Purification of human alpha-light class lymphotoxin to electrophoretic homogeneity

Abstract We have purified to homogeneity the α L -component (70,000–90,000) of the human LT cytolytic system. This lymphokine was purified ~ 10, 000-fold from supernatants of lectin-stimulated human tonsil and adenoid lymphocytes by molecular sieving, ion-exchange chromatography on DEAE-Sepharose, and preparative PAGE. The homogeneity of the radiolabeled molecule was confirmed both by electrophoresis and electrofocusing. The identity of the labeled peak with the lytic activity was demonstrated with a concomitant bioassay of the electrophoresed preparation; in addition, immunoprecipitation with a heterologous specific anti-α L antiserum showed simultaneous precipitation of the radiolabeled component and lytic activity. Immunological and biochemical evidence has previously shown this molecule to be a subunit of the Cx- and α H -LT forms. The latter LT classes are of intense interest because of their capacity for rapid selective cell lysis.

system.This lymphokinewas purified -lO,OOQ-fold from supernatants oflectin-stimulated human tonsil and adenoid lymphocytes by molecular sieving, ion-exchange chromatography on DEAE-Sepharose, and preparative PAGE.The homogeneity of the radiolabeled molecule was confirmed both by electrophoresis and electrofocusing.
The identity of the labeled peak with the lytic activity was demonstrated with a concomitant bioassay of the electrophoresed preparation; in addition, immunoprecipitation with a heterologous specific anti-cc, antiserum showed simultaneous precipitation of the radiolabeled component and lytic activity.Immunological and biochemical evidence has previously shown this molecule to be a subunit of the Cx-and a,-LT forms.The latter LT classes are of intense interest because of their capacity for rapid selective cell lysis:' INTRODUCTION LT was originally described as a weak, nonspecific growth-inhibitory or cell-lytic lymphokine (Ruddle & Waksman, 1968;Granger & Williams, 1968;Jeffes & Granger, 1975).Recent functional and biochemical studies on LT from human and animal species indicate that LT is in fact represented by a family of diverse mol.wt cytotoxins (Walker et al., 1976;Granger et al., 1978;Ross et al., 1978).The lower mol.wt forms (alpha-light, 70,000-90,000 d; beta, 35,00&50,000 d; gamma, 12,0~20,000 d) are capable only of protracted lysis of a limited number of continuous cell lines; the larger forms (complex, > 200,000 d; alpha-heavy, 120,00~160,000 d) are capable of causing rapid nonspecific lysis of a spectrum of allogeneic cells, and selective lysis when obtained from lectinstimulated alloimmune immune lymphocytes (Yamamoto et al., 1979;Hiserodt et al., 1978a).Both immunological and biochemical evidence suggests that these classes are comprised of related subunits (Yamamoto et al., 1978;Hiserodt et al., 19786).
The definitive assignment of a role for these mediators in cell-mediated immune reactions in oitro is as yet not possible.1976;Hiserodt & Granger, 1977;Sawada & Osawa, 1978;Ware & Granger, 1981).It is evident from these studies that antisera which recognize determinants expressed by these soluble toxins may block several types of cellmediated cytotoxic reactions in vitro.
Nevertheless, an analysis of the participation of these mediators in the molecular mechanisms of cell-mediated cytotoxicity will be greatly facilitated by the availability of these cytotoxins in a homogeneous form.The purification of LT has been very difficult, since these mediators are present in extremely low concentrations in lymphocyte supernatants.We have employed serum-free culture conditions to greatly reduce the amount and complexity of exogeneous protein (Lewis et al., 1976); furthermore, we have been able to monitor protein in the picomole range utilizing a mild radiolabeling technique (Klostergaard et al., 1980).As a result, we have been able to purify to electrophoretic homogeneity the human Q,subclass, an important subunit of the Complex and alpha-heavy classes.

MATERIALS AND METHODS
The methods involved in the preparation of human lymphocyte supernatants, in generating the various mol.wt LT classes by molecular sieving of the supernatants, and in resolution of the or,-LT class into its charge subclasses by ionexchange chromatography have been published previously in detail (Granger et al., 1978;Lewis et al., 1976) 1).The aLsubclasses were obtained from the cc,-class by ion-exchange chromatography on DEAE-Sepharose.The predominant activity, the Qsubclass (Table l), which was eluted on a salt gradient, was then subjected to a variety of purification schemes.Radioiodination of lymphocyte supernatants and LT preparations with "'1 (New England Nuclear, Boston, MA) was performed by the Iodogen (Pierce Chemical Co., Rockford, IL) technique as previously described (Klostergaard et al., 1980).
Preparative PAGE of radioiodinated c(,,-LT using a discontinuous buffer system was conducted in an apparatus designed by Furlong et al. (1973).The preparation was brought to 20% sucrose and then overlayed on a 3% acrylamide stacking gel (1 cm dia, 1 cm high); the separating gel was 7% acrylamide (1 cm dia, 3 cm high).The gel was run at constant current (11 MA) at 4°C.Fractions (20 drops) were collected at a flow rate of 60 drops/hr, employing a Shandon-type apparatus.
Fractions were tested for both radioactivity in a Biogamma counter (Beckman Instruments, Fullerton, CA) and for lytic activity as described later.

