Taste-independent detection of the caloric content of
sugar in Drosophila
Monica Dusa
, SooHong Mina,1
, Alex C. Keeneb,1
, Ga Young Leea
, and Greg S. B. Suha,2
a
Molecular Neurobiology Program, Skirball Institute for Biomolecular Medicine, Department of Cell Biology, New York University School of Medicine,
New York, NY 10016; and b
Department of Biology, New York University, New York, NY 10003
Edited* by David J. Anderson, California Institute of Technology, Pasadena, CA, and approved May 25, 2011 (received for review November 16, 2010)
Feeding behavior is influenced primarily by two factors: nutritional
needs and food palatability. However, the role of food
deprivation and metabolic needs in the selection of appropriate
food is poorly understood. Here, we show that the fruit fly,
Drosophila melanogaster, selects calorie-rich foods following
prolonged food deprivation in the absence of taste-receptor signaling.
Flies mutant for the sugar receptors Gr5a and Gr64a cannot
detect the taste of sugar, but still consumed sugar over plain
agar after 15 h of starvation. Similarly, pox-neuro mutants that
are insensitive to the taste of sugar preferentially consumed
sugar over plain agar upon starvation. Moreover, when given a
choice between metabolizable sugar (sucrose or D-glucose) and
nonmetabolizable (zero-calorie) sugar (sucralose or L-glucose),
starved Gr5a; Gr64a double mutants preferred metabolizable
sugars. These findings suggest the existence of a taste-independent
metabolic sensor that functions in food selection. The preference
for calorie-rich food correlates with a decrease in the two
main hemolymph sugars, trehalose and glucose, and in glycogen
stores, indicating that this sensor is triggered when the internal
energy sources are depleted. Thus, the need to replenish depleted
energy stores during periods of starvation may be met
through the activity of a taste-independent metabolic sensing
pathway.
Food quantity and quality can vary greatly in natural habitats. To
survive such variations, animals must be able to search for and
detect appropriate food sources under all conditions, especially
during times of food scarcity. Peripheral chemosensory neurons,
such as sugar taste neurons, allow animals to detect palatable foods
(1–5). Additional mechanisms may be necessary for the detection
of foods to meet acute nutritional needs. Indeed, animals learn to
positively associate a flavor paired with intragastric sugar infusion
(6). Recently, studies of Trpm5−/−
mice, which are insensitive to the
taste of sugar, also have revealed that these animals develop
a preference for a sugar solution on the basis of its caloric content
even in the absence of gustatory input (7). Unfortunately, the nature
of such mechanisms is currently unknown. It is also not clear
whether they function under starvation conditions.
To search for mechanisms by which animals can respond to
the caloric content of food independently of orosensory cues, we
studied the effect of starvation on food choice in Drosophila
mutants that are unable to taste sugar. Specifically, we sought to
determine whether food-deprived flies carrying mutations in
Gr5a and Gr64a (3–5), the sugar receptor genes, and in poxneuro
(poxn) (8–10), a gene that specifies chemosensory neurons,
develop a preference for the caloric content of sugars in the
absence of taste perception. We found that these mutant flies
demonstrated a preference for caloric food upon starvation and
that this preference correlated with the energy needs of the fly.
Furthermore, wild-type (WT) flies showed a shift in preference
to metabolizable sugars following prolonged periods of starvation
even though nonmetabolizable sugars induced similar
taste responses. Our findings suggest that starvation activates
a previously uncharacterized pathway that allows animals to
make feeding choices based on nutritional needs rather than
palatability.
Results
Flies Mutant for the Sugar Taste Receptors Gr5a and Gr64a Exhibit
a Preference for Sugars After Prolonged Periods of Starvation. Flies
mutant for Gr5a and Gr64a are unable to taste most sugars,
including sucrose, glucose, and trehalose (3–5). To determine
whether flies have a taste-independent pathway that enables the
detection of calorie-rich food after periods of food deprivation,
we asked whether starved Gr5a; Gr64a mutants develop a preference
for the sugar when given a choice between sucrose and
plain agar in two-choice feeding assays. In this assay, flies were
presented with two food substrates, each colored with a different
dye, and feeding preference was scored by dye accumulation in
the abdomen of individual flies (11) (Fig. S1 A and B). Thus,
this is a qualitative measurement of food choice that reflects the
preference for one food substrate over another. We tested Gr5a;
Gr64a mutants (Gr5a; Gr64a1
and Gr5a; Gr64a2
alleles) that had
been starved for 22 h. For controls, we used flies lightly fooddeprived
for 5 h. WT Canton-S (CS) and flies carrying single
mutations for Gr5a or Gr64a (Gr64a1 and Gr64a2
alleles) were
attracted to sucrose after either 5 or 22 h of starvation. However,
the majority of Gr5a; Gr64a double mutants did not eat when they
were lightly food-deprived for 5 h (Fig. 1A), and no dye accumulation
was observed in their crop and gut after dissection (Fig. 1B).
