Expression of mutant mRNA and protein in pancreatic cells derived from MODY3- iPS cells

Maturity-onset diabetes of the young (MODY) is a heterozygous monogenic diabetes; more than 14 disease genes have been identified. However, the pathogenesis of MODY is not fully understood because the patients’ pancreatic beta cells are inaccessible. To elucidate the pathology of MODY, we established MODY3 patient-derived iPS (MODY3-iPS) cells using non-integrating Sendai virus (SeV) vector and examined the mutant mRNA and protein of HNF1A (Hepatocyte Nuclear factor 1A) after pancreatic lineage differentiation. Our patient had a cytosine insertion in the HNF1A gene (P291fsinsC) causing frameshift and making a premature termination codon (PTC). We confirmed these MODY3-iPS cells possessed the characteristics of pluripotent stem cells. After we differentiated them into pancreatic beta cells, transcripts of HNF1A gene were cloned and sequenced. We found that P291fsinsC mutant transcripts were much less frequent than wild ones, but they increased after adding cycloheximide (CHX) to the medium. These results suggested that mutant mRNA was destroyed by nonsense-mediated mRNA decay (NMD). Moreover, we were not able to detect any band of mutant proteins in pancreatic lineage cells which were differentiated from MODY3-iPSCs by western blot (WB) analysis. A scarcity of the truncated form of mutant protein may indicate that MODY3 might be caused by a haplo-insufficiency effect rather than a dominant negative manner.


Introduction
Maturity-onset diabetes of the young (MODY) is an autosomal dominant form of monogenic diabetes that stems from one or more mutations in a single gene. To date, 14 [1,2]. Although MODY is considered to be a genetic a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 using SeV vectors (MBL, Nagano, Japan) overnight. These cells were washed and cultured in DMEM supplemented 10% fetal bovine serum (FBS) for 6 days. Then SeV-infected fibroblasts were seeded on mitomycin (MMC; Wako, Osaka, Japan) treated mouse embryonic fibroblast (MEF) feeder cells. The next day, the medium was replaced by hiPS medium (DMEM/F'12 supplemented with 20% Knockout serum replacement (KSR; GIBCO BRL, Palo Alto, CA, USA), 2 mM L-glutamine (Wako), 0.5x penicillin/ streptomycin (Wako), 1x non-essential amino acids (Wako), 55 μM 2-mercaptoethanol (Gibco) and 7.5 ng/ml FGF2 (Peprotech, Rocky Hill, NJ, USA)). Three to four weeks later, primary hiPS colonies appeared. We transferred each colony onto MMC treated SNL feeder cells (ECACC, Salisbury, UK). These hiPS colonies were maintained on MMC treated SNL feeder cells and passaged using CTK solution at 1:5-1:8 once a week as described previously [31]. Control -iPSCs had previously been established from healthy donor fibroblasts [30]. These experiments were carried out with the approval of ethical committees in Tokyo Women's Medical University and in National Center for Global Health and Medicine.

Differentiation protocol
For sequencing analysis, differentiation from hiPS cells into pancreatic beta cells was based on the Maehr's method [32] with minor modifications [30] using adherent culture.

CHX treatment
MODY3-iPSCs were treated with 0.25 mM cycloheximide (CHX) for 18hr at the beta stage.

Teratoma assay
Animal studies were conducted according to protocols approved by the Animal Care and Use Committee in the National Center for Global Health and Medicine. Eight-week old NOD-S-CID mice (JAPAN Clea) were used for experiments. Undifferentiated MODY3-iPS cells (one x 10 6 cells) were grafted into a testis (under the capsule) of SCID mice anesthetized with Sevoflurane. After 10-12 weeks, teratomas were extirpated and processed according to standard protocol for paraffin embedding and hematoxylin and eosin staining.

Karyotype analysis
Three different colonies of MODY3-iPSCs were picked up, and, over the 10th passage, each cell line containing 1 x 10 6 cells was sent to MBL Research Laboratories, Japan, where chromosomal G-band analysis was performed.