Analytical PAGE
Fractions containing lytic activity from preparative PAGE of a,,-LT were pooled, concentrated, radioiodinated and subjected to analytical discontinuous PAGE (Davis, 1964).Tube gels (3 mm dia) with a 1 cm stack of 3% acrylamide and a 7 cm running gel of 7% acrylamide were employed.Electrophoresis was conducted at 5 MA per gel at 4°C.After the run, gels were sliced in 2 mm sections and counted in a gamma counter; the slices were allowed to elute overnight at 4°C in 0.3 ml of media, and then the eluate was assayed on L-cells for lytic activity (see later).

Analytical isoelectricfocusing
Radioactive fractions containing lytic activity from analytical PAGE were subjected to isoelectricfocusing in tube gels.The acrylamide concentration was 4%, and the ampholine gradient was 1% from pH 4 to 8.After running, the gels were sliced (2 mm) and the radioactivity in each slice determined.

Neutralization and immunoprecipitation of lz51cr,,-LTwith rabbit anti-human a-,,-LTantiserum
Aliquots containing two to four units of radioiodinated a,,-LT (-1000 cpm) from analytical PAGE were incubated with &20 ~1 of a rabbit anti-human a,,-LT antiserum.This antiserum had been raised against electrophoretically purified a,,-LT (Yamamoto et al., 1978).After incubation at room temperature for 30 min, a sheep antiserum (&20 ~1) to rabbit serum was added and the mixture was allowed to incubate for an additional 30 min.The complexes were pelleted by centrifugation.
The supernatants were tested on L-cells for lytic activity (see later), and the pellets were counted in a Biogamma counter (Beckman Instruments, Fullerton, CA).As a control, a normal rabbit serum was used in the same manner as the immune serum.Radioactivity precipitated by the immune serum was corrected for the amount reacting nonspecifically with the normal serum (-15%).

LTassay
The detection and quantitation of LT activity in lymphocyte supernatants, column fractions, and in gel slices was carried out as previously described (Granger et al., 1978).Briefly, 1 x lo5 mitomycin-C (Sigma, St. Louis, MO) treated c1-L-929 cells were cultured for 24 hr in slant tubes in 1 ml RPMI-1640 with 3% newborn calf serum (GIBCO, Grand Island, NY).Aliquots (10-200 ~1) of supernatant, column fractions, or gel eluates were added to the cultures; after 1624 hr, the remaining viable adherent cells were determined in a Coulter Counter.A unit of killing was defined as that required to decrease the adherent cell count by 50%.
The human a,,-LT subclass was obtained by tandemized molecular sieving and ion-exchange chromatography of supernatants.
The uLZcomponent was first radiolabeled with "'1 by the solid-phase Iodogen technique, which preserved lytic activity, so that both protein and lytic activity could be followed in the course of determining optimum purification (Klostergaard et al., 1980).
Both lectin-and hydrophobic-affinity chromatography were initially used independently in purification, since the uL2component binds to Con A-Sepharose and to decyl-agarose.
Despite the fact that l&20-fold purification could be realized, in either case analysis on PAGE revealed that the adsorbent also retained contaminating proteins resident in the crude a,,-preparation (Klostergaard et al., 1980).It was apparent from these experiments that most of the contaminants in these preparations had distinctly higher electrophoretic mobilities than the actual LT moiety; this suggested that a preparative electrophoretic or electrofocusing step might be effective in our purification scheme.A very striking purification could in fact be achieved by preparative PAGE (Fig. 1).Most of the contaminating proteins migrate rapidly and appear in the early fractions where no lytic activity is present.Lytic activity appears in later fractions which contain very little protein.This electrophoretic step gives an about two orders of magnitude increase in specific activity (Table 1).
In order to obtain a,,-LT from preparative electrophoresis which was of suitable specific activity (radioactivity), fractions from preparative electrophoresis containing the lytic activity were pooled, concentrated about lo-fold on an Amicon PM-10 membrane, and radioiodinated.When analysed on analytical PAGE (Fig. 2A), multiple radioactive peaks are apparent; the peak with the lowest R, (I) coincides well with lytic activity.The faster moving peaks appear to be trailing peaks from preparative electrophoresis, since they may be substantially reduced when fractions from the lytic peak closest to the contaminating peaks are eliminated from the pool.  in each slice was determined as for Peak I from Fig. 2A appears to be a true electrophoretic component, migrating indistinguishably from lytic activity (Fig. 2B).Peak I from Fig. 2A was subjected to isoelectricfocusing in polyacrylamide gels.Each gel slice was assayed for radioactivity.
Fig. 3 indicates that peak I is extremely homogeneous, with a pI of 4.8.
Aliquots of peak I were reacted with increasing amounts of a rabbit antiserum raised against human a,,-LT which had been purified by native PAGE (Yamamoto et al., 1978).The immune complexes formed were precipitated by the sheep antiserum against rabbit serum: lytic activity remaining in the supernatant was evaluated on Lcells, and the radioactivity in the pellet was determined.
The results are seen in Fig. 4. As the amount of primary antiserum added is increased, lytic activity neutralized and radioactivity immunoprecipitated increases steadily until a level of 5 ~1 of antiserum is added in the experiment shown here.At this level, all lytic activity is blocked, and radioactivity is totally immunoprecipitated.
The coincidence of total blocking and immunoprecipitation of peak I at the same level of added antiserum is strong serological evidence that peak I is homogeneous and identical with the lytic moiety.

