the localization of retinol-binding protein in rat liver by ...

7 downloads 0 Views 2MB Size Report
Mar 20, 1975 - SUMMARY. The localization of immunoreactive retinol-binding protein (RBP) in rat liver was studied by immunofluorescence microscopy.
J. Cell Sci. 19, 379-394 ('975) Printed in Great Britain

379

THE LOCALIZATION OF RETINOL-BINDING PROTEIN IN RAT LIVER BY IMMUNOFLUORESCENCE MICROSCOPY A. R. POOLE, J. T. DINGLE, A. K. MALLIA AND DeW. S. GOODMAN Tissue Physiology Department, Strangetvays Research Laboratory, Cambridge, England and Department of Medicine, Columbia University College of Physicians and Surgeons, NewYork, NewYork 10032, U.S.A.

SUMMARY The localization of immunoreactive retinol-binding protein (RBP) in rat liver was studied by immunofluorescence microscopy. The study employed specific antisera to rat RBP prepared in a rabbit and in a sheep. The indirect, two-stage method of localizing tissue antigens was employed, and livers of both normal and vitamin A-deficient rats were examined. Fab' fragments of immunoglobulins were used, to minimize non-specific labelling of the frozen sections of liver. With these techniques, the specific immune staining of RBP was observed within liver parenchymal cells. This staining appeared as both particulate and diffuse within the cytoplasm of the parenchymal cells, and was not concentrated within one region of the liver cell or lobule. Staining for RBP was not observed in nuclei or in cells other than parenchymal cells. Similar particulate and diffuse immune staining for RBP was observed in liver sections from both vitamin A-deficient and normal rats. More intense immune staining appeared to be present in the sections of vitamin A-deficient animals, in good correlation with the expected higher levels of RBP in deficient as compared to normal liver. When liver sections were exposed to an antiserum to rat albumin, instead of one to rat RBP, immune cytoplasmic staining was observed which was entirely of a diffuse nature, and did not appear particulate or granular. T h e findings suggest that RBP, unlike albumin, is localized in part within cytoplasmic vesicles or granules which are large enough to be detected with immunofluorescence, and which are present in livers of both normal and vitamin A-deficient animals. The nature of these putative RBP-containing particles remains to be explored. INTRODUCTION

Vitamin A is normally transported in plasma as retinol bound to a specific transport protein, retinol-binding protein (RBP). RBP has been isolated from the serum of several species, including man (Kanai, Raz & Goodman, 1968), the Cynomolgus monkey (Vahlquist & Peterson, 1972), the rat (Muto & Goodman, 1972; Peterson, Rask, Ostberg, Andersson, Kamwendo & Pertoft, 1973), the dog (Muto, Smith & Goodman, 1973), the pig (Rask, 1974), and the chicken (Mokady & Tal, 1974). In all of these species RBP is a small protein of about 20000 Daltons weight, which binds one molecule of retinol per molecule of RBP, and which circulates in plasma as a protein-protein complex of higher apparent molecular weight. In man, the monkey and the rat, RBP has been shown to bind to plasma prealbumin, and to circulate in the form of an RBP-prealbumin complex (Kanai et al. 1968; Vahlquist & Peterson, 1972; Muto & Goodman, 1972; Raz, Shiratori & Goodman, 1970).

380

A. R. Poole, J. T. Dingle, A. K. Mallia and DeW. S. Goodman

In previous studies a monospecific antiserum was raised against purified rat RBP, and a radioimmunoassay was developed to measure RBP levels in rat serum and liver (Muto & Goodman, 1972; Muto, Smith, Milch & Goodman, 1972). When rats were made deficient in vitamin A the serum RBP levels declined dramatically, while the levels of immunoreactive RBP in the liver increased about four-fold (Muto et al. 1972). When graded amounts of vitamin A were administered to deficient rats, there was a rapid and dose-related increase in serum RBP, which was mirrored by a doserelated decrease in the level of RBP in the liver (Smith, Muto, Milch & Goodman, 1973). The rapid rise in serum RBP levels after the injection of vitamin A was not blocked by prior treatment with either puromycin or cycloheximide. These findings indicated that vitamin A deficiency primarily interferes with the secretion, rather than with the synthesis, of RBP by the liver and that the deficient liver contains a pool of previously formed apo-RBP which can be released rapidly into the serum as holoRBP when vitamin A becomes available. Some information is available about the subcellular localization of RBP in rat liver (Smith, Muto & Goodman, 1974). After gentle homogenization and differential centrifugation, approximately two-thirds of the RBP in liver homogenates from both normal and vitamin A-deficient rats were found to be associated with the microsomal fraction, and only about 10% with the soluble supernatant. The addition of 1 % sodium deoxycholate to the liver homogenates quantitatively released the RBP from the particulate fractions into the 105000-g supernatant. This suggests that RBP is present in liver associated with membranous subcellular structures, rather than existing as nascent protein chains still attached to ribosomes. The study reported here was designed to learn more about the localization of RBP within rat liver, using monospecific antisera to rat RBP and immunocytochemical methods. Livers from both normal and vitamin A-deficient rats were examined. These studies aim to extend our understanding of the mechanisms which regulate the storage of RBP in liver and its secretion into the blood. MATERIALS AND METHODS

