Subcellular localization and partial characterization of insulin proteolytic activity in rat liver

180

Biochirnica et Biophvsica Acta 840 (1985) 180-186

Elsevier BBA 22053

S u b c e l l u l a r l o c a l i z a t i o n and partial c h a r a c t e r i z a t i o n o f insulin p r o t e o l y t i c activity in rat liver B a l v i n d e r K . C h o w d h a r y , G e o f f r e y D . S m i t h * a n d T i m o t h y J. P e t e r s Division of Clinical Cell Biology, MRC Clinical Research Centre. Watford Road, Harrow, Middlesex ( U.K.)

(ReceivedOctober 19th, 1984) (Revised manuscript receivedMarch 11th. 1985)

Key words: Insulin proteolysis;Subcellularlocalization;(Rat liver)

Using (A 14.12s|)-insulin as a tracer, insulin proteolytic activity in rat liver was found to be localized both to the cytosol and the endoplasmic reticulum. The membrane-associated activity was highly latent (70-80%). Both cytosolic and particulate activities had similar Km values and M r of approx. 300000 by gel filtration. Both were strongly inhibited by diamide (90%), but were unaffected by leupeptin or pepstatin. A comparison of the subceilular distributions with various 12s 1-isomers of insulin as tracers showed that both particulate and cytosolic activities were highest with (A 14-12sI)-insulin.

Introduction The degradation of insulin by rat liver has been the subject of much study. It has been proposed that insulin is first cleaved by glutathione transhydrogenase, with reduced glutathione, into its A and B chains, followed by proteolysis of the individual peptides to lower molecular weight peptides [1]. Others have suggested that the glutathione-independent proteolysis of insulin is the only important degradative route in rat liver [2,3]. Substrate specificity studies with the rat skeletal muscle [2,4] and liver [5] enzymes have shown that t h e purified glutathione-independent enzyme has high specificity, utilizing only insulin and glucagon as substrates. The subcellular localization of the glutathione-dependent enzyme has been shown by Ansorge et al. [6] to be microsomal, and more specificially, by Chowdhary et al. [7] to be located to the endoplasmic reticulum. In contrast, the glutathione-independent enzyme has been localized to the cytosol [6]. However, Ansorge and * To whom correspondenceshould be addressed.

colleagues used differential centrifugation techniques and non-specifically labelled insulin as substrate. In the present study we describe the subcellular localization of a glutathione-independent insulin proteolytic activity by analytical subcellular fractionation techniques. The apparent subcellular distribution of insulin proteolytic activity with various isomers of 125I-insulin as substrate is shown. Preliminary characterization of the insulin proteolytic activity is also presented. The subcellular distribution of insulin proteolytic activity is compared with the distribution of internalized insulin.

Materials and Methods Chemicals

Zinc-free insulin was prepared from Monotard MC insulin (Novo Laboratories, Basingstoke, Hants). Insulin was iodinated primarily at the A14 position with lactoperoxidase (Sigma (London)) and purified by chromatography on QAE-Sephadex [8]. For the isolation of (A19-a2sI) -, ( B 2 5 125I)-, (A14-12sI) - and (B16-125I)-insulins the

0304-4165/85/$03.30 © 1985 ElsevierSciencePublishers B.V. (Biomedical Division)

181

labelled protein mixture was chromatographed on HPLC by an adaptation of the method described by Welinder et al. [9]. Samples (0.5 ml) were chromatographed on a 20 cm × 5 mm column of Hypersil butyl W300 (Shandon) and eluted with 23.5% acetonitrile in 0.25 mol/1 triethylaminephosphate buffer (pH 3.0) at a flow rate of 1 ml/min. All other chemicals were obtained from Sigma (London) or BDH Chemicals.