In contrast, most of the Gr5a; Gr64a mutants developed a strong
preference for sucrose after 22 h of starvation, as shown by the
presence of dye in their abdomens (Fig.1 C and D). This feeding
choice happened despite the inability of Gr5a; Gr64a mutants to
taste sugar. We asked whether Gr5a; Gr64a mutant flies did not
eat after 5 h of food deprivation because they have abnormal
responses to starvation. To test this, we measured circulating
glucose levels, glycogen stores, and starvation-induced sleep suppression
(12), a physiological manifestation of hunger, but we did
not detect any differences between Gr5a; Gr64a mutants and WT
flies (Fig. S1 C–G). Gr5a; Gr64a mutants therefore showed the
behavioral changes that normally accompany starvation.
To determine whether the preference of Gr5a; Gr64a mutants
for sucrose after prolonged starvation reflected the activity of an
unknown sugar taste receptor, we measured their proboscis extension
reflex (PER) (13) to a 100-mM sucrose solution after
22 h of starvation. Gr5a; Gr64a mutants failed to respond to the
taste of sucrose after starvation (Fig. 1E and Fig. S2), indicating
that they did not acquire gustatory responses to sucrose even
after periods of prolonged starvation.
To validate the existence of a taste-independent mechanism
for detecting sugar and to ensure that the preference for sucrose
over plain agar was not influenced by any other gustatory neurons
in the fly, we tested mutants for the poxn gene, which
Author contributions: M.D. and G.S.B.S. designed research; M.D., S.M., A.C.K., and G.Y.L.
performed research; M.D. and G.S.B.S. analyzed data; and M.D. and G.S.B.S. wrote the
paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
1
S.M. and A.C.K. contributed equally to this article.
2
To whom correspondence should be addressed. E-mail: greg.suh@med.nyu.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1017096108/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1017096108 PNAS Early Edition | 1 of 6
NEUROSCIENCE
have no chemosensory neurons in any of their gustatory organs
(8–10). We used two different alleles, poxnΔM22B5 and
poxnΔM22B5; full152 (a transgene that rescues poxnΔM22B5
CNS defects), as taste mutants and mutant animals bearing a full
genomic rescue construct, poxnΔM22B5; SuperA158, as controls.
The poxn mutations did not exhibit a PER response to sucrose
after 22 h of starvation, whereas the controls had the same
responses as WT flies (Figs. S2 and S3A). As was observed in
Gr5a; Gr64a mutant flies, the majority of 5-h food-deprived
poxnΔM22B5 and poxnΔM22B5;full152 flies did not eat, whereas
the controls, poxnΔM22B5; SuperA 158, chose sucrose over agar
(Fig. S3C). After 22 h of starvation, both poxn mutants and the
controls chose sucrose (Fig. S3D). Thus, poxn mutants, like Gr5a;
Gr64a mutants, are able to detect sucrose in the absence of
gustatory input. As a final control, we tested whether taste-blind
flies use olfactory cues to distinguish sucrose from agar. However,
Gr5a; Gr64a mutant flies still chose sucrose in the absence
of the antennae and maxillary palps, which house the olfactory
receptor neurons (Fig. S4). Taken together, these results suggest
the existence of a taste- and olfactory-independent pathway that
mediates a preference to sugar upon starvation.
A taste-independent metabolic mechanism would be expected
to take longer than direct peripheral sensory input to detect the
presence of food. If such a mechanism does direct Gr5a; Gr64a
mutants to select sucrose after 22 h starvation, we would expect it
to take longer for the sugar blind mutants to develop a preference
for sugar compared with WT flies. To test this, we gave flies
a choice of 100 mM sucrose versus plain agar after depriving
them of food for 22 h and then assessed their food choices at
different time points thereafter (Fig. 1F). Within 5 min after the
food was presented, the majority of WT flies had already chosen
sucrose, whereas most of the Gr5a; Gr64a mutants did not. After
45 min, the percentage of Gr5a; Gr64a mutant flies that consumed
sucrose was comparable to that of the WT flies (Fig. 1F).
poxnΔM22B5 mutants behaved similarly to Gr5a; Gr64a mutants
(Fig. S3B). Therefore, the selection of an appropriate food
choice is delayed, but not abolished, in the absence of gustatory
input. This delay is consistent with assessment of the food nutritional
value by a mechanism independent of direct peripheral
gustatory input, and instead dependent on metabolic evaluation.
