCs and hDPSCs media did notshow any peaks with m/z 5800 Da. This result suggests that after 14days of culture in differentiation media the two stem cell types arealready committed towards the pancreatic �-cells and are able toproduce and release insulin.

3.3. Insulin release assay and immunocytochemistry assay

To further clarify the function of differentiated hAFSCs and hDP-SCs, insulin concentrations in the culture media were quantitativelymeasured by using the Human Insulin ELISA kit after 0, 7, 14 and21 days of induction.

As shown in Fig. 4A, 7 days after differentiation the insulinsecretion from hAFSCs was significantly higher (p < 0.05) whencompared to the respective control (time 0). Furthermore, theinsulin release statistically increased after 14 and 21 days of differ-entiation (30.7 ± 5.5 and 42.0 ± 4.1 �UI/ml) p < 0.01 and p < 0.001respectively (Fig. 4A).

Likewise, after 7 days of induction the insulin amount in hDPSCsmedium was significantly higher in comparison to the undifferen-tiated hDPSCs (time 0; p < 0.05; Fig. 4A). Moreover, after 14 and21 days of differentiation the insulin concentration in the mediumincreased in significant manner (27.6 ± 4.7 and 42.8 ± 0.9 �UI/ml)p < 0.01 and p < 0.001 respectively (Fig. 4A).

Furthermore, at day 14 the expression of PDX-1, insulin andGlut-2 in undifferentiated and differentiated hAFSCs and hDPSCswas evaluated by immunofluorescence assay. The results showedthat differentiated hAFSCs and hDPSCs after 14 days of culturewere intensely labelled by anti-PDX-1, anti-insulin and anti-Glut-2Abs (Fig. 4B). Furthermore, undifferentiated hDPSCs express PDX-1showing a weak staining for anti-PDX-1 and anti-Glut-2 Abs, andare not labelled by anti-insulin Ab. The same results were found inundifferentiated hAFSCs (data not shown).

To confirm immunofluorescence data, the presence of PDX-1and Glut-2 was evaluated by Western blotting (WB) in hAFSCs andhDPSCs at three different times of differentiation (7, 14 and 21days). Densitometric analysis was carried out on the WB bands toobtain a semi-quantitative analysis. In differentiating culture, bothcell types expressed PDX-1 and Glut-2 with the immunoreactivebands corresponding to 31 kDa and 62 kDa respectively (Fig. 4C).Particularly, densitometric analysis revealed that in undifferenti-ated cells, the protein bands corresponding to PDX-1 and Glut-2were weakly expressed, while in differentiating hAFSCs and hDP-SCs the expression of PDX-1 and Glut-2 was significantly increased(p < 0.001).

3.4. Dithizone staining

After 21 days of induction islet-like structures were stainedby DTZ. Particularly, differentiated hAFSCs and hDPSCs culturesshowed the typical “crimson red” staining. Undifferentiated cellswere not stained by DTZ (Fig. 4D).

3.5. Effect of glucose on PDX-1 translocation and insulin secretion

In order to investigate the intracellular localization of PDX-1 andthe expression of insulin in response to the changes of glucose con-centration, islet-like structures formed by hAFSCs and hDPSCs, at

day 21, were incubated in media at two different glucose concen-trations: 5.5 mM and 25.0 mM. Immunofluorescence assay revealedthat in hAFSCs and hDPSCs maintained in 5.5 mM glucose PDX-1was mainly localized in the cytoplasm of the cells forming the islets

G. Carnevale et al. / Digestive and Liver Disease 45 (2013) 669– 676 673

Fig. 3. Morphological features and SELDI analysis of hAFSCs and hDPSCs �-cell differentiation. A: Phase contrast microscopy of the two cell types at different time ofdifferentiation. Bar 100 �m. B: Proteomic profile of media from differentiating cultures as indicated. Value indicated on the top of the peaks indicates the m/z ratio of thedetected protein.

(s

Pawi

stmPbua2iu

Fig. 5A,C; G,I). Under the same condition cells were very weaklytained by anti-insulin Ab (Fig. 5B,C; H,I).