Enzyme assay Insulin proteolytic activity was assayed essentially as described by Burhgen et al. [5]. Samples were incubated for about 20 min at 37°C with 0.6 /~M monocomponent insulin containing up to 900 Bq 125I-insulin in 0.1 M Tris-HC1 buffer (pH 7.0) containing 3 g / l bovine serum albumin (fraction V, Sigma (London)) in a total volume of 0.22 ml. The reaction was stopped by the addition of 0.5 ml 20% (w/v) trichloroacetic acid and centrifuged for 2 min in an Eppendorf microfuge. The percentage of degraded insulin was determined by counting the trichloroacetic acid-soluble radioactivity. Activity was expressed as mU, where 1 mU is 1 nmol insulin degraded/min. Substrate blanks were carried out in all experiments to correct for non-enzymic degradation of insulin during the incubation. The assay was linear with respect to incubation time and protein concentration, provided not more than 40% of the substrate was degraded. Glutathione-insulin transhydrogenase was assayed as described previously by Chowdhary et al. [7].

Analytical subcellular fractionation Rat liver (2 g) was homogenized in 10 ml 0.25 M sucrose, containing 1 mM EDTA and 20 mM ethanol. The homogenate was subjected to sucrose density gradient centrifugation in a Beaufay rotor. Homogenization, analytical subcellular fractionation and assay of marker enzyme activities, protein and DNA were as described by Smith and Peters [10] and Smith et al. [11]. Cytosol and microsomes were prepared for gel filtration and latency studies, by differential centrifugation. Liver (2 g) was homogenized in 10 ml 0.25 M sucrose, 5 mM imidazole buffer (pH 7.4). The homogenate was centrifuged for 6 • 103 gmin in a 4 × 50 ml swing out rotor. The supernate was removed and centrifuged for 4.105 groin in an 8 × 50 ml fixed-angle

rotor. The post-mitochondrial/lysosomal supernatant was then centrifuged for 6.106 gmin in a 10 × 10 ml angle rotor to obtain a microsomal pellet and cytosol (supernatant). The microsomal pellet was resuspended in 1-2 ml homogenization medium with five strokes of a loose-fitting pestle in a small Dounce homogenizer.

Gel filtration Molecular weight estimations were performed on a column (70 × 2.2 cm) of Sepharose-6B, preequilibrated with 0.1 M Tris-HC1 buffer (pH 7.6) containing 0.1 M NaC1 at 20 m l / h for 20 h. The void volume was determined with blue dextran, and the column was calibrated with a set of standards of known molecular weight (Pharmacia, high molecular weight kit). Results

Subcellular fractionation studies Rat liver was subjected to analytical subcellular fractionation as described in Materials and Methods. The distribution of insulin proteolytic activity is plotted as a frequency-density histogram in Fig. 1 and compared with the distribution of other "membrane marker enzymes. The distribution of insulin proteolytic acitivity is bimodal; one of the peaks coincides with that of lactate dehydrogenase, the marker enzyme for cytosol. The second peak lies in the particulate region of the gradient, and is similarly distributed to ~-glucosidase, the endoplasmic reticulum marker. However, there is considerable overlap with the distribution of 5'nucleotidase, the plasma membrane marker. The particulate activity was shown not to be due to simple adsorption of cytosolic activity to membranes, by homogenization in sucrose medium containing 0.1 M KC1 (thin line) which desorbed some of the lactate dehydrogenase activity, but had no effect on the distribution of insulin proteolytic activity. The selective membrane perturbants pyrophosphate and digitonin were used to distinguish between localization of the insulin proteolytic activity to endoplasmic reticulum or to plasma membrane. Homogenization in the presence of 15 mM sodium pyrophosphate followed by subcellular fractionation caused a decrease in the equilibrium

