In vitro conversion of proapoprotein A-I to apoprotein A-I. Partial characterization of an extracellular enzyme activity.

Previous studies have established that human hepatocellular carcinoma cells (Hep G2) secrete into serum-free medium the pro form of apolipoprotein A-I (proapo-A-I) suggesting that its conversion to mature apo-A-I occurs after secretion. In order to assess the mode and site of proapo-A-I to apo-A-I conversion, we incubated the medium from [3H]proline-labeled Hep G2 cells with either human plasma, serum, lymph, or fractions thereof obtained by density gradient ultracentrifugation. The conversion was monitored by two-dimensional gel electrophoresis and by Edman degradation. Human plasma, serum, or mesenteric lymph all induced proapo-A-I to apo-A-I conversion; this was time dependent, unaffected by the serine protease inhibitor phenylmethylsulfonyl fluoride and inhibited by EDTA. Purified radiolabeled proapo-A-I bound to lymph chylomicrons and plasma high density lipoproteins. The converting enzyme was associated with both of these particles. Activity was also found in the d greater than 1.21-g/ml fraction and may have been derived from high density lipoprotein after displacement by high salts and/or ultracentrifugal force. We conclude that the conversion of proapo-A-I to apo-A-I occurs extracellularly and is probably effected by a metallo-enzyme which may act at the amphiphilic surface of either chylomicrons or high density lipoproteins.

Previous studies on the biosynthesis and processing of human apo-A-I' have shown that this protein is synthesized * *his work was supported by Grants HL 18577, AM 30292, and AM 20407 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. as a preproapoprotein with a 24-amino acid NH2-terminal extension (1-3). The presegment, 18 amino acid residues long, is cleaved co-translationally by signal peptidase. The remaining proprotein containing an hexapeptide prosegment covalently linked to the NH, terminus of mature apo-A-I is secreted into the medium. Lack of intracellular proapo-A-I cleavage may be due to the presence of a Gln-Gln dipeptide at the COOH-end of this prosegment. This feature distinguishes it from the dibasic residues present at the COOH terminus of most vertebrate prosegments which are cleaved during secretion (4). We previously performed two-dimensional gel electrophoresis on the purified proapoprotein and found that it represented the more basic isoform 2. By contrast, mature apo-A-I in circulating plasma (2,5 ) consists primarily of the more acidic isoform 4. Based on these observations, we have suggested that a converting enzyme system is located extracellularly and plays an active role in the posttranslational proteolytic processing of proapo-A-I to mature apo-A-I. In order to test this hypothesis and better characterize this proteolytic activity, we have embarked on studies in which we have followed the events attending the incubation of radiolabeled proapo-A-I obtained from the culture medium of Hep G2 cells with human plasma, serum, lymph, or fractions thereof under various experimental conditions. The results of these studies are the subject of this report.

EXPERIMENTAL PROCEDURES
The techniques of isolation of human proapo-A-I from the medium of cultured Hep G2 cells have been described previously (1). Automated NH2-terminal Edman degradation analyses were performed using a 0.33 M Quadrol program (6, 7) and a Beckman 890 C Sequencer.
Two-dimensional Polyacrylamide Gel Electrophoresis-Radiolabeled proteins present in the incubation medium were analyzed by two-dimensional polyacrylamide gel electrophoresis using the conditions detailed by O'Farrel (8) and modified later by Lester et al. (9). Stained gels containing 3H-labeled proteins were subjected to fluorography.
Preparation of Samples for Incubation with Hep G2 Medium-Blood was collected from fasting normolipemic subjects in lithiumheparin coated tubes. The plasma was separated by centrifugation (10). Serum was obtained from the same fasted subjects following blood clotting using identical centrifugal conditions. Retroperitoneal lymph of mesenteric origin was collected during abdominal surgery on a normolipidemic patient with an aortic aneurism. Red cells were sedimented by centrifugation (10) and chylomicrons were separated by ultracentrifugation at 100,000 X g for 30 min at 4 "C. The top floating chylomicrons were removed, overlayed with 0.15 M NaC1, pH 7.0, d 1.006 g/ml, and refloated under the same conditions. In the case of plasma, top and bottom d 1.21-g/ml fractions were prepared by adjusting the density to 1.21 g/ml with solid NaBr and centrifuging at 220,000 x g for 24 h a t 10 "C. The floating top 1-ml fraction and the bottom 1-ml fraction were removed and dialyzed at 4 "C against 0.15 M NaCl, pH 7.0, before use.
Incubation with Hep G2 Medium-Hep G2 medium containing radiolabeled proapo-A-I was incubated with plasma, serum, lymph, or their ultracentrifugal fractions in a shaking water bath at 37 "C under the conditions specified under "Results." For two-dimensional gel electrophoresis, the reaction was stopped by adding one-half volume of lysis buffer (9). The incubated products were also subjected to Edman degradation following immunoadsorption with monospecific antisera (1). Apo-A-I was run as a marker protein with each sample to permit a comparison among gels.
Preparation of HDL3 and Apo-A-I"HDL3 was prepared by conventional ultracentrifugal methods (IO). Apo-A-I was purified by size

