Ammonium selectively inhibits the regulated pathway of protein secretion in two endocrine cell lines

Ammonium selectively inhibits the regulated pathway of protein secretion in two endocrine cell lines Jill J. Dyken and Athanassios Sambanis School o f Chemical Engineering, Georgia Institute o f Technology, Atlanta, Georgia

The effect of ammonium on protein processing and secretion from mouse insulinoma flTC3 cells and recombinant mouse pituitary AtT-20 cells was investigated. Protein stored in granules of the regulated secretion pathway was discharged from cells with secretagogues, and the addition of new protein was encouraged by recharging cells in serum-containing medium with or without added ammonium. Ammonium at 6 mm inhibits the addition of insulin-related peptides to intracellular stores in both cell lines; it is not clear whether this effect is dose dependent for concentrations between 0 and 6 ram. There is a slight increase in insulin-related proteins secreted during recharging of cells in the presence of ammonium. Using reverse-phase high-performance liquid chromatography to separate proinsulin from insulin, we found that extracts from flTC3 cells recharged in the presence of 6 mm ammonium contain significantly less insulin than control cells recharged in the absence of ammonium.

Keywords: Ammonium;protein secretion; endocrine cells

Introduction Mammalian cell culture is becoming increasingly attractive for the production of pharmaceutically important biochemicals. This is because animal cells contain the complex enzymatic machinery, less developed in microbial cells, necessary for correct posttranslational processing of endogenous and recombinant proteins. Animal cells currently used in production secrete proteins in a continuous fashion into culture medium. Specialized cells of the endocrine system, in addition to this generic constitutive secretion pathway of proteins, are able to concentrate and store proteins intracellularly, releasing them in a regulated fashion in response to secretion agonists, or secretagogues. A schematic of secretion events taking place in an endocrine cell is presented in Figure 1. At the trans-Golgi, sorting of proteins destined for constitutive or regulated secretion takes place: constitutive proteins are shuttled by vesicles passively and continuously to the plasma membrane, where the proteins are secreted essentially at the rate of expression. Proteins of the regulated path-

Address reprint requests to Dr. Sambanisat the Schoolof Chemical Engineering, Georgia Institute of Technology,Atlanta, GA 303320100 Received 5 February 1993; revised 23 July 1993 90

Enzyme Microb. Technol., 1994, vol. 16, February

way are actively packaged into vesicles coated with the protein clathrin; these vesicles go through a maturation process, becoming acidic and condensing the proteins within, and develop into storage granules, which accumulate intracellularly. 1'2 The maturation process has been shown to be the site of final proteolytic processing of some hormones3; this processing is often essential for biological activity. Exposing cells to a secretagogue, such as norepinephrine or a cyclic AMP (cAMP) analog, raises second messenger levels and induces the release of stored proteins.l The unique properties of regulated secretion give endocrine cells potential both industrially and medically. In production operations, cells can be cycled between rich, protein-supplemented "growth" medium, allowing the buildup of biologically active protein, and a small volume of protein-free "secretion" medium, containing secretagogues. The relatively high concentration and purity of the desired protein obtained in the secretion medium would allow the elimination and/or downscaling of expensive downstream processing and purification steps. 4'5 In tissue engineering, endocrine cells can be used for long-term control of hormonal deficiencies through the development of implantable bioartificial organs. Cells are isolated from the immune system of the host by being immobilized in a biopolymer with pores large enough to allow the free transfer of nutrients, metabolites, and desired hormone(s), but small enough to exclude antibodies and © 1994 Butterworth-Heinemann

Inhibition of protein secretion by ammonium: J. J. Dyken and A. Sambanis

SECRETIONONLYINTHE PRESENCEOFSECRETAGOGUES

@

S,OR,GE @

/

.EOOLATEO SECR'O"

1

o ACIDICSTEP, PROCESSING

O

/

CONSTITUTIVE (BASAL)