Glucose and trehalose are the two main sugars in the Drosophila
hemolymph (14). We tested whether these sugars induce a tasteindependent
feeding behavior in starved flies. As was observed for
sucrose, Gr5a; Gr64a mutants did not eat glucose or trehalose after
5 h of starvation (Fig. 2 A and C), but developed a robust preference
for both sugars after 22 h of starvation (Fig. 2 B and D). GR5a
is the only known receptor for trehalose, and Gr5a mutants lack
behavioral and electrophysiological responses to trehalose (15,
16). Unlike the WT and Gr64a controls, and similar to Gr5a; Gr64a
flies, Gr5a mutants did not eat after 5 h of food deprivation, but
preferred trehalose over plain agar after 22 h of starvation (Fig. 2 C
and D). We tested a nonhemolymph sugar, galactose, and observed
that Gr5a; Gr64a mutants also developed a preference to
the sugar after prolonged starvation (Fig. 2 E and F).
If taste plays a primary role in food-choice behavior of animals
that are sated or food-deprived for short times, then Gr5a; Gr64a
mutants should eat sugars that they can taste. Gr5a and Gr64a
are not required for detection of fructose (4). As expected, Gr5a;
Gr64a mutants showed robust PER responses to this sugar (Fig.
S5) and chose fructose over agar after 5 h of food deprivation
(Fig. 2G). To determine whether fructose can trigger the tasteindependent
food choice, we used poxn mutants, which lack
a PER response to fructose (Fig. S3A). After 5 h of light food
deprivation, the majority of poxnΔM22B5 and poxnΔM22B5; full
152 mutants did not eat fructose, whereas most of the controls
did (Fig. S3E). After 22 h of starvation, both poxn mutants chose
fructose over plain agar (Fig. S3F).
Taste-Independent Pathway Detects the Nutritional Content of
Sugars. The taste-independent food choice is triggered by different
sugars with diverse structures. Once ingested, these sugars
are soon metabolized. We asked whether this pathway indeed
responds to the nutritional value of sugars. To address this question,
we examined whether Gr5a; Gr64a mutants prefer the nonmetabolizable
sweetener sucralose to agar (Fig. 3A, Upper). WT
flies exhibited a robust PER to sucralose, but Gr5a; Gr64a1
mutants
A
C
E
B
D
F
CS Gr5a;Gr64a
5 hours
22 hours
CS Gr5a;Gr64a
Fig. 1. Gr5a; Gr64a mutants develop a
preference for sucrose after prolonged
periods of food deprivation. (A and C) Twochoice
preference assay with sucrose versus
agar after 5 h (A) and after 22 h (C) of
starvation. Flies of different genotypes
were given a choice between agar containing
100 mM sucrose versus plain agar.
The y axis shows the average percentage of
flies that ate sucrose, plain agar, or did not
eat. n = 4–8 with each trial comprising 50
flies. Error bars: SEM, *P < 0.001, two-way
ANOVA with Bonferroni post hoc test in
this figure and all subsequent figures unless
indicated otherwise. (B and D) Dissected
crop and gut from 5-h (B) and 22-h (D)
starved flies that were given a choice between
agar containing 100 mM sucrose
mixed with green dye versus plain agar
mixed with red dye (A and C). (E) PER of
22-h starved flies when their labellum was
stimulated by 100 mM sucrose. A full extension
was given a score of 1 and a partial
extension was given a score of 0.5. n = 24–
37. (F) Time course for the development of
sucrose preference. Flies were starved for
22 h and then given a choice between 100
mM sucrose versus agar for different
durations of time (x axis). n = 4. *P < 0.001
and **P < 0.05.
2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1017096108 Dus et al.
did not (Fig. 3B). In two-choice feeding assay, WT flies and singlemutant
controls preferred sucralose to plain agar after 22 h of
starvation (Fig. 3C), which suggests that sucralose is palatable to
flies, consistent with a recent report (12, 17). By contrast, the majority
of Gr5a; Gr64a mutant flies failed to choose either sucralose
or agar (Fig. 3C), but chose sucrose when given a choice between
sucrose and sucralose (Fig. 3D). We obtained the same result with
poxn mutant flies (Fig. S3 G and H). These results support our
hypothesis that the preference for a calorically rich sugar develops
independently of sensory cues.
To further investigate the possibility that the taste-independent
pathway detects the nutritional value of sugars, we carried out
two-choice assays with L-glucose versus D-glucose (Fig. 3A, Lower).
L- and D-glucose are stereoisomers, but only D-glucose can be
metabolized and can thus generate energy. WT flies exhibited
a robust PER to both glucose stereoisomers (Fig. 3E), whereas
Gr5a; Gr64a1
mutant flies responded to neither. In a two-choice
assay with L-glucose versus plain agar, WT and control flies chose
L-glucose over agar, whereas Gr5a; Gr64a mutants chose neither
(Fig. 3F). When given a choice between D-glucose and L-glucose,
all of the flies, including Gr5a; Gr64a mutants, selected D-glucose
over L-glucose after starvation (Fig. 3G). These findings provide
further support to our hypothesis that the taste-independent
pathway responds to the caloric content of sugars.
