Precursor Supply for Insect Juvenile Hormone I11 Biosynthesis in a Cockroach*

The biosynthesis of the sesquiterpenoid juvenile hor- mone 111 (JH 111) was studied using corpora allata of the cockroach Diploptera punctata incubated in vitro and a radiochemical assay for the hormone produced. The influence of several exogenous precursors such as glucose, trehalose, acetate, amino acids, and mevalon- ate on JH synthetic rates was studied. Glucose or trehalose were needed for an optimal rate of JH synthesis. Highest rates were achieved at trehalose concentrations below the normal hemolymph levels (35-40 mM). About one-third of the glucose utilized for the biosynthesis of JH 111 was metabolized through a pentose pathway, but acetyl-coA derived from glucose was significantly diluted by acetyl-coA from other sources. Amino acids provided both a source of carbon for JH 111 synthesis and a source of energy that allowed JH I11 synthesis from acetate and stimulated JH I11 syn- thesis from glucose. Acetate was a poor substrate, because it could not support JH I11 synthesis in long term incubations. The incorporation of exogenous mevalonate into JH I11 was dependent on the physiological state of the glands, but there was a significant dilution with endogenous mevalonate. This dilution reflected in part the poor penetration of mevalonate into the corpora allata cells, because JH synthesis in mevinolin-treated cells was not fully rescued by mevalonate.

spectrometry also showed the presence of only J H I11 (3, 4). Schooley et al. (5) first demonstrated that radioactivity from [2-14C]acetate and [2-'4C]mevalonate was incorporated into J H I11 biosynthesized by corpora allata of Manduca serta incubated in uitro, and further studies confirmed that JH 111 was synthesized according to a "classical" terpenoid pathway (1,6,7 ) . In addition to JH 111, the corpora allata from lepidopteran species produce higher homologs, such as J H I1 and JH I, in which one or two isoprene units have been replaced by a homoisoprene unit derived from 3'-homomevalonate (1,6,7 ) . However, despite such structural information, virtually nothing is known about the mechanisms controlling both the nature of the J H produced and its rate of production. A better understanding of the demands of the corpora allata cells for the necessary precursors and their utilization for J H synthesis is needed. In this paper, we have studied some aspects of precursor supply for J H I11 synthesis in female adults of D. punctata. This insect is particularly suitable for this type of study, because its corpora allata synthesize large amounts of J H I11 in vitro (3,4,8). Throughout our experiments, we have taken advantage of an in uitro radiochemical assay for J H synthesis, which is based on the stoichiometric incorporation of label from [ methyl-14C] -or [methyl-3H]methionine into the methyl ester moiety of the biosynthesized juvenile hormone (2, 7,9). To complement this known requirement of exogenous methionine for synthesis of the methyl ester group of JH, our aim in this paper was to clarify the metabolic source of carbon utilized for de nouo synthesis of the sesquiterpenoid carbon chain of J H 111.

EXPERIMENTAL PROCEDURES
Insects-The viviparous cockroach D. punctata was reared as described previously (10). The activity of the corpora allata in vitro is strictly dependent on the physiological state of the adult insect (3,9), and thus day-5 mated female adults were chosen to provide high activity corpora allata, and day-9 mated females as well as day-5 virgin females provided low activity glands.  [6-14C]glucose (59 mCi/mmol) were obtained from Amersham Corp. Labeled and unlabeled mevalonolactone (from Sigma) were converted to mevalonate by titration with 1 N NaOH (11). Mevinolin was a gift from A. W. Alberts (Merck Institute for Therapeutic Research, Rshway, NJ) and was used as its sodium salt.