Rat RBP Rat RBP was purified by affinity chromatography, using human prealbumin coupled to Sepharose-4B. Preliminary experiments showed that rat RBP binds to human prealbuminsubstituted Sepharose at high ionic strength and dissociates at low ionic strength. Human prealbumin (100 mg), purified as described previously (Raz & Goodman, 1969), was coupled to 15 g of CNBr-activated Sepharose 4B (Pharmacia Fine Chemicals Inc., Piscataway, New Jersey) according to the procedure described by the manufacturer (yield of coupled prealbumin about 93 %). RBP was isolated from 1 1. of rat serum purchased from Pel-Freez Bio-chemicals, Inc., Rogers, Arkansas. Rat serum proteins were initially fractionated by chromatography on columns of DEAE-Sephadex (Pharmacia) as described previously (Kanai et al. 1968; Muto & Goodman, 1972; Raz & Goodman, 1969). Fractions displaying retinol fluorescence (Muto & Goodman, 1972) were pooled, dialysed against distilled water and lyophilized. The protein pool containing the RBP obtained after DEAE-Sephadex chromatography was subjected to affinity chromatography on a human prealbumin-substituted Sepharose column (1-8x32 cm) equilibrated with 0-05 M Tris-HCl buffer, pH 7-4, containing 0-5 M NaCl. Eight hundred and fifty milligrammes of protein (part of the pool containing RBP) were dissolved in 5 ml of the starting buffer (005 M Tris-HCl, pH 74, 05 M NaCl) and

Localization of retinol-binding protein in liver

381

applied to the affinity column. The column was then washed with the starting buffer until no more protein appeared in effluent. RBP bound to the gel was eluted with .distilled, deionized water adjusted to pH 10 with NH 3 . Several affinity chromatographic runs were performed in order to isolate RBP from the entire sample obtained after DEAE-Sephadex chromatography. Final purification of RBP, obtained after affinity chromatography, was achieved by preparative polyacrylamide gel electrophoresis as described previously (Muto & Goodman, 1972).

Sera An antiserum to purified rat RBP was raised in a white rabbit and characterized as described previously (Muto & Goodman, 1972; Muto et al. 1972). The rabbit antiserum employed in these studies was the same antiserum which has been used in radioimmunoassay studies of rat RBP (Muto & Goodman, 1972; Muto et al. 1972, 1973; Smith et al. 1973, 1974). A second antiserum to rat RBP was raised in a sheep. The immunogen employed was RBP which had been purified from rat whole serum by affinity chromatography as described above. Purified rat RBP (0-7 mg) was dissolved in 2 ml isotonic NaCl, emulsified with 2 ml complete Freund's adjuvant (CFA) (Difco), and injected in 2 portions intramuscularly into a sheep. Booster injections of rat RBP were prepared in the same manner and given on days 18 and 32. The sheep was exsanguinated on day 50 and its serum was collected and studied. This antiserum was not monospecific, and displayed 2 major and 2 minor precipitin lines on immunodiffusion against normal rat serum. The major precipitin line containing RBP was recognized by a reaction-of-identity with the monospecific rabbit anti-rat-RBP serum. A more specific antiserum to rat RBP was raised in another sheep by injecting precipitin lines containing the RBP and sheep IgG. This more specific antiserum was used in the tissue localization studies carried out. A series of single precipitin lines was prepared by reacting the first sheep antiserum against solutions of purified rat RBP, using parallel linear wells cut in agarose gels on immunodiffusion plates. The plates were washed and the precipitin lines were collected and prepared for injection (including emulsification with CFA) as described by Weston (1969). A sheep was injected intramuscularly with the emulsified precipitin lines on days o, 13, and 28, and was exsanguinated on day 42. The resulting antiserum appeared monospecific on double immunodiffusion analysis when examined with darkground illumination and displayed a single precipitin line which exhibited a reaction of complete identity with the line produced by the specific rabbit antiserum to rat RBP when both sera were allowed to react in an immunodifTusion plate with normal rat serum. More extensive analysis of this sheep antiserum demonstrated, however, that it was not absolutely monospecific. Under some conditions a very faint minor line was just visible after staining the immunodifTusion plate with Coomassie brilliant blue, as shown in Fig. 2. The antibody titre of this almost monospecific sheep antiserum was estimated by comparing it to the rabbit antiserum against rat RBP using the radioimmunoassay procedure, as previously described (Muto et al. 1972). Identical amounts of l l 5 I-RBP were bound (60% bound) by the rabbit antiserum at a dilution of 1:3000 and by the sheep antiserum at a dilution of 1:2400. Samples of non-immune rabbit and sheep sera were used for control observations. Three separate preparations of control rabbit sera obtained from New Zealand white rabbits were studied. Two were isolated from rabbits in the colony at Strangeways Research Laboratory; the third was obtained from Wellcome Research Laboratories, Beckenham, Kent, England. Samples of non-immune sheep sera from 12 sheep were pooled and used as the control sheep serum. A monospecific rabbit antiserum to rat albumin, prepared by Cappel Laboratories (Lot 5978, Cappel Laboratories, Downington, Pa.) was purchased from Biocult Laboratories, Paisley, Scotland. A pig antiserum to rabbit Fab' was prepared by intramuscular injection into the hind legs on day o of 2 5 mg of Fab' emulsified in 1 ml isotonic NaCl with 1 ml CFA. On day 14 the pig received a similar set of injections into the fore-legs. After injections again into the hind legs on day 28, the animal was bled-out on day 38 by intracarotid catheterization. A pig antiserum to sheep Fab' was similarly raised by 3 intramuscular injections at 2-week intervals of a total of 5 mg of sheep Fab' emulsified with CFA. The animal was also bled-out 10 days after the final injection. The precipitating activities of the pig antisera to rabbit or sheep Fab' were •estimated by radial immunodiffusion as shown in Fig. 1.

