The Biosynthesis of Differentiation-Inducing Factor, a Chlorinated Signal Molecule Regulating Dictyostelium Development*

Differentiation-inducing factor (DIF)-1 is a chlorinated alkyl phenone released by developing Dictyostelium amoebae, which induces them to differentiate into stalk cells. A biosynthetic pathway for DIF-1 is proposed from labeling, inhibitor, and enzymological experiments. Cells incorporate 36 Cl 2 into DIF-1 during development, showing that the chlorine atoms originate from chloride ions; peak incorporation is at the first finger stage. DIF-1 synthesis can be blocked by cerulenin, a polyketide synthase inhibitor, suggesting that it is made from a polyketide. This is most likely the C 12 polyketide (2,4,6-trihydroxyphenyl)-1-hexan-1-one (THPH). Feeding experiments confirm that living cells can convert THPH to DIF-1. Conversion requires both chlorination and methylation of THPH, and enzymatic activities able to do this exist in cell lysates. The chlorinating activity, assayed using 36 Cl 2 , is stimulated by H 2 O 2 and requires both soluble and particulate components. It is specific for THPH and does not use this compound after O -methy-lation. The methyltransferase is soluble, uses S -adeno-syl- L -methionine as a co-substrate, has a K m for di- chloro-THPH of about 1.1 m M , and strongly prefers this substrate to close analogues. Both chlorinating and methyltransferase activities increase in development in parallel with DIF-1 production, and both are greatly reduced in a mutant strain that makes little DIF-1. It is proposed that DIF-1 is made by the initial assembly of a

DIF-1 1 regulates the central cell fate decision during Dictyostelium development. In suitable conditions, isolated cells are induced by DIF-1 to differentiate into stalk cells, whereas without it they become spores (1)(2)(3)(4)(5)(6). Likewise, when Dictyostelium aggregates develop on a substratum containing DIF-1, the proportion of stalk precursor cells (prestalk cells) increases, and the proportion of prespores decreases (7,8). DIF-1 levels rise strongly in development as prestalk cells first differentiate (9,10), and our current view is that these rising DIF-1 levels induce the most sensitive cells in the aggregate to differentiate into prestalk cells. These cells rapidly produce DIF-1 dechlorinase, which inactivates DIF-1 and prevents a further rise in levels, allowing the majority of cells to differentiate as prespores (11,12). To understand this process further requires a better knowledge of DIF-1 signaling and, to this end, I have attempted to discover the biosynthetic pathway for DIF-1.
DIF-1 is an unusual signal molecule, a chlorinated alkyl phenone (13), for which neither the biosynthetic pathway nor any of the biosynthetic enzymes are known. The methods used for elucidating the biosynthetic pathways of many natural products are difficult to apply to DIF-1. Genetic analysis has identified a number of mutants potentially defective in DIF-1 biosynthesis (4), but efforts to clone the mutated genes by complementation (14) have so far failed 2 ; nor have such mutants been isolated yet by insertional mutagenesis, 3 which would facilitate subsequent cloning of the mutated genes (15). DIF-1 is active at 10 Ϫ9 M and so is only present at concentrations orders of magnitude lower than many secondary metabolites; it took a massive effort to isolate 50 g for identification of the structure (13,16). Standard methods for determining the biosynthetic origin of the carbon backbone and oxygen substituents by stable isotopic labeling are therefore difficult. However, it has been possible to deduce a likely biosynthetic pathway for DIF-1 from labeling experiments with 36 Cl Ϫ , from inhibitor studies, and by searching biochemically for some of the predicted biosynthetic enzymes. The initial experiments were guided by clues provided by the structure of DIF-1 itself and by various biosynthetic precedents.
Since DIF-1 does not resemble any known intermediary metabolite, it is probably made by a dedicated biosynthetic pathway. The aromatic ring of DIF-1 could arise either from the shikimate pathway of aromatic amino acid biosynthesis (17) or from a polyketide. Of these alternatives, a polyketide origin for DIF-1 is the more likely, since it automatically explains the four alternating oxysubstitutions of the final molecule (18) and because this is the way that many simple aromatic metabolites are made, including acetylphloroglucinol (19), a homologue of the proposed polyketide precursor of DIF-1.