After 3 h of incubation in high glucose concentration (25.0 mM),DX-1 translocated from cytosol to nucleus. As a matter of fact,s showed in Fig. 5D and J the nuclei of both hAFSCs and hDPSCsere strongly stained by anti-PDX-1 Ab. Under these conditions the

ntracellular expression of insulin increased (Fig. 5E, F; K,L).Moreover, in order to confirm whether the amount of insulin

ecreted in the medium increased in response to glucose stimula-ion, ELISA test was performed on media from hAFSCs and hDPSCs

aintained in the two glucose concentrations (5.5 mM or 25.0 mM).articularly, as shown in Fig. 5M and N, the insulin release values inoth hAFSCs and hDPSCs were low at 5.5 mM glucose. Insulin val-es increased in a significant manner (46.2 ± 4.1 �UI/ml in hAFSCs

nd 47.6 ± 11.3 �UI/ml in hDPSCs) when cells were maintained in5.0 mM glucose for 3 h. On the other hand, insulin concentration

n media from undifferentiated hAFSCs and hDPSCs was very lownder both glucose concentrations.

In order to verify whether the differentiation protocol used inthis study influenced cell viability, Poly(ADP-ribose) polymerases(PARP) expression was evaluated in islet-like structures formedby both cell types. Fig. 5O shows WB analysis of PARP in hAFSCsand hDPSCs maintained in 5.5 mM or 25.0 mM glucose. Densito-metric analysis revealed that the native form of PARP was clearlyexpressed while the cleaved form (cPARP), involved in apoptoticprocess, was undetectable. Conversely, positive control (etoposidetreated HL60) clearly showed the band corresponding to cPARP,therefore demonstrating that cell death did not occur in hAFSCsand hDPSCs samples.

4. Discussion

The aim of this study is to explore the potential of hAFSCsand hDPSCs as a source of insulin-producing cells. Particularlywe developed a simple multistep method, through a non-viral

674 G. Carnevale et al. / Digestive and Liver Disease 45 (2013) 669– 676

Fig. 4. Characterization of hAFSCs and hDPSCs �-cell differentiation. A: Insulin release in the medium from 0 to 21 days differentiating cultures, detected by a specific ELISAassay (ANOVA followed by Dunnett’s test: *p < 0.05, **p < 0.01, ***p < 0.001, vs 0 days). B: Immunofluorescence staining of 14 days differentiating hAFSCs and hDPSCs by DAPI(blue), anti-PDX-1 (green), anti-Insulin (red) and anti-Glut-2 (green) Abs. Undifferentiated hDPSCs used as control show a scarcely detectable signal from anti-Insulin Ab.The same result was obtained for hAFSCs (not shown). Bar: 10 �m. C: Western blot analysis of PDX-1 and Glut-2 expression in �-cell differentiating hAFSCs and hDPSCs.A s is repu of ditha

gtptppdmtkdma2gMaupothp

ac

ctin bands were presented as control of the protein loading. Densitometric analysindiff hAFSCs; §§§p < 0.001 diff hDPSCs vs undiff hDPSCs). D: Phase contrast imagess indicated.

enomic reprogramming approach, to direct the differentiation ofhese two stem cell types as a source of glucose-regulated insulin-roducing cells. In the first step we used a medium that includedrans RA and BME. As demonstrated by Micallef et al., trans-RAromotes expression of the endogenous PDX-1 gene and thereforerovides a reliable marker of cells within the pancreatic endodermifferentiation pathway [22]. In the second step the low glucoseedium was replaced with the medium at high glucose concen-

ration and supplemented with NA, zinc and selenium. It is wellnown that high-glucose condition is a crucial factor for stem cellsifferentiation into insulin-producing cells [23]. Zinc is an essentialicronutrient required for insulin synthesis, secretion, signalling

nd packaging [24]. An integral part of insulin forms crystals for-Zn-insulin hexamer, as well as free ionized zinc in the extra-ranular space that acts as a reservoir for granular zinc [24–26].oreover, selenium stimulates pancreatic �-cell gene expression

nd enhances islet function [27]. In the second step NA was alsosed. NA is a PARP inhibitor that can induce islet formation fromancreatic progenitor cells, trans-differentiation and maturationf liver stem cells into insulin-producing cells [28,29]. NA is ableo induce the differentiation and increase �-cell mass in cultureduman foetal pancreatic cells [28], as well as to protect cells from

rolonged exposure to large amounts of glucose [29].

