Human Pancreatic Beta Cells Not Continually Expressed

The first rodent β cell lines were developed in the 1970s. They have been generated in different species: Rattus norvegicus (RIN and INS1 cells), derived from x-ray–induced insulinomas in rats (41, 42); Cricetus cricetus (HIT cells), obtained by simian virus 40 (SV40) transformation of adult hamster islet cells (43); and Mus musculus (βTCs and Min6 cells), produced by targeted oncogenesis in transgenic mice expressing an insulin promoter SV40 early region that contained the coding information for SV40 large T (SV40LT) and SV40 small t tumor antigens (44– 46). These cell lines have been very useful for detailed study of rodent β cells. In contrast, the generation of a functional human β cell line proved to be more complicated, and despite major investments, the undertaking remained frustrating. For example, the BetaLox-5 cell line, published in 1999, was derived from purified β cells of a human fetal pancreas at 24 weeks of development by transduction with retroviral vectors expressing the SV40LT antigen, the oncogene Hras^val12, and the human telomerase reverse transcriptase (hTERT) (47– 49). However, this line expressed low levels of insulin (INS) and was unstable, with INS expression decreasing even further with passages, which explains why it was not frequently used by the scientific community. In 2005, there was a seemingly major breakthrough in the field as a multinational team published a reversibly transformed human β cell line. The NAKT-15 cell line was derived by Moloney murine leukemia virus (MoMLV) long terminal repeat–driven expression of SV40LT and hTERT in islet cells. NAKT-15 cells were highly similar to primary human β cells in terms of expression of key transcription factors, insulin production, and glucose responsiveness. It was even possible to excise the SV40LT and hTERT oncogenes in order to obtain postmitotic human β cells (50). At the time of publication, the NAKT-15 cell line drew the attention of the scientific community, as it represented a powerful tool for diabetes research ( 51). However, some points in the work remain unclear. First, NAKT-15 cells were generated upon transduction of adult human islets with γ-retroviral vectors. It is well established that these vectors can only stably transduce proliferating cells ( 52). However, adult human β cells are mostly if not exclusively postmitotic. A recent study found no β cell proliferation, as determined by Ki-67 immunostaining, among a total of 37,845 β cells analyzed on pancreata from 18 nondiabetic subjects ( 53). Second, when expressed by adult human β cells, SV40LT is inefficient in activating markers of proliferation ( 11, 54). Finally, to our knowledge, the NAKT-15 cell line has only been used in two original publications ( 55, 56), and it did not find its way into diabetes laboratories around the world.

In the early 2000s, we decided to try to generate human β cell lines. Our protocol was based on the successes and failures of previous attempts (44– 46, 49, 50). We chose human fetal pancreata as starting material because they were thought to be more prone to immortalization than adult pancreas. We reasoned that a higher proliferation rate in fetal cells would render them more susceptible to both transduction and the pro-proliferative effect of the introduced oncogenes. Lentiviral vectors were used for transduction, and SV40LT and hTERT were chosen for immortalization. At that time this was a risky choice, since human cell transformation was considered to require the combined expression of SV40 early region, hTERT, and an oncogenic mutant of H-RAS (Hras^val12) (57). In some settings, however, hTERT alone has been used successfully to immortalize neural progenitor cells derived from human fetal spinal cord ( 58). In our approach, we included a 405-bp fragment of the rat insulin-2 promoter (RIP) to drive the expression of SV40LT and hTERT. This promoter fragment was well characterized and displays specificity and efficiency in both rodent and human pancreatic β cells ( 59– 61). Overall, our approach based on targeted oncogenesis in fetal pancreatic tissues resembled the one used to derive the βTC and Min6 cell lines from transgenic mice. As access to human fetal pancreatic fragments is limited, we validated each step of the process using rat fetal pancreata as starting material ( 62).