SECRETION

\

Figure 1 Constitutive and regulated protein secretion pathways in endocrine cells

cytotoxic cells. The ability of endocrine cells to release stored protein in response to physiological stimuli makes them ideal for proper control of blood hormone levels .6 The effects of stressful conditions that endocrine cells would inevitably experience in production systems and immunoprotected implants have not been methodically studied. It is known that accumulation of metabolic wastes eventually hinders cell growth and viability. 7-1° The regulated pathway of protein secretion may be more sensitive to such environmental factors than other cell functions, though, decreasing the amount of biologically active protein obtainable from cultures with significant but nontoxic metabolite levels. A metabolite c o m m o n to all mammalian cell cultures is ammonium ion, a product of the metabolism and degradation of glutamine. Most culture media contain this essential nutrient at concentrations up to 6 mM; metabolism results in the production of between 1 and 2 moles of ammonium per mole ofglutamine consumed. Depending on the cell line, batch cultures can accumulate ammonium up to concentrations of 3 to 8 mM in two or three days, and fed-batch cultures without complete medium replacement may be exposed to even higher levels. 1° Ammonium, in equilibrium with ammonia in culture medium, can inhibit cell growth at levels ranging from 3 to 20 mM, depending on the cell line, pH, and possibly other culturing conditions.7'8 This toxicity of ammonium is thought to be due to the neutralization of acidic cellular compartments, such as the lysosomes, by ammonia. 8 We postulate that acidic steps in the regulated pathway may also be sensitive to ammonia, resulting in faulty sorting, processing, or secretion of regulated proteins and a decrease in biologically active protein obtainable. Inhibition of protein trafficking processes may occur before decreases in growth or cell viability are observed. We have investigated the effects of elevated ammo-

nium on insulin processing, storage, and secretion in two endocrine cell lines, recombinant mouse pituitary AtT-20 cells and mouse insulinoma/3TC3 cells. Cells were discharged of their stored protein with secretagogue-containing medium, and then the stores were recharged in the presence or absence of ammonium. During recharging, the flux of protein into the regulated pathway is increased, and thus any effects of ammonium on the pathway are amplified. The amounts of stored and secreted insulin immunoreactive peptides measured by immunoassay were compared for cells recharged with or without added ammonium. To allow us to postulate more specific mechanisms of the effects of ammonium, the levels of bioactive peptide were estimated, using high-performance liquid chromatography (HPLC) to separate insulin from its less active precursor, proinsulin. Additionally, the amounts of total cell protein were measured to determine any effect of ammonium on overall protein accumulation.

Materials and methods

Cell lines and culture media Insulin-producing AtT-20 cells were obtained from the laboratory of Dr. Regis Kelly (Department of Biochemistry and Biophysics, University of California, San Francisco, CA) and cultured in complete Dulbecco's Modified Eagle's Medium (C-DMEM) consisting of DMEM (25 mM glucose) plus 10% fetal bovine serum (FBS) and additional L-glutamine to a total of 6 mM (all from Sigma Chemical Company, St. Louis, MO). A passage number of 1 was assumed for the cells received from Dr. Kelly's lab and was incremented by 1 at every subculturing. AtT-20 cells were subcultured approximately every 5 days and fed every 1-2 days. Secretion experiments were performed on cells of passage number 15-25 grown on flasks coated with 0.5 ml of 0.01% polyL-lysine (Sigma) per 25 cm 2 of surface area to strengthen attachment./3TC3 cells of passage number 20 were obtained from the laboratory of Dr. Shimon Efrat (Department of

Enzyme Microb. Technol., 1994, vol. 16, February

91

Papers Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY). They were cultured in the same DMEM described above, supplemented with 15% heat-inactivated horse serum (HIHS, obtained from Sigma) and 2.5% FBS (/3-DMEM). Passage number was incremented by 1 at every subculturing./3TC3 cells were subcultured every 7-12 days and fed every 2-3 days; cells of passage number 30-36 were used for experiments. All cultures were maintained at 37°C in a humidified 5% COJ95% air atmosphere. Cultures consistently tested negative for mycoplasma contamination using a nucleic acid hybridization-based mycoplasma test kit (GenProbe Incorporated, San Diego, CA).

sayed. Medium samples were chilled, cleared by centrifugation, split into aliquots, and stored at - 7 0 ° C until assayed.

Nuclei counts Cell numbers were estimated with nuclei counts. Cells were treated with a crystal violet/citric acid solution. Citric acid lysed the cell membranes, and crystal violet stained each cell's intact nucleus. Nuclei were then counted on a hemocytometer. This technique does not differentiate nuclei from viable and nonliving cells.