Polyketides are typically formed by condensing together acetate units, from malonyl CoA, onto a starter such as acetate in a reaction closely related to fatty acid biosynthesis (19,20). The initial carbonyl groups of the polyketide are either left intact or variously reduced to hydroxyl, alkenyl (after elimination of water), or fully to alkyl, depending on the programming of the particular polyketide synthase. The polyketide can be ring closed, aromatized, and substituted in various ways, allowing a great diversity of products to be made. Polyketide synthases usually combine the activities required for polyketide synthesis * This work was supported by the Medical Research Council and the Howard Hughes Medical Institute International Scholars Program. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This paper is dedicated to the memory of the late Dr. Mary Berks, good friend and colleague.
into a large complex, often with a single polypeptide combining several activities. The condensing enzymes of polyketide and fatty acid synthases share conserved sequence motifs around the active site (21), and the active site cysteine is normally the target for the covalent inhibitor, cerulenin (22)(23)(24). Inhibition by cerulenin therefore provides a diagnostic test for the proposed polyketide origin of DIF-1 (25,26).
The most likely polyketide precursor of DIF-1 is a hexaketide, which, after complete reduction of two carbonyls and ring closure, would yield (2,4,6-trihydroxyphenyl)-1-hexan-1one (THPH; see Fig. 8 for structures). THPH has the complete 12-carbon backbone of DIF-1 and only requires chlorination and O-methylation to form DIF-1. Chlorinations can be carried out by chloroperoxidase enzymes, which utilize hydrogen peroxide to oxidize chloride ions and chlorinate a variety of phenolic compounds (27,28). Methylations are normally carried out by methyltransferases, using S-adenosylmethionine as the methyl donor, and a number of O-methyltransferases have been identified in the biosynthesis of secondary metabolites (29 -31). Since THPH and a variety of other potential substrates have been synthesized (32,33), it was possible to devise biochemical assays for the predicted chlorinating and methylating enzymes, which were duly detected.
Cell Growth and Labeling-Cells were grown and developed at 22°C. Strain Ax2 was grown in axenic medium with shaking (35), and strain V12M2 was grown on nutrient plates in association with Klebsiella aerogenes (36) and washed free of bacteria in KK2 (20 mM K 1 K 2 PO 4 , 2 mM MgSO 4 , pH 6.2) before use.
V12M2 cells were also developed in submerged monolayers (10 ml of medium/9-cm diameter tissue culture plate) at a density of 4 ϫ 10 6 /ml in 50% DIFlab, 0.1 mM MgSO 4 , containing 100 g/ml streptomycin, 5 mM cAMP, 0.1 Ci/ml 36 Cl Ϫ with additions as indicated. After 16 h, the medium was taken off, and nonpolar compounds were extracted using a C18 SepPak cartridge (Waters).
The rate of 36 Cl Ϫ uptake by cells was determined by incubating disaggregated slug-stage cells (disaggregated by passing twice through a 20-gauge and once through a 23-gauge syringe needle) at 4 ϫ 10 7 cells/ml in 10% DIFlab, 0.1 mM MgSO 4 containing 0.1 Ci/ml 36 Cl Ϫ and 0.5 Ci/ml [ 3 H]inulin (Amersham). After different times of incubation, the cells from triplicate samples were pelleted, and the pellets were dissolved in 20 l of formic acid before dual channel scintillation counting. Uptake was calculated after correcting for extracellular volume, using inulin as the marker. Fig. 1 shows that uptake is half-maximal after about 5 min.
Chlorination Assay-100 l of sample in lysis buffer containing 0.1 Ci of 36 Cl Ϫ (equivalent to 2 mM Cl Ϫ ), 0.1 mM THPH, and 50 mM H 2 O 2 , unless otherwise stated, was incubated at 25°C, and the incubation was terminated at the appropriate times by the addition of 100 l of stop solution (90/10/2 ethyl acetate/hexane/acetic acid, containing 5 mg/ml butylated hydroxytoluene and 1 mg/ml 50% tocopherol as antioxidants). After centrifugation, the upper phase was taken off, and the lower phase was reextracted with 150 l of ethyl acetate. The combined organic phases were dried down in a Savant Speed-Vac and quantitatively loaded onto TLC plates.
