We examined the spectrum of intermediate cell types in the regenerating pancreas as duct epithelial cells progressed through their differentiation pathway to become mature endocrine cells. The model used was transgenic mice in which the pancreatic islets continue to grow during adulthood, unlike normal mice whose islet cell formation ceases early in life. Because the intermediate cells migrated into islet-like clusters at specific locations, we propose a specific pathway for islet development. Endocrine cells are derived from duct cells co-expressing a duct cell antigen, carbonic anhydrase II (CA II) and an exocrine enzyme, amylase. The CA II/amylase cells become amylase/endocrine intermediate cells as they exited from their lumenal location. The abluminal amylase/endocrine cells continue to differentiate to multihormone-bearing young endocrine cells, which migrated to form clusters with other differentiating endocrine cells.

Endocrine cells in the pancreas are believed to originate from its duct epithelial cells, which derive from two outpocketings of the primitive gut (Pictet and Rutter, 1972; Wessells and Cohen, 1967; Wessells and Evans, 1968). In mice, such endocrine cells initially appear individually in the duct wall, later form small clusters in the interstitial tissue, and finally organize into well-defined islet structures with centrally located beta cells surrounded by the other endocrine cell types (Deltour et al., 1991; Dubois, 1989; Orci, 1988). Each islet consists of endocrine cells arising from several epithelial progenitors (Deltour et al., 1991). This neogenesis in the islets of Langerhans continues throughout the neonatal period and ceases shortly after weaning (Deltour et al., 1991; Githens, 1988). Thereafter, no additional islets form, and the islet cells do not exhibit appreciable growth (Hellerstrom et al., 1988). The general inability of the pancreatic duct cells to proliferate and differentiate in adults is reflected by the fact that the beta cell mass, which is lost in insulin dependent diabetes mellitus, is not replenished by regeneration of islet cells.

Despite the cessation of islet cell growth, differentiated endocrine cells are known to be capable of undergoing mitosis even in adult animals under certain conditions (Davidson et al., 1989; Green et al., 1981; Korcakova, 1971; Nielsen et al., 1989). However, the pattern by which the ductal precursors generate endocrine cells in adults is unknown.

We previously described duct epithelial cell proliferation in adult transgenic mice carrying the interferon-g (IFN-g) gene linked to an insulin promoter (Gu and Sarvetnick, 1993). The pancreata of these mice underwent ductal hyperplasia and destruction of islets in association with lymphocytic infiltration. Simultaneously these mice exhibited extremely high proliferative activity in their pancreatic ducts, with a mitotic index sometimes reaching as high as 73% in a 12-hour period. Within the duct walls, single endocrine cells and clusters were found (Gu and Sarvetnick, 1993). We have hypothesized that the new islet cells originated from the proliferating duct cells and that the islet regeneration in this mouse strain might recapitulate embryonic islet neogenesis.

We describe the lineage of the new islet cells by studying the transitional cells possessing duct, endocrine and exocrine characteristics. Our studies revealed the presence of numerous intermediate cell types including cells simultaneously exhibiting both duct and exocrine phenotype or endocrine and exocrine features as well as single endocrine cells containing multiple islet hormones. The multiple hormone-containing cells have been observed in fetal islet neogenesis (Alpert et al., 1988; de Krijger et al., 1992; Herrera et al., 1991; Teitelman et al., 1993), and in in vitro differentiation of islet cells derived from tumors (Madsen et al., 1988; Madsen et al., 1986; Madsen et al., 1993; Philippe et al., 1987). These multiple hormone-bearing cells are indications of newly differentiated endocrine cells on their path to the final single hormone phenotype. Based on the appearance and the specific location of intermediate cells that transiently co-expressed CA II and amylase as well as amylase and hormones, we propose that islet cells are derived from ductal cells. Our results also suggest that endocrine and exocrine cells have a common origin (see review by Le Douarin, 1988).

Light and electron microscopy

For light microscopic examination, samples of tissue were fixed in 4% zinc-formalin or Bouin’s fixative, dehydrated in graded ethanol, cleared in xylene, and embedded in paraffin. Paraffin sections were cut at 5 μm thickness, deparaffined, and stained with H & E for routine observation.

Samples for electron microscopy were initially fixed in 1.5% glutaraldehyde and 2% paraformaldehyde in 0.1 M cacodylate buffer, postfixed in 1% OsO4, dehydrated in graded ethanol, cleared in propylene oxide and embedded in epoxy resin. Thin sections were stained with uranyl acetate and lead citrate and viewed with a Hitachi HU 12A electron microscope (Gu and Sarvetnick, 1993).