Assays Media used for discharging~recharging experiments The discharge medium for AtT-20 cells (DMEM-S) consisted of DMEM containing 5 mM of the cAMP analog 8-bromocyclic AMP. Discharge lasted 4 h. The recharge medium was C-DMEM. Recharge media with varying levels of added ammonium chloride were adjusted to the same pH of approximately 7.05 before filter sterilization. The discharge medium for/3TC3 cells (/3SM) was formulated to resemble DMEM, except that it contained 5 mM glucose, 1 mM of the phosphodiesterase inhibitor isobutylmethylxanthine (IBMX), and 1 /xM of the calcium ionophore carbachol. Discharge lasted from 2 to 2.5 h. Recharge medium (fiRM) was also formulated to resemble DMEM (25 mM glucose) with the following modifications: it contained lower levels of calcium (0.5 mM as opposed to 1.778 mM for normal DMEM) and potassium (1 mM instead of 5.365 mM), and was supplemented with 20 txM of the calcium ion blocker verapamil. These changes were made to hyperpolarize the membrane and reduce calcium-mediated secretion, as shown by Grampp. H,]2 The recharge medium was also supplemented with 15% HIHS and 2.5% FBS. Recharge media with varying levels of added ammonium chloride were adjusted to the same pH of approximately 7.05 before filter sterilization. For both AtT-20 and/3TC3 cells, the maximum ammonium concentration investigated was 6 mM. In preliminary experiments, both 6 and 12 mM ammonium were found to inhibit the regulated pathway of secretion in AtT-20 cells; the lower concentration was chosen for in-depth studies to highlight the sensitive nature of the regulated pathway and since it is much more likely to be reached in culturing operations.

Insulin was assayed with a double antibody radioimmunoassay kit, originally purchased from Ventrex (Portland, ME) and later from Binax (South Portland, ME). This assay utilizes competitive binding to antibody sites between the sample and ~25I-labeled insulin. The manufacturer reports a 33% crossreactivity of the antibody for proinsulin. The term insulin-related peptides (IRP) thus refers to both immunoreactive insulin and proinsulin reacting with the antibody according to their respective activities. In most cases, IRP values were normalized with a representative nuclei count. Separation of insulin and proinsulin was accomplished with the use of reverse-phase HPLC. The system used was an LDC Analytical (West Palm Beach, FL) Multiple Solvent Delivery System equipped with a UV Detector. The column was 10 cm long, 4.6 mm in diameter, and packed with 3 tzm ODS Microsorb particles (Rainin Instrument Corporation, New Haven, CT). Column temperature was controlled at 35°C, and relative UV absorbance was monitored at 200 nm. Samples were run using the gradient program described by Grampp, II developed from the technique of Halban et al.13 Two solvents were used. Buffer A consisted of 50 mN sodium perchlorate, 20 mM triethylamine, and 50 mM phosphoric acid, adjusted to pH 3. Buffer B was 90 vol % acetonitrile and 10 vol % water. The solvent composition was held isocratic at 36% B for 9 min, ramped to 39% B over the next 5 min and held constant at 39% B for 5 more min. The column was then washed with 60% B for 10 min and reequilibrated to the initial conditions. Fraction-collecting tubes contained 100/~1 of a

106:

Preparation of samples Cell extracts were prepared in either of two ways, specified for each experiment in Results. (1) Some cell extracts were prepared by shearing the cells with an ice-cold lysis buffer containing 66 mM Naz(EDTA) • H 2 0 , 1% Triton X- 100, 10 mM trizma HCI, 0.02% sodium azide, and 0.2 Trypsin Inhibition Units (TIU) ml -~ aprotinin in phosphate-buffered saline (PBS), adjusted to a pH of approximately 7.2. The resulting suspension was centrifuged and the supernatant split into aliquots and stored at - 70 ° C until assayed. (2) Other extracts were obtained by detaching the cell monolayer by treatment with trypsin/EDTA. Trypsin was neutralized with C-DMEM or fl-DMEM, and the suspension was centrifuged. The resulting pellet was resuspended in ice-cold sonication buffer consisting of PBS with 0.2 TIU ml-~ aprotinin and sonicated in 3 × 15-s bursts on a 4710 Series Ultrasonic Homogenizer (Cole Parmer, Chicago, IL). After centrifugation, the extract was filtered through a low-protein-binding 0.45-/~m membrane to remove any particles that could clog the HPLC column, split into aliquots, and frozen at - 7 0 ° C until as-

92

Enzyme

Microb.

Technol.,

1 9 9 4 , v o l . 16, F e b r u a r y

0 mM NH4CI . . . . -o . . . .