Methyltransferase Assay (Dichloro-THPH Methyltransferase)-A 50-l sample in lysis buffer was incubated with 50 M dichloro-THPH and 1 M AdoMet (including 0.5 Ci of [ 3 H]AdoMet) at 25°C, and the reaction was terminated by adding 50 l of stop solution. After centrifugation in a microcentrifuge, 35 l of the organic phase was loaded directly onto each lane of the TLC plate. Kinetic data was analyzed using the Enzyme Kinetics software of D.G. Gilbert.
Partial Purification of Dichloro-THPH Methylase-The 35-60% ammonium sulfate cut of a high speed supernatant from a lysate of 1.2 ϫ 10 10 slug stage cells was dialyzed into 20 mM MOPS, 2 mM EDTA, 10% glycerol, 1 mM dithiothreitol, pH 7.5 (MEG buffer) and loaded onto a POROS 20HQ strong anion exchange column (10 ϫ 0.46 cm). Proteins were eluted with a gradient of 0 -1 M KCl in MEG buffer over 5 min at 5 ml/min, and methyltransferase activity was located. The two steps gave a purification of 11.5-fold over the high speed supernatant.
TLC and HPLC-TLC was on activated Whatman LK6D silica plates developed with 60/40/2 ethyl acetate/hexane/acetic acid for the enzyme assays (R F values were as follows: THPH, 0.51; monochloro-THPH, 0.50; dichloro-THPH, 0.65; DIF-3, 0.58; DIF-1, 0.72) or 90/21/3/3 chloroform/methanol/acetic acid/water for whole cell extracts. Tritiated compounds were detected by autoradiography after spraying the plates with 3H-Enhance (NEN Life Science Products) and 36 Cl Ϫ -labeled compounds directly, either on film (Kodak XAR5) or on phosphor imaging plates (Eastman Kodak Co.) exposed in a lead box to reduce background. Tritiated compounds were quantitated by scintillation counting after scraping the labeled bands from the TLC plate and 36 Cl by reference to standard spots on the TLC plates, using Molecular Dynamics scanners for film and image plates as appropriate. The standard curve was linear from 0.5-16 cpm on film and over a much wider range on phosphor imaging plates.
HPLC was performed as previously on ODS columns using methanol/ water/acetic acid or acetonitrile/water/acetic acid gradients (40,41).

RESULTS
In Vivo Labeling-In vivo labeling with potential precursors was used to define the period of peak DIF-1 accumulation during development and to search for possible biosynthetic intermediates. Labeling with radioactive acetate or methionine (as a methyl donor) yielded strong incorporation into a number of lipids (42) but not detectably into DIF-1. Alternatively, the DIF-1 released into the medium by developing cells can be labeled with 36 Cl Ϫ (41). Since none of the accompanying labeled compounds are potential precursors (40), cell-associated compounds were examined. Developing cells were labeled with 36 Cl Ϫ and extracted with organic solvents, and the extracts were separated by TLC. The very low specific activity of 36 Cl Ϫ necessitated autoradiographic exposures of a few months on film or a few days on phosphor imaging plates. Fig. 2 shows that only two labeled compounds are detected in cell extracts. One is DIF-1, as shown by co-chromatography on two different TLC systems (not shown) and by the previous demonstration of DIF-1 in slug extracts by HPLC (43); the other compound (X) is unidentified. This compound runs in approximately the same place on TLC as dichloro-THPH, a proposed precursor of DIF-1 (Fig. 8), but further characterization was not attempted due to the low amounts present.
