Characterization of Non-hormone Expressing Endocrine Cells in Fetal and Infant Human Pancreas

Context: Previously, we identified chromograninA positive hormone-negative (CPHN) cells in high frequency in human fetal and neonatal pancreas, likely representing nascent endocrine precursor cells. Here, we characterize the putative endocrine fate and replicative status of these newly formed cells. Objective: To establish the replicative frequency and transcriptional identity of CPHN cells, extending our observation on CPHN cell frequency to a larger cohort of fetal and infant pancreas. Design, Setting, and Participants: 8 fetal, 19 infant autopsy pancreata were evaluated for CPHN cell frequency; 12 fetal, 24 infant/child pancreata were evaluated for CPHN replication and identity. Results: CPHN cell frequency decreased 84% (islets) and 42% (clusters) from fetal to infant life. Unlike the beta-cells at this stage, CPHN cells were rarely observed to replicate (0.2 ± 0.1 vs. 4.7 ± 1.0%, CPHN vs. islet hormone positive cell replication, p < 0.001), indicated by the lack of Ki67 expression in CPHN cells whether located in the islets or in small clusters, and with no detectable difference between fetal and infant groups. While the majority of CPHN cells express (in overall compartments of pancreas) the pan-endocrine transcription factor NKX2.2 and beta-cell specific NKX6.1 in comparable frequency in fetal and infant/child cases (81.9 ± 6.3 vs. 82.8 ± 3.8% NKX6.1+-CPHN cells of total CPHN cells, fetal vs. infant/child, p = 0.9; 88.0 ± 4.7 vs. 82.1 ± 5.3% NKX2.2+-CPHN cells of total CPHN cells, fetal vs. infant/child, p = 0.4), the frequency of clustered CPHN cells expressing NKX6.1 or NKX2.2 is lower in infant/child vs. fetal cases (1.2 ± 0.3 vs. 16.7 ± 4.7 clustered NKX6.1+-CPHN cells/mm2, infant/child vs. fetal, p < 0.01; 2.7 ± 1.0 vs. 16.0 ± 4.0 clustered NKX2.2+-CPHN cells/mm2, infant/child vs. fetal, p < 0.01). Conclusions: The frequency of CPHN cells declines steeply from fetal to infant life, presumably as they differentiate to hormone-expressing cells. CPHN cells represent a non-replicative pool of endocrine precursor cells, a proportion of which are likely fated to become beta-cells. Precis: CPHN cell frequency declines steeply from fetal to infant life, as they mature to hormone expression. CPHN cells represent a non-replicative pool of endocrine precursor cells, a proportion of which are likely fated to become beta-cells.

Context: Previously, we identified chromograninA positive hormone-negative (CPHN) cells in high frequency in human fetal and neonatal pancreas, likely representing nascent endocrine precursor cells. Here, we characterize the putative endocrine fate and replicative status of these newly formed cells.
Objective: To establish the replicative frequency and transcriptional identity of CPHN cells, extending our observation on CPHN cell frequency to a larger cohort of fetal and infant pancreas.