10 s

1 0 4

.

0

.

.

.

,

.

.

.

.

50

,

100

.

.

.

.

,

6 mM NH4CI

•

150

Time, hr

Figure 2 Effect of a m m o n i u m on growth of AtT-20 cells. Cultures were initiated identically on 25 c m 2 tissue culture flasks in C-DMEM with either 0 or 6 mM added a m m o n i u m chloride. Flasks were fed with fresh C-DMEM containing the same a m m o n i u m concentration every 1 to 2 days. At each time point, 1 or 2 flasks of each a m m o n i u m concentration were sacrificed for nuclei counts. Error bars represent standard deviations of counting; duplicate sacrifices showed that the uncertainty of counting was similar to that between flasks

Inhibition of protein secretion by ammonium: J. J. Dyken and A. Sambanis 14000"

Discharge ~

Recharge in CDMEM 0 mM added NH4CI

200

0." E

12000" 10000"

8 ~~

8000"

"5 : ~

6000'

"E

o cO

6 mM added NH4C[

4000.

--O 100 E

E

o

2000

=L

E

0

;

~ Time,

,'2

~6

2'0

hr

o

e~

0

if)

Figure 3 Effect of a m m o n i u m on recharging of AtT-20 cells. Cells were grown on 25 cm 2 tissue culture flasks coated with poly-L-lysine until each flask contained approximately 7 x 106 cells. Cells were exposed to DMEM-S for 4 h. At time zero, cells were switched to recharging C-DMEM with either 0 or 6 mM added a m m o n i u m chloride. At each time point, two or three flasks of each a m m o n i u m concentration were extracted with lysis buffer. Error bars represent standard deviations. The remaining flasks were fed with fresh recharging medium containing the same concentration of added a m m o n i u m 6 h into the recharge period

sodium borate/(BSA) solution (pH 9) to neutralize the elution buffer and minimize protein adsorption to tube walls. Standards of human insulin and human proinsulin, kindly donated by Dr. Ronald Chance (Eli Lilly, Indianapolis, IN), were injected to determine retention times. Insulin eluted between 4 and 5.5 min, and proinsulin eluted between 12 and 14 min. Insulin peaks were normalized either with the number of cells in cultures or the total protein assayed in cell extracts. Total cell protein was measured with a dye-binding assay purchased from Bio-Rad Laboratories (Richmond, CA). Results

Effect of ammonium on growth of AtT-20 cells AtT-20 cells were cultured in C - D M E M with 0 or 6 mM added ammonium. Approximately every 48 h,

Recharge in 13RM

t` "5

~

1800

1400 2 .

1000

0mM

6 mM added NH4CI E 600

I -2

2 4 mM Ammonium

0

1

800 n600

-8 E tt~ OLU

400

c o

E

-~

200

o L~

0 b

2

4

mM Ammonium

Figure 5 A m m o n i u m dependence of AtT-20 and/~TC3 recharging. AtT-20 and /3TC3 cells were cultured and discharged as described in Figures 3 and 4, respectively. Cells were then switched to recharging medium with various concentrations of a m m o n i u m between 0 and 6 mM. (a) Dependence of AtT-20 recharging on a m m o n i u m concentration. Cultures were fed with fresh recharging medium containing the same concentration of added a m m o n i u m 5.5 h into the recharge period, and t w o or three flasks of each a m m o n i u m concentration were extracted with lysis buffer 15.5 h into the recharge period (b) Dependence of/3TC3 recharging on a m m o n i u m concentration. Cultures were fed with fresh recharging medium containing the same concentration of added a m m o n i u m 5 h into the recharge period, and three flasks of each a m m o n i u m concentration were extracted with sonication 18 hours into the recharge period

J

-

o

cL ¢n

added N H 4 C I

0

a

, 2

.

, 6

.

Time,

, 10

.

, 14

.

, 18

hr

Figure 4 Effect of a m m o n i u m on recharging of/~TC3 cells. Cells were grown on 25 cm 2 tissue culture flasks until each flask contained approximately 2 × 107 cells. Cells were discharged for 2 h with/~SM; cells were then recharged in/3RM with either 0 or 6 mM added a m m o n i u m chloride. At each time point, two or three flasks of each a m m o n i u m concentration were extracted with lysis buffer; the remaining flasks were fed with fresh/3RM containing the same concentration of added a m m o n i u m 6 h into the recharge period

flasks from each ammonium concentration were sacrificed; Figure 2 shows the cell densities measured. Cell growth in ammonium-supplemented cultures was the same as in control cultures over a 6-day period. No differences in cell morphology were observed.

