Immunolocalization, Quantitation and Cellular Heterogeneity of Apolipoprotein B in Rat Hepatocytes JAMES P. CORSETTI, BARBARA A. WAY, CHARLES E. SPARKS AND JANET D. SPARKS Department of Pathology and Laboratory Medicine University of Rochester School of Medicine and Dentistry, Rochester, New York 14642

Hepatocyte autofluorescence represents a major Intrinsic autofluorescence is a significant problem problem in immunofluorescence studies with fluo- encountered when rat hepatocytes are immunostained rescein conjugates because of significant spectral with fluorescein conjugates. This problem is relatively overlap. We describe a method for immunostaining unique to hepatocytes because in most cultured cells hepatocytes with R-phycoerythrin (a fluorochrome autofluorescence is negligible (1).Additional problems with minimal overlap with autofluorescence) with arise because autofluorescence of unstained hepatocytes paraformaldehyde fixation and Triton X-100 permeabilization for better antibody penetration. This is variable from rat liver to rat liver and is heterogeneous method produced both perinuclear (presumed Golgi in hepatocytes isolated from the same rat. Autofluoresapparatus) and dispersed, reticular staining (pre- cence is due to significant amounts of fluorescent sumed endoplasmic reticulum) in rat hepatocytes in molecules, including flavins, normally present in hepaculture stained with a monoclonal antibody to rat tocytes. To overcome the problem of autofluorescence, apolipoprotein B. Treatment with brefeldin A resulted investigators have screened hepatocytes from several in loss of apolipoprotein B perinuclear staining and liver perfusions and used hepatocytes that demonstrate increased reticular immunofluorescence consistent minimal autofluorescence for immunostaining (2). This with known properties of brefeldin A (inhibition of is not a preferred method because of the animal protein transport within the secretory pathway by resources and the time and effort needed to perform liver dissolution of Golgi bodies). This suggests that apolipoprotein B epitopes are present in both Golgi bodies perfusions. In addition, exactly what condition correand endoplasmic reticulum. To demonstrate the utility sponds with minimal hepatocyte autofluorescence is not of the technique for quantitative studies, static cell known. For routine immunostaining of hepatocytes, the cytofluorometry of brefeldin A-treated cells was per- evaluation of hepatocyte autofluorescence compared formed, demonstrating increases in specific immuno- with specific immunofluorescence becomes critical to fluorescenceof apolipoprotein B correspondingclosely interpretation of staining patterns. to results estimated by monoclonal antibody radioimWe recently described an intracellular pool of apolimunoassays of cellular homogenates. The technique poprotein B (apo B) in rat hepatocytes (3) using a was then used with flow cytometry of single-cell sus- monoclonal antibody (MAb) that identifies an epitope of pensions of control rat hepatocytes derived from im- apo B mapped to the amino-terminal half of the apo B munostained primary cultures to reveal cell-to-cell heterogeneity of apolipoprotein B epitope expression molecule (4). We were interested in determining the manifested as apolipoprotein B-negative and positive intracellular location of the apo B epitopes in hepatopopulations. Results for brefeldin A-treated cells re- cytes under different metabolic conditions by immunovealed even clearer delineation of heterogeneity as staining with this antibody. We wanted to use immunoindicated by frank bimodality of the populations, along fluorescence because of the increased sensitivity assowith not only higher mean apolipoprotein B levels but ciated with the method. Although others have previously also a significantlyhigher proportion of apolipoprotein examined apo B in hepatocytes using fluorescein-labeled B-positive cells than in the control. (HEPATOLOGY polyclonal antibodies and screened hepatocytes to min1992;15:1117-1124.) imize autofluorescence (2), the use of an MAb has the

Received June 10,1991;accepted January 17, 1992. This work was supported by research grant HL 29837 (to Dr. Sparks) from the National Heart, Lung and Blood Institute of the National Institutes of Health and by American Heart Association Grant-In-Aid no. 880793 (to Dr. Corsetti). Address reprint requests to: James P. Corsetti, M.D., Ph.D., Department of Pathology and Laboratory Medicine, Box 608,University of Rochester School of Medicine and Dentistry, 601 Elmwood Ave., Rochester, NY 14642. 31/1/36833

advantage of localizing specific epitopes of apo B. Indirect immunostaining of hepatocyte apo B with MAbs and fluorescein conjugates was recently described (5); however, only a cytoplasmic staining pattern of fluorescence was observed, consistent with the presence of apo B in the endoplasmic reticulum (ER). Little specific fluorescence staining was observed in perinuclear regions consistent with Golgi apparatus apo B (11, even though the Golgi body is well known to contain very low density lipoprotein (6).

1117

1118

CORSETTI ET AL.