Immunocytochemistry

The procedures for immunolabeling and identification of proliferating cells by BrdU incorporation, colocalization of pancreatic hormones, and colocalization of insulin, carbonic anhydrase II (CA II), amylase, and BrdU were essentially the same as described by Gu and Sarvetnick (1993), except that the immunofluorescent markers FITC and rhodamine were also used. No frozen sections were used in the present study. For double immunofluorescent staining, the sequential staining technique was used. The first primary antibody was either anti-glucagon, anti-somatostatin, anti-pancreatic polypeptide, or anti-amylase. The first secondary antibody was fluorescence (DTAF, Dichlorotriazinyl amino) conjugated goat anti-rabbit. The second primary antibody, anti-insulin was applied. The second secondary antibody was rhodamine (TRITC) conjugated goat anti-guinea pig. For comparison, BALB/c pancreatic tissue served as a control and was used and processed like that from transgenic mice. To check the specificity of antibodies, control experiments consisted of replacement of each primary antiserum with a preimmune serum or with PBS/bovine serum albumin (BSA). Background staining was minimal, and no spectral overlap was observed. Furthermore, the presence of both single- and double-positive cells indicated the absence of cross-reactivity between the different hormone antibodies. Antibodies were obtained from the following sources: guinea pig antibodies against insulin were purchased from Dakopattes. Of the rabbit antibodies, those against glucagon came from Chemicon, those against somatostatin from Dakopatts, those against pancreatic polypeptide from ICN Immuno Biologicals, and those against amylase from Sigma. Sheep antibodies against CA II were purchased from The Binding Site. Monoclonal rat antibodies against BrdU were purchased from Sera Lab and Chemicon. The biotinylated goat anti-guinea pig IgG, biotinylated goat anti-rabbit IgG, biotinylated rabbit anti-sheep IgG, and biotinylated rabbit anti-rat IgG were purchased from Vector Labs. In addition, rhodamine conjugated goat anti-guinea pig IgG and fluorescence conjugated goat anti-rabbit IgG were purchased from Chemicon.

BrdU pulse-labeled chase experiment

27 mice were injected intraperitoneally with 100 μg/g body weight of BrdU. The maximum availability time of BrdU is 40-60 minutes (Boswald et al., 1990). The fates of the BrdU incorporated by duct cells, beta cells and amylase positive cells were counted at 6, 12 and 18 hours after BrdU administration.

Quantification of labeled cells and statistical analysis

The immunolabeled cells were scored on a 400× microscope field. At least 20 ducts or islet structures from 5 animals were examined for each test group. Each section was systematically scanned to avoid overlapping of the fields. The percentage of positive cells was expressed as the number of labeled cells divided by the total number of cells counted times 100. The statistical analyses were performed by Student’s t-test.

Evidence for primitive endocrine cells

Because multiple hormone-bearing cells are indicative of newly differentiated endocrine cells on their path to the single hormone of their mature state, we performed double immunolabeling experiments with antibodies to the four islet hormones, i.e., insulin, glucagon, somatostatin, and pancreatic polypeptide (B, A, D, and PP). We chose 4-month old transgenic mice for this co-expression of hormones studies, since the existing islets were destroyed and the regeneration of new islet structures was vigorous. The double-positive cells in the ducts and islet structures were then scored (Fig. 1; Tables 1, 2). These experiments demonstrated that approximately 16% of the insulin-containing duct cells in pancreata of the transgenic mice expressed an additional islet hormone. Of these double-positive cells, a high proportion coexpressed insulin and glucagon (see Table 1). On the contrary, in pancreata of control, non-transgenic mice, no endocrine cells expressed more than one detectable hormone in this test.

Table 1.

A survey of intermediate beta cells in the duct wall

A survey of intermediate beta cells in the duct wall
A survey of intermediate beta cells in the duct wall
Table 2.

A survey of intermediate beta cells in the islet structure

A survey of intermediate beta cells in the islet structure
A survey of intermediate beta cells in the islet structure
Fig. 1.

Demonstration of islet cells containing multiple hormones. Indirect sequential dual color immunofluorescence staining of paraffin embedded pancreas to show insulin (red) and glucagon (green) in the same cells. Note double-positive cells in both the ducts (A and B, long arrows) and islet structures (C and D, long arrows). Also note the single-positive glucagon cells (A and C, short arrows). Magnification, 1,000×.