182 Digitonin 15

Insulin protease

15

Succinate dehydrogenase

Pyrophosphate

5' Nucleotidase

/<

5 0

1

15

•

0

r

15 ~ o~ Glucosidase

gl

10 5

LL 0

15

1 c~

1.05

Glucosidase

1.15

l--

I

r

i

@

i

5' Nudeotidase

125

1.05

1.15

15

1.25

Fig. 1. Frequency-density distribution of insulin proteolytic activity in rat liver. The distribution profile of insulin proteolyric activity is compared with enzyme markers for the mitochondria (succinate dehydrogenase), cytosol (lactate dehydrogenase), lysosomes (N-acetyl-fl-glucosaminidase), endoplasmic reticulum (a-glucosidase) and plasma membrane (5'nucleotidase). The effect of homogenizing rat livers in sucrose medium containing 0.1 M KC1 (fine line) on the distribution of insulin proteolytic activity and lactate dehydrogenase is shown.

density of membrane-bound a-glucosidase, due to removal of the ribosomes from the rough endoplasmic reticulum (Fig. 2, right column). A similar shift to lower equilibrium density is shown for insulin proteolytic activity. Treatment of the homogenate with digitonin caused an increase in the equilibrium density of membrane-bound 5'nucleotidase, the plasma membrane marker (Fig. 2, left column), an effect not shown by a-glucosidase or insulin proteolytic activity. These results indicate that insulin proteolytic activity is located to the cytosol and endoplasmic reticulum. A comparison of insulin proteolytic activity in a sucrose density gradient was made with (A14-

125I)., (A19_125I)_ (B26-125I)- (B16-tzsI)-insulin

isomers. Fig. 3 shows the distribution of the prot e o l y t i c a c t i v i t y w i t h t h e s e f o u r i s o m e r s as t r a c e r s . Both cytosolic and particulate activities were markedly lower when assays were carried out with (A19-12sI)-, ( B26-125I) - a n d (B16-tzsI)-insulin a s

Insulim protease

1.05

1.15

1.25

1.05

1.15

1.25

Density (g.cm -3) Fig. 2. The effect of digitonin and pyrophosphate on the

frequency-density distribution of insulin proteolytic activity, a-glucosidase and 5'-nucleotidase. The distribution profiles are shown for control data (Fig. 1) (fine line) and from rat livers homogenized in the presence of 1 m g / m l digitonin or 15 mM sodium pyrophosphate (thick line).

substrates compared to (A 14-~2~I)-insulin. The relative proportion of both cytosolic and particulate activities remained constant for the B16- and 14A 19A . . . . . 8 --

26B - 16B . . . . .

g, N

01.05

1.10

1.15

1.20

1.25

1.30

Density (g.cm-3) Fig. 3. Comparison of the effect

(B26-125I) - and (B16-12~I)-insulin

(A14-1251)-, (AI9-12~AI)-,

isomers as substrates on the observed frequency-density distribution of insulin proteolytic activity in rat liver. The distribution profile of insulin proteolytic activity with (A19-125I)-, (B26-125I)- or (B16-125I)-insulin as substrates is shown relative to the distribution of insulin proteolytic activity with (A14-1251)-insulin.

183

B26-isomers but the particulate activity is greater

60 ~ o

affected in the A19-isomer.

Characterization of insufin proteolytic activity Some other biochemical characteristics of the insulin proteolytic activity from both subcellular localizations were examined in an attempt to determine whether the activities were due to the same enzyme protein. Fractions corresponding to the cytosol and endoplasmic reticulum fractions were used as the sources of enzyme activity. Kinetic studies showed particulate and soluble activities to have identical pH optima at pH 7.0 with both activities showing an apparent K m of between 5 and 7 #M. The activity of the particulate fraction was found to increase on freezing and thawing. The latency of the enzyme in a microsomal preparation was therefore investigated with various detergents and by sonication. The effects of various detergents on the activity of both the microsomal and soluble enzyme activities is shown in Table I. The detergents had significantly different effects on cytosolic and particulate activity. The effect of controlled sonication on the insulin proteolytic activity of microsomal and cytosol fractions, is shown in Fig. 4. Centrifugation of a sonicated microsomal preparation at 1.105 g for 1 h, resulted in approx. 80% of the total proteolytic activity being released into the supernatant. This indicates that the microsomal activity is free within the vesicles, i.e., not membrane associated.

