|Home | About | Journals | Submit | Contact Us | Français|
Shwachman-Diamond syndrome (SDS), a rare autosomal recessive disorder characterized by exocrine pancreatic insufficiency and hematopoietic dysfunction, is caused by mutations in the Shwachman-Bodian-Diamond syndrome (SBDS) gene. We created human pluripotent stem cell models of SDS by knock-down of SBDS in human embryonic stem cells (hESCs) and generation of induced pluripotent stem cell (iPSC) lines from two SDS patients. SBDS-deficient hESCs and iPSCs manifest deficits in exocrine pancreatic and hematopoietic differentiation in vitro, enhanced apoptosis and elevated protease levels in culture supernatants, which could be reversed by restoring SBDS protein expression through transgene rescue or by supplementing culture media with protease inhibitors. Protease-mediated auto-digestion provides a mechanistic link between the pancreatic and hematopoietic phenotypes in SDS, highlighting the utility of hESCs and iPSCs in obtaining novel insights into human disease.
Shwachman-Diamond syndrome (SDS) is a rare pediatric disorder characterized by exocrine pancreatic insufficiency, hematopoietic dysfunction, and predisposition to leukemia. Histological examination reveals fatty infiltration of pancreatic acini with spared ductal architecture (Shimamura, 2006). Neutropenia is the most frequent hematological abnormality with limited progression to aplastic anemia. An increased risk for myelodysplasia and acute myelogenous leukemia is the major source of mortality (Donadieu et al., 2005).
SDS is associated with inherited mutation in the Shwachman-Bodian-Diamond syndrome (SBDS) gene (Boocock et al., 2003). The majority are nonsense or splice-site mutations that result in reduced, but not absent, protein expression. No patients with biallelic null mutations have been identified, and targeted deletion of murine Sbds results in embryonic lethality, indicating that some Sbds expression is essential for viability (Zhang et al., 2006). The two most common mutations are rs113993993, which affects a donor splice site leading to premature protein truncation, and rs120074160, which encodes a premature stop codon (Boocock et al., 2003; Nakashima et al., 2004).
SBDS is implicated in ribosome biogenesis (Finch et al., 2011). SBDS protein concentrates in the nucleolus (Austin et al., 2005), and SBDS deficiency results in delayed ribosomal RNA processing (Ganapathi et al., 2007). Functional 80S ribosomes form through the assembly of 40S and 60S subunits. Eukaryotic initiation factor 6 (eIF6) prevents premature inter-subunit bridge formation during pre-60S subunit maturation (Gartmann et al., 2010); however, eIF6 must be released from the 60S subunit before the 80S ribosome can form. Sbds is a protein cofactor that participates in the mechanism by which elongation factorlike 1 (Efl1) releases eIF6 from the pre-60S subunit (Finch et al., 2011). Diminished Sbds expression impairs ribosomal subunit assembly in patients (Burwick et al., 2012).
Human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs) are valuable tools to study developmental dysfunction in genetic disease. Directed differentiation of hESCs/iPSCs into specific tissues enables detailed study of cell fate decisions and provides a scalable in vitro model of early human development.
In this study, we modeled SDS in hESCs by inducing SBDS protein deficiency with lentiviral RNAi, and derived iPSCs from fibroblast cell lines of two patients. SDS hESCs/iPSCs displayed defective exocrine pancreatic differentiation and impaired myeloid hematopoietic development. SDS pancreatic and hematopoietic cultures displayed prominent granular content, elevated protease activity, and increased cell death. Pharmacological inhibition of protease activity reversed the cellular phenotypes in vitro, suggesting that granule processing defects resulting in pathologic auto-digestion and cell death link the pancreatic and hematopoietic phenotypes.
We knocked down SBDS in hESCs through lentiviral delivery of shRNA and then isolated a stable knockdown population using fluorescent activated cell sorting (FACS) and manual selection of green fluorescent protein (GFP) positive colonies (SBDSi hESC). We maintained SBDSi hESC as a population to minimize phenotypic variation that occurs when selecting individual clones. In addition, we reprogrammed fibroblast cell lines from two SDS patients into induced pluripotent stem cells (SDS-iPSC-1, SDS-iPSC-2; (Park et al., 2008). We restored normal levels of SBDS protein expression through transgene rescue by infecting SBDSi hESC, SDS-iPSC-1, and SDS-iPSC-2 with a lentiviral vector expressing human SBDS cDNA, and selecting for stably corrected populations (hESC+SBDS, iPSC1+SBDS, iPSC2+SBDS). We generated control cell lines with an empty puromycin selectable vector (hESC-SBDS, iPSC1-SBDS, iPSC2-SBDS).
