Journal of Chemical Ecology, Vol. 19, No. 3, 1993
PARTIAL CHARACTERIZATION A N D HPLC ISOLATION OF BACTERIA-PRODUCED ATTRACTANTS FOR THE MEXICAN FRUIT FLY, Anastrepha ludens 1
D.C.
ROBACKER,
2'* W.C.
WARFIELD,
2 and R.F.
ALBACH 3
2Crop Quality and Fruit Insects Research, ARS, USDA 3Conservation and Production Systems Research, ARS, USDA 2301 South International Blvd. Weslaco, Texas 78596 (Received September 21, 1992; accepted November 3, 1992)
Abstract--Methods were developed to collect and isolate volatile chemicals produced by a Staphylococcus bacterium in tryptic soy culture that are attractive to protein-hungry adult Mexican fruit flies. Centrifugation of bacteria culture yielded a slightly attractive pellet containing most of the bacteria cells and a highly attractive supernatant. Supernatant filtered to remove the remaining bacteria was as attractive as the unfiltered supematant. Filtrate at pH 7 and above was much more attractive than filtrate at pH 5 and below. Most of the attractiveness was retained on strong cation exchange media under acidic conditions and eluted with base. Attractive principles could not be trapped on adsorbents such as Porapak Q or extracted with organic solvents from aqueous preparations, but they were easily collected by headspace sweeping with steam. The attractive components were efficiently concentrated by rotary evaporation of steam distillate at pH 5, but at higher pH much of the attractiveness distilled. A reverse-phase HPLC method using a negative counter-ion was developed to separate and collect attractive components of concentrated steam distillate. Attractive fractions collected using this method were concentrated and injected onto silica HPLC. Activity eluted from silica in two distinct bands. Results suggest that the most attractive components of the bacterial odor are highly polar, low-molecular-weight amines. Key Words--Attractants, Mexican fruit fly, Diptera, Tephritidae, Anastrepha ludens, bacteria, amines.
*To whom correspondence should be addressed. Diptera: Tephritidae.
543 0098-0331/93/0300-0543507.00/0 (c) 1993 Plenum Publishing Corporation
544
ROBACKER ET AL.
INTRODUCTION
Attractiveness of bacteria to fruit flies has been investigated for 40 years (Gow, 1954), but few of the chemicals responsible for the attractiveness have been identified. Ammonia is one by-product of bacterial action that has been identified. It has been used in some capacity as a fruit fly attractant since at least the 1930s (Jarvis, 1931; Hodson, 1943), but much disagreement exists concerning its attractiveness. Several authors have stated that attractiveness of bacteria and protein baits is due primarily to components other than ammonia (Gow, 1954; Mazor et al., 1987; Drew and Fay, 1988). Other studies have indicated that ammonia alone or in protein baits is an effective attractant if its release rate is within the optimum range (Bateman and Morton, 1981; Morton and Bateman, 1981; Wakabayashi and Cunningham, 1991). More recently, 2-butanone and butanol were identified from volatiles produced by Proteus species bacteria (Hayward et al., 1977) and shown to be attractive to Dacus tryoni (Froggatt) (Drew, 1987). Generally, other volatiles identified from bacteria (Hayward et al., 1977) or protein baits (Morton and Bateman, 1981; Buttery et al., 1983) have not been attractive. The general lack of success in identifying the probable attractive principles of bacteria and protein baits suggests that the problem is difficult and may require unusual approaches. Robacker et al. (1991a) isolated a bacterium (identified as Staphylococcus aureus) from the mouthparts of a female Mexican fruit fly (Anastrepha ludens Loew) that was highly attractive to adult Mexican fruit flies in laboratory experiments. That study and a subsequent one (Robacker, 1991) presented evidence that the attraction response of the flies to bacterial odor was motivated by hunger for protein. The attractant chemicals produced by the bacteria were not identified. This paper reports chemical properties and methods for isolation of the attractive principles of the odor of the Staphylococcus species (RGM-1) identified by Robacker et al. (1991a). Evidence is presented that the most important attractants are low-molecule-weight amines that cannot be handled by standard methods for collecting and isolating insect pheromones and other attractants.