Fig. 1.

Demonstration of islet cells containing multiple hormones. Indirect sequential dual color immunofluorescence staining of paraffin embedded pancreas to show insulin (red) and glucagon (green) in the same cells. Note double-positive cells in both the ducts (A and B, long arrows) and islet structures (C and D, long arrows). Also note the single-positive glucagon cells (A and C, short arrows). Magnification, 1,000×.

All double-positive cells, which were considered intermediate cells in the maturation process, were found in both the duct walls and the islet structures. However, the proportion of islet cells coexpressing two hormones was much higher in the ductal location. Since this characteristic is associated with immature endocrine cells, these findings suggested that the duct endocrine cells were in the early stages of differentiation and the islet structure contained a high proportion of newly formed endocrine cells.

Evidence for the ductal origin of islet cells

To study the origin of the young endocrine cells, we performed BrdU labeling experiments. The changes in percentage of BrdU-positive duct cells and islet cells were compared and analyzed at different time intervals. Mice were given injections of BrdU and killed at the time intervals indicated in Table 3. The number of BrdU-positive cells in the duct increased 15.3% by 6 hours after injection and reached a peak of 21.32% at 12 hours (a 1.4-fold increase). Clearly, mitotic activity of the duct cells was extremely high in this transgenic mouse strain. In contrast, the islet structures contained only 1.89% BrdUpositive cells after the first 6 hours of incubation. However, the number of labeled islet cells increased nearly 4.3-fold to 8.03% at the end of a 12-hour incubation. Thus, the rate of increase in the labeling index of the islet cells during the 6- to 12-hour period was significantly higher than that of duct cells (P<0.001). Since no additional stimulatory factors were present or administered to the mice to cause duct cells or islet cells to synthesize DNA at a later stage, this dramatic increase in the labeling index of islet cells during this 6- to 12-hour period suggests that the dividing duct cells could be the source of the extra BrdU-labeled cells in the islets. The percentage of BrdU-labeled cells dropped sharply after 18 hours in both ducts and islets, indicating that many cells containing BrdU were not detectable as the total cell population increased, thereby diluting the signal (data not shown).

Table 3.

Time course of proportion of total BrdU labeled cells and BrdU endocrine double positive cells in ducts and islet structures

Time course of proportion of total BrdU labeled cells and BrdU endocrine double positive cells in ducts and islet structures
Time course of proportion of total BrdU labeled cells and BrdU endocrine double positive cells in ducts and islet structures

To ascertain that the newly arrived islet cells indeed were differentiated endocrine cells, and not lymphocytes or fibroblasts, etc., we performed double labeling experiments to identify BrdU-incorporating endocrine cells in the ducts and in islet structures adjacent to the ducts. We determined that these BrdU and endocrine double positive cells represented 95% (40/42) and 97% (117/120) of the total BrdU incorporated cells in islets at the 6- and 12-hour incubation times, respectively. Table 3 depicts the changes in the percentage of the BrdU and endocrine double positive cells, including A, B, D and PP cell types at 6, 12 and 18 hours after BrdU administration. At the 6 hour time point, 2.34% of the islet endocrine cells were BrdU-incorporating endocrine cells, a percentage similar to 1.89% of BrdU single-labeled positive cells from total islet cell numbers (P>0.05). The labeling index of double-labeled endocrine cells in islet structures increased 3.7-fold in the 6to 12-hour period, which is significantly higher than the 1.4-fold increase in the ducts (P<0.001). This increase was slightly lower than the nearly 4.3-fold increase registered by single-labeled islet cells, possibly because some newly differentiated cells contained low levels of hormone and were not detected by the method used. Nevertheless, the large number of double-positive cells that appeared in the islet structures during this period most likely represented recent arrivals from the differentiating duct cells.

Evidence for the existence of ductal/exocrine intermediate cells

Interestingly, we frequently observed cells containing exocrine granules in the lumen of ducts in the transgenic mice (Fig. 2), but not in non-transgenic mice. To characterize these potential transitional cells, we performed immunomarking experiments employing antibodies to CA II and amylase. The CA II antigen is expressed in interlobular duct epithelial cells of both normal and transgenic mice, whereas the exocrine enzyme, amylase, was chosen because of its characteristically acinar localization.

Fig. 2.