Western blotting demonstrated reduced SBDS protein expression in hESC-SBDS, iPSC1-SBDS, and iPSC2-SBDS, and restoration of protein expression with transgene rescue (Figure 1a). iPSC1-SBDS showed trace SBDS expression only with overexposure of the Western blot; iPSC2-SBDS displayed reduced levels of SBDS relative to control hESC (Figure 1a). Fibroblasts used to generate SDS-iPSC-2 were from a patient who possesses two splice site mutations and expresses full length SBDS mRNA, although at lower levels, due to alternative splicing (Austin et al., 2005).
SBDSi hESC, SDS-iPSC-1, and SDS-iPSC-2 kept in culture for more than forty passages maintained hallmarks of human pluripotent cells regardless of gene correction. All cell lines showed >95% positivity for the hESC marker Tra-1-60, formed cystic tri-lineage teratomas after injection into immunodeficient mice (Figure S1a), and expressed levels of mRNA transcripts for the pluripotency markers NANOG, OCT4, SOX2, and KLF4 comparable to control hESC/iPSC lines (Figure S1b). After 16 days of differentiation in vitro, cells expressed specific markers reflecting commitment to endoderm (AFP, GATA4), mesoderm (MYOCD, RUNX1), and ectoderm (NCAM1, NES; Figure S1c). We observed no significant differences between SBDS-deficient and transgene rescued cell lines, indicating that reduced SBDS levels do not compromise pluripotency. Although SBDS deficiency can lead to abnormal mitoses and polyploidy (Austin et al., 2008), we found that no cell lines demonstrated greater than 5% polyploidy.
Cells from patients and animal models of SDS manifest SBDS-dependent reductions in ribosomal assembly, as reflected by depressed ratios of the 80S and 60S subunits relative to 40S (Burwick et al., 2012; Wong et al., 2011). We sedimented cell lysates from iPSC1+SBDS and iPSC1-SBDS through sucrose gradients, measured ribosomal peaks by UV absorbance, and determined relative levels of the ribosomal subunits after quantification of area-under-the-curve (Figure 1b). SBDS protein rescue reversed the depressed 80S:40S and 60S:40S ratios consistent with functional correction of the ribosomal subunit association defects. Thus our pluripotent stem cell models reflect defects in ribosomal assembly characteristic of the disease (Finch et al., 2011).
Human pluripotent stem cells can be differentiated into pancreatic tissue in a stepwise manner that recapitulates cell fate decisions of pancreatic organogenesis in utero (Cai et al., 2011; Chen et al., 2009); therefore, we used in vitro directed differentiation to model pancreatic development in SDS and assessed differentiation at various time points. Differentiation cultures from all cell lines at day 5 stained positive for FOXA2, a marker of definitive endoderm, in regions between pluripotent colonies (Figure S2a). We observed no significant differences in endoderm commitment regardless of gene rescue when evaluating expression of FOXA2 or SOX17, another endodermal marker (Figure S2b). We also performed qRT-PCR for FOXA2 and an extra-embryonic endodermal marker, AFP, and found no significant differences in gene expression regardless of transgene rescue. With the presence of definitive endoderm established, we then treated cultures with (−)-indolactam V (ILV), a compound that directs pancreatic progenitor development from definitive endoderm (Chen et al., 2009). Pancreatic progenitors adopted a streak-like morphology visualized by immunofluorescent (IF) staining for the pancreatic progenitor marker PDX1. These PDX1+ cells arose from definitive endoderm in the network of gaps between original hESC/iPSC colonies (Figure S2c). We found similar percentages of PDX1+ pancreatic progenitors regardless of transgene rescue (Figure S2d); similarly, qRT-PCR analysis revealed no significant differences in mRNA expression for PDX1 and the pancreatic transcriptional regulator HNF-6. Therefore, loss of SBDS does not impair definitive endoderm formation or generation of pancreatic progenitors from hESCs/iPSCs.