Materia&-~-[methyl-~H]Methionine
Incubation Medium-In some experiments, the incubation medium used was medium 199 (Gibco) with Hank's salts, L-glutamine, 25 mM HEPES at pH 7.2, and 2% Ficoll (Sigma). This medium contained 0.61 mM sodium acetate and 5.5 mM glucose and has been designated as "medium 199" throughout this paper. In other experiments, the medium was prepared from Gibco's MEM Select-amine kit. Minimal medium consisted of Hanks salts, phenol red, and 25 mM HEPES at pH 7.2, MEM vitamin mixture, and 2% Ficoll. The amino acids were Arg, Cys, Glu, His, Ile, Leu, Lys, Phe, Thr, Trp, Tyr, and Val at concentrations recommended in Gibco's MEM. When so indicated, glucose and/or sodium acetate were added at 5.5 and 0.61 mM, respectively. Minimal medium with amino acids and glucose has been designated as "complete M E M in this paper. Both medium 199 and complete MEM lacked L-methionine which was provided either as [methyl-"Clmethionine or as [methyl-3H]methionine at a final concentration of 0.05 mM. Preliminary experiments showed that the rate of juvenile hormone synthesis was equivalent in medium 199 and complete MEM.
Assay of Juvenile Hormone Synthesis-Dissection and incubation of the corpora allata in uitro was done essentially as described (12), except that dissection was done in Hank's salts rather than in Yeager's saline. Following incubation of the corpora allata in medium containing the appropriate substrates and methyl-labeled methionine, J H I11 synthesis was measured by an isooctane partition assay (12). Because the glands are removed from the incubation medium before extraction, the assay measured J H released into the medium rather than total J H synthesized. However, there is a linear correlation between release and synthesis over the whole range of activity of the corpora allata (3,9) and we have therefore designated the results as "synthesis" for the sake of clarity. In all double-label experiments the isooctane extract of the incubation medium was analyzed by thinlayer chromatography on plastic-backed silica plates in the solvent system hexane/ethyl acetate (3:1, v/v) and the J H I11 zone assayed by liquid scintillation spectrometry as described previously (12).
High Performance Liquid Chromatography-Analysis of the secretion products of the corpora allata by HPLC was done using a Perkin-Elmer model 400 liquid chromatograph, with a Rheodyne 7125 injector, Perkin-Elmer model LC95 variable wavelength UV detector (4.5-p1 flow cell), and LC1 100 integrator/plotter. Reversed-phase HPLC used a Supelcosil LC-18 column (250 X 4.6 mm, 5-pm particle size) and the solvent system acetonitrile/water (7030) at 1.5 ml/min. Normal phase HPLC used a supelcosil LC-Si column (150 X 4.6 mm, 5-pm particle size) with a solvent system hexane (propanediol saturated)/ether (96.4) at 1 ml/min. Reference J H I, 11, and 111 were obtained from Behring Diagnostics.
Molar Incorporation Ratio and Dilution Factor Calculations-For the incorporation of "C label from glucose, the following formula was used for the molar incorporation ratio (MIR), taking advantage of the stoichiometric incorporation of label from [methyl-3H]methionine: dpm[14C]/Glc specific activity dpm [3H]/Met specific activity = MIR[lclclc Because [1-"C]glucose can generate a labeled acetyl-coA only when metabolized through a glycolytic pathway, and assuming that the 2 triose phosphates generated by glycolysis are equivalent and that all acetyl-coA derived from glucose are generated through either glycolysis or a pentose pathway, then the fractions of glucose utilization through glycolysis (G) and pentose pathway (P) are given, according to Katz and Wood (13) by the following relations: The dilution factor of acetyl-CQA derived from exogenous glucose by acetyl-coA generated from other sources can then be calculated knowing that 9 acetyl-coA are needed for 1 J H I11 produced and that there are (2G + P) acetyl-coA generated for each molecule of exogenous glucose utilized. Thus, the dilution factor is: For the incorporation of mevalonate, the molar incorporation ratio was calculated using the specific activities of mevalonate and methionine as described above for glucose. Because the mevalonate used was labeled with 3H in the C5 position, the maximal incorporation ratio possible would be 2.0 (oxidation of farnesyl pyrophosphate to farnesoate would remove the 3H label from 1 of the 3 isoprene units). Thus, the dilution factor is: (2 -M I R [~~I~~~~~~) / M I R~~~J~~~~~~~.