If the nutritional value of food can be detected independently of
orosensory cues, it would be expected to take priority over food
palatability to ensure that the animal’s metabolic needs are met,
especially during periods of food scarcity. To test this, we identified
a concentration of L- and D-glucose at which four different fly
lines—CS, Oregon-R (Or-R), yw, and Harwich—showed no preference
for either sugar after 5 h of food deprivation in a twochoice
feeding assay (Fig. 3H, light stippled bars). We calculated
the preference index (PI) for D-glucose to determine whether the
food preference shifted to the metabolizable enantiomer during
starvation. All four Drosophila lines exhibited a marked preference
for D-glucose over the more concentrated L-glucose after 22 h
of starvation (Fig. 3H, dark stippled bars). This shift in preference
to the metabolizable compound during starvation indicates that
the mechanisms directing the choice for caloric versus noncaloric
foods are triggered when internal energy reserves are depleted.
Taste-Independent Food Choice Correlates with a Decrease in
Hemolymph Sugar Levels. We next sought to determine whether
the metabolism of carbohydrates influences the preference for
metabolizable over nonmetabolizable sugars during starvation.
We measured circulating sugar levels and glycogen stores in WT
flies after 5, 10, 15, and 22 h of starvation. We found that a drop
in the hemolymph glucose and trehalose levels occurs between
10 and 15 h of starvation (Fig. 4A). Glycogen stores also declined,
but before 10 h of starvation (Fig. 4B).
To characterize the timing of the induction of the tasteindependent
food choice in more detail, we observed the choice
made by Gr5a; Gr64a mutants and controls when given a choice
between sucrose and plain agar after 5, 10, 15, and 22 h of
starvation. There was no significant difference in the percentage
of Gr5a; Gr64a mutants that ate sucrose between 5 and 10 h of
starvation. However, more than half of the Gr5a; Gr64a mutants
chose sucrose by 15 h of food deprivation (Fig. 4C), suggesting
that the taste-independent food choice in Gr5a; Gr64a flies
occurs between 10 and 15 h of starvation. These data indicate a
strong relationship between the levels of sugar in the hemolymph
and the metabolic decision to choose metabolizable sugars.
If glucose and trehalose levels in the hemolymph are sensed by
a metabolic sensing pathway, we would expect that manipulating
their levels would alter the timing of the shift in preference to
metabolizable sugars. To test this prediction, we fed flies the
glucoprivic reagent 2-deoxyglucose (2DG), an inhibitor of glucose
metabolism, to decrease total hemolymph sugar levels and determined
whether this induces an early shift to a preference for
D-glucose. We selected 2DG because, like glucose, it is phosphorylated
by hexokinase, but unlike glucose, the phosphorylated
compound cannot undergo glycolysis (18). Thus, 2DG obstructs
the glycolytic pathway and disrupts sugar metabolism. After WT
flies were fed with D-glucose (no starvation), 2DG or starved on
agar for 10 and 15 h, their hemolymph glycemia was measured.
We found no significant differences in the levels of circulating
glucose and trehalose in flies fed with D-glucose or flies starved on
agar for 10 h (Fig. 4D). By contrast, flies fed with 2DG for 10 h
had lower levels of circulating glucose and trehalose (Fig. 4D). At
15 h, both 2DG-fed and agar-starved flies showed low circulating
sugar levels, whereas flies fed with D-glucose maintained high
hemolymph glycemia (Fig. 4E). These results demonstrated that
in flies, as in mammals, 2DG accelerates the depletion of circulating
sugar levels.
We then examined whether 2DG influences the timing of the
preference shift to D-glucose. WT flies fed with D-glucose or
starved on agar for 10 h showed equal preference for D- and
L-glucose (Fig. 4F). This is consistent with the findings that Gr5a;
Gr64a mutants do not develop a behavioral preference to sugars
and that hemolymph glycemia does not decrease after 10 h of
starvation (Fig. 4 A and C). By contrast, flies fed 2DG for 10 h
preferentially consumed D-glucose over L-glucose, suggesting that
2DG accelerates the ability of flies to select food on the basis of its
caloric content (Fig. 4F). Flies fed with 2DG for 15 h demonstrated
Fig. 2. The taste-independent sensor responds to different sugars. (A–H) Twochoice
preference assay with agar containing 200 mM glucose (A and B), 200 mM
trehalose (C and D), 400 mM galactose (E and F), or 200 mM fructose (G and H)
versus plain agar. Gr5a;Gr64a1
and Gr5a;Gr64a2
double mutants and control
flies—CS WT, Gr5a, Gr64a1
, and Gr64a2
single mutants—were tested after
5 h (A, C, E, and G) and after 22 h of starvation (B, D, F, and H). n = 3–6. *P < 0.001.
Dus et al. PNAS Early Edition | 3 of 6
NEUROSCIENCE
a strong preference for D-glucose, but were not significantly different
from flies starved for 15 h (Fig. 4G), which already preferred
the metabolizable sugar. In starved flies and flies treated with 2DG,
the shift in behavior correlated with a measurable decrease in
hemolymph glucose and trehalose levels. This observation suggests
that circulating glucose and trehalose levels control the functioning
of the metabolic sensing pathway.