INTRODUCTION
In humans, most beta cell growth and development occurs during gestation and early life (1)(2)(3). Beta-cells are first detected at 9 weeks gestation, with fractional beta-cell area increasing linearly throughout gestation accompanied by a readily detectable frequency of beta-cell replication (1). This implies a key role of replication in the prenatal expansion of beta-cells and in establishment of beta-cell mass. After birth, a high frequency of beta-cell replication in infancy further contributes to the expansion of beta-cell mass, but replication of beta-cells declines rapidly with increasing age and, after the age of 3-5 years, is negligible in most cases (2). By contrast, beta-cell apoptosis has been reported to be low during mid-gestation, rising during the perinatal period and falling again in infancy (3); although later studies are contradictory, detecting a higher frequency of apoptosis during gestation (1) and no perinatal rise in apoptosis in humans (2).
Recently, we reported a high frequency of chromograninApositive hormone-negative (CPHN) cells in fetal human pancreas, whose presence falls rapidly after birth, and these cells detectable only at very low frequency in adult non-diabetic human pancreas (4). Further, we have demonstrated that these cells are more frequent in the setting of both type 1 and type 2 diabetes, perhaps indicating an attempt, albeit insufficient, at regeneration of beta-cells lost to disease (4)(5)(6).
It has been established that endocrine cells arise from pancreatic and duodenal homeobox (PDX)-1-positive precursors during gestation (7,8), and that beta-cell replication then plays a major role in establishing the beta-cell complement (1,(9)(10)(11)(12)(13). Contributions from other mechanisms, such as neogenesis and transdifferentiation, have been proposed (9,14,15). Moreover, Sarkar et al. have suggested that replication of existing betacells is insufficient to account for the increase in beta-cell mass during human gestation and therefore that a significant proportion of endocrine expansion during fetal life must derive from replication of hormone-negative precursors (13).
CPHN cells may represent a pool of endocrine cells, largely expressing beta cell transcription factors and a proportion of which are likely fated to become beta-cells. While they are present in highest frequency during gestation, they only occur at low levels throughout life, and their numbers can increase in response to endocrine deficiency, as seen in diabetes (16)(17)(18). The purpose Abbreviations: CPHN, chromograninA positive hormone negative; NKX, homeobox protein family; PDX, pancreatic and duodenal homeobox; PDG, pancreatic duct glands. of this study was to quantify the frequency of CPHN cells in human fetal and infant pancreas, as well as to characterize this population in terms of their replicative capacity and markers of endocrine cell identity. Four Micrometer paraffin tissue sections from each subject were stained for chromograninA, insulin, glucagon, somatostatin, pancreatic polypeptide, and ghrelin. Standard immunohistochemistry protocols were used for fluorescent immuno-detection of various proteins in pancreatic sections, as previously described (5).

Design and Case Selection
In brief, one pancreas section obtained from the tail of pancreas was analyzed for each subject. Briefly, slides were incubated at 4 • C overnight with a cocktail of primary antibodies prepared in blocking solution (3% BSA in TBST) at the following dilutions: Slides were viewed using a Leica DM6000 microscope (Leica Microsystems, Deerfield, IL) and images were acquired using the 20x objective (200x magnification) using a Hamamatsu Orca-ER camera (C4742-80-12AG, Indigo Scientific, Bridgewater, NJ) and Openlab software (Improvision, Lexington, MA).

Morphometric Analysis
One section of the pancreas per subject was stained with appropriate primary and secondary antibodies. Fifty islets per subject were imaged at 20x magnification. An islet was defined as a grouping of four or more endocrine cells. A cluster was defined as a grouping of three or fewer chromogranin A positive cells. Islets were selected by starting at the top left corner of the pancreatic tissue section and working across the tissue from left to right and back again in a serpentine fashion, imaging all islets in this systematic excursion across the tissue section. Analysis was performed in a blinded fashion (ASMM, CM, and AEB), and all CPHN cells identified were confirmed by a second observer. The endocrine cells contained within each islet were manually counted and recorded as follows: (1) the number of cells staining for chromogranin A, (2) the number of cells staining for the endocrine hormone cocktail, and (3) the number of cells staining for insulin. Thus, cells staining for chromogranin A but not the other known pancreatic hormones (insulin, glucagon, somatostatin, pancreatic polypeptide, or ghrelin) were noted. At 200x magnification, using the Leica DM6000 with a Hamamatsu Orca-ER camera and a 0.7x C-mount, each field of view was calculated to be 0.292 mm 2 . Within the fields imaged to obtain the 50 islets per subject, all single endocrine cells and clusters of endocrine cells (two or three adjacent endocrine cells) were counted and recorded as outlined above.

Assessment of Replication and Endocrine Transcription Factors in ChromograninA Positive Hormone-Negative [CPHN] Cells
To determine replication and transcription factor expression in CPHN cells, we utilized our previously published protocol (5). Briefly, we developed and used a new immunohistochemical staining technique involving monovalent F(ab ′ )2 fragments to distinguish between the two mouse primary antibodies.
Slides were viewed and imaged as described above. CPHN cells (located in islets, clusters and in single cells) that express NKx6.1, NKx2.2, or Ki67 were identified (by following the similar procedure of detecting CPHN cells in the different compartments of pancreas) and documented blindly by two independent researchers (ASMM and CM).