Our results are in accordance to these findings. Particularly,fter 7 days of differentiation hAFSCs and hDPSCs morphologyhanged rapidly by forming islet-like structures after 14 and 21

orted in the bottom (ANOVA followed by Dunnett’s test: ***p < 0.001 diff hAFSCs vsizone staining (crimson red) of islet-like structures formed by hAFSCs and hDPSCs

days of induction. Moreover, after just one week of differentiationthe expression of PDX-1 and Glut-2 in hAFSCs and hDPSCs as wellas the secretion of insulin in the media was observed. As a matter offact, the transcription factor PDX-1 is the first molecular marker thattemporally correlates with the pancreatic commitment and playsa critical role in pancreas development, �-cells differentiation andmaintenance of �-cells function [5,30]. The expression of PDX-1as well as the expression of Glut-2 was maintained and temporallyincreased indicating that both cell types were committed to �-cellsdifferentiation and improved glucose receptor expression. Besides,also insulin concentrations increased throughout the time of dif-ferentiation, although both cell types were alternatively culturedat high glucose and low glucose concentrations. No cellular deathwas observed according to the possible mechanisms that explainthe protective effects of NA. Therefore, PARP inhibitors prevent �-cell death, induce and maintain insulin secretion [31]. The mainrole of pancreatic �-cells is the secretion of insulin in response toglucose and although we did not transplant the insulin-producingcells into experimental diabetic animals, we then studied whetheror not insulin-producing hAFSCs and hDPSCs were able to respondto the change of glucose concentration in vitro, by simulating thephysiological stimulus. Particularly, our results suggest that both

differentiated hAFSCs and hDPSCs are able to secrete insulin inresponse to a glucose change thus maintaining their function. Thesedata are supported by the nuclear translocation of PDX-1 after highglucose stimulation. Particularly, under low glucose culture, PDX-1

G. Carnevale et al. / Digestive and Liver Disease 45 (2013) 669– 676 675

Fig. 5. Effect of glucose on PDX-1 translocation and insulin secretion. A–L: Immuno-staining by DAPI (blue), anti-PDX-1 (green) and anti-Insulin (red) Abs of islet-like structuresformed by hAFSCs (A–F) and hDPSCs (G–L), exposed to 5.5 mM (A–C; G–I) and 25.0 mM (D–F; J–L) glucose. M–N: Insulin release in the medium from glucose exposed islet-likestructures formed by hAFSCs (M) and hDPSCs (N), detected by a specific ELISA assay (paired t-test: *p < 0.05, 25.0 mM glucose vs 5.5 mM glucose). O: Western blot analysis ofPARP in hAFSCs and hDPSCs cultured in 5.5 mM and 25.0 mM glucose. HL60, treated with etoposide, were loaded as positive control of the presence of cleaved PARP (cPARP).Actin bands were presented as control of the protein loading. Densitometric analysis is reported in the right side: no significant differences were observed in c-PARP levelscomparing glucose treated cells vs controls (ANOVA followed by Dunnett’s test).

6 nd Liv

wSto[leprpoihftcrt

aocbwcf

CN

A

tpcdR

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

76 G. Carnevale et al. / Digestive a

as mainly localized into the cytoplasm in both hAFSCs and hDP-Cs, while in response to high glucose stimulation translocated tohe nucleus. Previous studies indicated that nuclear translocationf PDX-1 occurs as a consequence of glucose concentration changes32,33]. In the nucleus PDX-1 binds to the insulin gene [34,35] thuseading to insulin gene transcription [36]. Moreover, nuclear PDX-1xpression prevents deleterious effect of glucose and consequentlyreserves survival and function of �-cells [33]. These observationsepresent a very important consideration for the future islets trans-lantation programme by using hAFSCs and hDPSCs. Even thoughther stem cell types, such as adipose derived stem cells and bil-ary tree stem cells [9,37], appear appropriate for this purpose,AFSCs and hDPSCs present the advantage to be easily collected

rom routine medical practice and to be rapidly expanded in cul-ure. Moreover the possibility to culture hDPSCs in human serumontaining media, maintaining an higher proliferative potential,epresents a further advantage for the application to human cellherapy [18].