Discussion
We have shown that fruit flies can develop a preference for the
caloric content of sugars. This is accomplished even in the absence
of sugar taste receptors or any known peripheral chemosensory
systems. This taste-independent sugar-sensing pathway has several
distinctive characteristics. First, this pathway is specifically
associated with a starved state; taste-blind flies execute foodchoice
behavior after prolonged food deprivation of between 10
and 15 h of starvation. This time frame coincides with the onset of
starvation-induced sleep suppression (12), indicating that these
two behaviors might share a common metabolic trigger. Second,
the taste-independent pathway operates on a different timescale
from the gustatory pathway. Whereas WT flies made a food
choice almost instantly, taste-blind flies chose sugars only after the
ingestion of food. Third, this pathway responds to the nutritional
content of sugars, but not to their orosensory value. Taste-blind
flies chose metabolizable sugars over nonmetabolizable sugars
and never consumed nonmetabolizable sugars. Furthermore, the
fact that WT flies failed to distinguish a metabolizable sugar from
a nonmetabolizable sugar, but shifted their preference to the
metabolizable sugar after starvation, indicates that the taste-independent
pathway is not an artifact associated with taste-blind
flies, but functions in WT flies. Finally, the ability to detect the
caloric content of sugars correlated under multiple experimental
conditions with drops in hemolymph glycemia.
These results demonstrate that starvation directs the selection
of nutrient-rich foods in the fly in the absence of the gustatory cues.
Thus, as previously suggested in mice, postingestive cues can drive
feeding behavior independently of gustatory information (6, 7).
The physiological factors that triggered the taste-independent
food choices in mice are, however, unknown. In Drosophila, the
internal energy state and carbohydrate metabolism play crucial
roles in the metabolic sensing of food according to our results.
A possible evolutionary purpose of taste-independent metabolic
sensing is to ensure that animals select calorie-rich foods to quickly
replenish energy, especially in times of food shortage.
How do starved sugar-blind flies preferentially ingest metabolizable
sugar over nonmetabolizable sugar? It is plausible that
sugar-blind flies are equally attracted to and feed on both sugars,
A
B
C
E
D
F
G
SucraloseSucrose
H
D-Glucose L-Glucose
Fig. 3. The taste-independent sensor responds to metabolizable sugars, but
not to nonmetabolizable sugars. (A) Molecular structures of sucrose and
sucralose (Upper panels) and D-glucose and L-glucose (Lower panels). (B) PERs
of Gr5a; Gr64a1
mutants and CS WT flies to 0.3 mM sucralose. *P < 0.001
with one-way ANOVA. n = 37–40. (C and D) Two-choice preference assay
with (C) 0.3 mM sucralose versus plain agar and (D) 0.3 mM sucralose versus
100 mM sucrose after 22 h of starvation. An amount of 0.3 mM sucralose
induces comparable PER responses to 100 mM sucrose (Fig. 1E). n = 6–8.
(E) PERs of Gr5a; Gr64a1
mutants and CS WT flies to 200 mM D-glucose and
200 mM L-glucose. The same concentration of D-glucose and L-glucose leads
to similar PER responses in WT flies. *P < 0.001 with one-way ANOVA. n =
22–36. (F and G) Two-choice preference assay with (F) 200 mM L-glucose
versus plain agar and (G) 200 mM L-glucose versus 200 mM D-glucose after 22
h of starvation. n = 4. (H) PI for D-glucose in 5-h (light stippled bars) and 22-h
(dark stippled bars) starved flies, which were given a choice between 200
mM L-glucose versus 50 mM D-glucose. Four independent WT strains—CS, OrR,
yellow white (yw), and Harwich—were used. A PI value of 0.5 indicates
that flies equally prefer D- and L-glucose whereas a PI value of 0.5–1 indicates
a preference for D-glucose. For the detailed calculation of PI (D-glucose), see
Materials and Methods. *P < 0.001, Student’s t test within each genotype;
n = 4 for each trial comprising 50 flies.
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1017096108 Dus et al.
but those on nonmetabolizable sugar resume foraging because of
the lack of nutritional value in this sugar. These foraging flies
are again equally attracted to both sugars, but those on nonmetabolizable
sugar continue to forage until they find the correct
food substrate. Food choice in this model is mediated by random
selection and “trapping” of the flies on the metabolizable sugar.
Alternatively, sugar-blind flies might readily detect the metabolizable
sugar without ingesting a large amount of food because
nutrient information is rapidly conveyed to the brain within
minutes of ingesting food. In this model, the flies select for
metabolizable sugar over nonmetabolizable sugar by a metabolic
sensor that operates on a fast timescale to mediate discrimination
between the two sugar substrates. Tracking and monitoring
the locomotor activity and feeding behavior that generates a
preference for metabolizable sugar will address this question.