Immunohistochemical Staining of Ductal Structures for Replication and Endocrine Cell Subtypes
Adjacent sections of pancreas were stained for (1) Ki67, insulin and glucagon and (2)

CPHN Quantification
Fifty islets per subject were imaged at 20x magnification. An islet was defined as a grouping of four or more chromograninA positive cells. A cluster was defined as a grouping of three or fewer chromograninA positive cells. The endocrine cells contained within each islet were manually counted as previously described (5). The mean number of endocrine cells counted within islets for the fetal group was 782 ± 50 cells per subject and for the infant group was 1,514 ± 151 cells per subject. The mean number of cells counted in clusters for the fetal group was 156 ± 32 cells per subject, and for the infant group was 81 ± 7 cells per subject. The mean number of chromograninA positive hormone-negative [CPHN] cells per individual identified in islets from the fetal subjects was 47.3 ± 11.4 cells per individual and from the infant group was 15.6 ± 1.9 cells per individual. The mean number of CPHN cells per individual identified in clusters from fetal subjects was 47 ± 11 and from infant subjects was 15 ± 2 cells per individual. At 200x magnification, using the Leica DM6000 with Hamamatsu Orca-ER camera and a 0.7x C-mount, each field of view was calculated to be 0.292 mm 2 . Within the fields imaged to obtain the fifty islets per subject, all clusters of endocrine cells (one, two, or three adjacent endocrine cells) were counted and recorded as outlined above.

Quantification of Replication and Expression of NKX6.1 and NKX2.2 in CPHN Cells
To investigate the potential endocrine cell lineage of CPHN cells in fetal and infant/children pancreas, we evaluated Ki67 as a replication marker (mouse anti-Ki67, 1:50; RRID:AB_2142367; catalog no. M7240; Agilent Technologies), NKX2.2 as a panendocrine transcription factor (mouse anti-NKX2.2, Frontiers in Endocrinology | www.frontiersin.org 1:50; RRID:AB_531794; catalog no. 74.5A5; DHSB), and NKX6.1 as a β-cell transcription factor (mouse anti-NKX6.1, 1:300;RRID:AB_532378; catalog no. F55A10; DSHB) CPHN cells were identified as described previously (20). To assess replication and the presence of the transcription factors NKX2.2 and NKX6.1 in CPHN cells, 20 fields were viewed with a Zeiss Axioskop 2 microscope (Carl Zeiss Microscopy, Thornwood, NY) and images acquired using an Axiocam MR3 camera and Axiovision 4.0 software (Carl Zeiss Microscopy, Thornwood, NY). Using this microscope, camera and software, each field of view has an area of 0.42 mm 2 . CPHN cells that express Ki67 or NKX6.1 or NKX2.2 were counted in three different compartments (within islets, as clustered cells or as single cells) of a pancreas section. Data in islets were expressed as number of Ki67 or NKX6.1 or NKX2.2 cells per islet sections and in case of clustered or single cells, data were expressed as number of cells per mm 2 area.

Quantification of Replication and Hormone Expressing Cells in Ductal Structures
Slides stained by immunohistochemistry were digitally scanned using Aperio ScanScope (Aperio Technologies, Vista, CA) and analyzed using Aperio ImageScope version 12.1.0.5029. Sections were examined and quantified in a blinded manner. Ductal structures embedded in the mesenchyme and having a pancreatic duct gland (PDG) compartment were identified as interlobular ducts. PDG compartments were identified as invaginations stemming from the interlobular ducts. The total number of cells in interlobular ducts and surrounding PDG compartments were counted by the Aperio software. The number of Ki67, insulin, glucagon, somatostatin and pancreatic polypeptide cells found in interlobular ducts and PDGs were counted manually. The mean number of interlobular duct cells counted in the fetal sections was 684 ± 200 (range 56-1,598) and in the infant/child sections was 2,122 ± 533 (range 232-11,057). The mean number of PDG cells counted in the fetal sections was 385 ± 152 (range 5-1,350) and in the infant/child sections was 927 ± 240 (range 42-3,593).