We then differentiated pancreatic progenitors into mature pancreatic cell fates. Exocrine acinar cells comprise the majority of naturally developed pancreas and pancreatic tissue differentiated in vitro from hESCs (Chen et al., 2009). At day 19 in pancreatic differentiation culture, all cell lines displayed greater than 95% amylase positivity irrespective of gene rescue, indicating that initial commitment to exocrine cell fate is not SBDS-dependent. Culture for an additional 7–10 days produces mature pancreatic acinar cells. At this later time point (day 25), SBDS-deficient hESC and SDS-iPSC-1 displayed deficits in amylase positivity by IF staining (Figures 2a and 2b) and lower levels of amylase and carboxypeptidase gene expression as measured by RT-PCR, relative to their gene rescued counterparts (Figure 2c). These data suggest that SBDS is essential to maintaining or preventing loss of exocrine cells at the later stages of pancreatic differentiation in vitro. We also observed continued loss in exocrine cell number from day 25 to 35 among SBDS-deficient hESC and SDS-iPSC-1 relative to gene rescued cells (Figure 2d), indicating that SBDS-dependent exocrine cell loss continues over time in culture. SBDS deficiency accelerates Fas-dependent apoptosis (Rujkijyanont et al., 2008). Consistent with this, we found elevated levels of the apoptotic marker Annexin V in SBDS-deficient hESC and SDS-iPSC-1 (Figure 2e).
Unlike in hESC and SDS-iPSC-1, the empty vector control for SDS-iPSC-2 (iPSC2-SBDS) did not display a reduction in exocrine cell number compared to transgene-rescued cells at day 25 of pancreatic differentiation. We speculated that this might be due to residual SBDS protein expression in this line. Thus, we further depleted SBDS protein by lentiviral RNAi and confirmed the reduction in SBDS expression by Western blot (Figure S2e). In iPSC2-SBDS cells with SBDS RNAi added, we detected a decrease in amylase positive cells (Figure 2f), a decrease in exocrine cells (Figure 2g), and an increase in IF staining and the percentage of cells positive for Annexin V (Figures 2h and 2i). These data argue that reduction of SBDS protein levels is associated with reduced survival of terminally differentiated exocrine pancreatic cells.
Gross examination of pancreatic differentiation cultures for SBDS-deficient cell lines suggested a marked disruption of cellular architecture. Therefore, we assessed ductal and endocrine cell fates among our pancreatic differentiation cultures. In transgene-rescued cultures, IF staining for DBA, a ductal marker, revealed that ductal tissue developed in an organized fashion, creating tube-like structures between colonies of acinar cells (Figure 3a). In differentiated cultures of SBDS-deficient hESC and SDS-iPSC-1, we observed a significant reduction in the percentage of cells staining positive for DBA (Figure 3b) and in gene expression for CK19, a duct-specific cytokeratin (Figure 3c), compared to their gene-rescued counterparts. Again, as seen in differentiated cultures of exocrine cells, iPSC2-SBDS showed no deficiency in ductal development, but further depletion of SBDS protein by addition of SBDS RNAi reduced the percentage of DBA+ staining cells (Figure 3b).
With regard to endocrine cell fate, we found only trace insulin positivity by IF staining and limited staining for c-peptide (Figure S3a). The presence of SBDS resulted in no significant difference in the percentage of cells staining positive for c-peptide (Figure S3b) or in gene expression for insulin when measured by qRT-PCR (Figure S3c). These data are consistent with prior reports showing low levels of endocrine commitment using comparable protocols (Chen et al., 2009), indicating that our culture conditions reflect differentiation predominantly towards ductal and exocrine cell fates.
In vitro pancreatic differentiation of normal hESC and iPSC produces acinar-like colonies of exocrine cells separated by interlobular septa and ductal tissue; these structures resemble two-dimensional pancreatic lobules. In contrast, differentiated cultures of SBDS-deficient hESC and SDS-iPSC-1 displayed thickening of the interlobular septa, disruption of the ordered acini-like lobules, and increased regions of cell death (Figures 3d/e), as confirmed by expression of Annexin V. Gene rescue of SBDS restored the appearance of organized pancreatic lobules and acini-like structures with reduced patches of dead cells in both SBDSi hESC and SDS-iPSC-1 differentiation cultures. Again, SDS-iPSC-2 cultures failed to manifest cell death phenotypes, irrespective of gene rescue.