To minimize the problem of autofluorescence of hepatocytes we substituted R-phycoerythrin (RPE)conjugated secondary antibody for the fluorescein derivative. This fluorochrome exhibits significantly less spectral overlap with hepatocyte autofluorescence than does fluorescein. To improve morphological preservation we used a gentle fixation procedure using paraf'ormaldehyde and permeabilization with Triton X-100 t o increase penetration of hepatocytes by antibodies. With

this method, the need for preliminary screening of hepatocytes for autofluorescence was unnecessary, and interference by autofluorescent elements on specific immunostaining patterns was minimized. Immunofluorescence staining patterns of normal rat hepatocytes for apo B epitopes with the MAb suggests the presence of apo B in both the ER and Golgi apparatus, consistent with electron microscopic studies reported previously (6). Furthermore, cells were also treated with brefeldin A (BFA), resulting in a change in staining pattern consistent with disruption of Golgi apparatus known t o occur with this agent. The staining method was also amenable t o quantitation by static-cell cytofluorometric study, which allowed the statistical evaluation of specific cellular fluorescence of immunostained hepatocytes with and without BFA. This technique was also useful in investigation of whether the hepatic synthesis of apo B was a heterogeneous process on a cell-by-cell basis, as is known t o be t h e case for many other metabolic functions in the liver (7). The staining method made possible cell-by-cell fluorescence analysis with flow cytometry with single-cell suspensions of rat hepatocytes prepared from the primary cultures after staining for apo B as described above t o address this question. These experiments did indeed demonstrate cell-to-cell heterogeneity in the level of apo B epitope expression both in control and BFAtreated cells.

MATERIALS A N D METHODS Studies were performed with 200- to 280-gm male SpragueDawley rats fed Rodent Lab Chow (Purina Mills, Inc., St. Louis, MO) ad libitum. Waymouth's MB 752/1, Dulbecco's modified Eagle's medium (DMEM) without phenol red, L-glutamine, HBSS, Triton X-100, benzamidine, gentamicin and RPE-conjugated goat antimouse IgG were from Sigma Chemical Co. (St. Louis, MO). FITC-conjugated rabbit antimouse IgG was from Zymed Laboratories (San Francisco, CAI. Penicillidstreptomycin solution was from Gibco Co. Inc. (Grand Island, NY). BSA was from Miles Laboratories (Elkhart, IN). BFA was from Epicentre Laboratories (Madison, WI). Paraformaldehyde was of electron microscopygrade from Polysciences Inc. (Warrington, PA). Hepatocytes were prepared for primary culture in serum-free medium as described previously (3).Animal protocols were approved by the University Committee on Animal Resources of the University of Rochester. The cells were either seeded onto 30-mm plastic dishes containing autoclaved glass coverslips (1 coverslip/dish) coated with rat-tail collagen or seeded onto 60-mm plastic dishes coated with rat-tail collagen. The dishes were incubated for 2 to 4 hr at 37" C in an atmosphere of 95% air/5% CO,. After this the medium and nonadherent cells were discarded, and adherent cells were

HEPATOLOGY

washed three times in HBSS containing 0.2% (wt/vol) BSA. The 60-mm dishes were reincubated overnight in Waymouth's MB 752/1 medium containing 0.2%iwt/vol) BSA (2 mV6O-mm dish) and 0.1 nmol/L insulin. The 30-mm dishes were reincubated overnight (12 to 14 hr) in DMEM without phenol red containing 0.2% (wt/vol) BSA (1 mV30-mm dish) and 0.1 nmoVL insulin. After overnight incubation, cells were washed in HBSS containing 0.2%(wt/vol) BSA and fresh medium was applied with or without added BFA dissolved in DMSO. DMSO concentration was 0.25% (vol/vol) in all cases, and BFA final concentration was 5pg/ml. Cellular protein was determined (8) after the cells were washed three times in HBSS. Cellular protein of 60-mm dishes ranged from 1.0 to 2.0 mg protein/dish. Medium was removed and adherent cells were washed twice with PBS. To immobilize hepatocytes on the substratum and to prepare them for permeabilization, cells were fixed for 15 min in 4% (wt/vol)paraformaldehyde. Cells were then washed once with PBS and incubated with 50 mmol/L ammonium chloride for 20 min at 4" C to stop the fixation process. All subsequent steps were performed at room temperature. For intracellular localization, hepatocytes were permeabilized by treatment with 0.2% (vol/vol)Triton-X 100 in PBS for 5 min. This was followed by a 30-min incubation with 5% (wt/vol) BSA/PBS to block nonspecific binding sites. Ascitic fluid containing the mouse MAb prepared against rat apo B and described in detail elsewhere (4) was diluted 1:250 in 1% (wt/vol)BSA/F'BS and incubated with the cells for 2 hr. Excess MAb was removed by incubation in PBS for 10 min with gentle agitation; this was repeated three times. The cells were then reacted with PE-conjugated goat antimouse IgG diluted 1 :40 in 5% (wt/vol) BSA/F'BS for 30 min, after which excess secondary antibody was removed by incubation in PBS as described above. For flow cytometric study, the stained cultures were trypsinized to provide single-cell suspensions. For microscopic study, coverslips were rinsed in distilled water, inverted and mounted on microscope slides with an aqueous mounting medium. The cells were examined at x 1,000 under oil with a Nikon AFX IIA microscope equipped with a fluorescent light source and a Nikon FX-35WA camera (Nikon, Tokyo, Japan). Cells were photographed with Ektachrome 400 daylight film for color slides (Eastman Kodak Co., Rochester, NY),and the time of exposure was held constant. Immunostained cells were evaluated for cellular fluorescence on a Leitz Wetzlar Aristophot Ortholux I1 static cell cytofluorometer (9) with an excitation wavelength of 488 nm and an emission wavelength of 515 nm for evaluation of FITC fluorescence. An excitation wavelength of 540 nm and an emission wavelength of 580 nm was used to evaluate RPE fluorescence. The excitation source was a 150-Wxenon lamp. Voltage, photomultiplier tube settings, gains and the field diameter assayed were kept constant for evaluation of cells in a given perfusion. Background autofluorescence of cells was the measured fluorescence of fixed and permeabilized cells treated as above but with no primary or secondary antibody. This value was similar to negative controls, which were stained with secondary antibody alone. To determine specific cellular fluorescence, the averaged background fluorescence of negative controls was subtracted from total cellular fluorescence. For quantitative analysis of hepatocytes grown on glass coverslips, hepatocytes in a field of defined area (7.8 x lo3 pm2) were excited and total cellular fluorescence was measured at 515 nm or 580 nm. Fields of this area were found to encompass approximately six nuclei or approximately four or five hepatocyte4field because 20% to 25% of rat hepatocytes