3 82

A. R. Poole,J. T. Dingle, A. K. Mallia and DeW. S. Goodman

Immunoglobulin G (IgG) and Fab' preparations Partially purified IgG was prepared from sera by precipitating twice at room temperature that fraction obtained between o and50 % saturation with (NH^SOj. Pepsin digests containing (Fab')j were prepared from partially purified IgG by digestion at pH 4-5 in 0-2 M sodium acetate buffer of 25 mg/ml IgG with 2 % (w/w) pepsin (Worthington, twice crystallized) at 40 °C for up to 48 h in the presence of 01 % sodium azide. Pepsin was inactivated by adjusting the pH to 8-6 and digests were then dialysed against phosphate-buffered saline (PBS, 145 mM NaCl, 9 mM Na2HPO4,1 mM NaH.POJ for 24 h. The dialysis residue was retained. This method is based on that described by Nisonoff (1964). 140

r

120

:er squarec

E E

100 80 60

V

E

40

5 20 25

50

75

100

Antiserum concentration, %

Fig. 1. Radial immunodiffusion estimations of the precipitating activities of the pig antiserum to rabbit Fab' (O — O), and the pig antiserum to sheep Fab' ( • — • ) . The gel containing the former contained 5-1 /tg/ml of purified rabbit IgG, and the gel containing the latter 15-6 /tg/ml of purified sheep IgG. The square of the mean diameter (in mm) of stained duplicate precipitin rings was plotted against the concentration of antiserum. Immunoglobulin G (IgG) was purified from partially purified IgG by a method based on that described by Aalund, Osebuld & Murphy (1965). The partially purified IgG was chromatographed on a column of cellulose ion-exchanger DEAE-52 (Whatman) which had been equilibrated with 50 mM Tris-HCl buffer pH 7-8. IgG was not absorbed to the column and was eluted. Its identity and purity were confirmed by using commercial antisera to whole serum proteins and to IgG of the species under investigation (prepared by Dakopatts A/S and obtained from Mercia, Watford, Herts., England). Purified IgG was digested with pepsin as described above. After dialysis, the rabbit or sheep Fab' was purified for use as an immunogen by gel filtration on Sephadex G-100 equilibrated with 1 mM dithiothreitol in PBS. Fab' was identified in purified preparations and in digests of partially purified IgG by the fact that although it will react with antigen it will not precipitate it. Hence, the precipitation of antigen with native divalent IgG antibody can be blocked by monovalent Fab'. Digests of purified or partially purified IgG exhibited no detectable precipitating activity. The protein contents of (FabOi preparations were estimated spectrophotometrically assuming E i s at 280 nm = 15 (Little & Donahue, 1968). (FabOj preparations were labelled with fluorescein isothiocyanate (FITC) as described by Th6 & Feltkamp (1970). Protein and FITC contents of conjugates were determined spectrophotometrically by recording E,7e and EJM and making reference to the nomograph presented by Goldman (1968). Immediately before use divalent (FabOa was reduced to monovalent Fab' with 5 mM cysteine in PBS for 30 min.

Localization of retinol-binding protein in liver

383

Immunodiffusion methods Both double immunodiffusion gels (DID) and radial immunodiffusion gels (RID) were prepared with 11 ml of 1 % agarose in 20 mM phosphate buffer, pH 7 3 containing 150 mM NaCl. Gels were poured on 325 x 3-25 in. (8-25 x 8-25 cm) Kodak slide coverglasses initially pre-coated with a layer of 0-2 % agarose in water which was then air dried. In the case of RID plates the antigen was added and mixed at a temperature of 55 C C. Five- and 2O-/d capacity wells were cut for RID and D I D plates respectively. Where necessary, additional volumes of reagents were added to D I D plates at the beginning of the diffusion period. After immunoreaction for 24 h (DID) or 48 h (RID), gels were washed for 72 h in phosphate-buffered saline (PBS) and then for 24 h in distilled water before dehydrating in ethanol and drying in air. Gels were stained for 24 h with 0 1 mg/ml Coomassie brilliant blue in a solvent mixture of 3 5 ml 98 % formic acid, 2 g sodium formate, 63 ml water and 33 ml ethanol. Excess stain was removed with the solvent before rinsing the plate in water and drying. Diameters of stained precipitin rings in RID gels were measured with a x 8 eyepiece micrometer.

Animals and diets Hooded rats were used from the Strangeways Research Laboratory animal colony. The normal rats were fed a stock diet described in detail by Moore & Holmes (1971), consisting of Rat Cake Diet No. 86 (North Eastern Farmers Ltd, Bannermill, Aberdeen, Scotland), supplemented with retinol at a level of about 2-4 /*g/g diet, and containing also considerable amounts of carotene. The normal rats were 2-5-3 months old and weighed 250-300 g each when used. Vitamin A-deficient rats were prepared at the Strangeways Research Laboratory by feeding the rats since weanling on a vitamin A-deficient diet supplemented with retinoic acid, as described by Moore & Holmes (1971). This diet consisted of white bread flour (80 %), vitaminfree casein (5 % ) , dried yeast (10 %), and arachis oil (5 %). Each 100 ml of arachis oil contained 1 mg vitamin D3, 60 mg a-tocopheryl acetate, 3 mg butylated hydroxytoluene, and 5 mg retinoic acid. The diet provided about 2 5 fig of retinoic acid per g diet. Rats on this diet were placed on a completely vitamin A-deficient diet (same diet without retinoic acid) for 10 days before being used for experiments. At the time of studies the rats were 3 months old and weighed 200-250 g. Portions of the livers of the normal and deficient rats were analysed for vitamin A by the trifluoroacetic acid method (Roels & Mahadevan, 1967), using non-saponifiable extracts of liver homogenates for assay. These assays showed that the normal rats had 60-90 fig vitamin A per g liver (wet weight), whereas the deficient rats had, as expected, non-detectable levels of vitamin A (less than 0-5 fig vitamin A per g liver).