Human Proapolipoprotein A-I Conversion
to Mature Apo-A-I exclusion high performance liquid chromatography as previously described (1 l).
Binding Studies of Purified Proapo-A-I with Serum or Isolated Lipoproteins-["HIProline-labeled proapo-A-I (10,000 dpm) was incubated with either whole serum or HDL, a t room temperature for 30 min with gentle stirring in 0.15 M NaCI, 1 mM EDTA, and 1 mM PMSF. In general, 1 to 3 mg of HDLR proteins or 0.5 to 1 ml of serum were utilized per incubation. Each sample was then adjusted to 1 ml with 0.15 M NaCI, 1 mM EDTA, pH 7.0, and separated by density gradient ultracentrifugation.
Density Gradient Ultracentrifugation-Density gradients were prepared as previously described (12). The gradients were collected in volumes of 0.4 ml/fraction and the absorbance monitored a t 280 nm. Aliquots of 100 pl were diluted with 10 ml of scintillation fluid and counted.  Fig. 1C was seen only occasionally. Its identity is unknown, but has also been observed previously in Hep G2 medium (13). Based on densitometric tracings of the fluorographs, the extent of conversion of isoforms 2 and 3 to 4 and 5 after a 1-and 6-h incubation with plasma was 8 and 25%, respectively. Automated sequential Edman degradation of ["Hlproline-labeled proapo-A-I after its immunoadsorption from the incubation mixture corroborated these results (Fig. 2). The distribution of ["Hlproline residues in the immunoadsorbed product obtained from the Hep G2 medium which had not been incubated with plasma, revealed peaks a t proline positions 9, 10, and 13. This agreed precisely with the known NH2-terminal proapo-A-I sequence (1,2). As shown at the bottom of Fig. 2, incubation of ["Hlproline-labeled Hep G2 medium with plasma for 6 h generated peaks of radioactivity a t cycles 3,4, and 7, as well as 9 and 10. The ["Hlprolines observed in positions 3, 4, and 7 correspond to the sequence of mature apo-A-I described previously (14). The ["Hlprolines appearing in positions 9 and 10 ( Fig. 2, bottom) are contributed by the proapo-A-I that remained unconverted. From these data, we conclude that conversion of isoforms 2 and 3 to 4 and 5 reflects proteolytic removal of the hexapeptide prosegment. Based on the ["Hlproline peaks a t cycles 3,4, and 7, compared to 9 and 10 there was 30% conversion. This is an agreement with data obtained from densitometric tracings of the fluorographs. The presence of 1 mM PMSF in the incubation reaction mixture containing Hep G2 medium and plasma had no effect on conversion. However, addition of 1 mM EDTA blocked the conversion completely (Fig. 1F). Like plasma, serum incubated with ["Hlproline-labeled Hep G2 medium for 6 h a t 37 "C resulted in the appearance of isoform 4 (Fig. IC).