Vol. 15, No. 6, 1992

1119

APO B IMMUNOSTAINING AND HETEROGENEITY IN RAT HEPATOCYTES

are binucleate. A minimum of 25 such fields (but usually 50 to 75 fields) were examined, depending on the density of growth on each cover-slip preparation. Specific cellular fluorescence was averaged for each test condition; Student's t test was used to compare means. The apo B content of detergent-solubilizedcells was assayed with an MAb in a solid-phaseRIA (3,4). The MAb used was the same for both immunostaining and RIA and was equally reactive against rat apo B of higher (apoBH)and lower (apo B,) weights (4).Cells from individual culture dishes (60-mm)were scraped with a rubber policeman into 0.5% (vol/vol) Triton X-100/0.05 mol/L barbital buffer (pH 8.6) containing2 mmoVL benzamidine and 0.5 mmol/L a-toluenesulfonyl fluoride.After brief sonication,debris was removed by centrifugation at 3,500 rpm for 20 min at 4" C, and the supernatant was used for apo B assay. For 35S-labeledmethionine labeling experiments, hepatocytes were cultured under basal conditionsin Waymouth's MB 752/1 medium containing 0.1 nmom insulin for 12 to 14 hr. After incubation, cells were washed three times and incubated in methionine-free RPMI medium containing 0.2% (wt/vol) BSA and 0.25% (vol/vol)DMSO with or without added BFA (7.5 pg/ml) for 15 min. After this, 35 S-labeledmethionine (25 pCi/ml and 5 pM L-methionine, final concentration) was added, and the cells were reincubated. After 3 hr, the cells were washed once in medium without added label and twice more with HBSS. Hepatocytes were then solubilized and media and cell lysates were immunoprecipitated with rabbit polyclonal antirat apo B and antirat albumin as described previously (3). Rabbit antirat serum albumin was obtained from Cappel Laboratories (Organon Teknika Corp., West Chester, PA). Rabbit antirat apo B was prepared in our laboratory. Flow cytometric studies were performed on an Epics C flow cytometer (Coulter Electronics,Hialeah, FL) equipped with an argon-ion laser. Excitation was 500 mW at 488 nm. All fluorescence data collection was done in linear mode. The technique for deconvolution of total histograms to approximate the stained and unstained distributions was by a modification of the method of Dux et al. (10). This method consists of a channel-by-channelstatistical test t o define an initial negative (NEG)region of the total histogram not significantly different than the corresponding control histogram. Next, a best fit in this NEG region in the least-squares sense is obtained with regard t o slight translations of the control histogram along the abscissa. Finally, the NEG region is redefined with the optimum translation in the abscissa. The positively staining cellular POS region lies above the NEG region, and the distribution of the stained cells is the difference between the total and control in the POS region. Our modification consisted of the inclusion of a least-squares optimization of the proportion of the contribution of the control histogram in the determination of the initial NEG region. All deconvolutions were performed on the raw histograms. A five-point moving average smoothing routine was then applied to the solution histograms.

RESULTS Use of FITC-antimouse IgG with our MAb to localize hepatocyte apo B resulted in low signal-to-noise levels, typically on the order of 1.4:1.0 (stained vs. autofluorescence), on static cytofluorometry. The low signal-tonoise ratio and the large field-to-field variability of fluorescence resulted in immunostained and autofluorescence means that were not statistically different. To determine a more suitable fluorochrome for immuno-

.. .... *... . . .... .... . .

0\

I I

",

I

360

420

460

500

\ \

.