Influence of Glucose and Trehalose Concentration on JH
Biosynthesis-Corpora allata from day-5 mated females (high activity glands) were incubated in complete MEM containing trehalose or glucose but no acetate. The rates of J H synthesis in short term incubations were compared to control rates in complete MEM. Fig. 1 shows that optimal rates of J H synthesis were obtained a t 5.5 mM glucose. No stimulation of J H synthesis was observed with higher concentrations, even a t 100 mM glucose. With low activity glands (from day-5 virgin females), no significant difference was found between 5.5 and 100 mM.
Because trehalose, not glucose, is the major carbohydrate circulating in insect hemolymph, it was necessary to test the influence of trehalose on J H synthesis in vitro in a similar way. Fig. 2 shows that essentially identical results were obtained with trehalose in the absence of glucose and acetate. These results imply that the cells of the corpora allata contain sufficient trehalase activity to generate the levels of glucose needed for J H synthesis.
In addition, the hemolymph trehalose levels in D. punctata were found to be well within the range of trehalose concentrations that were optimal for J H synthesis in uitro: Table I shows that hemolymph trehalose levels were remarkably constant between 35 and 40 mM. These results indicate that glucose levels are not limiting in standard incubations of corpora allata and that the difference in the rate of J H synthesis between high and low activity glands is not caused by a limited supply of trehalose. The results also justify the use of glucose instead of trehalose in in vitro assays of J H synthesis.
Pathways of Glucose Utilization-Corpora allata from day-5 mated females were incubated with either [l-'4C]glucose or

TABLE I1 Incorporation of radioactivity from [l-'4C]-and [6-'4CJglucose into JH III
In each experiment, two pairs of corpora allata from day-5 mated females were incubated in MEM medium with amino acids and 0.  [rr~ethyl-~HIMethionine was used as mass marker (1, 2) for the J H 111 biosynthesized. Table I1 shows that incorporation from glucose labeled in either position increased with incubation time, as the endogenous pool of carbohydrate was replaced with labeled glucose from the incubation medium. The dilution of acetyl-coA derived from exogenous glucose by endogenous acetyl-coA indeed decreased with time from a factor of 2.75 after 3 h to 1.84 after 11 h. The 14C/3H molar incorporation ratio from [6-'4C]glucose was significantly higher than that from [l-14C]glucose, even though the levels of J H I11 produced were similar in the two sets of experiments. This result showed that a large portion of glucose was metabolized through a pentose pathway. Assuming that the 2 triose phosphates generated by glycolysis were equivalent and that other pathways were not involved (these assumptions and their limitations have been discussed; 13, 14), we calculated that 29.8% of the glucose utilized for J H synthesis (as acetyl-CoA) was metabolized through a pentose pathway. Table I1 also shows that the addition of acetate (0.61 mM) slightly decreased the incorporation of label from glucose into J H 111. We calculated that the pentose pathway contribution remained, however, constant at 28.4%. This suggested that exogenous acetate could compete with glucose as source of the C2 units needed for J H I11 synthesis.
Acetate as Precursor for J H Biosynthesis-The experiment described above prompted us to examine the role of acetate in in vitro incubations. Acetate tested at concentrations up to 6.1 mM in the absence of glucose or trehalose could not sustain J H I11 synthesis at rates higher than 59.5 rt 7.5% of controls (complete MEM) in short term incubations (results not shown). This suggested that although corpora allata cells have some acetyl-coA synthetase activity, the efficient utilization of exogenous acetate is dependent on a source of ATP. Acetoacetate at concentrations up to 6.1 mM did not stimulate the rate of J H synthesis over the low rate observed in the absence of a carbon source (other than methionine). This suggested that acetoacetyl-CoA hydrolase or HMG-CoA lyase activities in corpora allata cells, if present, were virtually inactive in the direction of acetoacetate utilization (results not shown).
In long term experiments, corpora allata from day-5 mated females (high synthetic rate) were first incubated in complete MEM for 2 1-h incubation periods in order to establish the control rate of J H synthesis. The glands were then transferred to various media for up to 30 h and rates of JH synthesis were monitored at regular intervals (Fig. 3). In medium lacking carbon sources (other than methionine), J H synthesis drastically declined during the first hours and virtually stopped after 12 h. With acetate as sole carbon source, J H synthesis also stopped, although the glands initially utilized some acetate for J H synthesis. The addition of acetate to complete MEM did not enhance the rate of J H synthesis over the long term, although an initial and short term stimulation (25% for 3 h) was observed. These experiments show that the corpora allata do not normally utilize acetate as a precursor for J H synthesis.
