Fatty acyl amidases from Dictyostelium discoideum that act on lipopolysaccharide and derivatives. II. Aspects of substrate specificity.

The substrate specificities of two fatty acyl amidases partially purified from the slime mold Dictyostelium discoideum have been studied. The amidase act on lipopolysaccharide derivatives, such as (4'-O-phosphoryl-N-beta-hydroxymyristyl-D-glucosaminyl)-beta-(1 leads to 6)-N-beta-hydroxymyristyl-D-glucosamine-1-phosphate (III) in a sequential manner. Amidase-I removes the beta-hydroxymyristyl residue present on the amino group adjacent to the 1-phosphate and the product formed is a substrate for amidase-II; the latter removes the remaining beta-hydroxymyristyl residue from the distal amino group. Compound III itself is resistant to amidase-II. Removal of the C-1 or C-4 phosphate groups does not influence recognition by the amidases or their sequential action. Both amidases are specific for long chain fatty amide linkages. Thus, a formyl group on the glucosamine amino group adjacent to the C-1 phosphate is not hydrolyzed by amidase-I; however, this substituent does not hinder the action of amidase-II on the distal fatty acyl amide. The presence of the beta-hydroxyl group in myristyl-amide residues is not required for hydrolysis. Further, while amidase-I requires disaccharide structures for its action, amidase-II acts on monosaccharides as well. Finally, the effects of a variety of substrate analogs and divalent ions on the activity of the enzymes are reported.

linked to the hydrophobic component, lipid A, which is embedded in the outer membrane and is responsible for the endotoxicity of LPS. Lipid A of enteric bacteria has been shown to consist of a P-(l-6)-diglucosamine backbone that carries long chain fatty acid residues at the hydroxyl and amino groups, and one or more phosphate residues (Fig. 1).
The structures of the LPS from an Escherichia coli heptoseless mutant have been elucidated previously ( I , Fig. 1

) (3).
During this study, two fatty acyl amidases from the slime mold Dictyostelium discoideum were discovered which were shown to act sequentially on LPS derivatives (4). Thus, amidase-I cleaved one acyl amide group from compound I11 (Fig. 1) to form compound IV, and amidase-I1 then deacylated the latter to give compound V (Fig. 1).
This paper describes further partial purification of the amidases. The enzymes have been separated from each other and freed from other interfering activities. An accompanying paper describes the mode of action and studies on the substrate-specificities of these enzymes. Estemses -Esterases were d e t e m l n e d a t 3 f C ~n 24 mM c l t r a t e , 1 3.0 w j t h [14CILPS (330 m a t e , 1050 c m l n a n o m l e ) . The e n t i r e r e a c t I a n m i x t u r e was fractionated by TLC on s i l i c a P l a t e l and the free fattv acids Quantitated. ADDroximte mibse-I activity can be determined s i m~l t a n e~u l l y b y q i n t i t a t i d n of 0-hydmxymyrlltic acid released. AcldfPhmJ;$y

RESULTS
A. ~l c a t i a n r O f Amldare-l acld phoiphatdsel and erteraserbutlaclrigniflcant amdare-II activity. Therefore, use of the wlld type Itrlln slmpllfled PwlflCatlOn of amidase-I.
Extracts af veqetat<ve 0 dircaldeum NC4, the Wild type strain, contain amidase-I, Step I. Crude NC4 Extract  Step 11: Heat I n a c t l v a t i a n o f Contaminating Phosphatase Three enrymatlc x t l v l t l e s were monitored i n t h e c~lurnn eluate' amidase-1, a c i d phos-

Fatty Acyl Amidases from
Dictyostelium discoideum the amidase. The enzyne was concentrated i n an Anlcon U l t r a f l l t P l t l o n cell Wlth a ~~3 0 Column. pwsumably a consequence Of the interference of the phosphatase i n assays pmor to f i l t e r t o a volume O f 28 nl. Lre a c t l v i t y was reCOvePed than initially applled to the the chmmtography (Step Table I ) .  (Fig, 3) and concentrated in an Amlcon u l t r a f i l t r a t i o n c e l l t o a column of 2.14 m1. This procedure Increased the specific actlyity O f amidase-I 10 fold to 78 unitrlmg (Step 4 . Table I ) .

Gel F i l t r a t i o n
treatment, t w phosphatase p w k r Of a p p m x i m t e l y equal a c t i v i t y were resolved and the apparent anidare-1 activity decreased m r t e d l y i n C o n t r a s t t o t h e e x p e r l n e n t shown i n Fiq. Thus, although both phmphatlles apparently declined upon heat treatment, the actlvity ca-eluting with amidare-] was preferentially reduced by appmximtely

l n i t l a l heat
Of d r e a n d mjor p r o t e i n peak that coincided with amidase-11. t h u s r e s u l t i n g i n I mjor t r e l t m n t also affected the gel f i l t r a t i o n p w f i l e ( F i a . 3) by causing the disappearance ImpIovment i n purification of the enzyne.