Incubation Studies of ["HJProline-labeled
Incubation of ['Hlproline-labeled Hep G2 medium with mesenteric lymph for 1 h (Fig. 1H) and 6 h (Fig. 1Z) (Fig. 1J) and bottom (Fig. 1K) fractions from plasma showed that both were effective in producing isoform 4. The top (30 pg of protein) and bottom fractions (300 pg of protein) caused 19 and 10% isoform conversion, respectively. Isoform conversion was consistently found with the bottom fraction. With the top fraction, the conversion was observed in two out of three preparations studied; activity also was found with HDL:3 (Fig. 1L). Processing by chylomicron (Fig.  1M) was dose-dependent within the protein concentration range studied (77 to 350 pglincubation; data not shown).
Interaction of ("HJProline-labeled Proapo-A-I with Serum and HDL:I-All of the following studies were conducted in the presence of 1 mM EDTA to specifically inhibit the conversion of proapo-A-I to mature apo-A-I. Purified ["Hlproline-labeled proapo-A-I was incubated with serum and the products were separated by density gradient ultracentrifugation. Fig. 3A shows that the radioactivity profile only overlaps the absorption pattern of the HDL class. HDL, and HDL, account for over 90% of the radioactivity. Incubation of radiolabeled proapo-A-I with HDLs resulted in a product showing a close correspondence between radioactivity and the absorbance due to HDL, (Fig. 3B); 95% of the counts were recovered in this peak and less than 3% in the d > 1.21-g/ml fractions. These results show that proapo-A-I can interact with lipoproteins and preferentially with the HDL density class.
[3H]Proline-labeled proapo-A-I without lipids sedimented at density 1.28 g/ml. Incubation of proapo-A-I with lymph chylomicrons resulted in 15% of the counts being recovered in the top floating (d 1.006 g/ml) fraction (data not shown). This finding suggested that proapo-A-I also binds to chylomicrons.

DISCUSSION
The results of our studies show that when medium containing proapo-A-I secreted from an Hep G2 cell line is incubated with either plasma, serum, or lymph, further proteolytic processing occurs resulting in the conversion of proapo-A-I (isoforms 2 and 3) into mature apo-A-I (isoforms 4 and 5). This observation provides support for our earlier suggestion (1) that the proteolytic cleavage of proapo-A-I is an extracellular event. Under our experimental conditions, the kinetics of formation of apo-A-I, which was on the order of hours, is in agreement with the slow conversion process reported for other proproteins such as proinsulin to insulin and G34 gastrin to gastrin (3). However, we may not have chosen the optimal conditions for the conversion; a complete proapo-A-I to apo-A-I conversion was not achieved even after a 24-h incubation.2 Although we have not thoroughly defined the nature of the converting enzyme activity, it is apparent that it does not possess the general characteristics of a serine protease because of the lack of inhibition by PMSF. The fact that this activity was inhibited by EDTA suggests that we may be dealing with a metallo-enzyme. We observed no effect of exogenous Ca2+ (up to 10 mM) on the converting activity. However, the fluids that we examined, plasma, serum, or lymph, contain calcium. The effects of other metal ions and chelators have to be investigated.
Lymph chylomicrons and plasma HDL proved to be sources of the converting activity, also bound proapo-A-I. Based on the mass of protein present in the conversion assays, it appears that plasma HDL was more active than chylomicrons. From these findings, we may conclude that the interaction between the converting activity and proapo-A-I occurs at the amphiphilic interface of the lipoprotein particle. This conclusion is further supported by our observation that proapo-A-I combined with synthetic mixed discoidal particles of lecithin and cholesterol? However, we did find activity free from the HDL particles. This may be explained by a dissociation of this activity from the lipoprotein surface by the high salt concentrations and/or the ultracentrifugal gravitational force used. It could also account for our observation that top fraction mediated isoform conversion was occasionally absent. Alternatively, we may not be dealing with an ultracentrifugal artifact but rather with an activity which can also exist in a lipoprotein free state but capable of lipid binding when exposed to the Hep G2 medium. We may further speculate that at the time of secretion proapo-A-I is either already bound to lipids or becomes immediately associated with lipoproteins following its entry into the circulation, a step which may be necessary for efficient conversion. We have no information at this time on the source of the enzyme accounting for the converting activity. Its overall role in the events attending lipoprotein interconversions in plasma remains to be elucidated.

H u m a n Proapolipoprotein A-I Conversion to
Mature Apo-A-I 11433