We next used this protocol to generate an initial human β cell line. Drawings of this protocol can be seen in ref. 8 and ref. 12. First, we transduced fragments of human fetal pancreata (7–11 weeks of gestation) with a lentiviral vector expressing SV40LT. At this early stage of development, the human pancreas is mainly composed of progenitors and few insulin-containing cells ( 63– 65). Our hypothesis was that the SV40LT oncogene would integrate randomly into the pancreatic cells, but that it would only be expressed when those progenitors differentiated into INS-expressing β cells (the oncogene being under the control of the insulin promoter). Next, we transplanted the transduced human pancreatic fragments into immune-compromised (SCID) mice, which formed a permissive environment for human fetal pancreatic progenitor growth and differentiation (66, 67). In the case of transduction of a β cell progenitor, this resulted in the development of an insulinoma. Next, the insulinomas were transduced with a lentiviral vector expressing hTERT and again grafted into SCID mice to amplify the proliferating β cells. Finally, the insulinomas were surgically removed, dissociated into single cells, and expanded in culture as cell lines. Different candidate cell lines were obtained, all of which had a remarkably similar expression of many key β cell transcription factors, an almost constant expression of SV40LT, but a variable expression of INS. One of these cell lines was characterized in more detail and became the first human β cell line, EndoC-βH1 (11). EndoC-βH1 cells resemble human β cells in their expression of many specific genetic and epigenetic markers ( 68), in the limited expression of markers of other pancreatic cell types (some rare EndoC-βH1 cells contain somatostatin, but none contain glucagon; ref. 11), and in their ability to secrete insulin in response to glucose, the glucagon-like peptide 1 analog exendin-4, sulfonylureas such as glibenclamide, and branched-chain amino acids such as leucine ( 11). Electrophysiological properties of EndoC-βH1 also conform with human β cells ( 69).

The first generation of EndoC-βH cells was followed by a second generation. In these EndoC-βH2 cells, two loxP sites flank the integrated oncogenes, which allows their excision upon Cre recombinase expression (70). Next, we generated a third-generation EndoC-βH3 by stable integration of a tamoxifen-inducible form of Cre into the EndoC-βH2 cells, which makes them an easy-to-use excisable human β cell line ( 71). We also used the EndoC-βH2 line to generate human Fucci β cells, in which two fluorescent reporters alternate as the cells progress through the cell cycle ( 72). These human Fucci β cells can be used to study the human β cell cycle and differentiation. More recently, we used a similar approach to generate a new human β cell line, ECN90, derived not from fetal but from neonatal pancreas (4 months old) ( 73). There, we presumably directly transduced the β cells that are present at this early postnatal age and have a substantial replicative potential ( 74, 75). We are convinced that our approach does not work on postmitotic adult human β cells, even when effectively transduced by the lentiviral vectors ( 11). This situation is reminiscent of the outcome of experiments aimed at reactivating proliferation in postmitotic muscle cells or human β cells, namely a cell cycle block due to severe DNA damage, despite an initial triggering of DNA synthesis ( 76, 77). Accordingly, SV40LT and hTERT may be unable to reactivate proliferation in adult human β cells as they trigger an abortive cell cycle reentry; only replicating progenitor cells or immature neonatal β cells can be immortalized using our strategy.

Cell line misidentification, contamination, and poor annotation are recurrent problems that affect scientific reproducibility (78, 79). To allow the detection of such problems, we analyzed short tandem repeats on each cell line at early passage for authentication, and these data are available for comparison. We have now shared the EndoC-βH cell lines with more than 150 laboratories worldwide (Figure 1). Based on feedback of colleagues and on their use in more than 100 original publications, we think that they represent useful tools for the academic scientific community. A regularly updated list of publications that have used the human β cell lines we developed can be found at https://www.humanbetacelllines.com/ The cell lines have already proved to be useful in generating new knowledge on human β cells. Data from rat islets and rat β cell lines had shown, for example, that inflammatory cytokines induce iNOS expression and thereby increase the production of nitric oxide, resulting in β cell death (80). A similar mechanism was presumed in humans, but a recent study, using EndoC-βH1 cells, demonstrated that cytokines do not induce iNOS expression in human β cells ( 81). The EndoC-βH1 cells were also used to show that in human β cells the methyltransferases mixed-lineage leukemia 3 and 4 (MLL3/4) bind to the transcription factors MAFA and MAFB, and that the complexes formed thereby are necessary for proper glucose-induced insulin secretion ( 82). Another example for which the EndoC-βH1 cells were crucial is the demonstration of physical contact between the insulin promoter and diabetes susceptibility loci and the insulin promoter's involvement in the regulation of insulin transport and metabolism in human β cells ( 83). Peptidomic-based approaches have also used the human β cell lines to identify target epitopes processed and presented by β cells, thereby providing the first HLA-I peptidome catalog of human β cells ( 84). Finally, our group used EndoC-βH1 cells to develop models and identify markers of human β cell dedifferentiation ( 85, 86). All of the above examples depended on experiments that would have been difficult to perform with human islets because of limitations in cell purity or cell number and that would have produced different results in rodent β cell lines because of species differences. Interestingly, human β cell lines have now been validated by the pharmaceutical industry as a screening model to identify novel drug target candidates ( 87); to determine the effects of fatty acid esters of hydroxylated fatty acids on glucose-stimulated insulin secretion ( 88); to study the mechanism of secretagogin release by human β cells ( 89); and to further test a gastrointestinal peptide–based (GIP-based) dual incretin receptor agonist ( 90).