Ultrastructure of a newly formed duct. A low-powered electron micrograph showing part of duct wall that contains two acinar cells. These cells show typical well-differentiated exocrine cell morphology with abundant rough endoplasmic reticulum and zymogen granules (arrows) facing the lumen. The endocrine cell (asterisk) on the duct wall does not face the lumen. Magnification, 9,000×.

Fig. 2.

Ultrastructure of a newly formed duct. A low-powered electron micrograph showing part of duct wall that contains two acinar cells. These cells show typical well-differentiated exocrine cell morphology with abundant rough endoplasmic reticulum and zymogen granules (arrows) facing the lumen. The endocrine cell (asterisk) on the duct wall does not face the lumen. Magnification, 9,000×.

Immunolabeling studies with CA II revealed that all the positive cells resided within the duct walls, mainly in small ducts corresponding to the intercalated pancreatic ducts in the transgenic mice. Immunostaining with amylase showed that, in addition to acinar cells, numerous duct cells of the transgenic mice also contained amylase, as might have been predicted from our ultrastructural analysis. In comparison, only very few amylase-containing duct cells were observed in BALB/c mice. To identify transitional cells, we then studied the coexpression of the CA II antigen with amylase. Quite frequently, double-labeled cells were found within the ducts of our transgenic mice, apparently bordering the lumen. Since cells with these characteristics were not found in normal non-transgenic mice, we conjectured that they could represent transitional cells in our model of differentiation.

Since transitional cells in intermediate stages of this process should undergo mitosis, considering that epithelial cells of the pancreatic duct do proliferate in the adult transgenic mice we use, this possibility was tested further by double labeling for amylase and BrdU. The resulting mitotic index of these ductal amylase-positive cells was 2.3% at the 6 hour time point and increased to 6.8% at 12 hours, thereafter declining to 4.4% after 18 hours of incubation. A subsequent triple-labeling experiment, to determine whether CA II/amylase double positive cells incorporated BrdU, again showed that this intermediate cell type was mitotically active (Fig. 3).

Fig. 3.

Replicating duct/exocrine intermediate cells. Triple markings performed by indirect sequential immunofluorescence staining are for amylase (green) combined with immunoperoxidase for CA II (purple) and BrdU (red). A triple-stained cell is in the duct wall and lines the lumen (long arrows). A CA II/amylase stained cell (short arrows), which did not incorporate BrdU, is located above the triple-stained cell. Magnification, 1,000×.

Fig. 3.

Replicating duct/exocrine intermediate cells. Triple markings performed by indirect sequential immunofluorescence staining are for amylase (green) combined with immunoperoxidase for CA II (purple) and BrdU (red). A triple-stained cell is in the duct wall and lines the lumen (long arrows). A CA II/amylase stained cell (short arrows), which did not incorporate BrdU, is located above the triple-stained cell. Magnification, 1,000×.

Evidence for the existence of exocrine/endocrine intermediate cells

Electron microscopy (EM) previously revealed that the ductal endocrine cells do not border the lumen but are separated from it by duct cell cytoplasm (Gu and Sarvetnick, 1993; Higuchi et al., 1992). Consequently, the appearance of exocrine/endocrine intermediate cells would suggest that the differentiating cells had begun to exit from the lumen lining. Therefore, we immunostained duct cells for insulin and amylase to identify cells possessing both exocrine and endocrine characteristics. In control BALB/c mice, no cells co-expressed amylase and insulin at a detectable level. However, in pancreata of the transgenic mice, 1.6% (30/1838) of the amylase positive cells reacted with the insulin antibody. These cells frequently appeared within ducts and periductal regions and occasionally were visible in islets. In apparent agreement with EM observations, all of the double-positive amylase/insulin intermediate cells were not lining the lumen, while nearly all of the single stained amylase cells were at the lumen’s lining (Fig. 4). We considered that the double-positive cells might represent an intermediate stage between ductal cells and islet cells in our model.

Fig. 4.

Exocrine/endocrine intermediate cells. Double stainings performed by indirect sequential immunofluorescence staining are for insulin (red) and amylase (green). Double-positive cells (long arrows) in the duct wall do not border the lumen. Also note insulin containing cells (short arrow) and single amylase positive cells (open arrowhead). Magnification, 560×.

Fig. 4.

Exocrine/endocrine intermediate cells. Double stainings performed by indirect sequential immunofluorescence staining are for insulin (red) and amylase (green). Double-positive cells (long arrows) in the duct wall do not border the lumen. Also note insulin containing cells (short arrow) and single amylase positive cells (open arrowhead). Magnification, 560×.