Establishment of MODY3-iPSCs
Genetic profiles of the MODY3 patient are shown in Fig 1A. To establish transgene free MODY3-iPSCs, patient-derived skin fibroblasts were transduced with SOX2, OCT3/4, KLF4 and C-MYC using non-integrating Sendai virus. Four weeks later, MODY3-iPS colonies emerged ( Fig 1B) and we picked up several colonies and passaged them. Although the SeV genome was present in the cytoplasm of MODY3-iPS colonies at passage 3, no more SeV genome was detected at passage 10 by immunocytochemistry and RT-PCR (Fig 2M, 2N and 2R), indicating that these MODY3-iPSCs were transgene-free. Next, MODY3-iPSCs were characterized by immunocytochemistry and alkaline phosphatase staining. MODY3-iPSCs expressed the undifferentiated pluripotency markers OCT3/4, NANOG, SSEA4, TRA-1-60, and TRA-1-81 and had alkaline phosphatase activity (Fig 2B-2L). To assess the ability to form teratomas, three different cell lines from MODY3-iPSCs were injected into the testes of SCID mice. Large tumors developed three months later. Histological analysis revealed that all the tumors contained various tissues derived from all three germ layers; ectoderm, mesoderm and endoderm (Fig 2M-2O). We confirmed that the karyotype of MODY3-iPSCs was normal (Fig 2S).

Transcripts with PTC are destroyed in differentiated MODY3-iPSCs
In our MODY3 case, a P291fsinsC frameshift mutation of the HNF1A gene was the same as reported previously [26]. To detect the mutant mRNA in MODY3-iPS derived cells, we differentiated MODY3-iPSCs into pancreatic lineage cells, cloned mutant mRNA by RT-PCR and then sequenced them. Because the P291fsinsC mutation causes a frameshift by the insertion of cytosine, the genomic sequence data should have both the clear-cut signal of the wild type and a one-base-shifted same pattern signal from the mutation site to the 3' end. We confirmed this, as shown in Fig 3B. In contrast to the genomic sequence, the strong wild type signal and the faint one-base-shifted mutant mRNA signals were detected in pancreatic lineage cells from MODY3-iPSCs ( Fig 3C). These data indicate that the P291fsinsC mutant mRNA was destroyed in differentiated MODY3-iPSCs. We confirmed that the sequence of HNF1A mRNA derived from healthy-iPSCs had only clear wild type transcript signals (Fig 3E).

Restoration of mRNAs by inhibiting NMD with cycloheximide treatment
Degradation of transcripts with PTC in vivo by a NMD pathway known as RNA surveillance system has been suggested. Cycloheximide (CHX) is regarded as an inhibitor of NMD [34]. When Ullrich disease fibroblasts were treated with CHX, they restored collagen VI a2 mutant transcripts containing PTC [35]. We thought that the low amounts of MODY3 mutant mRNAs might be caused by NMD. To confirm this, differentiated MODY3-iPSCs were treated with CHX to inhibit NMD for 18hr before collection of the cells. CHX treatment clearly enhanced the sequence signal of P291fsinsC mutant mRNAs compared with non-treated samples ( Fig 3D). We then tried to quantify the expression level of mutant RNAs before and after CHX treatment. Although we designed several taq-man probes against P291fsinsC, this attempt was not successful, because none of the tested taq-man probes could distinguish between wild and mutant mRNAs.

MODY3-iPSCs were differentiated to pancreatic beta cells using suspension culture
As P291fsinsC mutant mRNA was destroyed by NMD, we further examined the mutant gene expression at the protein level. We first tried to check the HNF1A protein by western blot using a previously reported adherent culture protocol [30,31], but could not detect it. One possible reason is that differentiation efficiency was not enough for detection. We quite recently reported that suspension culture promoted differentiation toward definitive endoderm more efficiently than adherent culture and that iPS-derived beta cells in suspension culture effectively reversed hyperglycemia in diabetic mice [33,36], so we differentiated Expression of mutant mRNA and protein in pancreatic cells derived from MODY3-iPS cells MODY3-iPSCs into pancreatic beta cells using this suspension culture protocol (Fig 4A). qPCR showed that HNF1A mRNA expression gradually increased toward pancreatic beta cell differentiation (Fig 4B). We obtained MODY3-iPS beta cell spheroids most of which were 200 ±50 μm in diameter (Fig 4C). These MODY3-derived beta cells were positive for insulin c-peptide, PDX1, NKX6.1 and HNF1A at the protein level Fig 4D and 4E, indicating these cells had successfully differentiated into pancreatic beta cells.