Figure 1

Distribution of human β cell lines worldwide.

Notably, at the time we published EndoC-βH1, three additional human insulin-releasing cell lines (1.1B4, 1.4E7, and 1.1E7) were generated by electrofusion of human pancreatic β cells with the immortal human PANC-1 cell line that was established from a pancreatic carcinoma of ductal origin (91). These human insulin-releasing cell lines expressed the expected set of human β cell markers and were used to dissect human β cell function and survival ( 92– 95). However, their insulin content was extremely low, around 4 ng per million cells, which is less than a thousandth of the insulin content measured in primary human β cells and 150-fold less than in EndoC-βH1 cells.

EndoC-βH cells have been used recurrently for multi-omic profiling (68, 70, 96, 97). As a pure human β cell population, they can be used to validate the expression of genes of interest. Surprisingly, a number of proteins detected in primary human β cells by immunostainings are not detected at the transcript level in EndoC-βH cells or primary human β cells. For example, TLR4 ( 98) and the immunoregulatory antimicrobial peptide CRAMP ( 99) are not expressed in EndoC-βH cells. Notably, their mRNAs are also absent or at the lowest limit of detection in bulk transcriptional analyses of human islets ( 27, 100) and undetectable by single-cell transcriptomics ( 29). Such data contrast sharply with the immunostainings on human islet sections, which show very strong signals for TLR4 and CRAMP in β cells ( 98, 99). Another example is TGF-β–induced (βig-h3, or TGF-βI), a protein for which immunostainings have been shown in human β cells ( 101), but again for which the coding mRNA is not found in human β cell lines or human β cells at the single-cell level ( 29). βig-h3 mRNA was, however, detected in human islets ( 15), which suggests that it may be expressed by non-β islet cells. A final example concerns CFTR, a chloride channel mutated in patients with cystic fibrosis ( 102), which is detected in islet preparations ( 100). However, we know that duct cells, which always contaminate islet preparations ( 103), express high levels of CFTR (29). While multiple studies report CFTR expression and expand on its function in rodent β cells (104, 105), more recent work debates an intrinsic role of CFTR in β cells ( 106). We did not detect CFTR mRNA in any of the human β cells, and single-cell RNA-Seq on human islet preparations also showed that β cells barely express CFTR, whereas duct cells express CFTR at high levels (29). Overall, the above-described discrepancies may be due to insufficient antibody validation ( 107, 108) or to other hitherto unknown reasons that may be related to islet or pancreatic tissue preparation. Importantly, the above examples illustrate the complexity of working on primary human pancreata and islet preparations. In the study of human β cells, EndoC-βH cells can offer a robust and reproducible system that is particularly useful for screening before moving on to primary cells or for dissecting mechanistic aspects of experimental findings. The key advantages and limitations of EndoC-βH cells as compared with primary human β cells and stem cell–derived β cells are shown in Table 1.

Table 1

Key advantages and limitations of primary human β cells, stem cell–derived β cells, and EndoC-βH cell lines

futralbeforge.blogspot.com

Source: https://www.jci.org/articles/view/129484

Share this post