. \' : .i

\

540

.. '

500

620

(nm)

FIG. 1. Room-temperature emission spectra of native, unfixed rat hepatocytes (solid line) and FITC-conjugated (dashes) and RPEconjugated antimouse IgG (dots) obtained with a Perkin-Elmer 650-10s spectrofluorometer (Perkin-Elmer Cetus, Norwalk, CT) with excitation wavelength of 360 nm and uncorrected for instrumental response. For spectral analyses, hepatocytes were analyzed a t a density of 400,000 celldm1 PBS; FITC and RPE conjugates were diluted 1: 1,000 in PBS.

staining hepatocytes, the emission spectrum of native hepatocytes was compared to that of FITC and RPEconjugated secondary antibodies on excitation at 360 nm (Fig. 1).As seen in the figure, significant overlap of hepatocyte autofluorescence with the emission spectrum of the FITC conjugate occurred. In contrast, the RPE conjugate overlapped the hepatocyte autofluorescence to a much smaller extent. With excitation and monitoring appropriate for fluorescein, hepatocyte autofluorescence appeared as large, irregular and highly fluorescent vesicular elements, especially in perinuclear areas, against a homogeneous fluorescent background (Fig. 2A). This was a problem in evaluation of specific immunostaining patterns because of the potential to obscure Golgi apparatus and ER fluorescence patterns. With excitation and monitoring appropriate for RPE, autofluorescence was much less intense and less vesicular (more diffuse), indicating the superiority of RPE for this application (Fig. 2B). Indirect immunofluorescence with RPE secondary antibody conjugate was used to determine the pattern of fluorescence of cellular apo B with our MAb. Figure 2C shows the negative primary antibody control (staining with the secondary antibody alone). It is quite similar to the autofluorescence control of Figure 2B indicating minimal nonspecific staining. Specific immunofluorescence staining of control hepatocytes consistently demonstrated both an eccentric perinuclear pattern and a reticular staining pattern (Fig. 2D), indicating the presence of apo B epitopes in both the Golgi apparatus and the ER (1).In contrast to FITC, the signal-to-noise ratio (stained vs. autofluorescence) with RPE averaged 3.2:1, indicating a greater than twofold increase in sensitivity with RPE staining. To demonstrate the usefulness of the method in immunolocalization and quantitation of specific cellular immunofluorescence, we compared control hepatocytes and those treated with BFA, a n agent used to block

1120

CORSETTI ET AL.

HEPATOLOGY

FIG.2. Immunofluorescence microscopy of unfixed hepatocytes grown on glass coverslips. Autofluorescence of cultured rat hepatocytes was monitored (A) at 515 nm with excitation at 488 nm (FITC optics) and (B)at 580 nm with excitation at 540 nm (RPE optics). (C) Negative primary antibody control (staining with secondary antibody alone) with RPE optics. Immunostaining of hepatocytes was carried out as described in “Materials and Methods” for intracellular apo B localization in 3-hr control (D) and BFA-treated cells (E) and (F). For (D), (E) and (F), RPE optics were used. (E) Short exposure time to optimize subcellular detail. (F) Exposure time same as in (D) for emphasis of relative intensity change.

perinuclear staining became progressively more dispersed; by 3 h r it was entirely reticular in nature (Fig. 2E, 2F). This change in staining pattern with BFA was reproduced in hepatocytes derived from three separate rat liver preparations. With static-cell cytofluorometric study, we saw a progressive increase in specific cellular immunofluorescence over time (Fig. 3A); after 3 h r it resulted in an approximate 75% increase in specific cellular fluorescence over the 0 hr time point. The incremental increase in fluorescence with BFA treatment corresponded well to a similar increase in 0 1 2 3 0 1 2 3 4 IMMUNOFLUORESCENCE IMMUNOREACTIVITY cellular apo B immunoreactivity as determined by RIA TIME (hours) (Fig. 3B). These results suggest that apo B epitopes are normally present in Golgi apparatus and ER compartFIG.3. Comparison of the time course of the effect of BFA (5 pgiml) ments. With BFA treatment we see progressive loss of on apo B immunofluorescence evaluated (A) by static cell cytofluorometry and (B) on cellular apo B immunoreactivity as evaluated by perinuclear fluorescence (Golgi) and a n increase in RIA. Hepatocytes were cultured on collagen-coated glass coverslips for cytoplasmic, reticular fluorescence indicating the accuimmunostaining or on collagen-coated 60-mm dishes for RL4. After mulation of apo B epitopes in the ERs of hepatocytes. overnight incubation, hepatocytes were washed and reincubated in To demonstrate normal cell function and the expected medium containingDMS0 ( 0 )as control or in medium containing BFA inhibition of secretion of apo B in this system for (5 kg/ml) (0). Cells were terminated a t 0, 1, 2 or 3 hr. For comparison of the two methods, the results at each time point were calculated as validation of the previous results, 35S-labeled mepercentages of cellular apo B immunofluorescence or of immunoreac- thionine recovery experiments for both apo B and tivity present a t 0 time. Results are expressed as the mean albumin were performed; results are presented in Table percentage t S.E.M. of three separate rat liver experiments. 1. The results for albumin clearly demonstrate the *Significant difference between the time point and 0 time value at a preservation of the secretory function in the absence of probability level of at least p < 0.05 by Student’s paired t teat. BFA and the expected inhibition of secretion in the presence of BFA. The data for albumin also indicate the nearly equal recovery with or without BFA, attesting to protein secretion by disrupting Golgi structures (11-13). the absence of any effect on newly synthesized protein. After overnight incubation, hepatocytes were washed The data for apo B parallel these results closely. and reincubated for 1to 3 hr in medium with or without In performing the static cytofluorometry experiments, added BFA. During treatment of hepatocytes with BFA, it became evident that apparent heterogeneity existed in