To determine if the endocrine/exocrine intermediate cells exhibited mitotic activity, triple labelings for amylase, insulin and BrdU were performed (Fig. 5). As Fig. 5 shows, this intermediate cell type manifested mitotic activity.

Fig. 5.

Acinar/endocrine cells and their incorporation of BrdU. Indirect sequential immunofluorescence staining for amylase (green), insulin (purple), and BrdU (red), of paraffin embedded pancreas to reveal intermediate acinar/endocrine cells. (A) Bright-field micrograph of an islet structure confirmed its location at the duct lumen. An insulin and BrdU double-stained cell (long arrow) appears in the duct wall but not at the lumen’s lining. Single-stained BrdU and single insulin containing cells are also present (arrowhead and short arrow). (B) Immunofluorescence micrograph of the same field as A. Note that the doublestained insulin/BrdU cell shown in A is also stained with amylase (long arrow). An amylase single-stained cell (short arrow) is evident immediately above the triplestained cell. Magnification, 1,000×.

Fig. 5.

Acinar/endocrine cells and their incorporation of BrdU. Indirect sequential immunofluorescence staining for amylase (green), insulin (purple), and BrdU (red), of paraffin embedded pancreas to reveal intermediate acinar/endocrine cells. (A) Bright-field micrograph of an islet structure confirmed its location at the duct lumen. An insulin and BrdU double-stained cell (long arrow) appears in the duct wall but not at the lumen’s lining. Single-stained BrdU and single insulin containing cells are also present (arrowhead and short arrow). (B) Immunofluorescence micrograph of the same field as A. Note that the doublestained insulin/BrdU cell shown in A is also stained with amylase (long arrow). An amylase single-stained cell (short arrow) is evident immediately above the triplestained cell. Magnification, 1,000×.

Although in EM observations, no endocrine granules were found in cells lining the lumen we studied the coexpression of CA II and insulin to find ductal/endocrine intermediate cells. However, this dual labeling failed to locate any duct/endocrine intermediate. Evidently no cells of this type were present, at least not at a detectable level.

Evidence for the existence of exocrine/endocrine intermediate cells in the fetal pancreas

To determine whether the cells in the rudimentary pancreas in embryos also contain exocrine/endocrine intermediate cells, we immunostained BALB/c and transgenic E13E18 fetal pancreata for both amylase and insulin. First, we carried out single antibody labeling using antibodies to either amylase or insulin. The results of these single antibody staining experiments from BALB/c and transgenic pancreata were quite similar. The earliest stage we could detect insulin-containing cells was E13, in agreement with the published report by Alpert et al. (1988). We observed that amylase reactive cells first appeared on E14.5, in agreement with previous reports (Teitelman et al., 1987; Alpert et al., 1988). Interestingly, in double-labeling experiments, we could detect coexpression of amylase and insulin in the pancreas of transgenic embryos on E14.5 (Fig. 6). The double-stained cells were relatively rare, however, they could be clearly identified. We could not detect any double-labeled cells of this type in the pancreas of BALB/c embryos.

Fig. 6.

Exocrine/endocrine intermediate cells in the rudimentary pancreas of E14.5 transgenic embryos, indirect sequential double immunofluorecence staining for insulin (red) and amylase (green). Double-labeled cells containing both amylase and insulin appear in a cluster and in adjacent areas (long arrows). Single-labeled amylase containing cells are seen in the left of the micrograph (short arrow). Magnification, 400×.

Fig. 6.

Exocrine/endocrine intermediate cells in the rudimentary pancreas of E14.5 transgenic embryos, indirect sequential double immunofluorecence staining for insulin (red) and amylase (green). Double-labeled cells containing both amylase and insulin appear in a cluster and in adjacent areas (long arrows). Single-labeled amylase containing cells are seen in the left of the micrograph (short arrow). Magnification, 400×.

We have demonstrated that the pancreatic islet cells, which regenerate uniquely in adult IFN-g transgenic mice, originate from duct epithelial cells. This conclusion rests on two experimental results. First, the ductal endocrine cells from these mice expressed multiple hormones, a trait attributed to the newly differentiated endocrine cells. Second, the dividing ductal endocrine cells migrated to join the newly formed islet structures. That is, a disproportionately large number of BrdUlabeled endocrine cells homed to the islet structures, and intermediate endocrine cell types were identified within these islet structures. These combined results led us to conclude that the new islet cells arose from duct cells.