It is intriguing to speculate on the molecular nature of the
metabolic sensor. This sensor could be expressed in a subset of
neural, digestive, or other tissues. Among the organs and cells that
have been proposed for their involvement in feeding regulation in
the fly are the fat body (19–21), the insulin-producing cells (IPC)
(22), and the corpora cardiaca/allata complex (23). These cells
may respond to the metabolic value of sugars in circulation, as
seen with the glucose-excited and glucose-inhibited neuropeptide
neurons in the arcuate nucleus of the mammalian hypothalamus
(24, 25). A model that explains how changes in circulating glucose
levels alter the electrical and secretory properties of the hypothalamic
glucose-responsive neurons could also describe how
metabolizable sugars trigger the metabolic sensor. In mammals,
glucose-sensitive cells detect glucose availability by responding to
metabolites of glycolytic enzymes such as hexokinase (26, 27) or
the energy-sensing AMP-activated protein kinase (28).
Almost all crucial metabolic functions in mammals are also
conserved in Drosophila (22, 23, 29). During the past decade,
researchers using the fruit fly as a model system for studying
feeding behaviors and feeding-related disorders, including obesity,
have shed much light on the molecular mechanisms of
metabolism (29, 30). By revealing the possibility of a metabolic
sensing pathway in Drosophila, we have introduced the possibility
of understanding the molecular mechanism underlying this
pathway. Identification of the cellular and genetic nature of this
sensor might reveal the identity of the master switch that regulates
many hunger-driven behaviors.
Materials and Methods
Feeding Assays. Flies were reared in standard cornmeal medium. Male flies
(0–2 d old) in groups of 50 were collected under anesthesia and allowed to
recover in standard cornmeal food vials for at least 2 d before experiments.
After 4–8 d the male flies were then starved for 5, 10, 15, or 22 h in vials
containing 2 mL of milliQ water soaked in Kim-wipe tissue. For two-choice
preference assays, the groups of 50 male flies were cold-anesthetized,
transferred into 60-well microtiter plates (MicroWell MiniTrays with lids,
Fig. 4. Changes in the hemolymph glycemia correlate with the timing of the behavioral switch to metabolizable sugars. (A and B) The levels of (A) hemolymph
glucose and trehalose and (B) glycogen as a function of starvation time. *P < 0.001 in comparison with 5-h starvation, one-way ANOVA, Bonferroni
post hoc test; n = 7–10 for hemolymph glycemia and n = 9 for glycogen. (C) Time course for the induction of the taste-independent food choice behavior.
Gr5a;Gr64a1
, and controls were given a choice between agar containing 100 mM sucrose versus plain agar after 5, 10, 15, and 22 h of starvation. Gr5a; Gr64a1
mutants showed a preference for sucrose between 10 and 15 h of starvation. n = 4–6. (D and E) Concentrations of glucose and trehalose in the hemolymph of
flies fed 400 mM glucose, 400 mM 2DG, or plain agar after (D) 10 h or (E) 15 h of starvation. *P < 0.001 with one-way ANOVA. n = 10–12 for 10 h of starvation
and n = 9–10 for 15 h of starvation. (F and G) PI for D-glucose in flies fed 400 mM D-glucose, 400 mM 2DG, and plain agar for (F) 10 h and (G) 15 h. These flies
were then given a choice between D-glucose versus L-glucose. *P < 0.001 with one-way ANOVA. n = 4.
Dus et al. PNAS Early Edition | 5 of 6
NEUROSCIENCE
Nunc), and allowed to feed for 120 min. The flies were then scored by examining
the color of their abdomen. All feeding experiments were conducted
at the same time of the day, around zeitgeber time 6–7.
The agar substrate containing sugars was made of 1% agar and was colorlabeled
with 0.5–0.6% of green and red McCormick tasteless food dyes.
Color-labeled 1% agar without sugar did not produce a PER response and
did not generate a preference to either dye in WT flies. Preference index for
D-glucose was calculated as [(# flies ate D-glucose) + (0.5 × # flies ate both Dand
L-glucose)]/(total # flies that fed). Thus, a PI of 0.5 indicates no preference
whereas a PI of 0.5–1 indicates a preference. All tested sugars—sucrose,
sucralose, D-glucose, L-glucose, fructose, trehalose, galactose, and 2-deoxyglucose
at 99% purity—were purchased from Sigma-Aldrich.
Crop and Gut Dissections. Flies subjected to the two-choice preference assay
with sucrose versus plain agar were sorted into two groups, “sucrose eating”
and “no eating,” according to the color of their abdomen. Each group was
then dissected. Briefly, a fly was immobilized on a silicon plate using insect
pins (Fine Science Tools; #26002–10) and its legs and wings were removed
under a dissecting microscope. The cuticle of the thorax and abdomen was
peeled off in PBS using fine tweezers (Fine Science Tools; #11251–20) to
expose the crop and gut. Images were taken with a digital camera (Canon;
Powershot A450).