The time course shows that DIF-1 rises from very low, or undetectable, levels in the first 2 h of development to very low, but definite, levels in aggregation (e.g. 2.5 nM at 6 h) before increasing strongly as tips form to a peak in the fully elongated first finger. This is broadly in line with earlier work, using the bioassay to detect DIF (9, 10). The detection of DIF-1 during aggregation was surprising, because it has no clear role at this stage of development, but it was confirmed by HPLC fraction-ation of labeled compounds released after 5 h of development (41) (not shown). The expression of the DIF-1-induced mRNAs for ecmA and ecmB and the prespore marker mRNA, for pspA, is shown for comparison in Fig. 2. pspA mRNA levels rise coordinately with the major rise in DIF-1 levels; ecmA and ecmB mRNA levels rise with a delay of 1-2 h.
Since labeling of the intracellular pool with 36 Cl Ϫ is at equilibrium and saturating (equilibrium is reached in about 30 min, and doubling or halving the amount of label makes no difference to incorporation; see Fig. 1 and "Materials and Methods"), the specific activity of total Cl Ϫ is taken as that of the input label. With this assumption, and taking 1 ml of packed slug cells as 66.4 mg of protein, the concentration of DIF-1 in the slug is 62 Ϯ 30 nM (n ϭ 4).
These experiments indicate that chlorine is incorporated into DIF-1 from chloride ions, that the biosynthetic enzymes should be maximally active at the first finger and slug stages of development, and that dichloro-THPH could be a precursor of DIF-1. The low concentration of DIF-1 in cells helps to explain the difficulty in labeling it with other potential precursors.
Inhibition of DIF-1 Biosynthesis by Cerulenin-To test whether the polyketide synthase inhibitor, cerulenin, inhibits DIF-1 biosynthesis, cells were incubated in submerged culture with cAMP to stimulate their development and with 36 Cl Ϫ to label the DIF-1 produced. After 16 h, DIF-1 was extracted from the medium and resolved by TLC. Fig. 3A shows that cerulenin efficiently inhibits DIF-1 synthesis at concentrations similar to those inhibiting condensing enzymes in other organisms (25,26,44,45).
Cerulenin also inhibits the condensing enzyme of fatty acid synthase and is known to inhibit fatty acid synthesis in Dictyostelium (46). Although fatty acid synthesis would not be expected to be important during development, since the cells are starving, the cerulenin experiment still requires controls  36 Cl Ϫ and cerulenin as indicated. After 16 h, DIF-1 was extracted from the medium, separated by TLC, and label was quantitated using a PhosphorImager. Incorporation into DIF-1 by control cells was 21 cpm/TLC lane (equivalent to 8 ϫ 10 7 cells). B, stalk cell differentiation by cells incubated under conditions similar to those in panel A except that the density was reduced to 10 6 /ml; phase contrast micrographs were taken after 2 days. Cerulenin was 100 M, and DIF-1 was 100 nM. for specificity. Fig. 3B shows that when cells are allowed to develop in the absence of cerulenin, they eventually form stalk cells, having accumulated enough DIF-1 to induce their own differentiation. Stalk cell differentiation is inhibited by cerulenin, as expected from its inhibition of DIF-1 synthesis, but the cells remain amoeboid and move around actively. However, when DIF-1 is added to these inhibited cells, they efficiently turn into stalk cells, showing that cerulenin does not have a generally deleterious effect on the cells or compromise any specific function necessary to respond to DIF-1. As further controls, adding fatty acids to the medium restored neither stalk cell differentiation nor DIF-1 production to cerulenintreated cells (palmitic and stearic acids were tested, as well as hexanoic acid, the possible origin of the alkyl side chain of DIF-1; not shown).
Utilization of THPH by Cerulenin-treated Cells-The likeliest polyketide precursor for DIF-1 is the C 12 polyketide THPH, which is DIF-1 less the chlorines and methoxy group (see Fig.  8). It was synthesized, along with some related compounds, and fed to cells inhibited with cerulenin, to test whether it could be converted to DIF-1. Fig. 4 shows that it can be. DIF-1 synthesis is inhibited by cerulenin, as before, but it is fully restored when either THPH or monochloro-THPH is supplied to the cells. THPH is converted to compounds co-eluting with monochloro-THPH and dichloro-THPH, which are therefore likely intermediates in the biosynthetic pathway, and finally to DIF-1 itself. Monochloro-THPH is converted to dichloro-THPH and DIF-1 (the trace of label in the position of monochloro-THPH is most likely due to a trace contamination of THPH in the original substrate). The compounds running below monochloro-THPH are most likely metabolites of DIF-1, which would be made in these conditions (40).