Liver perfusion and tissue preparation Rats were anaesthetized with diethyl ether. A median ventral abdominal incision was made and the intestine was displaced to one side, exposing the inferior vena cava, from which blood samples were removed by venipuncture. The liver was then immediately perfused (at 30—37 °C) with a solution of 0-9 % NaCl containing 1 % gelatin which was injected slowly into the hepatic portal vein. The superior vena cava was sectioned to facilitate the washout perfusion. After several minutes, and when the liver lobes had visibly blanched, small pieces of liver 2-4 mm across were removed and placed in a solution of 7 % gelatin in 0-9% NaCl at about 35 °C which was inside a flat-bottomed plastic tube; they were immediately frozen as such for 90 s in liquid nitrogen. Tubes were sealed and samples were stored at —20 °C prior to sectioning.

Tissue sectioning Frozen sections 4-6 /»m thick were cut in a cryostat at a cabinet temperature of about — 25 °C, with a knife cooled with solid carbon dioxide. Good sections were obtained when 1 % gelatin was included in the liver perfusion solution, whereas poor frozen sections were obtained when

384

A. R. Pools, J. T. Dingle, A. K. MalUa and DeW. S. Goodman

livers were perfused with isotonic NaCl alone. Sections were immediately allowed to attach to microscope slides at room temperature and approximately 10 s later they were fixed at room temperature for periods ranging from 15 to 30 min, depending upon the experiment. The fixative was 4 % formaldehyde, freshly prepared from paraformaldehyde, in PBS (Graham 6 Karnovsky, 1966). Sections were then washed for 60 min in PBS containing 5 mM cysteine prior to treatment with Fab' preparations. The studies with rabbit Fab' were carried out with liver samples obtained from a total of 7 normal and 9 vitamin A-deficient rats. Approximately 50 sections each of normal and of vitamin A-deficient liver were examined in the several experiments conducted. For the study with sheep Fab', approximately 10 sections each of normal and of vitamin A-deficient liver were examined. Treatment of fixed sections with Fab' The indirect method of localizing tissue antigens was used in all these studies. Thus tissues were initially treated for 1 h in the presence of 5 mM cysteine with Fab' prepared from either an immune serum or a non-immune serum of the same species. They were then washed for 1 h in PBS containing 5 mM cysteine before treatment for 1 h with a preparation of Fab' directed against that used in the first step. Thus the following systems were employed: Method A Step 1. Rabbit anti-(rat RBP) Fab' or rabbit anti-(rat albumin) Fab' or non-immune rabbit Fab'. Step 2. Pig Fab' anti-(rabbit FabO labelled with FITC. Method B Step 1. Sheep anti-(rat RBP) Fab' or non-immune sheep Fab'. Step 2. Pig Fab' anti-(sheep Fab') labelled with FITC. After the final step, sections were washed again for 1 h before being rapidly rinsed in distilled water and mounted in a mixture of 1 volume 0-2 M Tris-HCl buffer, pH 8-6, and 9 volumes glycerol. Specimens were immediately examined by fluorescence microscopy as described previously (Dingle, Poole, Lazarus & Barrett, 1973) using a Reichart Zetopan transmitted darkground system and Zeiss planapochromat objectives (X40, X63, x 100) fitted with iris diaphragms. The following were used at a final concentration of 10 mg/ml: rabbit anti-(rat RBP) Fab'; rabbit anti-(rat albumin) Fab'; non-immune rabbit serum Fab'i, Fab'», and Fab'3 (from the 3 non-immune rabbit sera studied). Pig Fab' anti-(rabbit Fab') labelled with FITC was used at 3-2 mg protein/ml (molar ratio of FITC/(Fab / ), = 1-33/1-0). Sheep anti-(rat RBP) Fab' and non-immune sheep Fab' were used at a final concentration of 37-5 mg/ml. Pig Fab' anti-(sheep Fab') labelled with FITC was used at a final protein concentration of 5-1 mg/ml (molar ratio of FITC^FabOi = I-8I/I-O).

RESULTS

Sections of normal rat liver treated with pig Fab' anti-(rabbit Fab') labelled with FITC or pig Fab' anti-(sheep Fab') labelled with FITC did not stain at all and only a very weak dull brownish auto fluorescence was observed. Studies toith rabbit Fab' preparations

When sections of liver were treated in the first step with rabbit anti-(rat RBP) Fab', moderately intense particulate and weaker diffuse cytoplasmic staining was