Vol. 15, No. 6, 1992

1121

APO B IMMUNOSTAINING AND HETEROGENEITY IN RAT HEPATOCYTES

TABLE1. Cell and medium distribution and recovery of “S-labeled methionine-incorporated rat apo B and albumin in hepatocytes treated with BFA for 3 hr Apo B Radioactivity

Cell Media Recovery

Albumin

Without BFA

With BFA

Without BFA

With BFA

23,460“ 8,247 31,707

33,876 1,337 35,213

44,245 170,903 215,148

281,682 5,807 287,489

Hepatocytes were cultured under basal conditions in Waymouth’s medium containing 0.1 nmol/L insulin for 12 to 14 hr. After incubation, the cells were washed three times and incubated in methionine-free RPMI medium containing 0.2% (wt/vol) BSA and 0.256 (volivol) DMSO with or without added BFA (7.5 Kgjml) for 15 min. After this, 35S-labeledmethionine (25 pCi/ml, final concentration) and L-methionine (5 pmol/L, final concentration) was added, and the cells were reincubated for an additional 3 hr. After labeling, cells were washed once in medium without added label and twice more with HBSS. Hepatocytes were then solubilized and medium and cell lysates were immunoprecipitated with polyclonal rabbit antirat apo Band antirat albumin as described in “Materials and Methods.” Eluted radioactivity was assayed by beta counting. “Data expressed as counts per minute per milligram of cell protein.

cell-to-cell staining intensity. To investigate this further, flow cytometric studies were performed on single-cellsuspensions prepared by trypsinization of the stained hepatocyte cultures for the 0 hr time point and 3-hr BFA-treated cells. The results are given in Figure 4, which shows fluorescence histograms (cell number as a function of fluorescence intensity reported as fluorescence channel number) for the negative primary antibody control (staining as usual except for the apo B antibody) (Fig. 4A),the 0 time point cells (Fig. 4B) and the 3-hr BFA-treated cells (Fig. 4C). It should be noted that the distributions of the autofluorescence and negative primary antibody controls were virtually superimposable, indicating minimal nonspecific staining reflecting closely the immunofluorescence results of Figure 2. Unpaired Student’s t test comparison of the distribution means of the 3-hr BFA cells and 0 time point cells demonstrated statistically significant difference (p < 0.001) in the distributions. After we corrected for autofluorescence, mean fluorescence channel numbers were used to calculate the bulk ratio of staining intensity of the 3-hr BFA cells to the 0 time point cells. This results in a value of 1.5 in good agreement with the results of the static cell cytofluorometric study and RIA of Figure 3. As an additional control for flow cytometry, cells were treated with BFA for 3 hr at 22” C. This resulted in a distribution not significantlydifferent from the 0 time point cells (data not shown). The data of Figure 4B clearly demonstrate the heterogeneity of staining noted in the static cytofluorometry experiments. This is manifested as retention in the initial portion of the distribution of a clearly discernible peak corresponding to the negative control, indicating a population of the 0 time point cells not staining for apo B. The data of Figure 4C for the 3-hr BFA-treated cells are even more dramatic in this regard, with the actual resolution of a second peak corresponding to apo B-positive cells along with the peak of apo B-negative cells. The deconvolution algorithm described in “Materials and Methods” section was used to approximate the distributions of the unstained and stained populations. These are demonstrated in Figure 5A and 5B for the 0 time point cells and the 3-hr BFA-treated cells, respectively, with the dotted lines corresponding to unstained

cells and the solid lines to the stained cells. The mean fluorescence channel numbers result in a ratio of 1.75 for the staining intensity of the stained populations of the 3-hr BFA cells to the 0 time point cells after correction for autofluorescence. The relatively large value of the ratio in view of the modest difference in mean fluorescence channel numbers of the stained populations (117.40 vs. 109.30) is a result of the significantly greater proportion of stained cells in the 3-hr BFA-treated cells. That is, for the 0 time point cells, 59%were apo B negative and 41%were apo B positive. Of the 3-hr BFA-treated cells, 35% were negative and 65%were positive. This point is made even clearer with the data of Figure 5C and Figure 5D, which are transformations of the data of Figure 5A and Figure 5B, respectively. The transformation consisted of a channelby-channel multiplication of the channel number (a measure of fluorescence intensity) multiplied by the number of cells in that channel to give the contribution of that channel to the total fluorescence signal, which is then equal to the area under the curve for any of the distributions. The value of 1.75 for the ratio of the staining intensity of the apo B-positive populations of the 3-hr BFA cells to the 0 time point cells is much more readily apparent in this way. DISCUSSION Localization of apo B to specific cellular compartments is important in our understanding of apo B assembly with lipid and formation of very low density lipoprotein. In rat liver, apo B is synthesized and secreted in two forms (14-16),apo B, and apo B, (15). Recent evidence suggests the presence of a significant pool of apo B in rat hepatocytes, in which apo B, predominates (3). The cellular pool of apo B is altered in hepatocytes derived from streptozotocin-induced diabetic rats (17) and in control hepatocytes incubated with high levels of insulin (3). The size and location of the intracellular pool of apo B in hepatocytes may be important to our understanding of sites of lipoprotein assembly. For most cultured cells autofluorescence is negligible (1);however, this is not the case for hepatocytes making interpretation of fluorescence staining patterns difficult with antibodies employing FITC derivatives whose emission spectrum significantly overlaps that of hepa-

CORSETTI ET AL.