Amino Acids and JH Biosynthesis-The utilization of amino acids for J H synthesis was studied in long term incubations as described above for acetate (Figs. 3 and 4). With amino acids as the only carbon source, J H synthesis reached a steady state which was about one-tenth the control rate of synthesis. When both amino acids and acetate were supplied, J H synthesis stabilized at about one-third the rate of controls, thus indicating that the metabolism of amino acids generated sufficient energy for the utilization of acetate. Glucose as the only source of carbon was slightly less effective than in combination with amino acids and could maintain a rate of only about 80% of the controls. These results suggested that the corpora allata utilized amino acids both as a source of carbon (presumably HMG-CoA, acetoacetyl-CoA, and acetyl-CoA) and as a source of energy.   Fig. 3. The incubation media used were: minimal medium and Glc + AA as described in the legend to Fig. 3; AA, minimal medium + amino acids (see "Experimental Procedures" for composition); Glc, minimal medium + 5.5 mM glucose. The J H synthesized was extracted by isooctane partition, dried under a stream of nitrogen, taken up in acetonitrile, and an aliquot was analyzed by reversed-phase HPLC (4.6 X 250-mm Supelcosil LC18 column, solvent system acetonitrile/water (7030) at 1.5 ml/min). Fractions were collected every 24 s between 2 and 8 min and assayed for radioactivity by liquid scintillation spectrometry, The pattern of elution (UV absorbance at 225 nm) is also shown at the top of the figure and the major peak corresponds to the J H I11 synthesized.
In order to provide direct evidence of the metabolism of amino acids to utilizable J H I11 precursor(s), corpora allata were incubated in medium 199 in the presence of [4,5-3H] leucine. The JH I11 synthesized was analyzed by silica TLC and reversed-phase HPLC (Fig. 5) and both methods revealed that radioactivity from the amino acid had been incorporated into J H 111, although no determination of the incorporation rate was attempted.
Corpora allata incubated in complete MEM and 0.8 mM cycloheximide continued to produce J H 111 at a normal rate for at least 12 h (Fig. 6), and control experiments showed that this level of cycloheximide stopped incorporation of [~,EJ-~H] leucine into trichloroacetic acid-precipitable proteins in short term experiments. Fig. 6 also shows that corpora allata that stopped synthesizing J H because of a lack of adequate precursor supply almost immediately produced J H when transferred back to complete MEM. In fact a slight overshoot of J H synthesis was observed for the first 2 h following return to a normal medium, but the rate of J H synthesis then leveled off to the original control rate. These experiments indicated that the difference in synthetic rates between media contain- ing glucose and glucose + amino acids (Fig. 4) was not due to a possible decrease of an essential protein with a rapid turnover caused by the lack of amino acid precursors. Rather, this difference was due to the function of amino acids as energy and/or carbon sources.
Mevalonate Incorporation into JH ZZZ- Fig. 7 shows that, when present in the incubation medium at 10 mM, mevalonate stimulated J H I11 biosynthesis in the corpora allata from day-9 mated females, but not from day-5 mated or virgin females.  ratio calculated using the specific activities of each substrate was compared for glands incubated in the presence or absence of 1 PM mevinolin. Because mevinolin is a potent inhibitor of HMG-CoA reductase and J H synthesis in D. punctata corpora allata: it was expected to reduce the levels of endogenous mevalonate. Table 111 shows that the increase in mevalonate incorporation caused by mevinolin was similar for the three groups of corpora allata and virtually all J H I11 synthesized in the presence of mevinolin was derived from exogenous mevalonate. However, the level of inhibition was still substantial (about 50% for the three groups of glands) when compared to inhibition rates observed in the absence of exogenous mevalonate (typically 88% over a 3-h incubation period).' Thus we concluded that exogenous mevalonate, even at 10 mM, could not restore an adequate level of mevalonate within the cells of mevinolin-treated glands.