METHODS AND MATERIALS
Insects and Test Conditions. Flies were from a culture that had been maintained on laboratory diet for about 80-90 generations with no wild-fly introductions. Recent experiments indicated the culture flies were as vigorous as wild flies in mating competitiveness (Moreno et al., 1991) and had retained much of their natural courtship behavior (Robacker et al., 1991b). Mixed-sex groups of 180-200 flies were kept in 473-ml cardboard cartons with screen tops until used
BACTERIA-PRODUCED ATTRACTANTS
545
in tests. Flies were tested when 6-12 days old. Flies were deprived of protein as adults but were fed sucrose and water up until the time of attractiveness testing. All tests were conducted in the laboratory between 0830 and 1430 hr under a combination of fluorescent and natural light. Laboratory conditions were 22 _+ 2~ 50 +_ 20% relative humidity, and photophase from 0630 to t930 hr. Bacterial Preparations. Bacterial strain RGM-1 (Robacker et al., 1991a) was cultured in tryptic soy broth (Difco Laboratories, Detroit, Michigan) in a shaker for 144 hr at 30~ Numerous 100-ml samples were prepared as needed for the various experiments. Bacterial culture was centrifuged at 10,000 rpm for 20 min and separated into pellet and supernatant. Supernatant was filtered through 0.45-#m type HA aqueous filters (Millipore Corporation, Bedford, Massachusetts) followed by 0.22-t*m type GV aqueous/organic filters (Millipore). The pellet was suspended in water and centrifuged. The new pellet was suspended again in water. The centrifugation and filtration of the supernatant were done to remove bacteria cells. Resuspended pellets (second centrifugation only), supernatants (first centrifugation only), and filtrates were monitored for attractiveness to flies using cage-top bioassays (described below). Test quantities applied to filter papers were 10 #1. Controls were 10-/A quantities of tryptic soy broth. Attractiveness of resuspended pellets was also evaluated against a water control (10 txl). Data were analyzed as paired t tests to compare total counts at treatment papers to total counts at control papers. Cage-Top Bioassay Procedure. Bioassays were conducted by placing four filter paper triangles (3 cm/side), two containing test samples and two containing appropriate controls, near the corners on the top of an inset cage (30 cm/side, aluminum-screened). The numbers of flies beneath each filter paper were counted once each minute for 10 min. The two papers containing test chemicals were positioned diagonally from each other on two corners of the cage top, and the two papers containing controls were positioned diagonally from each other on the other two corners. The filter papers were raised 5 mm above the cage top using plastic tings to ensure that olfaction and not contact chemoreception was solely responsible for the response of the flies. Two cartons of 180-200 flies were used in each bioassay cage. Effects of pH on Attractiveness of Bacterial Filtrate. Bacterial filtrate had a pH of 7.9. The pH of aqueous bacterial filtrate was raised to 8, 9, 10, 11, and 12 with saturated sodium hydroxide and lowered to 2, 3, 4, 5, 6, and 7 with 85 % phosphoric acid. Each pH treatment was bioassayed for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 10 #1. Controls were 10-/xl quantities of water. Bioassays were conducted in a randomized complete block experiment. Ten replications of the experiment were conducted. Data used for analysis of variance were differences between total counts at treatment papers and controls.
546
ROBACKER ET AL.
Retention on Strong Cation Exchange Media. The pH of aqueous bacterial filtrate was lowered to 4.9 with 85 % phosphoric acid. The resulting solution (1 ml) was loaded onto a SCX PrepSep extraction column (Fisher Scientific, Fair Lawn, New Jersey) after preparing the column with successive rinses with 5 ml of 0.01 M monobasic sodium phosphate (NaHzPO4) (pH 4.9) and 1 ml of water. After loading the 1 ml of bacterial filtrate, the column was eluted with successive rinses with 5 ml of 0.01 M monobasic sodium phosphate (pH 4.9), 5 ml of water, and 1 ml of 0.05 M dibasic sodium phosphate (NazHPO4) (pH 9.0). Since passage through the acidic column always lowered the pH of the 1 ml of 0.05 M dibasic sodium phosphate to about 7-8, one drop of saturated sodium hydroxide was added to the eluate and it was again passed through the column. The attractiveness of all eluates was evaluated with cage-top bioassays after raising their pH to 10 or greater with sodium hydroxide. Bioassay test quantities were 10/zl for the void volume eluting during loading of the bacterial filtrate and for both of the 0.05 M dibasic sodium phosphate eluates, and 50/zl for the water eluate and for the 0.01 M monobasic sodium phosphate eluate. Controls were 10- or 50-/A quantities, as appropriate, of water containing one drop of saturated sodium hydroxide per milliliter. Experimental design