.
P u r i f r c a t i o n o f Amidase-11 When crude extract (Step 1) Was fractionated on OE-52 cellulare Without prior heat

2.
Amdare-11 i s secreted into the medium Of axenic Cultures aloon-er acid hydrolases includlng acid phosphatase, but no amidase-I activity was detected. The steps used i n puri.
f l c a t i o n are shown i n Table 2 .
overall 28% a Units defined in Material a n d m Step  Table 2 ) . Storage of the Amidases-Purified amidase preparations tolerate freezing and storage at -80 "C poorly and were almost completely inactive after 1 month under these conditions, although crude lysate and extracellular protein preparations retained their activity. Storage at 0 "C in ice preserved the activity of both amidases at purer stages. Therefore, both crude and purified preparations were stored in this way.
Expression of Amidase-I in D. discoideum AX3-As mentioned above, amidase-I activity was totally lacking among the secreted hydrolases of axenically cultured AX3. To determine the intracellular presence of amidase-I, approximately 8 X lo6 myxamoebae from axenic culture were lysed as described for NC4 cells. Although amidase-I abounds in NC4 amoebae, no activity was found in an identically prepared lysate of axenically growing AX3 cells. To test whether expres-  appropriately. Amidase-I activity in AX3 cells grown in liquid culture on A. aerogenes was comparable to that observed in NC4 amoebae cultured on lawns of E. coli. Clearly, growth on bacteria induced amidase-I. Sensitivity of Phosphatases to Fluoride-Acid phosphatase activity in D. discoideum NC4 extracts was only partially inhibited by up to 200 m~ K F while the extracellular activity in axenic cultures of AX3 was fully inhibited even by 5 mM KF. Only one of the two isozymes is extracellular. It was isolated during fractionation of AX3-derived extracellular protein on DE-52 cellulose (Fig. 5) and was examined for its susceptibility to fluoride by the p-nitrophenyl phosphate assay. This isozyme was highly sensitive to fluoride and was completely inhibited above 200 PM KF. Thus, the fluorideresistant activity in NC4 extract belongs to the second isozyme that co-elutes with amidase-I on DE-52 cellulose.
LPS-cleaving Fatty Acyl Esterases-Although gel fitration failed to separate the remaining fatty acyl esterase activity from amidase-I (Fig. 4), it appears that this esterase activity is distinct from amidase-I. First, the elution profies of the enzyme activities are dissimilar. To better compare their elution patterns, both amidase and esterase activities were assayed across a Sephacryl S200 column simultaneously with ['4C]LPS; amidase-I was assayed by release of P-hydroxymyristic acid, and esterase by the release of other fatty acids. The two elution peaks overlapped almost completely but had different shapes. Second, the stability of the two enzyme activities was different. At least one preparation, stored for 6 months, lost its esterase activity but retained amidase-I activity (data not shown).
Kinetic Properties of the Amidases-When compound 111 ( Fig. 1) was subjected to treatment with the purified phosphatase-free amidase-I preparation at 37 "C, the rate of hydrolysis remained constant up to 40% deacylation (Fig. 6A). Thus, below 40% hydrolysis of the substrate, product inhibition did not contribute significantly. The initial lag observed appears to be due to a drop in the temperature as cold enzyme solution was added to the buffered substrate. No delay was observed if the enzyme was preincubated in the assay buffer prior to the addition of substrate to start the reaction. The rate of hydrolysis of compound IV by purified amidase-I1 also remained relatively constant up to 30-40% deacylation (Fig.  6 B ) .
Michaelis constants for both enzymes were determined from assays at varying concentrations of the respective substrates with constant quantities of enzyme. From reciprocal plots of average rates and average substrate concentrations, K,,, values were estimated by linear regression analysis (Fig. 7 ) . Since initial rates could not be measured accurately at the lower substrate concentration, average rates were used. The validity of basing K, estimations on average values has been demonstrated by Lee and Wilson (11). The apparent K,,, of amidase-  I at pH 3.0 was found to be 3.7 p~ with compound I; for amidase-I1 it was estimated at 20 p~ with compound IV.
Critical Micelle Concentration of Compound I-The critical micelle concentration of compound I was determined as described under "Experimental Procedures" with the dye rhodamine 6G. A broad transition beginning at 15 p~ was observed. The profile of dye absorbance versus compound I concentration remained the same at pH 3.0 and 6.7. The khetics of amidase-I (Fig. 7A) showed no singularity in that concentration range, and the Lineweaver-Burk plot was linear. Thus, amidase-I is insensitive to the state of aggregation of its substrate.
Effect of Detergents-Both amidases were active in the presence of various non-ionic detergents, tested at 6 times their critical micelle concentrations (Nonidet (1.7 m~) , Brij 58 (0.46 mM), and Tween 20 (0.36 mM)) (13). Tween 20 markedly suppressed amidase-I activity, nonidet less so. However, Brij 58 slightly increased this enzyme's activity with respect to a control without detergent. As for amidase-11, both Brij 58 and Nonidet slightly stimulated its activity, while Tween 20 inhibited it by 30% under these assay conditions. The amidases also remained active in up to 40% ethylene glycol (data not shown). Sodium dodecyl sulfate (0.12%) in the assay mixture completely inactivated the enzymes. The stability of amidase activity in several detergents is consistent with the insensitivity of amidase-I to the state of aggregation of its substrate, as discussed above.