DIF-3, DIF-1 itself, 4-methoxy-THPH, and various fatty acids are all ineffective at restoring DIF-1 synthesis. The lack of chlorination of 4-methoxy-THPH and of the monochlorinated form of this compound (i.e. DIF-3) is significant; it indicates that once THPH is methylated, it cannot then be chlorinated and therefore that chlorination precedes methylation in the biosynthetic pathway. This supposition is supported by the substrate specificities of the chlorinating and methylating enzymes, described below. At 0.2 M THPH, DIF-1 is the major product, but at higher THPH concentrations, the chlorinated but unmethylated products accumulate, suggesting that in these conditions activity of the methyltransferase becomes rate-limiting (not shown).
These experiments therefore provided strong evidence for the polyketide origin of DIF-1 and encouraged a search for the predicted methylating and chlorinating enzymes.
Detection of a Specific Methyltransferase in Cell Lysates-Cell lysates were made from slugs, when DIF-1 synthesis is near maximal, and tested for the presence of a methylating enzyme by incubating them with S-adenosyl-[methyl-3 H]methionine as methyl donor and a number of potential DIF-1 precursors. After incubation at 25°C, nonpolar products were extracted from the reaction mixes with ethyl acetate/hexane and analyzed by TLC. Fig. 5 shows that, although lysates make a background of methylated compounds from endogenous substrates, there is a massive incorporation of label into a new product when dichloro-THPH is supplied. TLC and HPLC coelution with authentic DIF-1 shows that this product is DIF-1 (not shown). Monochloro-THPH can also be methylated to make DIF-3 (which coincides with a background band in Fig.   FIG. 4. Conversion of THPH to DIF-1 by living cells. V12M2 cells were incubated at 4 ϫ 10 6 /ml in tissue culture dishes in medium containing 5 mM cAMP, 36 Cl Ϫ , and additives (100 M cerulenin (Cer); 1 M THPH or Cl-THPH; 100 M hexanoic, palmitic, or stearic acids) as indicated. Labeled products were extracted and detected after TLC using a PhosphorImager. Each lane is the extract of 8 ϫ 10 7 cells; 575 cpm were incorporated into the combined products in the cerulenin plus THPH lane, compared with 47 cpm in the control lane. Labeled compounds were identified by reference to authentic standards run on the same TLC plates; the arrowed compounds have not been identified but may be DIF-1 metabolites (33,40).

FIG. 5. Demonstration of dichloro-THPH methyltransferase in the high speed supernatants of slug lysates.
Samples were incubated with 1 M [ 3 H]AdoMet and substrates as indicated, and labeled products were extracted, separated by TLC, and detected by autoradiography. In the solvent system used, DIF-3 is not resolved from an endogenous band, but its production is visible as increased incorporation in the two lanes with Cl-THPH; DIF-3 production is quite clear with further purified enzyme preparations, where the endogenous bands are eliminated. The faint band due to 2-methoxy-DIF-1 is not visible at this exposure. 5), and 2-methoxy DIF-1 is also utilized, albeit poorly (not visible in Fig. 5 but apparent with purified enzyme preparations).
The methylating activity resides in the high speed supernatant of cell lysates (S300) and is therefore presumably cytosolic. It has a broad pH optimum from pH 6.5 to 7.5 and is indifferent to the presence in the assay of EDTA, MgSO 4 , or CaCl 2 up to 5 mM (not shown). The methyltransferase was partially purified by ammonium sulfate precipitation and anion exchange chromatography to eliminate background incorporation, and its kinetic properties were investigated (see "Materials and Methods"). Double reciprocal plots were linear. The preferred substrate is dichloro-THPH (K m 1.1 M), although monochloro-THPH and the 2-methoxy isomer of DIF-1 show some activity (Table I, Fig. 5). However, there is no detectable incorporation with the bare polyketide, THPH, nor with DIF-1 or DIF-3, both of which have two hydroxyls available. The K m for AdoMet is about 4 ϫ 10 Ϫ7 M, which is in the physiological range (47), and activity can be inhibited by the reaction product, S-adenosyl homocysteine (K i about 2 ϫ 10 Ϫ6 M with 10 Ϫ6 M AdoMet and 10 Ϫ4 M dichloro-THPH; not shown).