1122

""

100

A

Y .

HEPATOLOGY

I!;+ A

IB

B

Lu 50 0

8 . LT W

m

I

3 50

.

C

z

0

0

100

200

FLUORESCENCE INTENSITY (CHANNEL NUMBER)

FIG. 4. Effect of BFA on apo B immunofluorescence of rat hepatocytes as evaluated by flow cytometric study. After overnight culture on 30-mm collagen-coated dishes, cells were washed and reincubated in medium containing BFA (5 Fglml for 3 hr at 37" C), fixed, permeabilized and immunostained as described. Single-cell suspensions for flow cytometric study were prepared by trypsinization. The figures show flow cytometric fluorescence histograms (number of cells as a function of fluorescence intensity). There were 3,000 cells counted for each histogram. (A) Negative primary antibody control (cells not treated with BFA and stained as usual except for the absence of the apo B antibody step); (B)0 time point cells (cells not treated with BFA but with complete staining); (C) 3-hr BFA-treated cells (complete staining). Results in the figure are from one liver, although experiments were performed on cells from two different livers on separate occasions with virtually identical results.

tocyte autofluorescence. Because of the advantage of the sensitivity of fluorescence staining, investigators have attempted to overcome the problem of autofluorescence by screening hepatocytes before immunostaining. For routine staining without screening hepatocytes, it becomes essential that proper negative controls are evaluated for correct interpretation of immunofluorescence staining patterns. We have found that RPE-conjugated secondary antibody is the preferred fluorochrome over FITC conjugates for use in immunostaining of rat hepatocytes because it minimizes interference from hepatocyte autofluorescence and allows a more than twofold increase in signal-to-noise ratio for quantitative purposes. In addition, with static cell cytofluorometry, total cellular fluorescence and background fluorescence can be statistically evaluated. This allows the quantitation of specific immunofluorescence of hepatocytes incubated with various agents as a time course and for quantitative evaluation of cellular immunofluorescence of hepatocytes under specific metabolic disturbances. For immunolocalization, RPE conjugates are also superior to FITC conjugates because there are fewer autofluorescent vesicular elements, which tend to obscure immunostaining patterns at the excitation wavelength of RPE. Many antibodies to human apo B are now commercially available (polyclonal as well as monoclonal);

255

255

FLUORESCENCE INTENSITY (CHANNEL NUMBER)

FIG. 5. Histograms of stained (apo B-positive) and unstained (apo %negative) populations of 0 time point and 3-hr BFA-treated cells determined by deconvolution as described. (A) 0 time point cells: unstained cells (dotted line) and stained cells (solid line). (B) 3-hr BFA-treated cells: unstained cells (dotted line) and stained cells (solid line). Transformations of the previous histograms were made as follows: channel-by-channel multiplication of channel number (measure of fluorescence intensity) times the number of cells in that channel so that total fluorescence is equal to the area under the curve for any of the distributions. (C) 0 time point cells: unstained cells (dotted line) and stained cells (solid line). (D)3-hr BFA-treated cells: unstained cells (dotted line) and stained cells (solid line).

however, few antibodies to rat apo B are available. The MAb used in this study has been mapped to the amino-terminal half of apo B and is reactive with apo B, and apo B, on a molar basis. Our finding of immunostaining patterns consistent with the presence of apo B in both ER and Golgi compartments differs from the recent report of Davis et al. (5), who also used MAbs to rat apo B for immunolocalization. In their studies, the immunofluorescence pattern obtained was localized throughout the cytoplasm, characteristic of an ER pattern, with little perinuclear Golgi apparatus staining observed. Their method differs substantially from ours. In their procedure, fixation and permeabilization of cells were performed with methanol/acetone. Because morphological preservation is better with formaldehyde than with organic solvents (1) and because organic solvents tend to denature and precipitate proteins in cells, we used paraformaldehyde fixation and Triton X-100 for permeabilization. In the methods used by Davis et al., the secondary antibody was a fluorescein conjugate. Because of the problem of hepatocyte autofluorescence, we employed the RPE-conjugated secondary antibody. To validate that our method could be employed in a quantitative manner and to distinguish ER and Golgi apo B immunostaining patterns, we used BFA, which specifically disrupts the Golgi apparatus (11-13) and causes accumulation of hepatocyte secretory proteins in cells (11).Preliminary 35S-labeledmethionine recovery experiments for apo B and albumin revealed normal secretory function in the absence of BFA, the expected