We used an EB-based in vitro hematopoietic differentiation protocol to investigate the effects of SBDS-deficiency on blood phenotypes. We differentiated all cell lines, disaggregated EBs at various time points, and analyzed by qRT-PCR, flow cytometry, and methylcellulose colony-forming unit (CFUs) assays. Time course studies from day 10 onwards identified days 18–22 as a plateau of maximal hematopoietic activity in normal cell lines.
At day 20 of in vitro hematopoietic differentiation, hESC and SDS-iPSC-2 cultures without transgene rescue displayed a decreased percentage of CD45+ cells measured by flow cytometry (Figure 4a), reduced hematopoietic CFUs (Figure 4b), and reduced mRNA expression for hematopoietic transcription factors (TAL1, SPI1, and GATA2) (Figure 4c). Three hESC and two iPSC normal control lines displayed comparable CD45 positivity to SBDS transgene rescued hESC and SDS-iPSC-2 (%CD45: 33.2 +/− 4.3), indicating that transgene-rescue does not augment hematopoietic potential above normal levels. Instead, SBDS-deficiency correlates with reduced hematopoietic differentiation in vitro.
Among the hematopoietic colonies detected in our differentiation cultures, we observed no significant lineage skewing as assessed by CFU assay (CFU-Granulocyte, CFU-Macrophage, CFU-Granulocyte/Macrophage), flow cytometry for markers of erythroid (glycophorin A), neutrophil (CD15), or monocyte/macrophage lineages (CD11b), or by gene expression analysis of proteins highly expressed in erythrocytes (HGB1), neutrophils (ELANE), and macrophages (CSF1R) (Figures S4a–c). To exclude a general defect in mesodermal lineage specification, we interrogated endothelial (TEK, CDH5) and cardiac (NKX2-5, myocardin) markers by qRT-PCR and found no significant differences irrespective of transgene rescue (Figure S4d). Moreover, analysis of SBDSi hESC and SDS-iPSC-2 derived teratomas showed no generalized loss of mesodermal structures.
Surprisingly, SDS-iPSC-1 displayed minimal hematopoietic activity upon differentiation, regardless of transgene rescue, generating less than 1% CD45+ hematopoietic cells at day 20 of hematopoietic differentiation, and never developing hematopoietic CFUs in 6 separate experiments. The diminished hematopoietic activity of SDS-iPSC-1 may reflect the known clonal variation in the differentiation potential of human pluripotent stem cells (Osafune et al., 2008). Given the absence of SBDS-dependent hematopoietic phenotypes in SDS-iPSC-1, we excluded these cells from subsequent analyses.
The exocrine pancreas and cells of the myeloid lineage, including neutrophils, are characterized by large numbers of cytotoxic granules. Production of proteases in the pro-enzyme form in exocrine pancreas and appropriate targeting and packaging of proteases in myeloid cells protect against autodigestion. Routine microscopic analysis of exocrine cells in day 25 pancreatic differentiation cultures demonstrated different granule morphologies between SBDS-deficient and transgene rescued cells. Histologic staining and electron microscopy of SBDS-deficient SDS-iPSC-1 cultures confirmed both increased granule size and number per cell compared to transgene rescued counterparts (Figures 5a and 5b, Table S1). The SBDS-deficient iPSC cultures possessed cellular phenotypes similar to observations in patient samples (Figure 5a). These differences may reflect immaturity of the granules, as zymogen granules lose volume during maturation through tighter aggregation of secretory proteins or may reflect a failure of other regulated steps in the granule trafficking pathway. Similarly, histological analysis of day 20 EB-derived hematopoietic cells differentiated from SBDS-deficient hESC revealed an increased staining of primary azurophilic granules within promyelocytes and monocytes compared to transgene-rescued SBDSi hESC (Figure 5c).