Localization of retinol-binding protein in liver

385

observed in all the parenchymal liver cells (Fig. 3 A, B). The particulate staining appeared as bright spots scattered throughout the cytoplasm and was not consistently concentrated in any one region within the hepatic cell. Nuclei were unstained. The amount of particulate staining was variable from one zone to another, but there was no detectable relationship to lobular architecture. In some regions cells appeared to be densely packed with fluorescent 'granules', whereas in other regions much less particulate staining was observed. Staining for RBP was not detected in cells other than parenchymal cells. In tissue sections initially exposed to rabbit anti-(rat albumin) Fab', moderately intense diffuse cytoplasmic staining was observed. In contrast to the findings with anti-(rat RBP) Fab', this staining was entirely of a diffuse nature, and did not appear particulate or granular (Fig. 5 A, B). In some sections from some animals more intense, diffuse, bright staining was observed at or close to the cell peripheries (e.g. Fig. 5 A as compared with Fig. 5B). It was thought that such bright peripheral staining probably represented albumin associated with the cell surface, perhaps in part reflecting a less complete washout perfusion of the liver before freezing. Staining was apparently restricted to all hepatic parenchymal cells, and cells from different lobular zones exhibited similar staining characteristics. Nuclei were again unstained. Control sections of liver initially treated with Fab'x or Fab'3 preparations prepared from non-immune rabbit sera exhibited very weak diffuse cytoplasmic fluorescein fluorescence which faded rapidly on inspection (Fig. 4 A, B). A little weak non-specific staining was sometimes seen on the surface of the section. One of the control sera (the Fab'2 preparation, made from a non-immune serum sample from a rabbit in the Strangeways animal colony) did, however, produce some particulate and diffuse cytoplasmic staining when liver sections were treated with it at the first step. This staining differed from that seen with the rabbit anti-(rat RBP) Fab', in that the particulate staining was of a different quality and of lesser quantity and intensity with the non-immune Fab'2 preparation, and was more variable from experiment to experiment. Sections of liver treated with non-immune Fab'2 showed far fewer bright particulate spots, of lesser intensity (brightness) and larger apparent size, than did the sections treated with the anti-(rat RBP) Fab'. Nevertheless, the finding of some particulate staining with a control, non-immune rabbit serum Fab', raised the question of the specificity of the findings seen with the rabbit anti-(rat RBP) preparation, and stimulated us to conduct further studies with an antiserum against rat RBP prepared in another species (the sheep) (see below). Studies were conducted with sections of liver from vitamin A-deficient rats, as well as with livers from normal rats. (See Figs. 3-5 for illustrative comparisons.) There were no gross differences in the kind of staining observed in normal and deficient liver for albumin, and with non-immune control preparations Fab'j and Fab'3. The results obtained with the non-immune Fab'2 preparation were somewhat different in that particulate staining was relatively inapparent with deficient compared with normal livers. The localization of RBP in sections of vitamin A-deficient liver was generally similar to that seen with normal liver (see Fig. 3). Some quantitative differences in RBP localization were, however, suggested by direct comparison of

386

A. R. Poole, J. T. Dingle, A. K. Mallia and DeW. S. Goodman

sections of deficient with those of normal liver in 2 experiments. In these experiments the overall intensity of staining and particularly the diffuse cytoplasmic staining, appeared somewhat greater in sections of liver from deficient than from normal rats. There also appeared to be a greater density of particulate staining, and the bright spots (the stained 'granules') appeared to be smaller in deficient liver. An attempt was made to compare quantitatively the sizes of the stained granules in deficient as compared to normal rats, by measuring the diameters of 120 granules of each group from photographs of sections from 4 normal and from 3 deficient rats. The mean diameter of the stained granules was larger in normal (0-64 /tm) than in deficient (0-49 /im) liver. These comparisons, between the localization of RBP in deficient and normal rat liver, should, however, be considered as only tentative, and will require more extended and definitive observations in order to be confirmed. Studies with sheep Fab' preparations

Sections of liver first treated with non-immune (control) sheep Fab' exhibited no cytoplasmic staining, but there was some weak diffuse nuclear staining (Fig. 6B). The use of sheep anti-(rat RBP) Fab' in the first step resulted in cytoplasmic staining of moderate to low intensity which was mainly made up of fine particulate staining (Fig. 6A). Since this staining was relatively weak it was more difficult to observe and to photograph than that obtained with the rabbit antiserum to rat RBP. Weak nuclear staining was sometimes also observed in liver sections initially treated with the immune sheep serum. The immune (anti-rat RBP) particulate cytoplasmic staining was clearly observed in the liver sections of one normal rat out of a total of three animals examined. In a study of the livers of 3 deficient rats, however, particulate immune staining was observed in all sections (Fig. 6 A).

DISCUSSION

This study was undertaken to examine the localization of immunoreactive RBP in rat liver. Previous biochemical studies have demonstrated that rat liver homogenates contain significant amounts of immunoreactive RBP, largely present in association with the microsomal fraction after differential homogenization (Muto et al. 1972; Smith et al. 1973, 1974). The studies reported here employed antisera to rat RBP prepared in a rabbit and in a sheep. The indirect, 2-stage method of localizing tissue antigens was employed in order to increase the sensitivity of the immunocytochemical study, and Fab' fragments of immunoglobulins were used to minimize non-specific labelling of sections of liver and to ensure penetration of fixed tissue sections (Poole, 1973). Using these techniques the specific immune staining of RBP was observed within liver parenchymal cells. This staining appeared to be both particulate and diffuse within the cytoplasm of the parenchymal cells, and was not concentrated within one region of the liver cell or lobule. Staining for RBP was not observed in nuclei or in cells other than parenchymal cells. The findings reported here suggest that RBP, unlike albumin, is localized at least