Vol. 15, No. 6 , 1992

APO B IMMUNOSTAINING AND HETEROGENEITY IN RAT HEPATOCYTES

inhibition of secretion with BFA and little effect on protein synthesis with BFA, attesting to the validity of the system. The incremental increase in total apo B of cellular homogenates with our monoclonal immunoassay corresponded well with the incremental increases in specific cellular fluorescence of hepatocytes treated with BFA. We observed changes in immunostaining patterns of apo B and increased intensity of fluorescence during BFA treatment. Our results are consistent with blockage of apo B transport from the ER t o the Golgi apparatus and accumulation of freshly synthesized apo B in the ER of hepatocytes. The loss of the perinuclear staining pattern could also have been due in part t o secretion of apo B epitopes present in the trans-Golgi network, which is believed to be BFA resistant (18). Flow cytometric studies were performed in this system to investigate heterogeneity of apo B staining on a cell-to-cell basis. The flow cytometric study results were used to calculate the bulk ratio of fluorescent intensities of the 3-hr BFA-treated cells to the 0 time point cells for comparison to static cytofluorometric study and RIA, both bulk assays. The results agreed closely, supporting the validity of this approach. Cellto-cell heterogeneity of apo B staining was clearly demonstrated from the histogram of the 0 time point cells, which showed a clearly discernible peak corresponding to the distribution of the negative primary antibody control along with positive staining, but especially by the histogram of the 3-hr BFA-treated cells demonstrating bimodality with the actual resolution of the positively staining peak along with the negative peak. The possibility that these results were artifacts resulting from cellular viability considerations is remote, as demonstrated by the reproduction of the BFA-induced increase in cellular apo B epitope immunoreactivity seen in static cytofluorometric study and RIA, and that the cells were put in suspension only after fixation and staining of the cells in culture, where viability was known to be excellent. Our finding of hepatocellular heterogeneity in the expression of apo B epitope immunoreactivity is not totally unexpected because heterogeneity of metabolic function on a cellular basis in the liver is a well-known phenomenon, as reviewed recently by Jungermann and Katz (7). Although this has been especially well documented for carbohydrate metabolism, relatively few studies demonstrate cellular heterogeneity in hepatic lipoprotein metabolism (7). Nevertheless, it would be surprising if at least some aspects of hepatic lipoprotein metabolism did not also demonstrate hepatocellular heterogeneity. Indeed, in a recent report (19) it was demonstrated that there is a greater capacity for free fatty acid esterification into VLDL with subsequently greater VLDL secretion by pericentral hepatocytes in rat liver. Deconvolution of the histograms was accomplished by an approximation algorithm to generate the separated distributions for the apo B epitope-negative and apo B epitope-positive populations. This algorithm depends on

1123

the assumption that neither negative nor positive cells correlate with autofluorescence and nonspecific staining. That is, there are no regions in the histogram of the negative primary antibody control more representative of apo B epitope-negative or apo B epitopepositive cells. This condition is apparently closely approached in this case because the negative portions of both the 0 time point and 3-hr BFA-treated histograms approximate the shape of the negative primary antibody control. This implies that the positive cells in the unlabeled state also approximate a distribution of the same shape as the negative primary antibody control, thus fulfillingthe initial assumption. If this were not the case, the negative portion of the 0 time point and 3-hr BFA-treated histograms would not so closely resemble the negative primary antibody control. The deconvolutions demonstrate a modest increase in the mean staining intensity of the 3-hr BFA-treated positive cells so that individual cells are generally stained more intensely. This is a result of the inhibition of secretion of apo B by BFA. The deconvolutions also demonstrate for the 0 time point 59%negative cells and 41%positive cells and for the 3-hr BFA-treated cells 35%negative cells and 65% positive cells, giving an approximate 50%increase in the number of apo B epitope-positive cells in the BFA-treated case. This increase in mean staining intensity and especially the increased number of positive cells results in a value of 1.75 for the ratio of fluorescence intensities of the positive populations of the 3-hr BFA-treated cells to the 0 time point cells. This is close to the values determined in the bulk assays. The 50% increase in the number of positive cells in the BFAtreated case was somewhat surprising because we presumed that with BFA the positive cells would simply become more positive. It is suggestive of the possibility of recruitment of cells in the expression of the apo B epitope from the negative population or inclusion of a population of cells with low secretion rates in the positive population only after amplification of signal with BFA. It may also be speculated that some “negative” cells may in fact be heavy secreters of apo B-containing particles normally maintaining relatively low cytoplasmic levels of apo B, which would accumulate significantly in the setting of Golgi disruption by BFA. Further experiments are needed to more fully clarify these issues. The remaining negative cells are probably “true” negatives in the sense of synthesis and secretion of apo B because they had neither significant apo B at 0 time nor did they accumulate apo B with BFA. A future avenue for exploration is whether these cells may be recruited into an apo B synthetic population under different pathological and metabolic states. In summary, we have demonstrated that RPEconjugated secondary antibody is preferred over FITC conjugates for quantitative immunostaining of rat hepatocytes because of the minimization of interference from autofluorescence allowing an increased signal-tonoise ratio. For qualitative purposes, RPE is superior to FITC because of the presence of fewer and less intense autofluorescent vesicular elements, which tend to ob-

CORSETTI ET AL.