Proteases are a major component of both pancreatic zymogen and myeloid azurophilic granules. We reasoned that presence of such granules might correlate with aberrant enzymatic activation and that release of granule enzymes leading to auto-digestion might contribute to SDS pathology. We subjected media from day 25 pancreatic and day 20 hematopoietic cultures (containing mostly neutrophil progenitors and monocytes/macrophages) to a protease activity assay. We studied SBDS-deficient SDS-iPSC-1 for pancreatic differentiation and SBDS-deficient hESC for hematopoietic differentiation, as these lines displayed the most dramatic phenotypes. Both these pancreatic and hematopoietic cultures displayed significantly elevated levels of protease activity compared to transgene rescued lines (Figure 5d). At baseline, SBDS-deficient SDS-iPSC-1 day 25 pancreatic cultures show significantly increased cell death compared to transgene rescued counterpart (Figures 5e/f). To test if the elevated protease levels had functional consequences that account for cell death, we administered a pharmacological blockade of potentially cytotoxic enzymes within granules. We treated SBDS-deficient SDS-iPSC-1 pancreatic cultures with the serine protease inhibitor aprotinin, a broad-spectrum protease inhibitor cocktail, a lipase inhibitor, and a pan-caspase inhibitor. Treatment occurred from day 13 onwards, coincident with the formation of mature pancreatic cell types with prominent granules. Aprotinin treatment significantly reduced cell death (Figures 5g), and also normalized colony morphology, reducing thickening of the interlobular septa in two-dimensional culture and reducing regions of cell death (Figure 5f). Treatment with the protease inhibitor cocktail initially demonstrated similar improvements as aprotinin alone, but was toxic to cells and eventually led to increased cell death (Table S2).
We performed parallel experiments targeting primary azurophilic granules in hematopoietic cultures. Irrespective of gene rescue, day 20 EB-derived cells from SBDS-deficient hESC displayed robust proliferation with minimal cell death (<2%) in liquid cultures with 20% serum and hematopoietic cytokines. However, stressing cells with a 12-hour exposure to low-serum (1%) conditions exposed significantly higher rates of cell death in SBDS-deficient hESC hematopoietic cells compared to the transgene rescued cell line (Figure 5h), suggesting stressinduced apoptosis may contribute to the hematopoietic phenotype. In low-serum conditions, we treated SBDS-deficient hESC hematopoietic cells with inhibitors including aprotinin, the broad-spectrum protease inhibitor cocktail, neutrophil elastase inhibitor sivelestat, a myeloperoxidase inhibitor, and a pan-caspase inhibitor. The pan-caspase inhibitor dramatically increased cell survival to levels above the transgene corrected SBDSi hESC hematopoietic cultures (Table S3), which likely reflects inhibition of baselines levels of apoptotic cell death in normal cultures. Amongst the granule enzyme inhibitors, only the broad-spectrum protease inhibitor cocktail significantly reversed the increased cell death in SBDS-deficient hESC hematopoietic cultures (Table S3), which may reflect the importance of non-serine proteases such as cathepsin B and cathepsin D in myeloid granules, or the synergistic effect of blocking a number of protease types.
SDS is a rare, autosomal recessive disorder characterized by developmental dysfunction of the exocrine pancreas, bone marrow failure, and predisposition to infection and leukemia. In this study, we present a novel human model of SDS based on pluripotent stem cells. Using lentiviral SBDS RNAi in hESCs and direct reprogramming of SDS patient fibroblasts, we generated hESCs and iPSCs that meet the widely accepted standards of pluripotency (Maherali & Hochedlinger, 2008), but also manifest dysfunction of ribosomal assembly, a biochemical feature of the class of bone marrow disorders characterized by ribosomopathy. Upon in vitro pancreatic differentiation, SBDS-deficient hESCs/iPSCs displayed an SBDS-dependent, progressive loss of exocrine tissue with increased cell death and impaired architecture of the selforganizing acinar-like structures that develop in culture. Hematopoietic development from our hESCs/iPSCs is also compromised. SDS hESC/iPSCs displayed a reduced percentage of CD45+ cells, a decreased CFU production, and a gene expression profile reflecting impaired myeloid development without general loss of mesodermal potential. We did not observe lineage skewing or maturation defects upon prolonged culture; however, EB-derived hematopoietic cells (mostly neutrophil progenitors and monocytes) had significantly increased levels of cell death in low-serum conditions, suggesting that cellular-stress induced apoptosis contributes to neutropenia (Yamaguchi et al., 2007). Our hESC/iPSC system enables a scalable model of human material for investigating SDS.