Effect of ammonium on recharging of intracellular protein stores AtT-20 cells were induced to release their intracellular protein stores by exposing them to D M E M - S for 4 h. Cells were then switched to recharging medium with either 0 or 6 mM added ammonium. T h r o u g h o u t the

Enzyme Microb. Technol., 1994, vol. 16, February

93

Papers

7. 3 6 8.41

INSULIN

¢ /

P tlI I

i\

~ 17.57

Figure 6 Extract from never-discharged ~TC3 cells. Cells were grown on 25 cm 2 tissue culture flasks until each flask contained approximately 2 × 107 cells. Cells were extracted with sonication, and the extract was subjected to HPLC. The resulting UV absorbance plot is shown. The peak at 8.41 min corresponds to insulin

experiment, there were no visible differences in morphology or medium color between cultures with and without ammonium. A typical time course of intracellular IRP during the experiment is shown in Figure 3. Cultures recharged in the presence of ammonium had significantly lower IRP levels at all points during the recharging period compared to cultures recharged in the absence of ammonium. Both cultures showed increases in intracellular IRP from the predischarge value. Part of this increase may be due to a similar increase in cell number in the two cultures over the 15.5-h recharge period. The ammonium inhibition effect was reproducible and was statistically significant using the student's t test (c~ = 0.05). A similar experiment was performed on/3TC3 cells. The recharge medium contained membrane hyperpolarizing agents to suppress secretion. Again, there were no apparent differences in cell morphology or medium

94

Enzyme Microb. Technol., 1994, vol. 16, February

color throughout the experiment. Typical results are shown in Figure 4. As with AtT-20 cells, intracellular IRP levels were reduced during the recharge period for cultures in the presence of ammonium (~ = 0.05). This result has also been reproduced. The doubling time of /~TC3 cells is approximately 35 h, 11 as compared to about 28 h for ART-20 cells, ¿4 so a smaller increase in cell number over recharging is expected. The suppression of secretion during recharging of /3TC3 cells was not complete, since relatively high IRP levels were measured in spent recharging medium. IRP secretion during the/~TC3 recharging experiment was slightly increased in the presence of ammonium (c~ = 0.1, data not shown). A similar increase in secretion was observed for AtT-20 cells. The net amount of IRP synthesized in each case was lower in the presence of ammonium due to the large differences in intracellular stores.

Inhibition of protein secretion by ammonium: J. J. Dyken and A. Sambanis

/

7.29

INSULIN

[

? ) I !

j

16.

56

76737

a Figure 7 Extracts from flTC3 cells recharged in the absence and presence of ammonium. Cells were grown on 25 cm 2 tissue culture flasks until each flask contained approximately 2 × 107 cells. Cells were exposed to flSM for 2 h, then switched to fiRM with 0 or 6 mM added ammonium chloride. Cultures were fed with 10 ml of fresh fiRM containing the same concentration of added ammonium 5 h into the recharging period; extracts obtained by sonication after 18 h of recharge were subjected to HPLC. (a) UV absorbance plot for cells recharged in the absence of ammonium. The peak at 7.29 min corresponds to insulin. The peak at 16.56 rain is due to the verapamil in the recharging medium, which ends up in the soluble extract after sonication. (b) UV absorbance plot for cells recharged in the presence of 6 mM ammonium. The peak at 7.68 min corresponds to insulin, and the peak at 16.70 rain is from verapamil. The peak at 6.98 is a background peak observed also when fresh sonication buffer was injected

Dependence of recharging on ammonium concentration Dosage experiments were performed on AtT-20 and /3TC3 cells by discharging the cells and recharging them in varying concentrations of ammonium from 0 to 6 mM. Secretion during recharging was suppressed in the /3TC3 experiment. Intracellular IRP levels after 15.5and 18-h recharges of AtT-20 and flTC3 cells, respectively, are shown in Figure 5a and b, respectively. For both cell lines, a reduction in accumulated IRP was always observed at 6 mM ammonium relative to control cultures (o~ = 0.05), but the apparent dosage effect for

intermediate concentrations was not always reproducible, and it is not statistically significant in the results shown. There were no differences seen in cell morphology or medium color between cultures of different ammonium concentration in each experiment. In the flTC3 dosage experiment, total cell protein was measured for each flask. There were no statistical differences in the total cellular protein levels of cultures exposed to any of the ammonium concentrations tested. For example, cultures incubated in the absence of ammonium had 2.05 +-- 0.35 mg total protein/flask, and cultures incubated in the presence of 6 mM ammonium had 1.87 - 0.35 mg total protein/flask.