The methyltransferase is barely detectable in lysates from growing cells and remains at low levels through aggregation before increasing strongly from 8 h of development in parallel with DIF-1 accumulation and in advance of expression of DIF-1 dechlorinase, a marker that is induced by DIF-1 (Fig. 6).
Detection of a Chlorinating Activity in Cell Lysates-A chlorinating activity was detected in slug cell lysates using 36 Cl Ϫ as the radioactive tracer and testing for substrate-dependent incorporation in much the same way as for the methyltransferase. Fig. 7 shows that THPH supports the formation of a chlorinated product, whose production is greatly stimulated by 50 mM H 2 O 2 . Such high concentrations of H 2 O 2 may be necessary due to the presence of a strong catalase activity in these lysates (48).
The chlorinated compound co-elutes with monochloro-THPH on two different TLC systems, and at lower substrate concentrations it can be further chlorinated to give a compound coeluting with dichloro-THPH (not shown). Indeed, with some lysates, there appears to be methylation of this compound as well, resulting in the production of DIF-1 (not shown).
The chlorinating activity has essential components in both the cytosolic and the organelle/membrane fractions of the cell; activity disappears from both fractions when they are separated by high speed centrifugation of a crude lysate but can be restored when they are re-mixed (Fig. 7). The nature of these separate components is not understood at the moment. The activity was therefore characterized in crude lysates, where it was quite unstable, and with the added handicap of a relatively insensitive assay, due again to the low specific activity of 36 Cl Ϫ . However, within these limitations it is clear that the chlorinat-ing enzyme is highly specific; THPH and monochloro-THPH are the preferred substrates, and there is no detectable activity against 4-methoxy-THPH, nor against phenols such as phloroglucinol or 5-methoxyresorcinol (Fig. 7). The K m for THPH is 2 ϫ 10 Ϫ6 M or lower with 50 mM H 2 O 2 . This substrate specificity is the converse of that of the methyltransferase, supporting the idea that the putative polyketide is first chlorinated and then methylated. Chlorinating activity is very low in vegetative cells (but definitely detectable) and rises at the end of aggregation along with dichloro-THPH methyltransferase (Fig. 6).
DIF-1 Biosynthesis in Mutant Strains-The "DIFless" mutant strain HM44, which arrests in development as a tight mound, has been widely used to examine the effects of DIF-1 on gene expression because it makes very little DIF-1 but remains fully responsive to it (4). DIF-1 biosynthesis by this mutant was therefore investigated. It only accumulates about 2.5% as much  DIF-1 as its parental strain, HM27, in submerged culture (Table II), confirming the original observations. Although DIF-1 production is stimulated by THPH, it is still less than 10% of wild type. Both the chlorinating and methylating enzymes are greatly reduced in lysates from HM44, compared with HM27, at the time of DIF-1 synthesis. Similar results were obtained with two other DIFless strains, HM42 and HM43, which have a less severe phenotype than HM44 (not shown). These results are consistent with the methylating and chlorinating enzymes being involved in DIF-1 biosynthesis, since their activities are greatly reduced in the mutants. However, since both activities are affected, it seems likely that the underlying mutations in these strains are regulatory and not in the structural genes of the biosynthetic enzymes. DISCUSSION The experiments described in this paper make a cumulative argument that DIF-1 is synthesized by the pathway shown in Fig. 8. In this pathway, a C 12 polyketide skeleton is first assembled by a polyketide synthase and then chlorinated and methylated by specific enzymes. The biosynthesis of a minor stalk cell-inducing activity, DIF-2, which has a C 4 alkyl side chain instead of the C 5 of DIF-1 (49), can also be accounted for by the same pathway, if it is assumed that a propionyl group is incorporated into the polyketide instead of two acetyls, resulting in one fewer carbon atom.