In this model, the differentiation process appears to initiate within duct structures and terminate within islet structures. Although we observed ductal cells and ductal/acinar intermediate cells on the duct lumen, the exocrine/endocrine and endocrine cells were found exclusively in the ‘periluminal’ location. Such compartmentalization of cell type and locality suggests that movement from a ductal to periductal location occurs as a developmental progression. Furthermore, our analysis of the transitional cells identified in this study demonstrated the presence of duct/acinar and acinar/endocrine cells but not duct/endocrine cells. Therefore, duct cells evidently progress to duct/acinar to acinar/endocrine to endocrine cells. Indeed, occasional cells within islet structures contained amylase, suggesting the down regulation of this enzyme during the migration of the developing islet cells into islet structures.

Some of the ductal cells examined here contained both insulin and amylase. Thus, these cells exhibited phenotypes of both acinar and endocrine cells. Cells having these characteristics have been reported in both normal animals and in the animals under certain metabolic stress (Melmed et al., 1972; Melmed, 1979). However, in our transgenic mice, these acinar/endocrine cells were located primarily in the duct wall and were mitotically active. The presence of ductal intermediate acinar/beta cells raises the possibility that beta cells and acinar cells may have a common progenitor, at least in this system. A noteworthy observation was that these acinar/beta intermediate cells were destined to become endocrine cells, not acinar cells, since we found no evidence for the neogenesis of acinar tissue in the pancreata of transgenic mice. This conclusion is in agreement with the general belief that restoration of the exocrine gland is achieved by replication of acinar cells not by differentiation of duct cells (De Lisle et al., 1990; Elsasser et al., 1986; Githens, 1988; Rao et al., 1990). We do not know whether the exocrine/endocrine intermediate cells occur during fetal life. The presence of such cells has been hypothesized, but not directly demonstrated (Alpert et al., 1988; Teitelman, 1987). Our own efforts to demonstrate such cells in the normal fetal pancreas of non-transgenic mice have proved unsuccessful, but we have identified putative exocrine/endocrine cells in our transgenic embryos. This allows us to speculate that exposure to the cytokine during fetal development induces the expansion of primitive dedifferentiated cells. If this theory proves to be true, our model will be useful for identification and characterization of these primitive cells.

The origin of the duct cells is also a subject of speculation. Owing to the mitotic activity exhibited by our duct cells, we concluded that the dividing duct cells can give rise to dedifferentiated duct/exocrine cells, as discussed above. Alternatively, the duct cells can be derived from acinar cells. In fact, the transformation of acinar cells to duct cells has been reported in acinar carcinoma derived cell lines (De Lisle et al., 1990; De Lisle and Logsdon, 1990; Pettengill et al., 1993). We have observed acinar cells in small ducts that appeared fully differentiated. Possibly these represent the transformation of acinar cells to duct cells in the exocrine tissue near the islet lesions. These acinar cells could dedifferentiate to duct/acinar cells and then eventually redifferentiate to islet cells.

Islet cell regeneration in the pancreas of adult transgenic mice progresses through the following steps. The duct wall contains duct/acinar cells with extremely high mitotic activity. When duct cells divide at a right angle to the duct wall, one daughter cell exits from the lumen, coexpressing amylase and hormone (insulin). The acinar/endocrine cells eventually stop expressing amylase while still retaining multihormone phenotypes as they move further away from the lumen. The newly differentiated endocrine cells initially appear as single cells dotting the duct wall and then migrate to form clusters in the islets. This regeneration process parallels islet neogenesis in embryos and provides a model system for studying factors influencing islet cell differentiation.

The authors would like to thank Dr Cheng-ming Chang, Electron Microscopy Laboratory, The Scripps Research Institute, for his kind help with electron microscopic studies and Dr John Polich for performing statistical analyses. We thank Dr Howard Fox for critical comments on the manuscript, Phyllis Minick for critical reading, and for help with the preparation of this manuscript. D. G. was supported by a postdoctoral fellowship from the California Myasthenia Gravis Foundation and from the Juvenile Diabetes Foundation. M.-S. Lee was supported by an advanced postdoctoral fellowship award from the National Multiple Sclerosis Society. N. S. was supported by a career development award from the Juvenile Diabetes Foundation. This work was supported by NIH grant DK41801-02 and HD29764. This is manuscript number 8286-NP. from The Scripps Research Institute.