Enzyme Microb. Technol., 1994, vol. 16, February

95

Papem

J

S (~7.

6e

~

INSULIN

)

\

]

16.-/B

1 7 . IE~

b

Figure 7 (Continued)

Processing of proinsu/in to insulin in J3TC3 cells The results obtained by radioimmunoassay show a definite reduction in the accumulation of IRP in the presence of ammonium. Next we studied the effect of ammonium on accumulation of the bioactive peptide, insulin. To do this, we used reverse-phase HPLC to separate insulin from its unprocessed precursor, proinsulin. We identified insulin peaks by comparison with human insulin standards and radioimmunoassay of collected fractions; insulin from /3TC3 cells eluted between 7 and 9 rain. This is later than the 4- to 5-min elution time measured for human insulin standards, probably due to differences between murine and human insulin. An extract obtained by sonication of/3TC3 cells never discharged was subjected to HPLC to separate insulin and proinsulin. The resulting UV absorbance plot is shown in Figure 6. The major peak (8.41 rain) corresponds to insulin; there is no distinct proinsulin

96

Enzyme Microb. Technol., 1994, vol. 16, February

peak. This result is expected, since the majority of IRP present in never-discharged /3TC3 cells is stored in granules in the processed form. The proinsulin expected to be present in the RER, Golgi, and constitutive vesicles appears to be minimal. Next,/3TC3 cells were exposed to/3SM for 2 h, and spent medium was subjected to HPLC. A large, narrow peak corresponding to insulin was observed at 7.43 min (data not shown). No significant proinsulin peak was observed; this is consistent with the release of mainly stored, processed protein upon secretagogue stimulation. After discharging,/3TC3 cells were switched to recharge medium with 0 or 6 m ~ added ammonium. The medium contained hyperpolarizing agents and reduced calcium to suppress secretion during recharging. After recharging for 18 h, extracts obtained from sonication of these cultures were subjected to HPLC. All experimental conditions were the same for the two cultures, and the amounts of total protein loaded onto the column

Inhibition of protein secretion by ammonium: J. J. Dyken and A. Sambanis were within 10%. Figure 7 shows insulin peaks at 7.29 and 7.68 min for the cultures recharged in 0 and 6 mM ammonium, respectively. The insulin peak from the culture exposed to ammonium was fivefold smaller than the one from the ammonium-free culture. The large peak seen in both cultures at 16 to 17 min is caused by the verapamil in the recharge medium, which remains associated with the soluble fraction of the extract and absorbs strongly at the wavelength observed. Since it is possible that proinsulin coeluted with this agent, we are unable to determine proinsulin levels in these samples. After 18 h, spent medium from cells recharged in the presence of ammonium was subjected to HPLC. There was no insulin peak visible (data not shown). The results described above have been reproduced in two ways. Duplicate samples from the same experiment were run, and samples from a duplicate experiment were also assayed. The inhibition of insulin accumulation in the presence of ammonium was observed in all cases.

Discussion Endocrine cells in unstimulated growth have a relatively constant level of intracellular proteinS; there is only a low turnover of stored protein. Thus, it is difficult to see changes in intracellular protein resulting from inhibition of the regulated pathway. However, if cells are discharged of most of their protein stores with secretagogues and then allowed to build them back up during a recharge period, there is an increased flux of newly synthesized regulated proteins to the storage granules. Since the majority ofintraceilular protein will have been newly shuttled through maturing vesicles into granules, it is easier to detect the effects of different agents on steps leading to intracellular storage. We have exploited the above considerations to amplify the effect of ammonium on storage of protein in the regulated pathway. We found that ammonium inhibited some aspect(s) of the pathway, resulting in a lower level of intracellular IRP during and after recharging for both AtT-20 and/3TC3 cultures. This was accomplished by a slight increase in IRP secretion during the recharge period. In/3TC3 cells discharged and recharged in medium with agents blocking secretion, the presence of ammonium resulted in the accumulation of less insulin. The growth rate of AtT-20 cells was not affected by the presence of 6 mM ammonium, the maximum investigated thoroughly in this study. Recent results have shown that long-term growth of~TC3 cells, however, is inhibited by the presence of 6 mM ammonium. During the course of the recharging experiments, there were no perceptible changes in cell viability or morphology at the concentrations of ammonium for which we observed relatively immediate effects on the regulated pathway of both AtT-20 and/3TC3 cells. These results indicate that ammonium selectively inhibits accumulation of processed protein in regulated stores, since no other cell functions were inhibited by