The polyketide origin of DIF-1 was first suggested by its structure, with an aromatic ring and alternating oxy-substitutions (50), and is strongly supported by the inhibition of DIF-1 biosynthesis by cerulenin, a general polyketide synthase inhibitor. The conversion of the proposed polyketide (THPH) into DIF-1 by living cells further supports this view. Recent small scale DNA sequencing projects have identified Dictyostelium genes with good homology to a polyketide condensing enzyme and to a reductase, indicating that Dictyostelium probably does possess polyketide synthases (pksA and pksB), 4 although it is not known if these particular ones are involved in DIF-1 biosynthesis.
It is assumed in Fig. 8 that acetate is the starter for the polyketide, which is then extended by the addition of another five acetate groups from malonyl CoA. Two carbonyls would be fully reduced to form the alkanone tail of THPH, and there would be ring closure to form the aromatic ring. An alternative possibility, based on the precedent of aflatoxin biosynthesis, is that the starter is a hexanoyl group that eventually forms the alkanone tail of THPH and is itself made by a specialized fatty acid synthase (51,52). This starter would then be extended by three more acetate units to form the aromatic ring of THPH.
Both the methylating and chlorinating activities appear in cell lysates at the expected time of development, and their levels are greatly reduced in the HM44 mutant, which makes 4 W. F. Loomis, personal communication.

TABLE II
Characterization of DIF-1 production by strain HM44, a DIFless mutant, and its parent HM44 is a mutant strain that only accumulates a low percentage of wild-type DIF-1 levels, arresting development as a mound; its parent is strain HM27. DIF-1 production was determined by incubating cells for 16 h in submerged monolayers, with 5 mM cAMP, 0.1 Ci/ml 36 Cl-and 1 M THPH as indicated, exactly as described under "Materials and Methods" and shown in Fig. 4. THPH chlorinase and dichloro-THPH methyltransferase were assayed in whole cell lysates or their high speed supernatants, respectively. Cells were allowed to develop on KK2 agar for the indicated times before samples were frozen for later assay. Each value is the mean of duplicate samples. little DIF-1 (4). This and their substrate specificities leave little doubt that these enzymes are dedicated to converting the polyketide THPH to DIF-1. The chlorinating activity utilizes THPH or monochloro-THPH directly, but the biochemistry of this activity is still ill defined. It appears more complex than other chlorinating enzyme described to date, which are simple soluble enzymes, utilizing H 2 O 2 as oxidant (27,28,53,54) and which do not have the essential soluble and membranous/ organellar components of the Dictyostelium activity. The methyltransferase uses AdoMet as methyl donor at about physiological concentrations (47) and has a substrate specificity converse to that of the chlorinating activity; it does not utilize the polyketide directly but requires it to be chlorinated first. It is also specific for the 4-hydroxyl over the 2,6hydroxyls of the aromatic ring.
In principle, DIF-1 could be made from THPH either by chlorination followed by methylation or by these reactions in the reverse order. The results show that chlorination must precede methylation in the pathway. Cells will accept THPH or monochloro-THPH for chlorination, but not the methylated version of either compound; the chlorinating activity of cell lysates has precisely the same specificity. Conversely, the methyltransferase will not use THPH but requires it to be chlorinated first.
This work lays the foundation for further elucidating the role of DIF-1 in Dictyostelium development. The most important next step is to identify mutants that are defective in DIF-1 biosynthesis and determine the consequences of this defect for development. The proposed biosynthetic pathway requires a minimum of about nine activities, any of which is a potential mutational target: condensing enzyme, acyl carrier protein, two acyltransferases, ketoreductase, dehydratase, and enoylreductase of the polyketide synthase plus the chlorinating and methyltransferase enzymes. It should now be possible to use the biochemical assays and feeding experiments to identify DIF-1 biosynthetic mutants among existing collections. Alternatively, mutants could be created by reverse genetics if potential biosynthetic genes are identified in the DNA sequence data bases or if one of the biosynthetic enzymes can be purified and cloned. The biochemical assays can also be used to learn more of how DIF-1 production is regulated and which cells make it. Finally, cerulenin provides a new tool for inhibiting DIF-1 biosynthesis.