In the absence of mevinolin, the rates of mevalonate incorporation were lower, but the molar incorporation ratio was significantly higher for day-9 mated females when compared to day-5 mated or virgin females. Taken together, those results suggested that the lack of J H stimulation by mevalonate in corpora allata from day-5 mated females was not caused by a different ( i e . even lower) rate of mevalonate penetration. This would have been reflected in mevinolin-treated glands by a higher rate of inhibition of J H synthesis and/or a lower incorporation ratio.

DISCUSSION
The structural aspects of J H biosynthesis are relatively well understood (1). J H I11 appears to be synthesized from acetyl-CoA through the classical isoprenoid pathway to farnesyl pyrophosphate. These early steps are shared with other pathways, such as the biosynthesis of cholesterol, ubiquinone, dolichols, etc. The later steps, oxidation of farnesyl pyrophosphate to farnesoate, methylation, and epoxidation, are specific to the JH 111 pathway. However, the origin of acetyl-coA has not been addressed experimentally, nor have the quantitative aspects of J H synthesis been pursued in detail. There are theoretically three sources of readily available hemolymph precursors capable of generating acetyl-coA for J H I11 biosynthesis: carbohydrates, i.e. mainly trehalose, free amino acids present in high concentrations in insect hemolymph, * R. Feyereisen and D. E. Farnsworth, unpublished results. and lipids carried by the lipoprotein lipophorin.
Matthews et al. (15) have determined that the glucose concentration in the hemolymph of the cockroach Periplaneta americana is about 0.15 mM and this has to be compared to trehalose levels of 45-53 mM in that cockroach (14), and 35-40 mM determined in this study. Obviously trehalose is the major carbohydrate available to the corpora allata. Because the dose-response for glucose and trehalose were comparable and beacuse there was no significant difference in the response of corpora allata of low and high activity, we conclude that the levels of carbohydrate available and the trehalase activity of the gland cells are not limiting factors for J H I11 synthesis. Radioactivity from [ l-'4C]glucose is incorporated into methyl farnesoate and J H I11 (10) and glycolytic degradation of glucose to acetyl-coA was thus presumed to occur in D. punctata corpora allata. However, the possible involvement of a pentose pathway needed to be tested in view of the demand for NADPH for J H synthesis, in particular by HMG-CoA reductase and methyl farnesoate epoxidase (17). We have shown here that about one-third of the glucose degraded to acetyl-coA for J H I11 synthesis was metabolized through a pentose pathway. These calculations of pentose pathway contribution are subject to a number of assumptions (13,14) but it nonetheless appears that the pentose phosphate pathway is very important in corpora allata cells. Its contribution to glucose metabolism is higher in corpora allata than in most vertebrate tissues including rat adipose tissue that is specialized in lipid synthesis (14). It is still unknown whether the share of total glucose metabolism contributed by the glycolytic and nonglycolytic pathways is dependent on glucose concentration or on the physiological age of the glands. We have also shown that glucose, when present alone, was not able to maintain an optimal rate of J H I11 synthesis over a long term incubation. In fact, considerable dilution of glucose-derived acetyl-coA was observed. An optimal rate was obtained only in the presence of glucose and amino acids.
These observations suggested that amino acids might provide a substantial source of carbon for JH 111 synthesis.
Incorporation of radioactivity from [4,5-3H]leucine into J H I11 and the demonstration that amino acids sustained a low, but significant level of J H synthesis support this view. Conversion of some amino acids to acetyl-coA (directly or through pyruvate), acetoacetyl-CoA, and 3-hydroxy-3-methylglutaryl-CoA (leucine) that can be used in JH I11 biosynthesis thus appears to be an important aspect of corpus allatum biochemistry that is linked to the oxidative degradation of amino acids.
Our study does not address the possible role of lipids as a source of energy or as an acetyl-coA precursor. The available lipid source in the hemolymph is a lipoprotein called lipophorin. Cockroach lipophorin carries about 50% of its weight as lipids, mainly as hydrocarbons, diacylglycerol, and phosphatidylcholine (18). In preliminary experiments, we have purified lipophorin from D. punctata using a KBr gradient flotation technique (19). This lipophorin preparation did not stimulate JH synthesis by corpora allata incubated in the absence of glucose and amino acids. Equally unsuccessful were incubations with palmitic acid, dipalmitin, or dipalmitoylphosphatidylcholine.' Future experiments with a lipophorin preparation loaded with radiolabeled lipids should more clearly establish whether lipophorin can provide precursors of acetyl-coA for J H I11 synthesis.