was a randomized complete block and data analysis was handled as in the previous experiment. Paired t tests also were conducted to compare counts at treatments with counts at controls. Collection of Volatiles from Bacterial Filtrate. Volatiles were collected using several methods. The first series of methods was to put bacterial filtrate into a glass aeration flask and pull air through the flask and subsequently through a trap. Bacterial filtrate was put into the aeration flask either as 250 ml of liquid or as an incompletely dried residue from 50 ml of the liquid on 2500 cm 2 of filter paper. Both forms of the bacterial filtrate were highly attractive to flies. Four different types of traps were used to collect volatiles. Three employed chemical adsorbents in glass tubes: 250 mg of Tenax-GC (60/80) (Alltech Associates, Inc., Deerfield, Illinois); 250 mg of Porapak Q (50/80) (Alltech Associates); or 20 mg of Super Q (Alltech Associates). Procedures and glassware for the collections using Tenax-GC and Porapak Q were similar to those described in Robacker et al. (1990). Apparatus for collections using Super Q was that described by Heath and Manukian (1992). Aerations were for at least 6 hr. The adsorbent traps were cleaned with 100 ml of 5 % acetone in pentane (v/v) before use and extracted after aeration with successive rinses of 1 ml each of 5% acetone in pentane, acetone, and water to obtain bacteria volatiles. The fourth was a cold trap in which the odor-laden air was pulled through a glass flask immersed in a Dry Ice-acetone bath. Volatiles were obtained from the collection flask in the ice that formed during the aeration. All trap extracts were evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter paper were 100 /~1. Controls were 100 tzl of
BACTERIA-PRODUCED ATTRACTANTS
547
solvents used for extraction (chemical adsorbents collections) or water (cold trap collection). The final volatile collection method was a headspace sweep with steam. Steam was generated from vigorous boiling of Milli-Q (Millipore) grade water in a two-neck 500-ml flask. The steam was vented into a second two-neck 500-ml flask containing 100 ml of bacterial filtrate heated gently. The steam entered the second flask about 5 cm above the liquid level so as to prevent excessive foaming of the bacterial filtrate. The steam distillate was collected into a 500-ml flask following condensation in a cold-water condenser. The procedure was carried out until 400 ml of steam distillate was collected. Both the steam distillate and the bacterial filtrate residue (the portion of bacterial filtrate that was not swept into the condenser) were evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 10 tzl. Controls were 10 tzl of water. Paired t tests were used to compare counts at treatments with counts at controls. A Student's t test was used to compare responses to steam distillate with responses to bacterial filtrate residue. Data for this test were differences between counts at treatment papers and controls. Solvent-Solvent Extraction. Steam distillate and bacterial filtrate were solvent-solvent extracted in separate experiments with hexane, dichloromethane, and ethyl ether. Quantities of the organic solvents were 10-fold greater than the aqueous distillate or filtrate. Ten-microliter quantities of the aqueous and 100 #1 of the organic phases were tested for attractiveness to flies using cage-top bioassays. Controls were 10-/xl quantities of water for testing aqueous phases and 100/A of appropriate organic solvent for testing the organic phases. Concentration of Bacteria-Produced Attractants. Steam distillate was concentrated using three methods. The first was lyophilization using a Speed Vac concentrator (Savant Instruments, Inc., Farmingdale, New York) connected in series to a refrigerated condensation trap (Savant Instruments, Inc.) and a high vacuum pump (Savant Instruments, Inc.). Steam distillate (100 ml) was concentrated to 1 ml. Dilutions of 1:10 and 1:100 with water were made, and 10-#1 samples of the concentrate and the two dilutions were evaluated for attractiveness using cage-top bioassays. Controls were 10 #1 of water. The second method was rotary evaporation using a Rotavapor RE-111 instrument (Buchi-Brinkman Instruments, Inc., Westbury, New York). Rotary evaporation was aided by heating the steam distillate while pulling vacuum on the system. In the first attempt, 250 ml of steam distillate was concentrated to 10 ml. The concentrate, distillate, and 1 : 10 dilutions of each were evaluated for attractiveness using cage-top bioassays. Test amounts were 10 #1 each. As a second attempt, the pH of 250 ml of steam distillate was lowered to 5 with phosphoric acid before rotary evaporation. The concentrate was evaluated when it reached a volume of 10 ml and again after it was evaporated to dryness and redissolved in 10 ml of water. The distillate and dilutions of the concentrates
548
ROBACKER ET AL.