In conclusion, the present study confirms the potential and thebility of the two stem cell types, isolated from human amni-tic fluid and dental pulp, to differentiate into insulin-producingells. Furthermore, no substantial differences have been observedetween the two stem cell types along the differentiation path-ays, thus offering another non-pancreatic, low-invasive source of

ells for islet regeneration and a possible new therapeutic targetor the treatment of diabetes mellitus.

onflict of interest statemento conflicts of interest exist.

cknowledgements

Authors are grateful to Dr. Emanuela Monari for SELDI assis-ance. This work was supported by grants from “Progetto Strategicoer lo sviluppo nella sede di Reggio Emilia della Facoltà di Medi-ina e Chirurgia” Prot: 2010 0007725, Arcispedale S. Maria Nuovai Reggio Emilia and MIUR FIRB Accordi di Programma 2010 Prot:BAP10Z7FS.

eferences

[1] Gores PF, Najarian JS, Stephanian E, et al. Insulin independence in type Idiabetes after transplantation of unpurified islets from single donor with 15-deoxyspergualin. Lancet 1993;341:19–21.

[2] Bluestone JA, Herold K, Eisenbarth G. Genetics: pathogenesis and clinical inter-ventions in type 1 diabetes. Nature 2010;464:1293–300.

[3] Steffes MW, Sibley S, Jackson M, et al. Beta-cell function and the development ofdiabetes-related complications in the diabetes control and complications trial.Diabetes Care 2003;26:832–6.

[4] Sachs DH, Bonner-Weir S. New islets from old. Nature Medicine 2000;6:250–1.[5] Ramiya VK, Maraist M, Arfors KE, et al. Reversal of insulin-dependent diabetes

using islets generated in vitro from pancreatic stem cells. Nature Medicine2000;6:278–82.

[6] Soria B, Roche E, Berná G, et al. Insulin-secreting cells derived from embry-onic stem cells normalize glycemia in streptozotocin-induced diabetic mice.Diabetes 2000;49:157–62.

[7] Voltarelli JC, Couri CE, Stracieri AB, et al. Autologous nonmyeloablativehematopoietic stem cell transplantation in newly diagnosed type 1 diabetesmellitus. Journal of the American Medical Association 2007;297:1568–76.

[8] Furth ME, Atala A. Stem cell sources to treat diabetes. Journal of Cellular Bio-chemistry 2009;106:507–11.

[9] Cardinale V, Wang Y, Carpino G, et al. Multipotent stem/progenitor cells inhuman biliary tree give rise to hepatocytes, cholangiocytes, and pancreaticislets. Hepatology 2011;54:2159–72.

10] Vija L, Farge D, Gautier JF, et al. Mesenchymal stem cells: stem cell therapyperspectives for type 1 diabetes. Diabetes and Metabolism 2009;35:85–93.

[

er Disease 45 (2013) 669– 676

11] Lechner A, Habener JF. Bone marrow stem cells find a path to the pancreas.Nature Biotechnology 2003;21:755–6.

12] Govindasamy V, Ronald VS, Abdullah AN, et al. Differentiation of den-tal pulp stem cells into islet-like aggregates. Journal of Dental Research2011;90:646–52.

13] Trovato L, De Fazio R, Annunziata M, et al. Pluripotent stem cells isolated fromhuman amniotic fluid and differentiation into pancreatic beta-cells. Journal ofEndocrinological Investigation 2009;32:873–6.

14] Li B, Wang S, Liu H, et al. Neuronal restrictive silencing factor silencing induceshuman amniotic fluid-derived stem cells differentiation into insulin-producingcells. Stem Cells and Development 2011;20:1223–31.

15] De Coppi P, Bartsch Jr G, Siddiqui MM, et al. Isolation of amniotic stem cell lineswith potential for therapy. Nature Biotechnology 2007;25:100–6.

16] Maraldi T, Riccio M, Resca E, et al. Human amniotic fluid stem cells seededin fibroin scaffold produce in vivo mineralized matrix. Tissue Engineering2011;17:2833–43.

17] Riccio M, Resca E, Maraldi T, et al. Human dental pulp stem cells produce min-eralized matrix in 2D and 3D cultures. European Journal of Histochemistry2010;54:205–13.