PER. PER assay was performed according to the protocol of ref. 13 with some
modifications. Flies starved for 22 h (in the presence of water) were tested
with water before the experiment and only flies that did not respond to water
were used. The taste bristles on the labellum or the legs were stimulated by
a Kim-wipe thread soaked in tastant solution. PER responses were scored as
follows: no extension = 0, half-extension = 0.5, and full extension = 1.
Hemolymph Glycemia Measurement. Hemolymph glucose and trehalose concentrations
were measured as previously described (20). Briefly, 10 flies
starved on agar or fed 2DG or D-glucose for 5, 10, 15, or 22 h were decapitated
and their hemolymph was drawn with a capillary pipette. A total
of 0.5 μL of hemolymph was mixed with 100 μL of the Glucose (HK) Assay Kit
(Sigma; GAHK20) adjusted to pH 6.8. A total of 10 μL of pig kidney trehalase
(Sigma) per 5 mL of glucose reagent was added to the mixture, incubated at
37 °C for 16 h, and measured with a fluorescent plate reader using quantitative
NADH fluorescence (excitation wavelength of 375 nm and emission
wavelength of 465 nm). Standard curves were generated from D-glucose and
trehalose (0–1,000 mg/mL) standards for each trial.
Glycogen Measurement. Total glycogen was measured as previously reported
(19). Briefly, 10 male flies were homogenized in 250 μL of lysis buffer [10 mM
KH2PO4, 1 mM EDTA (pH 7.4)] and centrifuged at 2,000 × g for 2 min. A total
of 1.5 μL of the supernatant was mixed with 2.5 μL of 0.1 unit/μL amyloglucosidase
(Sigma; A1602-25MG) and 221 μL of the peroxidase/glucose
oxidase (PGO) enzymes reaction solution. The mixture was incubated for
30 min at 37 °C, and its absorbance was measured at 450 nm. G0885-1G
glycogen (Sigma) standard was used to plot a standard curve.
Sleep Analysis. Three- to four-day-old males were placed in Drosophila Activity
Monitors (Trikinetics) in tubes containing fly food. Following 24 h of
baseline recording, flies were transferred to agar containing 100 mM sucrose
(or 5% sucrose for Fig. S1 F and G), agar containing 0.3 mM sucralose, plain
agar, or left in regular fly food (Fig. S1 F and G) (12). Sleep was monitored
continuously for 2–5 d and analyzed using an Excel-based macro as previously
described (31).
Statistical Analysis. The results from the two-choice preference assay were analyzed
by using two-way ANOVA that compared genotype and food choice for
parametrically distributed data. For the results from the PER assay and for the
glycemia and glycogen measurements, one-way ANOVA was used. Following
ANOVA analysis,weusedtheBonferroni posthoctesttodeterminesignificances.
When only two groups were compared, we performed Student’s t test.
Note. While our manuscript was under revision, two papers claiming that
Drosophila can learn to associate an odorant paired with nutritious food
were published in Current Biology (32, 33).
ACKNOWLEDGMENTS. We thank Ulrike Heberlein for poxn mutants; John
Carlson for Gr5a, Gr64a1
, and Gr64a2
mutants; and Minhee Kim for help during
the initial stage of this project. We are indebted to Eric Rulifson, Justin DiAngelo,
Linda Partidge, and Pankaj Kapahi for providing protocols. This manuscript
was aided by comments from Niels Ringstad, Robert Froemke, Fabiola
Rivas, and Greg Hannon. This work was supported by the Hilda and Preston
Davis Postdoctoral Fellowship in Eating Disorders (to M.D.), National Research
Service Award 5F32GM086207 (to A.C.K.), the Klarman Family Foundation
for Eating Disorders (G.S.B.S.), the Alfred P. Sloan Foundation (G.S.B.S.), the
Whitehall Foundation (G.S.B.S.), the Whitehall Presidential Award (to G.S.B.S.),
the Hirschl/Weill-Caulier Trust Award (to G.S.B.S.), and National Institutes of
Health Grant 1RO1GM089746 (to G.S.B.S.).
1. Wang Z, Singhvi A, Kong P, Scott K (2004) Taste representations in the Drosophila
brain. Cell 117:981–991.
2. Amrein H, Thorne N (2005) Gustatory perception and behavior in Drosophila melanogaster.
Curr Biol 15:R673–R684.
3. Dahanukar A, Lei YT, Kwon JY, Carlson JR (2007) Two Gr genes underlie sugar
reception in Drosophila. Neuron 56:503–516.
4. Slone J, Daniels J, Amrein H (2007) Sugar receptors in Drosophila. Curr Biol 17:1809–1816.
5. Jiao Y, Moon SJ, Montell C (2007) A Drosophila gustatory receptor required for the
responses to sucrose, glucose, and maltose identified by mRNA tagging. Proc Natl
Acad Sci USA 104:14110–14115.