Our results suggest a mechanistic link between exocrine insufficiency and hematopoietic dysfunction in SDS. Granule size and number were increased in SBDS-deficient SDS-iPSC-1 pancreatic cultures relative to transgene-corrected cells, and likewise the azurophilic primary granules of promyelocytes, myelocytes, and monocytes were more prominent in SBDS-deficient EB-derived hematopoietic cultures. These granular changes parallel the significantly higher protease activity levels we detected in both cell types. Proteases such as trypsin should not normally be active because they are stored and secreted as inactive proenzymes such as trypsinogen. However, our data correlate high protease levels with autodigestion and cellular death in vitro, suggesting that autodigestion due to granule abnormalities is a pathologic link between the pancreatic and hematopoietic dysfunction observed in SDS. Other tissues with high digestive granule content have also been shown to be defective in patients with SDS, including osteoclasts (Toiviainen-Salo et al., 2007), intestinal cells (Shah et al., 2010), and parotid glands (Stormon et al., 2010), suggesting a common pathophysiologic mechanism of digestive granule dysfunction in affected tissues, presumably due to vacuolar processing defects.
In our model, pharmacological inhibition of proteases reversed the aberrant cellular phenotypes in vitro. Aprotinin was previously FDA approved (Trasylol™; Bayer) and marketed to reduce transfusion requirements in patients undergoing cardiac bypass surgery, but its use was discontinued because of safety concerns. It is tempting to speculate that drug treatment might reduce or prevent acinar cell death or ameliorate neutropenia in SDS patients, if it can be tolerated. In addition to inhibiting trypsin directly, aprotinin can inhibit autophagy (Chau et al., 2003). In hematopoietic cell cultures, we also observed a reduction in cell death after treatment with an anti-proteolytic cocktail suggesting additional benefit in exploring a wider range of anti-proteolytics, as blocking non-serine proteases, such as cathepsin B, may eliminate direct and synergistic protease effects of an auto-digestive process.
Our study highlights both the advantages and limitations of pluripotent stem cell based models of human pathology. In both pancreatic and hematopoietic differentiation assays, variability existed in the different isolates of SBDS-deficient hESCs and iPSCs. Most notably, SDS-iPSC-2 displayed a discernible hematopoietic but not an exocrine pancreatic phenotype. SDS-iPSC-2 retained the largest residual SBDS expression (~30% of normal, whereas SBDSi hESC expressed ~7% and SDS-iPSC-1 <1%). Phenotypic variability may reflect residual SBDS function, as suggested in previous patient-based observations (Minelli et al., 2009). To confirm a relationship between phenotype and SBDS expression, we further depleted SBDS protein by RNAi knock-down in SDS-iPSC-2 cells and observed decreased pancreatic exocrine function, consistent with the correlation between phenotype and protein levels. However, genetic background may also account for phenotypic variation. 11–29% of SDS patients lack any known SBDS genetic variant (Boocock et al., 2003; Nakashima et al., 2004; Woloszynek et al., 2004); these patients exhibit severe hematopoietic disease with very mild pancreatic manifestations (Hashmi et al., 2011), supporting genotype-phenotype relationships. Complex genetic backgrounds likely explain the combination of manifestations, pancreatic or hematologic, observed in SDS patients and may distinguish related syndromes such as SDS, Johanson-Blizzard syndrome, Pearson marrow-pancreas syndrome, dyskeratosis congenita, and Fanconi anemia. Importantly however, there is considerable variability in the differentiation propensity of distinct pluripotent stem cells clones (Osafune et al., 2008), but the underlying mechanisms of this heterogeneity remains unclear. Therefore, assessment of phenotypic reversion for transgene corrected clones is imperative for concluding a link between genotype and phenotype in pluripotent stem cell models of disease.
hESCs/iPSCs provide an in vitro model of target organ tissues for discerning the relationship between complex genetic backgrounds and cellular phenotypes. For example, we confirmed the presence of the 258+2T>C (rs113993993) and 621+1G>A mutations in SDS-iPSC-2. 258+2T>C has been detected in patients with acquired aplastic anemia with early presentation, poor outcomes, and absence of the pancreatic manifestations characteristic of SDS (n=4) (Calado et al., 2007). 621+1G>A, a rare variant, has been observed in an inherited bone marrow failure patient who developed severe neutropenia but no pancreatic insufficiency (Tsangaris et al., 2011). SDS-iPSC-2 pancreatic culture, which carries these two mutations, did not exhibit the expected decrease in pancreatic exocrine cell numbers compared to its transgene rescued counterpart, consistent with the lack of pancreatic manifestations described above. While this patient met clinical criteria for SDS, his pancreatic involvement was limited to low-level serum trypsinogen. Therefore, the clinical presentation of the iPSC donor is reflected in our model. Perhaps 183–184TA>CT (rs120074160), which can be found in most patients including the source of SDS-iPSC-1 but not SDS-iPSC-2, contributes to the pancreatic phenotype by lowering SBDS protein levels below the threshold needed for healthy pancreatic function. Additional analyses with iPSC lines from patients with known clinical histories and genotypes may associate genetic variants with functions to personalize diagnosis and therapeutic decisions.