_

~

.~ 3c 2o 10 I

I

I

012

I

I

I

Effect of metal ions The effects of the addition of the divalent cations, Ca 2÷ and Mg 2+ was also examined on both particulate and cytosolic activities. The effect of increasing Ca 2+ concentration is shown in Fig 5. Clearly, Ca 2 + has a stimulatory effect on the cytosolic activity, and an inhibitory effect on the par-

120

~

g particulate

100 85 10 87 59

100 350 64 93 100

~ _ - - - - o

160

The detergents were added to give a final concentration in the assay of 0.1% (w/v), except for digitonin which was 0.3% (w/v). Activity is expressed as a percentage of the control sample (no addition). Activity (%)

20

Determination of the molecular weight of both cytosolic and released microsomal enzyme by gel permeation chromatography, as described in Materials and Methods, gave an M r 300000 for the activity from each source.

200

cytosolic

I

5 10 15 Sonicati0n (no. of los pulses)

Fig. 4. The effect of sonication on microsoma] (11) and cytosolic ( O ) activities. Sonication was carried out at an amplitude of 26 /tm in an 150 watt sonicator (MSE, Crawley, Sussex). There was a rest period of 30 s on ice in between each 10-s pulse of sonication. Assays were carried out in isotonic conditions in 0.01 M Tris-HCl buffer (pH 7.0) containing 0.25 M sucrose.

E F F E C T OF D E T E R G E N T S ON T H E CYTOSOLIC A N D P A R T I C U L A T E I N S U L I N P R O T E I N A S E ACTIVITIES

None Brij 35 Deoxycholate Digitonin Triton X-100

e

~. 4o

TABLE I

Addition

50

80 40

0

I

0

I

0.5

I

I

I

1 CaCI2(mM)

1.5

2

Fig. 5. Effect of Ca 2+ concentration on insulin proteolytic activity in cytosolic ( © ) and particulate fractions (e).

184

ticulate activity. The inhibition of Ca 2+ was also observed on insulin proteolytic activity, released from a particulate fraction by sonication. Mg 2+ had effects similar to those of Ca 2+, but to a lesser extent, showing 16% inhibition of the particulate activity and 48% stimulation of the soluble activity at 10 mM MgC12. 10 m M ZnCI 2 produced virtually complete inhibition of both forms of the proteolytic activity.

Effect of proteinase inhibitors The effect of some inhibitors of proteinase activity were also examined. Only diamide (1 raM) was a potent inhibitor of insulin proteolytic activity, showing 95% inhibition of both cytosolic and particulate activities. Leupeptin and pepstatin A, known inhibitors of cathepsins B and D, respectively, did not inhibit either form of the activity.

Comparison with the distribution of internalized insulin A rat was injected via the portal vein with 3/~Ci of 125I-labelled insulin. The liver was removed after 2.5 rain, and a 2 g portion homogenized and fractionated on a continuous sucrose density gradient as previously described. A comparison of the distribution of trichloroacetic acid-precipitable iodide ('intact insulin') and trichloroacetic acidsoluble iodide ('degraded insulin') with the distribution of insulin protease activity is shown in Fig. 6. Both intact insulin and iodinated degradation products are found predominantly at a density of 15

8" c::

Acid precipitable ~='~l

Insulin protease

Acid soluble 1251

0~-glucosidase

0

g

L'~- 10

@ 1.05

1.15

125

1,05

1.15

1.25

Density (g,cm 3)

Fig. 6. Fractionation of rat liver following portal vein injection of 1251-labelled insulin. The distribution of trichloroacetic acid-precipitable and -soluble 1251is compared with the distribution of insulin proteinase activity and the endoplasmic reticulum marker, a-glucosidase.

1.12 g. cm -~, a completely different locus to that of insulin proteinase activity.