Localization of retinol-binding protein in liver

387

in part within cytoplasmic vesicles or granules which are large enough to be detected with immunofluorescence, and which are present in livers of both normal and vitamin A-deficient rats. These RBP-containing cytoplasmic particles were vividly observed in all experiments conducted with the monospecific rabbit antiserum to rat RBP. Particulate fluorescent staining was not observed in similar experiments with an antiserum to rat albumin, or with 2 non-immune control rabbit sera. A third sample of serum collected from an unimmunized rabbit did, however, show some particulate staining in rat liver sections treated with it at the first step. Although this staining was of less intensity and somewhat different appearance than that seen with the antiserum to rat RBP, its occurrence raised the question of the specificity of the findings seen with the specific rabbit antiserum to rat RBP. Accordingly, a second specific antiserum to rat RBP was prepared in another species (the sheep) and the immunofluorescence experiments were repeated. These experiments again demonstrated particulate and diffuse immune staining (Fig. 6). Therefore, since comparable results were obtained with antisera to RBP prepared in 2 different species, the results strongly suggest that the particulate and diffuse fluorescent staining seen (Figs. 3, 6) represents the specific immune staining of RBP within parenchymal cells of rat liver. This conclusion is supported by the results with livers from vitamin A-deficient rats. Such livers are known to contain higher quantities of immunoreactive RBP than does normal rat liver (Muto et al. 1972; Smith et al. 1973, 1974). With the rabbit antiserum to rat RBP, more intense immune staining appeared to be present in liver sections of vitamin A-deficient animals. With the sheep antiserum to rat RBP, particulate immune staining was observed in all sections of liver from vitamin A-deficient rats, but only in some liver sections from normal rats. These findings correlate well with the expected higher levels of RBP in deficient as compared to normal liver. The localization of albumin has already been examined by immunofluorescence (Hamashima, Harter & Coons, 1964), and diffuse cytoplasmic staining was observed with no evidence for detectable localization in granules. The results reported here confirm these observations. The immunobiochemical studies of Judah, Gamble & Steadman (1973) and Geller, Judah & Nicholls (1972) have suggested that a precursor of albumin which is immunologically identical with albumin constitutes the bulk of the protein precipitated with antisera to albumin, and that this 'proalbumin' is converted to albumin before it is secreted from rat liver. This precursor, if it exists, may account for much of the cytoplasmic staining observed in liver. The intracellular transport of albumin (or proalbumin) has also been studied (Peters, Fleischer & Fleischer, 1971), and it was concluded that this protein(s) was transported from the rough-surfaced endoplasmic reticulum (ER) to the smooth-surfaced ER and then into the Golgi apparatus. Information is not available about the manner in which albumin (or proalbumin) is transported to the plasma membrane. The putative RBP-containing particles in rat liver remain to be further characterized. It would be of interest to explore whether the RBP-containing vesicles or granules contain other plasma proteins (particularly prealbumin) as well, and to try to isolate and examine them. The possibility also exists that immunoreactive RBP in liver is present in the form of a precursor protein of some sort, which is converted to RBP

388

A. R. Poole, J. T. Dingle, A. K. Mallia and DeW. S. Goodman

just prior to or during secretion from the liver cell. The question of the origin of the RBP-containing particles is also of interest. It is known, for example, that peroxisomes (which are particles enriched in catalase and some other enzymes) probably arise as dilations of the ER (De Duve, 1973; Novikoff & Novikoff, 1973). Evidence has also been reported indicating the formation of primary lysosomes from smooth ER (Holtzman, Novikoff & Villaverde, 1967; Topping & Travis, 1974). It is tempting to speculate that RBP-enriched particles may also arise as dilations or vesicle pockets of the ER. Other, alternative possibilities of course exist, and must be considered. Many exportable proteins in secretory cells have been found to have a granular localization, from which they are released to an extracellular site when the granule membrane fuses with the plasma membrane (Smith & Farquhar, 1966; Palade, 1959). It is possible that RBP is secreted in this manner from the parenchymal liver cell into the blood. Mature secretory granules in mammatrophic hormone-producing cells in the rat anterior pituitary gland are formed by aggregation of several smaller Golgiderived packets (Smith & Farquhar, 1966). A generally similar aggregation process seems to occur during the formation of azurophil granules in rabbit polymorphonuclear leukocytes (Bainton & Farquhar, 1966). A sequence of events somewhat resembling these may occur in the maturation of an RBP-rich 'granule', in view of our tentative finding that the particles in liver of vitamin A-deficient rats were smaller than those in normal liver. In the deficient-liver secretion of RBP is arrested and the protein accumulates in the liver. This phenomenon might involve the accumulation of immature granules, which would be anticipated to be of smaller size if they must fuse to form a more mature vesicle before secretion from the cell. The secretion of RBP from deficient rat liver occurs rapidly when vitamin A becomes available (Smith et al. 1973). The nature of the 'signal' which stimulates the rapid release of RBP into the circulation remains to be defined. One possibility is that retinol stimulates necessary conformational or chemical changes in an RBP precursor protein in liver, which leads to release of RBP from its membrane-bound location and secretion from the cell. Another possibility is that retinol is more directly involved with the secretory process itself. If RBP is present in the cell in association with a vesicle or granule, its secretion might involve the fusion of the granule membrane with the plasma membrane, in analogy with the secretion of other proteins contained within secretory granules (Smith & Farquhar, 1966; Palade, 1959). It has been demonstrated that retinol has the capacity to induce membrane fusion (Ahkong, Fisher, Tampion & Lucy, 1973); moreover, an excess of retinol stimulates the secretion of lysosomal enzymes which is thought to be due to a fusion of lysosomes with the plasma membrane (Dingle, 1969; Lucy, 1969). Thus, vitamin A may influence the secretion of RBP by interacting with the plasma membrane and/or the secretory vesicle membrane, causing them to fuse and so discharge the RBP into the plasma. Furthermore, since liver selectively secretes the retinol-RBP complex, it is possible that retinol binds to granule-bound RBP, resulting in a change in the delimiting membrane of this granule specifically favouring its fusion with the plasma membrane. Further investigation will be required to explore these and other various possibilities.