Alpert
,
S.
,
Hanahan
,
D.
and
Teitelman
,
G.
(
1988
).
Hybrid insulin genes reveal a developmental lineage for pancreatic endocrine cells and imply a relationship with neurons
.
Cell
53
,
295
308
.
Boswald
,
M.
,
Harasim
,
S.
and
Maurer-Schultze
,
B.
(
1990
).
Tracer dose and availability time of thymidine and bromodeoxyuridine: application of bromodeoxyuridine in cell kinetic studies
.
Cell Tissue Kinet
.
23
,
169
181
.
Davidson
,
P. M.
,
Campbell
,
I. L.
,
Oxbrow
,
L.
,
Hutson
,
J. M.
and Harrison,L. C
. (
1989
).
Pancreatic beta cell proliferation in rabbits demonstrated by bromodeoxyuridine labeling
.
Pancreas
4
,
594
600
.
de Krijger
,
R. R.
,
Aanstoot
,
H. J.
,
Kranenburg
,
G.
,
Reinhard
,
M.
,
Visser
,
W. J.
and
Bruining
,
G. J.
(
1992
).
The midgestation human fetal pancreas contains cells coexpressing islet hormones
.
Dev. Biol
.
153
,
368
375
.
De Lisle
,
R. C.
,
Grendell
,
J. H.
and
Williams
,
J. A.
(
1990
).
Growing pancreatic acinar cells (postpancreatitis and fetal) express a ductal antigen
.
Pancreas
5
,
381
388
.
De Lisle
,
R. C.
and
Logsdon
,
D. C.
(
1990
).
Pancreatic acinar cells in culture expression of acinar and ductal antigens in a growth-related manner
.
Eur. J. Cell Biol
.
51
,
64
75
.
Deltour
,
L.
,
Leduque
,
P.
,
Paldi
,
A.
,
Ripoche
,
M. A.
,
Dubois
,
P.
and
Jami
,
J.
(
1991
).
Polyclonal origin of pancreatic islets in aggregation mouse chimaeras
.
Development
112
,
1115
1121
.
Dubois
,
P. M.
(
1989
).
Ontogeny of the endocrine pancreas
.
Horm. Res
.
32
,
53
60
.
Elsasser
,
H-P.
,
Lutcke
,
H.
and
Kern
,
H. F.
(
1986
).
Acinar and duct cell replication and regeneration
. In
The Exocrine Pancreas: Biology, Pathobiology, and Diseases
(ed.
V. L. W.
Go
), pp.
45
53
.
New York
:
Raven Press
.
Githens
,
S.
(
1988
).
The pancreatic duct cell: proliferative capabilities, specific characteristics, metaplasia, isolation, and culture. J
.
Pediatr. Gastroenterol. Nutr
.
7
,
486
506
.
Green
,
I. C.
,
Seifi
,
S. E.
,
Perrin
,
D.
and
Howell
,
S. L.
(
1981
).
Cell replication in the islets of Langerhans of adult rats: effect of frequency, ovariectomy and treatment with steroid hormones
.
J. Endocr
.
88
,
219
224
.
Gu
,
D.
and
Sarvetnick
,
N.
(
1993
).
Epithelial cell proliferation and islet neogenesis in IFN-g transgenic mice
.
Development
118
,
33
46
.
Hellerstrom
,
C.
,
Swenne
,
I.
and
Andersson
,
A.
(
1988
).
Islet cell replication and diabetes
. (ed.
P. J.
Lefebvre
), pp.
141
-
170
. Berlin Heidelberg: Springer-Verlag.
Herrera
,
P-L.
,
Huarte
,
J.
,
Sanvito
,
F.
,
Meda
,
P.
,
Orci
,
L.
and
Vassalli
,
J.-D.
(
1991
).
Embryogenesis of the murine endocrine pancreas; early expression of pancreatic polypeptide gene
.
Development
113
,
1257
1265
.
Higuchi
,
Y.
,
Herrera
,
P.
,
Muniesa
,
P.
,
Huarte
,
J.
,
Belin
,
D.
,
Ohashi
,
P.
,
Aichele
,
P.
,
Orci
,
L.
,
Vassalli
,
J.-D.
and
Vassalli
,
P.
(
1992
).
Expression of a tumor necrosis factor in murine pancreatic B cells results in severe and permanent insulitis without evolution towards diabetes
.
J. Exp. Med
.
176
,
1719
1731
.
Korcakova
,
L.
(
1971
).
Mitotic division and its significance for regeneration of granulated B-cells in the islets of Langerhans in alloxan-diabetic rats