Discussion We have shown by isopycnic sucrose density gradient centrifugation a dual localization of insulin degrading activity in rat liver. Using the above technique and (A14-~25I)-insulin as substrate, up to 50% of the activity is found in the cytosol, while the rest is localized to the particulate region of the gradient. Use of selective membrane perturbants confirms that the particulate activity is localized to the endoplasmic reticulum. No significant activity was located to ligandosomes [12] or lysosomes. The position of ~25I label on the insulin molecule is crucial in determining the insulin proteolytic activity, since the assay is based on the release of trichloroacetic acid-soluble radioactive fragments of insulin. Both soluble and particulate activities are drastically less with AI9-, BI6- or B26-insulin compared to the most physiologically active form labelled on the A 14-tyrosine. The endoplasmic reticuhim activity showed a latency of about 70-80% as shown by the use of Brij 35 and sonication. Triton and deoxycholate did not reveal a full latency due to their inhibitory effect on the released insulin proteolytic activity. Although digitonin did not significantly inhibit the cytosolic insulin proteolytic activity it did not reveal latency of endoplasmic reticulum activity. The activity was shown to be free within vesicles. Kinetic studies have revealed both cytosolic and endoplasmic reticulum activities to have neutral p H optima and similar K m values. Both activities have an apparent molecular weight of 300 000 as determined by their behaviour on gel filtration. Although the cytosolic and endoplasmic reticulum activities show differential behaviour with metal ions, this is likely to be due to the presence of E D T A in the homogenization medium. The effect of thiols on insulin proteinase activity is somewhat varied (Chowdhary, unpublished results). Thus, cysteine has no effect on the particulate activity but inhibits the soluble activity. Dithiothreitol stimulates the cytosol activity but inhibits the particulate activity. This variation is probably due to the presence of essential thiol groups at the active site and structural disulphide

185 bridges in both enzyme and substrate. Inhibition by diamide of insulin degradation by intact hepatocytes, has been taken as evidence of involvement of glutathione-insulin transhydrogenase in the degradation process [13]. However, in the present study, the marked inhibition by diamide of both particulate and soluble insulin proteolytic activities, would suggest that this is not a definitive test. Lack of inhibition by leupeptin and pepstatin A suggest that the activity is not due to cathepsins B or D. We have shown that neutral insulin proteolytic activity exists in approximately equal amounts in both the cytosol and endoplasmic reticulum. These results are in contrast to those obtained by Burghen et al. [5], who found 96% of the activity in the cytosol. This mainly cytosolic distribution was also found in other studies on skeletal muscle [14], kidney [15], adipocytes [15], heart [17] and rat liver [18]. It is possible that this disparity arises from the use of heterogeneous iodinated insulin tracers in many of these other studies. However, this implies that the activities are showing different substrate specificity and thus are different enzymes. However, the assay for insulin proteinase is rather non-specific, reflecting only release of trichloroacetic acid-soluble material, which may be the result of the combined activity of several enzymes. Thus it is easy to envisage a change in the ratelimiting enzyme with the different tracers. A difference in rate of proteinase activity has already been shown for the different isomers by Brush et al. [19]. There is also the possibility that in studies showing the majority of activity to be cytosolic, that the activity has been 'solubilized' by damage to the microsomes during preparation. This is possible since we have shown the particulate activity to be latent but not membrane bound. Thus, although the majority of biochemical characteristics of the two activities are similar, circumstantial evidence indicates that different proteins may be involved in the overall effect. These studies do not support a physiological role for insulin proteinase activity in the subcellular processing of insulin by the liver. It has been shown in this laboratory [12,20] and in others [21,22] that insulin binds to the plasma membrane and is internalized to a population of low density membranes. The distribution profiles in Fig. 6

show quite clearly that insulin proteolytic activity has a completely different subcellular distribution to both intact insulin and its degradation products. It has been shown that insulin degradation to small peptides does not occur on binding to the plasma membrane [23,24] and the lack of a suitable insulin proteinase on the plasma membrane is confirmed in this present work. Preliminary studies from this laboratory [25] have shown that isolated vesicles of density 1.10-1.14 g - c m -3, containing internalized insulin, will degrade the insulin ex vivo. The studies on the subcellular processing of insulin do not suggest that internalized insulin comes in contact with either the cytosol or the endoplasmic reticulum and thus it is difficult to see how the insulin proteinase activity in these compartments can have access to the insulin.