(1:6 and 1:25) also were evaluated for attractiveness. For these tests, attractiveness was evaluated only after raising the pH to 9 or greater with sodium hydroxide. Test quantities were 10/zl. Controls for all of these tests were 10/zl of water. Reverse-Phase HPLC. A Waters (Milford, Massachusetts) high-performance liquid chromatograph (HPLC) was used. Detection was at 254 nm (model 490 Programmable Multiwavelength Detector, Waters Associates). Steam distillate was concentrated 200:1 at pH 5 by rotary evaporation. Concentrated steam distillate (pH 5) was fractionated using ion-pairing on a C-18 column (Partisil ODS2, 10-~m particle size, 25 cm x 4.6 mm) (Alltech Associates). Mobile phase was a gradient beginning with 20% methanol and 80% aqueous 0.004 M lauryl sulfate (Sigma Chemical Co., St. Louis, Missouri; purity 99%; pH 3.0) at 1 ml/min. The percentage of methanol was increased to 100% with a linear gradient from 1 to 15 min. The mobile phase remained at 100% methanol for the remainder of the analysis. Injection volume was 100/zl. Consecutive 1-min fractions were collected from 0 to 30 min. One drop of concentrated sodium hydroxide was added to each fraction to raise the pH to 9 or greater. Each fraction was evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 100 /zl. This amount represents a 5 : 1 concentration over the original bacterial filtrate, assuming no losses during the various preparatory steps. Controls were 100/zl of water containing one drop of saturated sodium hydroxide per 1 ml. Fractions were tested in random order during one day constituting one replication of a randomized complete block experiment. Ten replications of the entire procedure from fraction collection to bioassay were conducted. Silica HPLC. The HPLC described above was used to further fractionate three fractions collected from the reverse-phase ion-pairing method that were found to be attractive by cage-top bioassays. The fractions were those that eluted at 16-17, 17-18, and 18-19 min. For each of the three fractions, 20 collections of 1 ml each were made using the same HPLC method described above. Collections were combined and concentrated to near dryness by rotary evaporation under vacuum with heat. The dried concentrate was redissolved in 2 ml of methanol, and 0.2 N sodium hydroxide in methanol was added until pH 8 was obtained. Each concentrated fraction was analyzed on a silica column (Lichrosorb SI-60, 5-/zm particle size, 25 cm x 4.6 ram) (Alltech Associates). Mobile phase was a gradient beginning with 5 % methanol in pentane. The percentage of methanol was increased to 100% with a linear gradient from 1 to 15 rain. The mobile phase remained at 100 % methanol for the remainder of the analysis. Injection volume was 100/~1. Consecutive 1-min fractions were collected from 0 to 30 min. One drop of 0.2 N sodium hydroxide in methanol was added to each fraction. Each fraction was evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 100 /zl. This amount
BACTE~A-PRODUCED ATTRACTANTS
549
represents a 1:2 dilution compared to the original bacterial filtrate. Controls were 100/xl of water containing one drop of 0.2 N sodium hydroxide in methanol per 1 ml. Experimental design was the same as for testing of the fractions from the reverse-phase column.
RESULTS AND DISCUSSION
Attractiveness of Bacterial Preparations. Supernatant obtained by centrifugation of RGM- 1 culture and filtrate of the supernatant (containing no bacteria cells) were both significantly more attractive to flies than tryptic soy broth controls (Table 1). The pellet containing the bacterial cells after centrifugation was not as attractive as tryptic soy broth. However, the pellet was significantly more attractive than water. Previous research (Robacker et al., 1991a) demonstrated that RGM-1 filtrate is at least as attractive as cultures containing living RGM-1 bacterial cells. The results also indicate that the bacteria cells are somewhat attractive in the absence of nutrient broth, as has been reported for several bacteria species attractive to Bactrocera (Dacus) dorsalis Hendel (Jang and Nishijima, 1990) and Rhagoletis pomonella Walsh (MacCollom et al., 1992). Effect of pH on Attractiveness of Bacterial Filtrate. Attractiveness of bacterial filtrate was greatly affected by its pH (Figure 1). Bacterial filtrate was most attractive above pH 6, suggesting that the attractive compounds contain nitrogen capable of accepting protons. According to this interpretation, most of the attractant compounds exist as nonvolatile cations below pH 6 and as volatile neutral molecules above pH 6. The decline in attractiveness above pH 10 is more difficult to interpret but also may relate to ionization. Bacterial filtrate was slightly attractive below pH 6, as demonstrated by paired t tests. Except for pH 5, counts of flies at bacterial filtrate were significantly higher than counts at TABLE 1. MEAN COUNTS OF A. ludens AT FILTER PAPERS CONTAINING CENTRIFUGE FRACTIONS AND FILTERED BACTERIAL CULTURES COMPARED TO PAPERS CONTAINING TRYPTIC SOY BROTH (TSB) OR WATER AS CONTROLS ~
Test
N
Test sample
Control
Pellet vs. TSB Supematant vs. TSB Filtrate vs. TSB Pellet vs. Water
6 6 6 6
85.3 211.7 218.3 95.3
143.7 102.0 109.7 45.3
Means of test samples are significantly different from controls in every test by paired t tests (P
PARTIAL CHARACTERIZATION A N D HPLC ISOLATION OF BACTERIA-PRODUCED ATTRACTANTS FOR THE MEXICAN FRUIT FLY, Anastrepha ludens 1
D.C.