DISCUSSION
This report documents the purification to homogeneity of an important member of the human LT family, the a,,-component.This lymphokine is thus one of the very few convincingly purified to homogeneity.The purification scheme surmounted considerable difficulty, since despite beginning with serumfree, low-protein supernatants, it achieved an increase in specific activity of about lO,OOO-fold.
It is very clear from our data (  and the radioactivity in the pellets was determined (e-0).
The results from each level of immune serum were compared to those for the same level of normal rabbit serum, and the difference plotted.
sensitivity to demonstrate homogeneity in gel systems.In particular, two previous reports dealing with the purification of human c(,,-LT (Granger et al., 1973: Russel et ul., 1972) could not demonstrate a protein moiety truly migrating with lytic activity in PAGE.In fact, Russel et al. (1972), using a scheme apparently identical to that in this report, demonstrated in their most purified preparation significant stainable contaminants migrating in PAGE very near to regions containing lytic activity which were not stainable.We believe that their purest LT preparation was probably at least an order of magnitude from homogeneity.We are convinced that the criteria for purity presented here comprise a very solid basis for the statement that we have in fact purified c(,,-LT to homogeneity. First of all, the labeled component is indistinguishable from the lytic moiety in native PAGE, migrating as a single component (Fig. 2A and B).Secondly, the homogeneity of this component is further verified, since it focuses as a sharp peak in isoelectricfocusing (Fig. 3).It is highly unlikely that another component has coelectrophoresed with and has the same p1 as the lytic molecule, and that the sharp peaks seen in both electrophoretic systems are due to multiple components.
The serological studies further discount this remote possibility.The rabbit antiserum was raised against human clLZ which was purified by PAGE.As seen from the scheme presented in this paper (Table 1, Figs 1 and 2A), this is a fairly pure antigen.Even if the antiserum is not entirely monospecific, the fact remains that exactly the same levels of antiserum are required to completely neutralize the lytic activity of peak I (Fig. 2A) as to completely immunoprecipitate the radioactivity in the peak.In totality, the biological, biochemical and serological evidence presented here strongly support the homogeneity of the purified IX,,-LT preparation.
Purification of the a,,-LT component has brought certain goals closer to realization.In our own laboratory, we are now examining its peptide subunit composition.Since a previous report demonstrating high-affinity receptors for LT on ,sensitive targets and low-affinity receptors on resistant targets was apparently based on the use of an impure LT preparation (Tsoukas et al., 1976), we will reexamine this question with the pure mediator.

Fig
Fig. 2A (top panel).Analytical discontinuous PAGE (77" acrylamide) of human cc,,-subclass LT purified by preparative PAGE (fractions 18-23 from Fig. 1).Those peak fractions were pooled, concentrated and radioiodinated.Following analytical PAGE, the radioactivity in each gel slice was determined (to).Each slice was eluted in buffer overnight; the eluate was then assayed on L-cells (0-0).B (bottom panel).Analytical discontinuous PAGE (7% acrylamide) of peak I from Fig. 2A.Radioactivity (O-O) and lytic activity (O-O)in each slice was determined as for

Fig
Fig. 4. Blocking of lytic activity and simultaneous immunoprecipitation of radiolabeled c(,,-LT.Aliquots of peak I from Fig. 2A were reacted with increasing levels of a iabbit anti-human G(~ antiserum: resulting complexes were precipitated with a sheep antiserum to rabbit serum.Supernatants were tested for lytic activity on LT cells (0-0)and the radioactivity in the pellets was determined (e-0).The results from each level of immune serum were compared to those for the same level of normal rabbit serum, and the difference plotted.
these peptide maps will be compared to the map of the cc,-form, to examine the subunit relationship of c(~ and aH at this level.In light of the report by Lee & Lucas (1976), we will also test the pure a,-LT preparation for to normal cells, to the cytotoxic action of pure lymphotoxins is to be tested in our laboratory.

Table 1 .
and will only be briefly summarized Estimated specific activity and yield of human aLz -LT subclass during purification Class is the principal form in the supernatant.L a,-Subclass is the major component in the cc-class.
Detroit, MI), were obtained and concentrated as previously described (Lewis et al., 1976).The 5-day cultures were supported by a heat-stable fraction of bovine serum; this novel serum substitute reduces the amount of exogenous protein in the supernatant by about two orders of magnitude.Concentrated supernatants were subjected to molecular sieving on Ultrogel AcA 44 to resolve the various LT mol.wt classes.The a,-class (7O,OO(r90,000 d) was thereby separated from much of the contaminating protein (Table