Acknowledgements We wish to thank Mr. P. White for excellent technical assistance and Mrs. S. Kingsley for typing this manuscript. This work is supported by a grant from the British Diabetic Association.

References 1 Varandani, P.T. (1973) Biochim. Biophys.Acta 304, 642-659 2 Duckworth,W.C., Heinemann, M. and Kitabchi, A.E. (1975) Biochim. Biophys. Acta 377, 421-430 3 Brush,J.S. and Jering, H. (1979) Endocrinol. 104, 1639-1643 4 Yokono, K., Imamura, Y., Shii, K., Sakai, H. and Baba, S. (1981) Endocrinol. 108, 1527-1532 5 Burghen, G.A., Kitabchi, A.E. and Brush, J.S. (1972) Endocrinol. 91,633-642 6 Ansorge, S., Bohley, P., Kirschke, H., Langner, J. and Hanson, H. (1971) Eur. J. Biochem. 19, 283-288 7 Chowdhary, B.K., Smith, G.D., Mahler, R. and Peters. T.J. (1983) Biosci. Rep. 3, 3-329 8 Jorgensen, K.H. and Larsen, U.D. (1980) Diabetologia 19, 546-554 9 Welinder, B.S., Linde, S, and Brush, J.S. (1983) J. Chromatogr. 251, 162-165 10 Smith, G.D. and Peters, T.J. (1980) Eur. J. Biochem. 104, 305-311 11 Smith, G.D., Evans, W.H. and Peters, T.J. (1980) FEBS Lett. 120, 104-106 12 Smith, G.D. and Peters, T.J. (1982) Biochim. Biophys. Acta 716, 24-30 13 Poole, G.P., O'Connor, K.J., Lazarus, N.R. and Pogson, C.I. (1982) Diabetologia 23, 49-53

186 14 Yokono, K., lmamura, Y., Sakai, H. and Baba, S. (1979) Diabetes 28, 810-817 15 Duckworth, W.C. (1976) Biochim. Biophys. Acta 437, 518-530 16 Goldstein, B.J. and Livingstone, J.N. (1980) Biochem. J. 186, 351-360 17 Croall, D.E. and DeMartino, G.N. (1983) J. Biol. Chem. 258, 5660-5665 18 Yokono, K., Imamura, Y,, Sakai, H. and Baba, S. (1979) Endocrinol. 11l, 1102-1108 19 Brush, J.S., Sonne, O. and Gliemann, J. (1983) Biochim. Biophys. Acta 757, 269-273

20 Pease, R.J., Sharp, G.A., Smith, G.D. and Peters, T.J. (1984) Biochim. Biophys. Acta 774, 56-66 21 Ward, W.F, and Mortimore, G.E. (1980) Biochim. Biophys. Res. Commun. 93, 66-73 22 Khan, M.N., Posner, B.I., Khan, R.J. and Bergeron, J.J.H. (1982) J. Biol. Chem. 257, 5969-5976 23 Freychet, P., Kahn, R., Roth, J. and Neville, D.M. (1972) J. Biol. Chem. 247, 3953-3961 24 Wisher, M.H., Dron, D.I., Sonksen, P.H. and Thomas, J.H. (1977) Biochem. Soc. Trans. 5, 313-315 25 Smith, G.D., Christensen, J.R. and Peters, T.J. (1984) Biochem. Soc. Trans. 12, 1082-1083