ROBACKER,
2'* W.C.
WARFIELD,
2 and R.F.
ALBACH 3
2Crop Quality and Fruit Insects Research, ARS, USDA 3Conservation and Production Systems Research, ARS, USDA 2301 South International Blvd. Weslaco, Texas 78596 (Received September 21, 1992; accepted November 3, 1992)
Abstract--Methods were developed to collect and isolate volatile chemicals produced by a Staphylococcus bacterium in tryptic soy culture that are attractive to protein-hungry adult Mexican fruit flies. Centrifugation of bacteria culture yielded a slightly attractive pellet containing most of the bacteria cells and a highly attractive supernatant. Supernatant filtered to remove the remaining bacteria was as attractive as the unfiltered supematant. Filtrate at pH 7 and above was much more attractive than filtrate at pH 5 and below. Most of the attractiveness was retained on strong cation exchange media under acidic conditions and eluted with base. Attractive principles could not be trapped on adsorbents such as Porapak Q or extracted with organic solvents from aqueous preparations, but they were easily collected by headspace sweeping with steam. The attractive components were efficiently concentrated by rotary evaporation of steam distillate at pH 5, but at higher pH much of the attractiveness distilled. A reverse-phase HPLC method using a negative counter-ion was developed to separate and collect attractive components of concentrated steam distillate. Attractive fractions collected using this method were concentrated and injected onto silica HPLC. Activity eluted from silica in two distinct bands. Results suggest that the most attractive components of the bacterial odor are highly polar, low-molecular-weight amines. Key Words--Attractants, Mexican fruit fly, Diptera, Tephritidae, Anastrepha ludens, bacteria, amines.
*To whom correspondence should be addressed. Diptera: Tephritidae.
543 0098-0331/93/0300-0543507.00/0 (c) 1993 Plenum Publishing Corporation
544
ROBACKER ET AL.
INTRODUCTION
Attractiveness of bacteria to fruit flies has been investigated for 40 years (Gow, 1954), but few of the chemicals responsible for the attractiveness have been identified. Ammonia is one by-product of bacterial action that has been identified. It has been used in some capacity as a fruit fly attractant since at least the 1930s (Jarvis, 1931; Hodson, 1943), but much disagreement exists concerning its attractiveness. Several authors have stated that attractiveness of bacteria and protein baits is due primarily to components other than ammonia (Gow, 1954; Mazor et al., 1987; Drew and Fay, 1988). Other studies have indicated that ammonia alone or in protein baits is an effective attractant if its release rate is within the optimum range (Bateman and Morton, 1981; Morton and Bateman, 1981; Wakabayashi and Cunningham, 1991). More recently, 2-butanone and butanol were identified from volatiles produced by Proteus species bacteria (Hayward et al., 1977) and shown to be attractive to Dacus tryoni (Froggatt) (Drew, 1987). Generally, other volatiles identified from bacteria (Hayward et al., 1977) or protein baits (Morton and Bateman, 1981; Buttery et al., 1983) have not been attractive. The general lack of success in identifying the probable attractive principles of bacteria and protein baits suggests that the problem is difficult and may require unusual approaches. Robacker et al. (1991a) isolated a bacterium (identified as Staphylococcus aureus) from the mouthparts of a female Mexican fruit fly (Anastrepha ludens Loew) that was highly attractive to adult Mexican fruit flies in laboratory experiments. That study and a subsequent one (Robacker, 1991) presented evidence that the attraction response of the flies to bacterial odor was motivated by hunger for protein. The attractant chemicals produced by the bacteria were not identified. This paper reports chemical properties and methods for isolation of the attractive principles of the odor of the Staphylococcus species (RGM-1) identified by Robacker et al. (1991a). Evidence is presented that the most important attractants are low-molecule-weight amines that cannot be handled by standard methods for collecting and isolating insect pheromones and other attractants.