Localization of retinol-binding protein in liver

389

We thank Miss C. Camus and Mr D. Buttle for their expert technical assistance. D r Poole is a Nuffield Research Fellow, and Dr Dingle is a member of the external staff of the Medical Research Council of Great Britain. This work was supported in part by Grants AM-05968 and HL-14236 (SCR) from the National Institutes of Health, Bethesda, Md., and was performed in part while Dr Goodman was on leave from Columbia University as a Fellow of the John Simon Guggenheim Foundation. Dr Poole's present address is: Division of Immunology, Department of Pathology, University of Cambridge, England. Correspondence and reprint requests should be addressed to Dr Goodman in New York.

REFERENCES AALUND, O., OSEBULD, J. W. & MURPHY, F. A. (1965). Isolation and characterization of ovine

gamma globulins. Archs Biochcm. Biophys. 109, 142-149. AHKONG, Q. F., FISHER, D., TAMPION, W. & LUCY, J. A. (1973). The fusion of erythrocytes

by fatty acids, esters, retinol and a-tocopherol. Biochem.J. 136, 147-155. BAINTON, D. F. & FARQUHAR, M. G. (1966). Origin of granules in polymorphonuclear leukocytes. Two types derived from opposite faces of the Golgi complex in developing granulocytes. J. Cell Biol. 28, 277-301. DEDUVE, C. (1973). Biochemical studies on the occurrence, biogenesis, and life history of mammalian peroxisomes. J. Histochem. Cytochem. 21, 941-948. DINGLE, J. T . (1969). The extracellular secretion of lysosomal enzymes. In Lysosomes in Biology and Patlwlogy, vol. 2 (ed. J. T . Dingle & H. B. Fell), pp. 421-436. Amsterdam: North Holland Publishing. DINCLE, J. T., POOLE, A. R., LAZARUS, G. L. & BARRETT, A. J. (1973). Immunoinhibition of

intracellular protein digestion in macrophages. J. exp. Med. 137, 1124-1141. GELLER, D. M., JUDAH, J. D. & NICHOLLS, M. R. (1972). Intracellular distribution of serum

albumin and its possible precursors in rat liver. Biochein. J. 127, 865-874. GOLDMAN, M. (1968). Fluorescent Antibody Methods. New York: Academic Press. GRAHAM, R. C , JR. & KARNOVSKY, M. J. (1966). The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mouse kidney. Ultrastructural cytochemistry by a new technique. .7. Histochem. Cytochem. 14, 291-302. HAMASHIMA, Y., HARTER, J. G. & COONS, A. H. (1964). The localization of albumin and

fibrinogen in human liver cells. J. Cell Biol. 20, 271-279. HOLTZMAN, E., NOVIKOFF, A. B. & VILLAVERDE, H. (1967). Lysosomes and GERL in normal

and chromatolytic neurons of the rat ganglion nodosum. J. Cell Biol. 33, 419—435. JUDAH, J. D., GAMBLE, M. & STEADMAN, J. H. (1973). Biosynthesis of serum albumin in rat

liver. Evidence for the existence of 'proalbumin'. Biochem.J. 134, 1083-1091. KANAI, M., RAZ, A. & GOODMAN, DeW. S. (1968). Retinol-binding protein: the transport protein for vitamin A in human plasma. J. din. Invest. 47, 2025-2044. LITTLE, J. R. & DONAHUE, H. (1968). Spectral properties of proteins and small molecules of immunological interest. In Methods in Immunology and Immuno-Chemistry, vol. 2 (ed. C. A. Williams & M. W. Chase), p. 343. New York: Academic Press. LUCY, J. A. (1969). Lysosomal membranes. In Lysosomes in Biology and Pathology, vol. 2 (ed. J. T. Dingle & H. B. Fell), pp. 313-341. Amsterdam: North Holland Publishing. MOKADY, S. & TAL, R. (1974). Isolation and partial characterization of retinol-binding protein from chicken plasma. Biochim. biophys. Acta 336, 361-366. MOORE, T . & HOLMES, P. D. (1971). The production of experimental vitamin A deficiency in rats and mice. Lab. Animals 5, 239-250. MUTO, Y. & GOODMAN, D E W . S. (1972). Vitamin A transport in rat plasma: isolation and characterization of retinol-binding protein. J . 610/. Chem. 247, 2533-2541. MUTO, Y., SMITH, F. R. & GOODMAN, D E W . S. (1973). Comparative studies of retinol transport in plasma. J. Lipid Res. 14, 525~S32MUTO, Y., SMITH, J. E., MILCH, P. O. & GOODMAN, D E W . S. (1972). Regulation of retinol-

binding protein metabolism by vitamin A status in the rat. J. biol. Chem. 247, 2542255°-

390

A. R. Pooh, J. T. Dingle, A. K. Mallia and DeW. S. Goodman

A. (1964). Enzymatic digestion of rabbit gamma globulin and antibody and chromatography of digestion products. In Methods in Medical Research, vol. 10 (ed. H. N. Eisen), pp. 134-141. Chicago: Year Book Medical Publishers. NOVIKOFF, A. & NOVIKOFF, P. M. (1973). Microperoxisomes. J. Histochem. Cytochem. ai, 963966. PALADE, G. E. (1959). Functional changes in the structure of cell components. In Subcellular Particles (ed. T. Hayashi), pp. 64-83. New York: Ronald Press. PETERS, T., JR., FLEISCHER, B. & FLEISCHER, S. (1971). The biosynthesis of rat serum albumin. IV. Apparent passage of albumin through the Golgi apparatus during secretion. J. biol. Chem. 246, 240-244. PETERSON, P. A., RASK, L., OSTBERC, L., ANDERSSON, L., KAMWENDO, F. & PERTOFT, H. (1973). Studies on the transport and cellular distribution of vitamin A in normal and vitamin Adeficient rats with special reference to the vitamin A-binding plasma protein. J. biol. Client. 248, 4009-4022. POOLE, A. R. (1973). A study of the intracellular and extracellular localizations of the lysosomal protease cathepsin D using immunohistochemical methods. In Electron Microscopy and Cytochemistry (ed. E. Wisse, W. Th. Daems, R. Molenaar & P. van Duijn), pp. 159-162. Amsterdam: North Holland Publishing. RASK, L. (1974). The vitamin A transporting system in porcine plasma. Eur.J. Biochem. 44,