In addition to carbohydrates, amino acids (and lipids?) as sources of "early" precursors or intermediates, plant-derived farnesol may be found in the hemolymph of phytophagous species. Farnesol has been shown to stimulate J H I11 biosyn-thesis by D. punctata corpora allata in uitro (8), but it is likely that farnesol from plant sources would be carried in the hemolymph by lipophorin, and that the concentrations achieved would be to low to affect J H I11 synthesis in uiuo.
Exogenous acetate has been recognized as a precursor for J H biosynthesis in in vitro systems (l), and stoichiometric incorporation of [2-14C]acetate into J H I11 could be observed in the absence of any other exogenous carbon source, although rates of J H I11 synthesis sustained in those short term experiments were abnormally low (10). Here we show that acetate should be considered as an artificial substrate capable of entering the J H I11 biosynthetic pathway in uitro. Acetyl-coA synthetase although present in the corpora allata, did not appear to play an important role in J H I11 biosynthesis under normal, in uiuo conditions. Indeed, acetate alone did not support J H I11 synthesis beyond a few hours in uitro and even in the presence of adequate levels of glucose, acetate did not stimulate the rate of J H synthesis over long term incubations.
Just as acetate has been used as an acetyl-coA precursor, propionate has been used as propionyl-CoA precursor for synthesis of the higher homologs of JH I11 in Lepidoptera corpora allata incubations (1). The dilution of [l-14C]propionate upon incorporaton into J H I1 by Manduca sexta corpora allata was very low in the experiments reported by Schooley et al. (20). The reasons for this low dilution have not been studied, but in view of the results obtained with free acetate, we think that it is not free propionate but rather propionyl-CoA that is the normal substrate for synthesis of J H I1 and J H I in Lepidoptera.
Our studies on the precursor supply for J H I11 synthesis in the phylogenetically ancient cockroaches are relevant to the evolution of J H biosynthesis. The demonstration of the importance of amino acid metabolism in JH I11 synthesis suggests that these pathways have been conserved in insects and specialized in Lepidoptera. Indeed the metabolism of threonine, isoleucine, methionine, and valine may lead in several steps to propionyl-CoA. Another potential source of propionyl-CoA, such as succinyl-CoA via methylmalonyl-CoA is probably minor if operating at all, in view of the very low levels in insects of vitamin BIZ, the cofactor of methylmalonyl-CoA mutase (21,221. Exogenous mevalonate has been shown to serve as J H I11 precursor in the corpora allata of several species including D. punctata (1,8). Stimulation of low activity glands and lack of stimulation of high activity glands by exogenous mevalonate was observed previously (8) and has been confirmed, in part, by the present study. We show here that low activity corpora allata from day-5 virgin females were not stimulated by mevalonate, whereas a significant stimulation was observed with low activity glands from day-9 mated females. The stimulation of J H I11 synthesis by exogenous precursors such as farnesol and farnesoate gives useful insights into the regulation of the corpora allata (1,2,7) because it is indicative of the degree of saturation of the enzymes (dehydrogenases, methyl transferase, and epoxidase) situated beyond the entry of those precursors. However, the use of exogenous mevalonate as a probe of the physiology of the corpora allata is made difficult by the limited penetration of mevalonate into the gland cells. We showed that mevalonate was not able to restore J H I11 synthesis to normal levels even when it was virtually the sole precursor available for synthesis (i.e. when endogenous mevalonate synthesis was blocked by the HMG-CoA reductase inhibitor mevinolin). In spite of this low penetration of mevalonate, J H synthesis is stimulated in corpora allata from day-9 insects. Thus, the enzymes of J H biosynthesis situated after HMG-CoA reductase have a lower overall elasticity coefficient (23) with respect to mevalonate in day-5 mated or virgin females when compared to day-9 mated females.