ammonium during the course of the experiments. It is unlikely that this effect is due to the inhibition of proinsulin expression by ammonium. AtT-20 cells have a constitutive promoter which should not be affected by ammonium at the concentrations tested, and changes in /~TC3 expression documented until now have been due only to changes in glucose.~5 Also, other researchers have found that ammonium does not affect expression in several cell lines. 7'16'17 Finally, inhibition of expression would likely occur for a wide range of cellular proteins; we observed no differences in total cell protein between ammonium-free and ammonium-supplemented/3TC3 cultures. There are several possible mechanisms for the actions of ammonium which may together or separately account for the results presented. First, the sorting of proinsulin into the regulated pathway may be disrupted in ammonium-supplemented cultures, resulting in constitutive release of the molecule. A second hypothesis is that proinsulin is correctly sorted into the regulated pathway, but the presence of ammonium disrupts conversion of clathrin-coated vesicles to storage granules. This could result in either continuous release of unprocessed protein or accumulation of proinsulin-containing vesicles, due to the shifting of maturing vesicles from the optimal pH for enzyme activity. Since proinsulin has a lower crossreactivity to the antibody used in the RIA than insulin, the overall IRP measured will decrease. Discriminating unequivocally between the foregoing possibilities would necessitate probing the intracellular location of proinsulin and insulin at the ultrastructural level, e.g., with immuno-gold electron microscopy? However, from the results presented here, one may speculate that the inhibitory effect of ammonium is the result of a combination of the above mechanisms. Although the data showing lower levels of IRP and insulin in cells recharged in the presence of ammonium are consistent with all of the above hypotheses, the additional observation of greater IRP secretion from the cells indicates that some mis-sorting or continuous release of immature vesicles of the regulated pathway is occurring. It is also consistent with previous findings that another weak base, chloroquine, inhibited regulated secretion of adrenocorticotropic hormone (ACTH) from AtT-20 cells while increasing the continuous release of ACTH precursors. TM Since the uncertainty inherent in the techniques used in this work is too large to measure slight differences in protein levels, though, it is also possible that unprocessed proinsulin accumulates either in vesicles of the regulated pathway or in constitutive vesicles blocked from secretion by membrane hyperpolarizing agents. Each of the mechanisms for the effect of ammonium may have different consequences on long-term cell function. There is a well-documented overturn of protein within the regulated pathway of secretion. 19If misprocessed hormone accumulates, cells may be able to recognize its presence in storage granules and selectively degrade it. Eventually, the misprocessed protein would be replaced with correctly processed hormone,

Enzyme Microb. Technol., 1994, vol. 16, February

97

Papers since enzymatic activity would be merely diminished by ammonium. If instead, ammonium interferes with sorting or regulated vesicle maturation so that proinsulin is continuously released, the cell will sense nothing wrong with storage granules. If normal degradation continues, intracellular stores may be diminished, and long-term cell function will be seriously impaired. The experiments reported here show a clear inhibition of the regulated secretion pathway by ammonium at experimentally relevant concentrations. Although the molecular nature of this effect remains in question, the inhibitory nature of ammonium needs to be taken into consideration in the design of both production schemes and bioartificial organs. Cell aggregates, either free or immobilized in biocompatible polymers, experience mass transfer limitations which may result in the buildup of ammonium to levels interfering with efficient processing and/or secretion. Many feeding strategies rely on the indicator of cell viability or growth to determine allowable ammonium levels; this work clearly shows the fallacy of such thought. The benefit of obtaining a biologically active product may outweigh added medium costs from more frequent feedings.

Acknowledgements The authors wish to acknowledge financial support from NIH Training Grant GM08433-01, the Georgia Tech Biomedical Research Support Committee, and the Emory/Georgia Tech Biomedical Technology Research Center. J.J.D. was supported by DuPont and Amoco Fellowships. The authors would also like to thank Dr. Shimon Efrat and Dr. Regis Kelly for donating cells and Dr. Ronald Chance for donating human insulin and proinsulin standards.