DISCUSSION
The fatty acyl amidases of D. discoideum have been resolved from each other and also separated from other LPShydrolyzing activities, except for a low level of a fatty acyl esterase that co-purifies with amidase-I. The resolution of the two amidases from each other was facilitated by identifying strains of D. discoideum, NC4 and AX3, each of which produces predominantly a single amidase. Amidase-I has been purified about 110-fold from a crude cell lysate of the strain NC4, while amidase-I1 has been purified about 12-fold from extracellular proteins of AX3, thus avoiding contamination by intracellular proteins. The two amidase preparations, which are free of interfering activities, are now satisfactory for the specific removal of the amide-linked fatty acids in LPS for structure-function studies.
The affiiity of the amidases of LPS derivatives proved to be very high. Their apparent Michaelis constants, 3.7 PM for amidase-I and 20 p~ for amidase-11, are relatively low when compared to those of other lysosomal hydrolases isolated from D. discoideum, which are in the millimolar range (10). Since LPS tends to aggregate, it would appear that the enzyme activities might be affected by the state of aggregation. However, although compound I11 undergoes a monomer to micelle transition near 15 p~, no abnormality in the substrate concentration-dependent kinetics of amidase-I was observed (Fig. 7). This observation stands in contrast to the behavior of phospholipases that undergo activation at interfaces when their substrates are inserted into micelles or aggregate above their critical micelle concentrations. Therefore, the enzymes appear to be very efficient at degrading ingested LPS and act independently of the state of aggregation of the de-esterified molecule.
A number of differences between the two enzymes with regard to developmental regulation are noteworthy. Amidase-I, present in vegetative myxamoebae of D. discoideum NC4, is absent in the same strain after aggregation and subsequent stages of development. In this respect, amidase-I behaves like other developmentally regulated enzymes. It is essential to digestion of bacteria only during vegetative growth and, therefore, is probably degraded and not produced when no longer needed. Synthesis of amidase-I appears to require signals from nutrient Gram-negative bacteria. Axenic AX3 cells, grown on bacteria-free medium, do not secrete or contain amidase-I, but those grown on A. aerogenes do contain the enzyme. Growth conditions also influence the expression of other lysosomal hydrolases in D. discoideum (14). AX3 produces amidase-I1 but not amidase-I, implying that their synthesis is independently regulated. NC4, on the other hand, contains relatively low levels of amidase-I1 activity. Another interesting observation is that AX3 secretes only one of the two acid phosphatases found in the lysate of vegetative NC4. Other D. discoideum lysosomal enzymes have been shown to have isozymes, only some of which are secreted.
A number of interesting applications of the two fatty acyl amidases are now possible. By allowing selective and sequential removal of the two amide-linked chains, the amidases provide a tool for exploring the biological and structural roles of LPS and, in particular, the role of P-hydroxymyristic acid. Although the function of P-hydroxymyristic acid is not presently known, certain studies (19, 20) suggest that the fatty acid may be required for mitogenic activation of B lymphocytes by LPS. It is now possible to substitute other fatty acyl groups to explore structure-function relations, or even photolabeled fatty acids for studies of the LPS receptor. The enzymes would also be of use in the study of LPS biosynthesis since they facilitate the preparation of defined precursors. Specific release of P-hydroxymyristic acid by the amidases could be utilized for improved detection and quantitation of LPS. Finally, amidases are members of a family of specialized enzymes which includes lipases and ceramide hydrolases. Their sequential mode of action make the LPS-cleaving amidases a very interesting enzymologicd system to explore.