Statistical Analysis
Statistical analysis was performed using the Student's ttest, two-way ANOVA or non-linear correlation analysis (where appropriate) with GraphPad Prism 6.0 software (GraphPad Software, La Jolla, CA). Data in graphs and Supplementary Tables (19) are presented as means ± SEM. Findings were assumed statistically significant at P < 0.05.

Pancreas Morphology During Development
During gestation, endocrine cells are present in high density but occur mostly in small clusters rather than in well-formed islets; a very high frequency of replication is present in both endocrine and exocrine compartments (1). After birth, the density of endocrine cells is decreased, largely as a consequence of growth of the exocrine pancreas, and the majority of endocrine cells are now found in well-recognizable islets. The frequency of replication, whilst still high in infancy, is lower than in the fetal tissue (2). During early childhood, the frequency of replication decreases in the exocrine pancreas, and becomes negligible in the endocrine compartment after the age of 3 years, in most cases. With age, the morphology of the pancreas matures, with wellformed islets and a lobular exocrine structure as in the adult pancreas (Supplementary Figures 1-3).

ChromograninA Positive Hormone-Negative (CPHN) Cells Are Abundant in Late Fetal Life but Decrease During Infancy
During gestation small clusters of pancreatic endocrine cells begin to assemble into islets that include a high proportion of CPHN cells (Figure 1A), confirming our earlier findings (4). This aggregation process of endocrine cells in the pancreas continues in the neonatal and early infancy period, where CPHN cells are still frequent (Figures 1B,C) and by 5.0 years of age the majority of endocrine cells in the pancreas are assembled into islets ( Figure 1D) (2). For this analysis, cases already described in a previous publication (4) were also included here (Fetal cases 1-4 and infant cases 1-10) (19). CPHN cells were prevalent as single cells and in small clusters scattered throughout the exocrine pancreas in the fetal pancreas (Figure 1). However, the abundance of CPHN cells decreased during the transition from fetal to infant life both in islets (r = −0.35) (8.8 ± 3.9 vs. 1.4 ± 0.4%, fetal vs. infant, p < 0.01) ( Figure 1E) and small clusters (r = −0.60) (34.5 ± 6.2 vs. 20.5 ± 2.2%, fetal vs. infant, p < 0.01) ( Figure 1F). The steep decline in CPHN cells with age potentially indicates that CPHN complete the process of differentiation, becoming mature pancreatic endocrine cells during infancy.

CPHN Cells Do Not Replicate in Fetal and Infant Pancreas
Endocrine cells expand by replication during late fetal and early neonatal life in humans (2). Since CPHN cells may represent a precursor to fully differentiated endocrine cells, we quantified replication of CPHN cells in pancreatic tissue from fetuses and infants by use of Ki67, chromograninA and a cocktail of islet endocrine hormones (insulin, glucagon, somatostatin, pancreatic polypeptide, and ghrelin) (5). CPHN cells were rarely observed to replicate (0.2 ± 0.1 vs. 4.7 ± 1.0%, CPHN vs. islet hormone positive cell replication, p < 0.001), indicated by the lack of Ki67 expression in CPHN cells whether located in the islets or in small clusters, and with no detectable difference between fetal and infant groups (Figures 2A,B). Only in one fetal and one infant case were we able to find a CPHN cell expressing Ki67 [Supplementary Figures 4A,B, (19)].