Our observation of increased granule size and number in exocrine pancreatic cells derived in vitro from SDS patients contrasts with the cellular phenotype of a recently described murine model which manifests pancreatic features similar to human SDS. These mice were derived from a Sbds conditional knockout allele and/or a point mutation generated by site-directed mutagenesis (Tourlakis et al., 2012). In these mice, microscopy revealed normal structure but a marked reduction in number of acinar zymogen granules, whereas our human model showed an increase in granule number and size for both acinar zymogen granules and azurophilic granules in hematopoietic cells. These varied results could reflect an inherent difference between human and mouse disease models or between in vitro and in vivo systems; it could also reflect an unrelated variable such as granule maturation, time between feeding and tissue procurement, since feeding influences zymogen granule number, or lack of a unknown systemic factor, such as a paracrine signal, not present in one or both systems.
Our work establishes a new human cell-based model system for SDS that demonstrates distinctive developmental phenotypes not easily studied in other contexts. Future studies of inherited bone marrow failure syndromes can benefit from hESC/iPSC models that reveal genotype-phenotype relationships. Furthermore, our finding of elevated protease levels and rescue by pharmacological blockade in vitro in both pancreatic and hematopoietic lineages suggests a common functional pathway and potential therapeutic strategies in SDS.
We maintained hESC and iPSCs as undifferentiated cells on inactivated primary MEFs (Millipore). We reported derivation of iPSCs from patients with SDS previously (Park et al., 2008).
For siRNA vector, we cloned three oligonucleotides encoding stem-loop structures targeting SBDS under the control of the human U6 promoter in the pLentilox vector (Rubinson et al., 2003). The targeting sequences within SBDS are AACATGCTGCCATAACTTAGATT, AAGCTTGGATGATGTTCCTGATT, and AAGGAAGATCTCATCAGTGCGTT.
We produced lentiviral supernatants and infected hESCs. We enriched infected hESCs by FACS and manual picking of GFP positive colonies. We infected SBDSi hESC lines with a mixture of three knockdown viruses with different target sequences.
We cloned human SBDS cDNA into pSin-EF2-Nanog-Puro (Addgene), replacing the NANOG cDNA. We then transduced hESCs/iPSCs on Matrigel (BD) in MEF conditioned media by single round infections for 24 hours, followed by expansion onto MEFs and two rounds of puromycin selection (1 µg/ml) for 72 hours. We verified gene correction by Western blotting for SBDS (Abcam).
We injected NOD/SCID mice subcutaneously with 1–2×106 hESCs/iPSCs. We excised teratomas and analyzed them histologically in a blinded fashion for tissues from all three embryonic germ layers. We fixed pancreatic differentiation cultures for 20 minutes in 4% paraformaldehyde (PFA, Sigma-Aldrich) and stained with hematoxylin and eosin (H&E). We stained cytospins of erythroid body (EB) derived hematopoietic cells and liquid cultures with Wright-Giemsa. We used a Nikon Eclipse E400 microscope at 100X magnification and ImageJ software (http://rsbweb.nih.gov/ij/index.html).
We extracted iPSCs in polysome buffer, loaded extracts onto 5–50% sucrose density gradients (20 mM HEPES pH 7.4, 100 mM KCl, 10 mM MgCl2, 100 µg/ml cycloheximide, 200 µg/ml heparin), and centrifuged for 2 hours at 4°C in a SW41 rotor. We collected gradient fractions with a Teledyne Isco UA-6 with continuous UV monitoring at 254 nm. We quantitated areas under the curve for the 40S and 60S subunits and 80S ribosomal peaks using digitally captured A254nm absorbance readings (Burwick et al., 2012). Since we cultured SDS-iPSC-1 over a mouse embryonic fibroblast (MEF) layer, we harvested MEFs without iPSCs for polysome profiling and detected minimal background signals.