P291fsinsC mutant proteins were minimally translated into MODY3-iPSbeta cells
To detect the truncated form of P291fsinsC mutant proteins, we made an artificial construct which contained mutant cDNA for HNF-1a without any intron sequences, as shown in Fig 5A.  When mutant mRNA is transcribed from expression vector, splicing doesn't occur due to the lack of introns. Since PTC of mutant mRNA is recognized as original stop codon, mutant mRNA is supposed to be stabilized, and mutant proteins are stably translated (Fig 5A). On the other hand, in vivo, P291fsinsC mutant mRNA is transcribed from genomic DNA and spliced to remove introns (Fig 6). This splicing is regarded critically important for distinguishing between PTC and the original stop codon. Once PTC is recognized, PTC-bearing mRNA is destroyed by NMD, and mutant protein is minimally translated from mutant mRNA (Fig 6). Expression of mutant mRNA and protein in pancreatic cells derived from MODY3-iPS cells Therefore, it is generally thought that the amount of mutant protein translated from PTC-bearing mRNA is less than that of wild type protein in vivo.
To compare the amount of mutant and wild protein, we tried to detect those which were translated from expression vectors or genome of MODY3-iPSCs by western blot. P291fsinsC mutant protein was shorter than wild type protein due to C-terminal truncation, but the N-terminal sequence was conserved between wild and mutant proteins (Fig 5B). Hence, we performed western blot analysis using N-terminal region recognizing antibody to detect and Expression of mutant mRNA and protein in pancreatic cells derived from MODY3-iPS cells distinguish wild (632 aa) and mutant (315 aa) proteins by molecular weight. We were able to detect strong bands of wild and mutant protein in the lysates of LentiX293T cells transfected with wild or mutant HNF1A expression vectors, respectively. In contrast there are no bands of mutant proteins in the lysates of pancreatic lineage cells differentiated from MODY3-iPSCs ( Fig 5C). We previously reported that establishment of MODY5 (R177X)-iPSCs and R177X mutant RNA was also destroyed by NMD due to PTC [30,37,38]. Therefore, we analyzed protein levels for these MODY5-iPSCs by WB in the same way. Definite bands of wild and mutant proteins were detected in transfected cells. On the other hand, in pancreatic lineage cells differentiated from MODY5-iPSCs, a clear-cut signal of wild type protein was detected, but no mutant protein (S1 Fig). These data indicate that P291fsinsC and R177X mutant proteins, both of which were produced from PTC-bearing mRNAs, barely existed in the pancreatic lineage cells differentiated from MODY-iPSCs