METHODS AND MATERIALS
Insects and Test Conditions. Flies were from a culture that had been maintained on laboratory diet for about 80-90 generations with no wild-fly introductions. Recent experiments indicated the culture flies were as vigorous as wild flies in mating competitiveness (Moreno et al., 1991) and had retained much of their natural courtship behavior (Robacker et al., 1991b). Mixed-sex groups of 180-200 flies were kept in 473-ml cardboard cartons with screen tops until used
BACTERIA-PRODUCED ATTRACTANTS
545
in tests. Flies were tested when 6-12 days old. Flies were deprived of protein as adults but were fed sucrose and water up until the time of attractiveness testing. All tests were conducted in the laboratory between 0830 and 1430 hr under a combination of fluorescent and natural light. Laboratory conditions were 22 _+ 2~ 50 +_ 20% relative humidity, and photophase from 0630 to t930 hr. Bacterial Preparations. Bacterial strain RGM-1 (Robacker et al., 1991a) was cultured in tryptic soy broth (Difco Laboratories, Detroit, Michigan) in a shaker for 144 hr at 30~ Numerous 100-ml samples were prepared as needed for the various experiments. Bacterial culture was centrifuged at 10,000 rpm for 20 min and separated into pellet and supernatant. Supernatant was filtered through 0.45-#m type HA aqueous filters (Millipore Corporation, Bedford, Massachusetts) followed by 0.22-t*m type GV aqueous/organic filters (Millipore). The pellet was suspended in water and centrifuged. The new pellet was suspended again in water. The centrifugation and filtration of the supernatant were done to remove bacteria cells. Resuspended pellets (second centrifugation only), supernatants (first centrifugation only), and filtrates were monitored for attractiveness to flies using cage-top bioassays (described below). Test quantities applied to filter papers were 10 #1. Controls were 10-/A quantities of tryptic soy broth. Attractiveness of resuspended pellets was also evaluated against a water control (10 txl). Data were analyzed as paired t tests to compare total counts at treatment papers to total counts at control papers. Cage-Top Bioassay Procedure. Bioassays were conducted by placing four filter paper triangles (3 cm/side), two containing test samples and two containing appropriate controls, near the corners on the top of an inset cage (30 cm/side, aluminum-screened). The numbers of flies beneath each filter paper were counted once each minute for 10 min. The two papers containing test chemicals were positioned diagonally from each other on two corners of the cage top, and the two papers containing controls were positioned diagonally from each other on the other two corners. The filter papers were raised 5 mm above the cage top using plastic tings to ensure that olfaction and not contact chemoreception was solely responsible for the response of the flies. Two cartons of 180-200 flies were used in each bioassay cage. Effects of pH on Attractiveness of Bacterial Filtrate. Bacterial filtrate had a pH of 7.9. The pH of aqueous bacterial filtrate was raised to 8, 9, 10, 11, and 12 with saturated sodium hydroxide and lowered to 2, 3, 4, 5, 6, and 7 with 85 % phosphoric acid. Each pH treatment was bioassayed for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 10 #1. Controls were 10-/xl quantities of water. Bioassays were conducted in a randomized complete block experiment. Ten replications of the experiment were conducted. Data used for analysis of variance were differences between total counts at treatment papers and controls.
546
ROBACKER ET AL.
Retention on Strong Cation Exchange Media. The pH of aqueous bacterial filtrate was lowered to 4.9 with 85 % phosphoric acid. The resulting solution (1 ml) was loaded onto a SCX PrepSep extraction column (Fisher Scientific, Fair Lawn, New Jersey) after preparing the column with successive rinses with 5 ml of 0.01 M monobasic sodium phosphate (NaHzPO4) (pH 4.9) and 1 ml of water. After loading the 1 ml of bacterial filtrate, the column was eluted with successive rinses with 5 ml of 0.01 M monobasic sodium phosphate (pH 4.9), 5 ml of water, and 1 ml of 0.05 M dibasic sodium phosphate (NazHPO4) (pH 9.0). Since passage through the acidic column always lowered the pH of the 1 ml of 0.05 M dibasic sodium phosphate to about 7-8, one drop of saturated sodium hydroxide was added to the eluate and it was again passed through the column. The attractiveness of all eluates was evaluated with cage-top bioassays after raising their pH to 10 or greater with sodium hydroxide. Bioassay test quantities were 10/zl for the void volume eluting during loading of the bacterial filtrate and for both of the 0.05 M dibasic sodium phosphate eluates, and 50/zl for the water eluate and for the 0.01 M monobasic sodium phosphate eluate. Controls were 10- or 50-/A quantities, as appropriate, of water containing one drop of saturated sodium hydroxide per milliliter. Experimental design