We cultured hESCs/iPSCs on MEF layers until 50% confluent, generally 1–2 days after splitting and differentiated as described previously (Chen et al., 2009) and repeated in detail in supplementary data. For endocrine development, we used a published protocol (Nostro et al., 2011).
For immunostaining, we fixed cells in 4% PFA for 20 minutes at room temperature and immunostained using the following primary antibodies: anti-FOXA2 (Millipore, 07–633; 1:500), anti-SOX17 (R&D system, AF1924, 1:500), anti-PDX1 (R&D Systems, AF2419; 1:500), anti-amylase (Novus Biologicals, NBP1–21420; 1:200), anti-DBA (Vector Labs, B-1035; 1:400), and anti-insulin (Abcam, ab7842; 1:100). We used Alexa-488–, Alexa-555–, and Alexa-647–conjugated secondary antibodies at 1:500 (Life Technologies) and obtained images on a BD Pathway using a 7×7 montage automatic capture to allow for unbiased acquisition. We scored images using ImageJ software and confirmed counts manually of test regions.
We obtained single cell suspensions from EB-differentiated hESCs/iPSCs as described previously (Chadwick et al., 2003). We stained for the markers CD45, CD15, CD11b (R&D Systems), and Glycophorin A (BD) for 30 minutes on ice followed by two washes with PBS and analyzed using a FACSCalibur (BD). We plated EB-differentiated hESC/iPSC populations (1×104–2×105) into methylcellulose H4434 (Stem Cell Technologies) and incubated at 37°C and 5% CO2 for 14 days.
We obtained multiple, independent biological samples for RNA isolations and extracted total RNA using the RNAEasy Plus kit (Qiagen). We prepared cDNAs using the Superscript VILO cDNA synthesis kit (Life Technologies) according to manufacturer's instructions. We performed RT-PCR with SYBR green reagent kits (Agilent Technologies) on a Stratagene Mx3005P quantitative PCR machine using primers from Integrated DNA Technologies. Primer sequences are listed in Table S4.
We used the EnzCheck Protease Assay Kit (Life Technologies) to measure protease activity from supernatants of day 25 pancreatic and day 20 EB hematopoietic cultures. Briefly, this kit involves incubating samples for 1 hour with intra-molecularly quenched substrate (BODIPY FL casein) and detecting fluorescently cleaved products on a microplate reader. The EnzCheck Protease Assay is broadly sensitive with the capacity to detect serine, acid, sulfhydryl, and metallo-proteases. We selected controls of media from undifferentiated hESCs and standardized against titrations of trypsin activity.
We treated pancreatic differentiation cultures with the following inhibitors every 2 days beginning on day 13 of differentiation: aprotinin (Sigma-Aldrich, A6103), protease inhibitor cocktail (Sigma-Aldrich, P1860), pan-caspase fmk inhibitor Z-VAD (R&D Systems, FMK001), and orlistat (Sigma-Aldrich, O4139). We resuspended day 20 EB-derived hematopoietic cells in Iscove’s modified Dulbecco’s media (IMDM) + 1% FBS and placed on a shaker for 12 hours with the following inhibitors: aprotinin, protease inhibitor cocktail, pan-caspase fmk inhibitor, sivelestat (Sigma-Aldrich, S7198-5MG), and myeloperoxidase inhibitor-I (Santa Cruz Biotechnologies, sc-204107). We calculated cell viability by trypan blue exclusion and performed annexin V and propidium iodide (PI) analyses of pancreatic cultures using AlexaFluor 488 Annexin V / Dead Cell Apoptosis Kit (Life Technologies).
We used student’s t test for pair-wise statistical analysis. p values are presented in Table S5.
We thank Shuibing Chen (Weill Cornell Medical College) and Doug Melton (Harvard University) for help with pancreatic differentiation protocols and Roderick Bronson (Harvard University) for his histopathological expertise. Harry Kozakewich (Boston Children’s Hospital [BCH]) provided SDS patient slides. The work was supported by the Manton Center for Orphan Disease Research (BCH), NIH P30 DK056465-10, NIH R01 HL079582-11, NIH R24 DK092760, NIH RC4 DK09091, NIH U01 HL100001, NIH UL1 DE019582, and the Roche Foundation for Anemia Research. GQD is an investigator of the Howard Hughes Medical Institute.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The authors declare no conflicts of interests.