Discussion
In this study, we successfully established MODY3-iPSCs from Japanese patients by using SeV. It is advantageous to use SeV for generating iPS cells because unlike retro or lentivirus, SeV doesn't have the ability to insert transgenes into a host genome [39,40] and preservation of the genomic structure is guaranteed for investigating the pathogenesis of hereditary disease. Initial establishment of Caucasian MODY(1, 2, 3, 5 and 8)-iPS cells employed Cre-excisable polycistronic lentivirus [10]. In that system, transgene-free MODY-iPSCs can be obtained by removing the genome-integrated transgenes surrounded by two loxP sites using Cre recombinase. However, 5' and 3'viral LTR sequences outside the loxP sites cannot be excised, so that these viral LTR sequences remain in the genomes of these MODY-iPSCs even after the removal of the transgenes. Another group established Caucasian MODY2-iPSCs using non-integrating SeV and examined glucokinase activity of the cells [11]. Since the disease gene, HNF1A, of our MODY3-iPSCs is categorized as a transcription factor, while glucokinase in MODY2-iPSCs is categorized as an enzyme, it is suggested that different mechanisms may contribute to the pathophysiology of MODY.
As mentioned earlier, various mutations of several disease genes can develop MODY [16][17][18]. Possible effects of a mutant protein have been suggested by in vitro experiments [19,21,[41][42][43][44][45]. In those studies, mutant proteins were examined in various cell lines. including MIN6, HepG2 and Caco-2 cells, and suggested to have decreased, non-functional, or dominant negative activity. However, these expression vector constructs used for in vitro experiments have one critical problem when examining the effects of PTC. Although a 3'-intron flanking exon after PTC sequence and splicing are required for the recognition of PTC in the NMD pathway [27,28], these expression vector constructs have only one exon; therefore, it remains unclear whether the PTC-bearing mRNA is correctly destroyed by NMD in those assays.
It is estimated that PTC is responsible for 30% of known disease-associated mutations [28,46]; in the case of MODY3, it is reported that about 30% of mutations have PTC [47]. Importantly, in vitro studies of some truncated proteins have not always correlated with the mutant phenotype in vivo [48]. Additionally, escaping the ablation of PTC-bearing mRNA from NMD causes severe clinical phenotypes, such as peripheral demyelinating neuropathy, central dysmyelinating leukodystrophy, and Waardenburg syndrome, Hirschsprung disease (PCWH), or beta-thalassemia [48,49]. In the case of EJC-dependent NMD including P291fsinsC and R177X, only one molecule mutant protein can be translated from each mutant mRNA in phosphorylation of SURF complex, which leads to mRNA degradation. NMD; non-sense mediated mRNA decay, PTC; premature termination codon, EJC; exon junction complex. https://doi.org/10.1371/journal.pone.0217110.g006 Expression of mutant mRNA and protein in pancreatic cells derived from MODY3-iPS cells principle, because pioneer round translation occurs once before mRNA degradation (Fig 6). Therefore, the putative amount of mutant protein should be very small, if any. Given these facts, a proper assay system which examines disruption of transcripts with PTC in the NMD pathway and amount of mutant protein is important for an elucidation of the molecular mechanism of pathogenesis.
In our previous and present research, we investigated the behavior of the mutant genes using MODY-iPS cells: MODY3 (P291fsinsC) and MODY5 (R177X) [30]. Although we found that sequence signals of P291fsinsC and R177X mutant transcripts were weaker than that of wild type transcripts during the differentiation progression from MODY-iPSCs to beta cells, these mutant transcripts increased with CHX treatment (Fig 3), [30], indicating that the degradation of mutant mRNA was brought by NMD. Moreover, we demonstrated by western blot analysis that amounts of these mutant proteins were markedly less than that of wild type proteins in pancreatic lineage differentiated MODY-iPSCs (Fig 5, S1 Fig). In our previously established-MODY5 case, the nonsense mutation of R177X in the HNF1B gene produced nonfunctional or dominant negative protein, dependent on the cell lines used in the in vitro assay [43][44][45], suggesting that this MODY might be caused by a haplo-insufficiency or a dominantnegative pattern. It was reported that P291fsinsC mutant proteins acted as dominant-negative mutant in vitro assays, and these MODY might be caused by dominant-negative effects [19,21]. Our data showed that mutant proteins didn't exist within the detectable range, which indicates very low translation, strongly suggesting mutant mRNA degradation by NMD. From these data, the existence of a dominant negative effect caused by truncated protein seems less likely.
On the other hand, mutant HNF1A mRNA is expressed in hepatic cell as well as pancreatic cell and it is possible that it must also be degraded in hepatic cell by NMD, although we did not examine the expression of mutant mRNA and protein in hepatic lineage cells. In this patient, disfunction of insulin secretion from pancreatic beta cells was observed, but functions of hepatic cells were clinically almost normal. From the clinical standpoint of view, we think some compensation may occur in hepatic cells even if HNF1A mRNA is degraded, which should be proved in the future. Taken together, our results suggest that MODY3 (P291fsinsC) and MODY5 (R177X) are caused by the haplo-insufficiency effect rather than a dominant negative manner.
This approach is useful for studying other genetic diseases caused by a mutant transcript with PTC that is expressed in unavailable tissue like the central nervous system. Differentiating disease specific hiPSCs into previously unavailable tissues enabled us to examine the stability of a mutant mRNA with PTC and to compare the amount of mutant and wild type protein, leading to a more accurate understanding of pathogenesis.