was a randomized complete block and data analysis was handled as in the previous experiment. Paired t tests also were conducted to compare counts at treatments with counts at controls. Collection of Volatiles from Bacterial Filtrate. Volatiles were collected using several methods. The first series of methods was to put bacterial filtrate into a glass aeration flask and pull air through the flask and subsequently through a trap. Bacterial filtrate was put into the aeration flask either as 250 ml of liquid or as an incompletely dried residue from 50 ml of the liquid on 2500 cm 2 of filter paper. Both forms of the bacterial filtrate were highly attractive to flies. Four different types of traps were used to collect volatiles. Three employed chemical adsorbents in glass tubes: 250 mg of Tenax-GC (60/80) (Alltech Associates, Inc., Deerfield, Illinois); 250 mg of Porapak Q (50/80) (Alltech Associates); or 20 mg of Super Q (Alltech Associates). Procedures and glassware for the collections using Tenax-GC and Porapak Q were similar to those described in Robacker et al. (1990). Apparatus for collections using Super Q was that described by Heath and Manukian (1992). Aerations were for at least 6 hr. The adsorbent traps were cleaned with 100 ml of 5 % acetone in pentane (v/v) before use and extracted after aeration with successive rinses of 1 ml each of 5% acetone in pentane, acetone, and water to obtain bacteria volatiles. The fourth was a cold trap in which the odor-laden air was pulled through a glass flask immersed in a Dry Ice-acetone bath. Volatiles were obtained from the collection flask in the ice that formed during the aeration. All trap extracts were evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter paper were 100 /~1. Controls were 100 tzl of
BACTERIA-PRODUCED ATTRACTANTS
547
solvents used for extraction (chemical adsorbents collections) or water (cold trap collection). The final volatile collection method was a headspace sweep with steam. Steam was generated from vigorous boiling of Milli-Q (Millipore) grade water in a two-neck 500-ml flask. The steam was vented into a second two-neck 500-ml flask containing 100 ml of bacterial filtrate heated gently. The steam entered the second flask about 5 cm above the liquid level so as to prevent excessive foaming of the bacterial filtrate. The steam distillate was collected into a 500-ml flask following condensation in a cold-water condenser. The procedure was carried out until 400 ml of steam distillate was collected. Both the steam distillate and the bacterial filtrate residue (the portion of bacterial filtrate that was not swept into the condenser) were evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 10 tzl. Controls were 10 tzl of water. Paired t tests were used to compare counts at treatments with counts at controls. A Student's t test was used to compare responses to steam distillate with responses to bacterial filtrate residue. Data for this test were differences between counts at treatment papers and controls. Solvent-Solvent Extraction. Steam distillate and bacterial filtrate were solvent-solvent extracted in separate experiments with hexane, dichloromethane, and ethyl ether. Quantities of the organic solvents were 10-fold greater than the aqueous distillate or filtrate. Ten-microliter quantities of the aqueous and 100 #1 of the organic phases were tested for attractiveness to flies using cage-top bioassays. Controls were 10-/xl quantities of water for testing aqueous phases and 100/A of appropriate organic solvent for testing the organic phases. Concentration of Bacteria-Produced Attractants. Steam distillate was concentrated using three methods. The first was lyophilization using a Speed Vac concentrator (Savant Instruments, Inc., Farmingdale, New York) connected in series to a refrigerated condensation trap (Savant Instruments, Inc.) and a high vacuum pump (Savant Instruments, Inc.). Steam distillate (100 ml) was concentrated to 1 ml. Dilutions of 1:10 and 1:100 with water were made, and 10-#1 samples of the concentrate and the two dilutions were evaluated for attractiveness using cage-top bioassays. Controls were 10 #1 of water. The second method was rotary evaporation using a Rotavapor RE-111 instrument (Buchi-Brinkman Instruments, Inc., Westbury, New York). Rotary evaporation was aided by heating the steam distillate while pulling vacuum on the system. In the first attempt, 250 ml of steam distillate was concentrated to 10 ml. The concentrate, distillate, and 1 : 10 dilutions of each were evaluated for attractiveness using cage-top bioassays. Test amounts were 10 #1 each. As a second attempt, the pH of 250 ml of steam distillate was lowered to 5 with phosphoric acid before rotary evaporation. The concentrate was evaluated when it reached a volume of 10 ml and again after it was evaporated to dryness and redissolved in 10 ml of water. The distillate and dilutions of the concentrates
548
ROBACKER ET AL.