18] Pisciotta A, Riccio M, Carnevale G, et al. Human serum promotes osteogenicdifferentiation of Human Dental Pulp Stem Cells in vitro and in vivo. PLoS ONE2012;7:e50542.

19] Riccio M, Maraldi T, Pisciotta A, et al. Fibroin Scaffold repairs critical size bonedefects in vivo supported by Human Amniotic Fluid and Dental Pulp Stem.Tissue Engineering 2012;18:1006–13.

20] Gibellini L, De Biasi S, Pinti M, et al. The protease inhibitor atazanavirtriggers autophagy and mitophagy in human preadipocytes. AIDS 2012;26:2017–26.

21] Shiroi A, Yoshikawa M, Yokota H, et al. Identification of insulin-producing cellsderived from embryonic stem cells by zinc-chelating dithizone. Stem Cells2002;20:284–92.

22] Micallef SJ, Janes ME, Knezevic K, et al. Retinoic acid induces Pdx1-positive endoderm in differentiating mouse embryonic stem cells. Diabetes2005;54:301–5.

23] Tang DQ, Cao LZ, Burkhardt BR, et al. In vivo and in vitro characteriza-tion of insulin-producing cells obtained from murine bone marrow. Diabetes2004;53:1721–32.

24] Kelleher SL, McCormick NH, Velasquez V, et al. Zinc in specialized secretorytissues: roles in the pancreas, prostate, and mammary gland. Advances in Nutri-tion 2011;2:101–11.

25] Dodson G, Steiner D. The role of assembly in insulin’s biosynthesis. CurrentOpinion in Structural Biology 1998;8:189–94.

26] Chausmer AB. Zinc, insulin and diabetes. Journal of the American College ofNutrition 1998;17:109–15.

27] Campbell SC, Aldibbiat A, Marriott CE, et al. Selenium stimulates pancre-atic beta-cell gene expression and enhances islet function. FEBS Letters2008;582:2333–7.

28] Yang L, Li S, Hatch H, et al. In vitro trans-differentiation of adult hepaticstem cells into pancreatic endocrine hormone-producing cells. Proceedingsof the National Academy of Sciences of the United States of America2004;99:8078–83.

29] Otonkoski T, Beattie GM, Mally MI, et al. Nicotinamide is a potent inducer ofendocrine differentiation in cultured human fetal pancreatic cells. Journal ofClinical Investigation 1993;92:1459–66.

30] Melloul D. Transcription factors in islet development and physiology: role ofPDX-1 in beta-cell function. Annals of the New York Academy of Sciences2004;1014:28–37.

31] Takasawa S, Okamoto H. Pancreatic beta-cell death, regeneration and insulinsecretion: roles of poly(ADP-ribose) polymerase and cyclic ADP-ribose. Inter-national Journal of Experimental Diabetes Research 2002;3:79–96.

32] Campbell SC, Macfarlane WM. Regulation of the pdx1 gene promoter inpancreatic beta-cells. Biochemical and Biophysical Research Communications2002;299:277–84.

33] Elrick LJ, Docherty K. Phosphorylation-dependent nucleocytoplasmic shuttlingof pancreatic duodenal homeobox-1. Diabetes 2001;50:2244–52.

34] Ardestani A, Sauter NS, Paroni F, et al. Neutralizing interleukin-1beta (IL-1beta)induces beta-cell survival by maintaining PDX1 protein nuclear localization.Journal of Biological Chemistry 2011;286:17144–55.

35] Macfarlane WM, McKinnon CM, Felton-Edkins ZA, et al. Glucose stimulatestranslocation of the homeodomain transcription factor PDX1 from the cyto-plasm to the nucleus in pancreatic beta-cells. Journal of Biological Chemistry1999;274:1011–6.

36] Wu H, MacFarlane WM, Tadayyon M, et al. Insulin stimulates pancreatic-duodenal homoeobox factor-1 (PDX1) DNA-binding activity and insulin

promoter activity in pancreatic beta cells. Biochemical Journal 1999;344:813–7.

37] Moshtagh PR, Emami SH, Sharifi AM. Differentiation of human adipose-derivedmesenchymal stem cell into insulin-producing cells: an in vitro study. Journalof Physiology and Biochemistry 2012 [Epub ahead of print].