.
Material and Methods XIX, 24-30
.
Le Douarin
,
N. M.
(
1988
).
On the origin of pancreatic endocrine cells
.
Cell
53
,
169
171
.
Madsen
,
O. D.
,
Andersen
,
L. C.
and
Michelsen
,
B.
(
1988
).
Tissue-specific expression of transfected human insulin genes in pluripotent rat insulinoma lines induced during passage in vivo
.
Proc. Nat. Acad. Sci. USA
85
,
66526656
.
Madsen
,
O. D.
,
Larsson
,
L. I.
,
Rehfeld
,
J. F.
,
Schwartz
,
T. W.
,
Lernmark
,
A.
,
Labrecque
,
A. D.
and
Steiner
,
D. F.
(
1986
).
Cloned cell lines from a transplantable islet cell tumor are heterogeneous and express cholecystokinin in addition to islet hormones
.
J. Cell Biol
.
103
,
2025
2034
.
Madsen
,
O. D.
,
Serup
,
P.
,
Karlsen
,
C.
,
Petersen
,
H. V.
,
Blume
,
N.
,
Andersen
,
F. G.
,
Jensen
,
J.
and
Michelsen
,
B. K.
(
1993
).
A tumour model for the study of islet cell differentiation
.
Biochem. Soc. Trans
.
21
,
142
146
.
Melmed
,
R. N.
,
Benitez
,
C. J.
and
Holt
,
S. J.
(
1972
).
Intermediate cells of the pancreas 1. Ultrastructural characterization
.
J. Cell Sci
.
11
,
449
475
.
Melmed
,
R. N.
(
1979
).
Intermediate cells of the pancreas. An appraisal
.
Prog. Gastroent
.
76
,
196
201
.
Nielsen
,
J. H.
,
Linde
,
S.
,
Weilinder
,
B. S.
,
Billestrup
,
N.
and
Medsen
,
O. D.
(
1989
).
Growth hormone is a growth factor for the differentiated pancreatic beta cell
.
Mol. Endocrinol
.
13
,
165
173
.
Orci
,
L.
(
1988
).
Macro- and micro-domains in the endocrine pancreas
.
Diabetes
31
,
538
565
.
Pettengill
,
O. S.
,
Faris
,
R. A.
,
Bell
,
R. H. J.
,
Kuhlmann
,
E. T.
and
Longnecker
,
D. S.
(
1993
).
Derivation of ductlike cell lines from a transplantable acinar cell carcinoma of the rat pancreas
.
Am. J. Pathol
.
143
,
292
303
.
Philippe
,
J.
,
Chick
,
W. L.
and
Habener
,
J. F.
(
1987
).
Multipotential phenotypic expression of genes encoding peptide hormones in rat insulisnoma cell lines
.
J. Clin. Invest
.
79
,
351
358
.
Pictet
,
R.
and
Rutter
W. J.
(
1972
).
Development of the embryonic endocrine pancreas
. In
Handbook of Physiology. Vol. 1
. (ed.
D.
Steinerand
), pp.
25
66
.
Baltimore, MD
:
William and Wildins
.
Rao
,
M. S.
,
Yeldandi
,
A. V.
and
Reddy
,
J. K.
(
1990
).
Stem cell potential of ductular and periductular cells in the adult rat pancreas
.
Cell Diff. Dev
.
29
,
155
163
.
Teitelman
,
G.
,
Alpert
,
S.
,
Polak
,
J. M.
,
Martinez
,
A.
and
Hanahan
,
D.
(
1993
).
Precursor cells of mouse endocrine pancreas coexpress insulin, glucagon and the neuronal proteins tyrosine hydroxylase and neuropeptide Y, but not pancreatic polypeptide
.
Development
118
,
1031
1039
.
Teitelman
,
G.
,
Lee
,
J. K.
and
Alpert
,
S.
(
1987
).
Expression of cell type-specific markers during pancreatic development in the mouse: implications for pancreatic cell lineages
.
Cell Tissue Res
.
250
,
435
439
.
Wessells
,
N. K.
and
Cohen
,
J. H.
(
1967
).
Early pancreas organogenesis: morphogenesis, tissue interaction, and mass effects
.
Dev. Biol
.
15
,
237
270
.
Wessells
,
N. K.
and
Evans
,
J.
(
1968
).
Ultrastructural studies of early morphogenesis and cytodifferentiation in the embryonic mammalian pancreas
.
Dev. Biol
.
17
,
413
446
.