DISCUSSION
During human fetal pancreas development, insulin expression is first detected at 8-9 weeks gestation (1, 10) and fractional beta cell area increases linearly until birth (1). Our data show that the frequency of CPHN cells declines during fetal to postnatal transition (4), possibly supporting the concept that these cells are programmed to mature into functional endocrine cells. A similar synaptophysin-positive, hormone-negative cell type has been reported in fetal human pancreas (9). These studies suggest the existence of hormone-negative precursor cell type prevalent in fetal human pancreas. Islet formation in human pancreatic development involves a multistep rearrangement of endocrine cells, such that endocrine cells appear in the form of islets, as well as groups of small, scattered clusters in the exocrine tissue (1,10,23). Overall, ∼12-13% of fetal endocrine cells are CPHN cells, distributed in islets and small endocrine cell clusters. Within islets, ∼9% of the endocrine cells are CPHN cells, while a much higher proportion (∼35%) of the scattered endocrine cells are CPHN cells at this stage of development. These small foci of endocrine cells are reminiscent of the pancreatic phenotype in pregnant humans where the increased beta-cell mass is coincident with an increase in scattered foci of beta-cells not derived by replication (24). Taken together, this supports the hypothesis that CPHN cells could possibly be precursors to endocrine cells.
Given the high rates of endocrine cell replication in late fetal pancreas development, we examined whether, as a potential precursor pool, CPHN cells would display a high replicative index. Our data show that CPHN cells replicate very infrequently compared to differentiated endocrine cells as well as the potential precursor PDG compartment. The low replication frequency of CPHN cells is comparable to the Ngn3-positive endocrine progenitors, which typically have a very low replication index (25,26). If they do indeed represent a partially differentiated precursor population, the CPHN cells may not inherently be replication competent, perhaps to protect such a population from replicative stress. The susceptibility of partially differentiated endocrine cells to replicative stress has been widely documented, e.g., in the context of diabetes pathogenesis (27). Therefore, alternatively, these cells could represent a partially differentiated, misguided population that may fail to develop a mature endocrine identity and be fated for elimination via programmed cell death. One limitation of our study is that our fetal pancreatic samples are aged 19 weeks gestation onwards, such that the first time-point at which these cells emerge in the endocrine differentiation program is unclear.
NKX2.2 and NKX6.1 are key transcription factors that direct beta-cell development. NKX2.2 is required for endocrine cell specification and differentiation, and is present in beta-, as well as certain alpha-, and PP-cells in adults (28), while NKX6.1 is essential for beta-cell formation during embryonic development (29). The majority of CPHN cells in fetal and infant life are positive for NKX2.2 and NKX6.1, suggesting that most of these cells could possibly represent endocrine precursor cells and, if so, may differentiate into beta-cells. NKX6.1 lies downstream of NKX2.2 in the major pathway of beta cell formation (29). Moreover, while NKX2.2 appears in human embryos only after endocrine differentiation, both NKX6.1 and NKX2.2 are expressed in human fetal beta cells (30). Therefore, the high frequency of clustered or single NKX6.1 and NKX2.2 positive CPHN cells in human fetal pancreas suggests that the majority of CPHN cells may be programmed for a beta-cell lineage. In adults, it is equally possible that these cells underwent the maturation process and then lost some of their maturation signature through a dedifferentiation process, similar to what has been observed in stressed β cells (31,32). However, the high frequency of clustered CPHN cells positive for NKX6.1 or NKX2.2 in fetal and infant cases in association with their limited replication capacity suggests that the CPHN cells might represent a partially differentiated cell type with distinct features of endocrine lineage rather than de-differentiated cells.

AUTHOR CONTRIBUTIONS
ASMM, CM, KZ, MN, SK, and AEB performed the studies, undertook the microscopy with assistance from MC, ASMM, and SD, and performed the morphological analysis. AEB, SD, HF, JM, RR, and MA researched data, wrote, reviewed and edited the manuscript, and contributed to the discussion.

ACKNOWLEDGMENTS
This research was performed with the support of the Network for Pancreatic Organ Donors with Diabetes (nPOD), a collaborative type 1 diabetes research project sponsored by JDRF. Organ Procurement Organizations (OPO) partnering with nPOD to provide research resources are listed at http://www.jdrfnpod. org//for-partners/npod-partners/. The publication of this article was funded by the Qatar National Library.

SUPPLEMENTARY MATERIAL
The