4 5

6 7 8 9

10

11 12

13

14

15

16

References 1 2 3

98

Mains, R. E., Cullen, E. I., May, V. and Eipper, B. A. The role of secretory granules in peptide biosynthesis. Ann. N.Y. Acad. Sci. 1987, 493, 278-291 Burgess, T. L. and Kelly, R. B. Constitutive and regulated secretion of proteins. Ann. Rev. Cell Biol. 1987, 3, 243-293 Orci, L., Ravazzola, M., Storch, M.-J., Anderson, R. G. W., Vassalli, J.-D. and Perrelet, A. Proteolytic maturation of insulin

Enzyme Microb. Technol., 1994, vol. 16, February

17 18 19

is a post-Golgi event which occurs in acidifying clathrin-coated secretory vesicles. Cell 1987, 49, 865-868 Grampp, G. E., Sambanis, A. and Stephanopoulos, G. N. Use of regulated secretion in protein production from animal cells: An overview. Adv. Biochem. Eng. Biotech. 1992, 46, 35-62 Sambanis, A., Stephanopoulos, G. N. and Lodish, H. F. Multiple episodes of induced secretion of human growth hormone from recombinant AtT-20 cells. Cytotechnology 1990, 4, 111-119 Goosen, M. F. A. Insulin delivery systems and the encapsulation of cells for medical and industrial use. CRC Critieal Reviews in Biocompatibility 1987, 3, 1-24 Adema, E. Ammonium toxicity in mammalian cell culture. Ph.D. Thesis, Massachusetts Institute of Technology, 1989 McQueen, A. and Bailey, J. E. Growth inhibition of hybridoma cells by ammonium ion: Correlation with effects on intracellular pH. Bioproc. Eng. 1991, 6, 49-61 Kurano, N., Leist, C., Messi, F., Kurano, S. and Fiechter, A. Growth behavior of chinese hamster ovary cells in a compact loop bioreactor. 2. Effects of medium components and waste products. J. Bioteeh. 1990, 15, 113-128 Glacken, M. W., Fleischaker, R. J. and Sinskey, A. J. Reduction of waste product excretion via nutrient control: Possible strategies for maximizing product and cell yields on serum in cultures of mammalian cells. Biotech. Bioeng. 1986, 28, 1376-1389 Grampp, G. E. Controlled Protein Secretion in Animal Cell Culture. Ph.D. Thesis. Massachusetts Institute of Technology, Cambridge, MA, 1992 Grampp, G. E. and Stephanopoulos, G. N. Development and scale-up of controlled secretion processes for improved product recovery in animal cell culture. Ann. N. Y. Aead. Sei. (Biochem Eng VII) 1992, 665, 81-93 Halban, P. A., Rhodes, C. J. and Sheolson, S. E. High-performance liquid chromatography (HPLC): A rapid, flexible, and sensitive method for separating islet proinsulin and insulin. Diabetologia 1986, 29, 893-896 Sambanis, A., Stephanopoulos, G. N., Sinskey, A. J. and Lodish, H. F. Use of regulated secretion in protein production from animal cells: An evaluation with the AtT-20 model cell line. Biotech. Bioeng. 1990, 35, 771-780 Nagamatsu, S. and Steiner, D. F. Altered glucose regulation of insulin biosynthesis in insulinoma cells: Mouse/3TC3 cells secrete insulin-related peptides predominantly via a constitutive pathway. Endocrinology 1992, 130, 748-754 McQueen, A. and Bailey, J.E. Effect of ammonium ion and extracellular pH on hybridoma cell metabolism and antibody production. Biotech. Bioeng. 1990, 35, 1067-1077 Glacken, M. W., Adema, E. and Sinskey, A. J. Mathematical descriptions of hybridoma culture kinetics: I. Initial metabolic rates. Bioteeh. Bioeng. 1988, 32, 491-506 Moore, H.-P., Gumbinger, B. and Kelly, R. B. Chloroquine diverts ACTH from a regulated to a constitutive pathway in AtT-20 cells. Nature 1983, 302, 434-436 Halban, P. A. Structural domains and molecular lifestyles of insulin and its precursors in the pancreatic beta cell. Diabetologia 1991, 34, 767-778