6. Sclafani A (2001) Post-ingestive positive controls of ingestive behavior. Appetite 36:
79–83.
7. de Araujo IE, et al. (2008) Food reward in the absence of taste receptor signaling.
Neuron 57:930–941.
8. Nottebohm E, et al. (1994) The gene poxn controls different steps of the formation of
chemosensory organs in Drosophila. Neuron 12:25–34.
9. Awasaki T, Kimura K (1997) pox-neuro is required for development of chemosensory
bristles in Drosophila. J Neurobiol 32:707–721.
10. BollW,NollM(2002)TheDrosophilaPoxneurogene:controlofmalecourtshipbehaviorand
fertility as revealed by a complete dissection of all enhancers. Development 129:5667–5681.
11. Tanimura T, Isono K, Yamamoto MT (1988) Taste sensitivity to trehalose and its
alteration by gene dosage in Drosophila melanogaster. Genetics 119:399–406.
12. Keene AC, et al. (2010) Clock and cycle limit starvation-induced sleep loss in
Drosophila. Curr Biol 20:1209–1215.
13. Shiraiwa T, Carlson JR (2007) Proboscis extension response (PER) assay in Drosophila.
J Vis Exp 3):193.
14. Wyatt GR, Kale GF (1957) The chemistry of insect hemolymph. II. Trehalose and other
carbohydrates. J Gen Physiol 40:833–847.
15. Ueno K, et al. (2001) Trehalose sensitivity in Drosophila correlates with mutations in
and expression of the gustatory receptor gene Gr5a. Curr Biol 11:1451–1455.
16. Dahanukar A, Foster K, van der Goes van Naters WM, Carlson JR (2001) A Gr receptor
is required for response to the sugar trehalose in taste neurons of Drosophila. Nat
Neurosci 4:1182–1186.
17. Gordesky-Gold B, Rivers N, Ahmed OM, Breslin PA (2008) Drosophila melanogaster
prefers compounds perceived sweet by humans. Chem Senses 33:301–309.
18. Friedman MI, Tordoff MG, Ramirez I (1986) Integrated metabolic control of food
intake. Brain Res Bull 17:855–859.
19. Xu K, Zheng X, Sehgal A (2008) Regulation of feeding and metabolism by neuronal
and peripheral clocks in Drosophila. Cell Metab 8:289–300.
20. Kohyama-Koganeya A, Kim YJ, Miura M, Hirabayashi Y (2008) A Drosophila orphan G
protein-coupled receptor BOSS functions as a glucose-responding receptor: Loss of
boss causes abnormal energy metabolism. Proc Natl Acad Sci USA 105:15328–15333.
21. Colombani J, et al. (2003) A nutrient sensor mechanism controls Drosophila growth.
Cell 114:739–749.
22. Rulifson EJ, Kim SK, Nusse R (2002) Ablation of insulin-producing neurons in flies:
growth and diabetic phenotypes. Science 296:1118–1120.
23. Kim SK, Rulifson EJ (2004) Conserved mechanisms of glucose sensing and regulation
by Drosophila corpora cardiaca cells. Nature 431:316–320.
24. Levin BE (2006) Metabolic sensing neurons and the control of energy homeostasis.
Physiol Behav 89:486–489.
25. Burdakov D, Luckman SM, Verkhratsky A (2005) Glucose-sensing neurons of the
hypothalamus. Philos Trans R Soc Lond B Biol Sci 360:2227–2235.
26. Matschinsky F, et al. (1993) Glucokinase as pancreatic beta cell glucose sensor and
diabetes gene. J Clin Invest 92:2092–2098.
27. Kang L, et al. (2006) Glucokinase is a critical regulator of ventromedial hypothalamic
neuronal glucosensing. Diabetes 55:412–420.
28. Hardie DG, Hawley SA, Scott JW (2006) AMP-activated protein kinase: Development
of the energy sensor concept. J Physiol 574:7–15.
29. Buch S, Pankratz MJ (2009) Making metabolic decisions in Drosophila. Fly (Austin) 3:
74–77.
30. Melcher C, Bader R, Pankratz MJ (2007) Amino acids, taste circuits, and feeding
behavior in Drosophila: Towards understanding the psychology of feeding in flies and
man. J Endocrinol 192:467–472.
31. Pitman JL, McGill JJ, Keegan KP, Allada R (2006) A dynamic role for the mushroom
bodies in promoting sleep in Drosophila. Nature 441:753–756.
32. Fujita M, Tanimura T (2011) Drosophila evaluates and learns the nutritional value of
sugars. Curr Biol 21:751–755.
33. Burke CJ, Waddell S (2011) Remembering nutrient quality of sugar in Drosophila. Curr
Biol 21:746–750.
6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1017096108 Dus et al.