NISONOFF,

i-5-

A. & GOODMAN, DEW. S. (1969). The interaction of thyroxine with human plasma prealbumin and with the prealbumin-retinol-binding protein complex. J. biol. Client. i.\\, 3230-3237. RAZ, A., SHIRATORI, T. & GOODMAN, DEW. S. (1970). Studies on the protein-protein and protein-ligand interactions involved in retinol transport in plasma. J. biol. Client. 245,

RAZ,

1903-1912.

ROELS, O. & MAHADEVAN, S. (1967). Vitamin A. In The Vitamins, vol. 6, 2nd edition (ed. P. GySrgy & W. N. Pearson), pp. 156-157. New York: Academic Press. SMITH, J. E., MUTO, Y., MILCH, P. O. & GOODMAN, DEW. S. (1973). The effects of chylomicron vitamin A on the metabolism of retinol-binding protein in the rat. J. biol. Chem. 248, 15441549SMITH, J. E., MUTO, Y. & GOODMAN, DEW. S. (1974). Tissue distribution of retinol-binding protein (RBP) in the rat. Fedn Proc. Fedn Am. Socs exp. Biol. 33, 688 (abstr.). SMITH, R. E. & FARQUHAR, M. G. (1966). Lysosome function in the regulation of the secretory process in cells of the anterior pituitary gland. J. Cell Biol. 31, 319-347. Twt, T. H. & FELTKAMP, T. E. W. (1970). Conjugation of fluorescein isothiocyanate to antibodies. II. A reproducible method. Immunology 18, 875-881. TOPPING, T. M. & TRAVIS, D. F. (1974). An electron cytochemical study of mechanisms of lysosomal activity in the rat left ventricular mural myocardium. J. Ultrastruct. Res. 46, 1-22. VAHLQUIST, A. & PETERSON, P. A. (1972). Comparative studies on the vitamin A transporting protein complex in human and Cynomolgus plasma. Biochemistry, N.Y. 11, 4526-4532. WESTON, P. D. (1969). A specific antiserum to lysosomal cathepsin D. Imnmnology 17, 421-428-

{Received 20 March 1975)

Localization of retinol-binding protein in liver

391

Fig. 2. Double immunodifFusion plate showing the precipitin reactions of the antisera to rat RBP when tested against normal rat serum. The central well (unlabelled) contained normal rat serum. The wells marked R contained rabbit anti-(rat RBP) serum and those marked S contained sheep anti-(rat RBP) serum. Antiserum wells contained 20-//.I volumes, and a total of 60 /i\ (20 ml x 3 applications) of normal rat serum was used in the central well. The plate was left for 24h at room temperature before washing and staining with Coomassie brilliant blue (see Methods). Magnification x 2 5 . The figure shows a reaction of identity between the rabbit and sheep antisera. The arrow points to a very faint contaminant precipitin line, seen occasionally with the sheep antiserum, in addition to the major single line against rat RBP.

C E I.

19

392

A. R. Poole, J. T. Dingle, A. K. Mallia and DeW. S. Goodman

Figs. 3-5. Fluorescence micrographs to show the localization of immunoreactive RBP and albumin in 4-/tm-thick frozen sections of rat liverfixedfor 30 min and treated with Fab' preparations as indicated. Sections labelled A (Figs. 3 A, 4A, 5 A) were from normal rats and sections labelled B (Figs. 3B, 4B, 5B) were from vitamin A-deficient rats. Fig. 3 A, B. Liver sections treated with rabbit anti-(rat RBP) Fab' and then with pig anti-(rabbit Fab^ Fab' labelled with FITC. Intense green particulate and diffuse cytoplasmic staining was seen in parenchymal cells, A, X68O; B, x 708. Fig. 4A, B. Liver sections treated with rabbit Fab' t isolated from non-immune serum and then with pig anti-(rabbit Fab^ Fab' labelled with FITC. Sections exhibited very little staining which was diffuse and cytoplasmic. Both x 708. Fig. 5 A, B. Liver sections treated with rabbit anti-(rat albumin) Fab' and then with pig anti-(rabbit FabO Fab' labelled with FITC. Diffuse staining of moderate intensity was observed in parenchymal cells. More intense staining was noted at or close to cell peripheries in some sections (A as compared with B). A, X 704; B, X 785.

Localization of retinol-binding protein in liver

393

394

A. R. Pook,J. T. Dingle, A. K. Mallia and DeW. S. Goodman

Fig. 6A, B. Fluorescence micrographs to show the localization of immunoreactive RBP in 4-/im-thick sections of vitamin A-deficient rat liver fixed for 30 min and then treated with Fab' preparations as follows: Fig. 6 A. Sheep anti-(rat RBP) Fab' and then pig anti-(sheep Fab') Fab' labelled with FITC. Fig. 6B. Sheep Fab' isolated from non-immune serum and then pig anti-(sheep FabO Fab' labelled with FITC. Green particulate and diffuse cytoplasmic fluorescence of moderate intensity was seen in parenchymal cells in Fig. 6 A. Nuclei were often weakly stained. In Fig. 6B the sections exhibited weak nuclear staining and a little weak diffuse cytoplasmic staining. Both x 1000.