1124

scure intracellular architecture. The technique has been used with flow cytometry to investigate cell-to-cell heterogeneity of expression of newly synthesized apo B with single-cell suspensions of rat hepatocytes derived from immunostained primary cultures. These experiments revealed two populations of hepatocytes in control cells: one apo B-negative and one apo B-positive. Further experiments with BFA-treated cells revealed an even clearer delineation of the cell-to-cell heterogeneity manifested as frank bimodality in the fluorescence histograms. Additionally, the expression of total apo B was higher, and we saw a significantly higher proportion of apo B-positive cells in the BFA-treated case compared with the control.

Acknowledgments: We thank Mary Bolognino and Cecelia DeFranco for expert technical assistance. We also thank Dr. Leon L. Wheeless for help with static cytofluorometry . REFERENCES 1. Willingham MC, Pastan I. An atlas of immunofluorescence in cultured cells. New York Academic Press, 1983:ll-12. 2. Keller GA, Glass C, Louvard D, Steinberg D, Singer SJ. Synchro-

nized synthesis and intracellular transport of serum albumin and apolipoprotein B in cultured rat hepatocytes as studied by double immunofluorescence. J Histochem Cytochem 1986;34:1223-1230. 3. Sparks JD, Sparks CE. Insulin modulates the intrahepatic metabolism of apolipoprotein B. J Biol Chem 1990;265:8854-8862. 4. Sparks JD, Bolognino M, Trax PA, Sparks CE. Production and utility of monoclonal antibodies to rat apolipoprotein B-lipoproteins. Atherosclerosis 1986;61:205-211. 5. Davis RA, Prewett AB, Chan DCF, Thompson J J , Borchardt RA, Gallaher WR. Intrahepatic assembly of very low density lipoproteins: immunologic characterization of apolipoprotein B in hepatic membrane fractions and its intracellular distribution. J Lipid Res 1989;30:1185-1196. 6. Alexander CA, Hamilton RL, Have1 RJ. Subcellular localization of B apoprotein of plasma lipoproteins in rat livers. J Cell Biol 1976;69:241 -243.

HEPATOLOGY

7. Jungermann K, Katz N. Functional specialization of different hepatocyte populations. Physiol Rev 1989;69:708-764. 8. Markwell MK, Haas SM, Bieber LL, Tolbert NE. A modification of the Lowry procedure to simplify protein determination in membrane and lipoprotein samples. Anal Biochem 1978;87: 206-210. 9. Wheeless LL, Bahr GF, Wied GL, Patten SF. Computerized cytofluorometric research system. In: Wied GL, Bahr GF, eds. Automated cell identification and cell sorting. New York: Academic Press, 1970:161-175. 10. Dux R, Kindler-Rohrborn A, Lennartz K, Ragewsky MF. Calibration of fluorescence to quantify antibody binding surface determinants of cell subpopulations by flow cytometry. Cytometry 1991;12:422-428. 11. Misumi Y, Misumi Y, Miki K, Takatsuki A, Tamura G, Ikehara Y. Novel blockade by brefeldin A of intracellular transport of secretory proteins in rat hepatocytes. J Biol Chem 1986;261: 11398-1 1403. 12. Lippincott-Schwartz J , Yuan LC, Bonifacino JS, Klausner RD. Rapid redistribution of Golgi proteins into the ER in cells treated with brefeldin A: evidence for membrane cycling from Golgi to ER. Cell 1989;56:801-813. 13. Lippincott-Schwartz J, Glickman J , Donaldson JG, Robbins J , Kreis TE, Seamon KB, Sheetz MP, et al. Forskolin inhibits and reverses the effects of brefeldin A on Golgi morphology by acAMP-independent mechanism. J Cell Biol1991;112:567-577. 14. Krishnaiah KV, Walker LF, Borensztajn J , Schonfeld G, Getz GS. Apolipoprotein B variant derived from rat intestine. Proc Natl Acad Sci USA 1980;77:3806-3810. 15. Sparks CE, Marsh JB. Metabolic heterogeneity of apolipoprotein B in the rat. J Lipid Res 1981;22:519-527. 16. Elovson J , Huang YO, Baker N, Kannan R. Apolipoprotein B is structurally and metabolically heterogeneous in the rat. Proc Natl Acad Sci USA 1981;78:157-161. 17. Sparks JD, Sparks CE, Bolognino M, Roncone AM, Jackson TK, Amatruda JM.Effect of non-ketotic streptozotocin diabetes on apolipoprotein synthesis and secretion by primary cultures of rat hepatocytes. J Clin Invest 1988;82:37-43. 18. Chenge NW, Pfeffer SR. Compartmentation of the Golgi complex: brefeldin-A distinguishes trans-Golgi cisternae from the trunsGolgi network. J Cell Biol 1990;111:893-899. 19. Guzman M, Castro J. Zonation of fatty acid metabolism in rat liver. Biochem J 1989;264:107-113.