(1:6 and 1:25) also were evaluated for attractiveness. For these tests, attractiveness was evaluated only after raising the pH to 9 or greater with sodium hydroxide. Test quantities were 10/zl. Controls for all of these tests were 10/zl of water. Reverse-Phase HPLC. A Waters (Milford, Massachusetts) high-performance liquid chromatograph (HPLC) was used. Detection was at 254 nm (model 490 Programmable Multiwavelength Detector, Waters Associates). Steam distillate was concentrated 200:1 at pH 5 by rotary evaporation. Concentrated steam distillate (pH 5) was fractionated using ion-pairing on a C-18 column (Partisil ODS2, 10-~m particle size, 25 cm x 4.6 mm) (Alltech Associates). Mobile phase was a gradient beginning with 20% methanol and 80% aqueous 0.004 M lauryl sulfate (Sigma Chemical Co., St. Louis, Missouri; purity 99%; pH 3.0) at 1 ml/min. The percentage of methanol was increased to 100% with a linear gradient from 1 to 15 min. The mobile phase remained at 100% methanol for the remainder of the analysis. Injection volume was 100/zl. Consecutive 1-min fractions were collected from 0 to 30 min. One drop of concentrated sodium hydroxide was added to each fraction to raise the pH to 9 or greater. Each fraction was evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 100 /zl. This amount represents a 5 : 1 concentration over the original bacterial filtrate, assuming no losses during the various preparatory steps. Controls were 100/zl of water containing one drop of saturated sodium hydroxide per 1 ml. Fractions were tested in random order during one day constituting one replication of a randomized complete block experiment. Ten replications of the entire procedure from fraction collection to bioassay were conducted. Silica HPLC. The HPLC described above was used to further fractionate three fractions collected from the reverse-phase ion-pairing method that were found to be attractive by cage-top bioassays. The fractions were those that eluted at 16-17, 17-18, and 18-19 min. For each of the three fractions, 20 collections of 1 ml each were made using the same HPLC method described above. Collections were combined and concentrated to near dryness by rotary evaporation under vacuum with heat. The dried concentrate was redissolved in 2 ml of methanol, and 0.2 N sodium hydroxide in methanol was added until pH 8 was obtained. Each concentrated fraction was analyzed on a silica column (Lichrosorb SI-60, 5-/zm particle size, 25 cm x 4.6 ram) (Alltech Associates). Mobile phase was a gradient beginning with 5 % methanol in pentane. The percentage of methanol was increased to 100% with a linear gradient from 1 to 15 rain. The mobile phase remained at 100 % methanol for the remainder of the analysis. Injection volume was 100/~1. Consecutive 1-min fractions were collected from 0 to 30 min. One drop of 0.2 N sodium hydroxide in methanol was added to each fraction. Each fraction was evaluated for attractiveness using cage-top bioassays. Test quantities applied to filter papers were 100 /zl. This amount
BACTE~A-PRODUCED ATTRACTANTS
549
represents a 1:2 dilution compared to the original bacterial filtrate. Controls were 100/xl of water containing one drop of 0.2 N sodium hydroxide in methanol per 1 ml. Experimental design was the same as for testing of the fractions from the reverse-phase column.
RESULTS AND DISCUSSION
Attractiveness of Bacterial Preparations. Supernatant obtained by centrifugation of RGM- 1 culture and filtrate of the supernatant (containing no bacteria cells) were both significantly more attractive to flies than tryptic soy broth controls (Table 1). The pellet containing the bacterial cells after centrifugation was not as attractive as tryptic soy broth. However, the pellet was significantly more attractive than water. Previous research (Robacker et al., 1991a) demonstrated that RGM-1 filtrate is at least as attractive as cultures containing living RGM-1 bacterial cells. The results also indicate that the bacteria cells are somewhat attractive in the absence of nutrient broth, as has been reported for several bacteria species attractive to Bactrocera (Dacus) dorsalis Hendel (Jang and Nishijima, 1990) and Rhagoletis pomonella Walsh (MacCollom et al., 1992). Effect of pH on Attractiveness of Bacterial Filtrate. Attractiveness of bacterial filtrate was greatly affected by its pH (Figure 1). Bacterial filtrate was most attractive above pH 6, suggesting that the attractive compounds contain nitrogen capable of accepting protons. According to this interpretation, most of the attractant compounds exist as nonvolatile cations below pH 6 and as volatile neutral molecules above pH 6. The decline in attractiveness above pH 10 is more difficult to interpret but also may relate to ionization. Bacterial filtrate was slightly attractive below pH 6, as demonstrated by paired t tests. Except for pH 5, counts of flies at bacterial filtrate were significantly higher than counts at TABLE 1. MEAN COUNTS OF A. ludens AT FILTER PAPERS CONTAINING CENTRIFUGE FRACTIONS AND FILTERED BACTERIAL CULTURES COMPARED TO PAPERS CONTAINING TRYPTIC SOY BROTH (TSB) OR WATER AS CONTROLS ~
Test
N
Test sample
Control
Pellet vs. TSB Supematant vs. TSB Filtrate vs. TSB Pellet vs. Water
6 6 6 6
85.3 211.7 218.3 95.3
143.7 102.0 109.7 45.3
Means of test samples are significantly different from controls in every test by paired t tests (P
