Microtubule-assisted mechanism for functional
metabolic macromolecular complex formation
Songon Ana, Yijun Denga, John W. Tomshoa, Minjoung Kyoungb, and Stephen J. Benkovica,1
aDepartment of Chemistry, Pennsylvania State University, University Park, PA 16802; andbDepartment of Molecular and Cellular Physiology, Stanford
University, Stanford, CA 94305
Contributed by Stephen J. Benkovic, June 15, 2010 (sent for review April 29, 2010)
the purinosome, that participates in de novo purine biosynthesis to
form clusters in the cytoplasm of living cells under purine-depleted
conditions. Here we identified, using fluorescent live cell imaging,
distribution of purinosomes in the cytoplasm. Application of a cell-
based assay measuring the rate of de novo purine biosynthesis
confirmed that the metabolic activity of purinosomes was signifi-
cantly suppressed in the absence of microtubules. Collectively, we
propose a microtubule-assisted mechanism for functional purino-
some formation in HeLa cells.
metabolism|protein complex|purine biosynthesis
novo purine biosynthetic pathway (Fig. 1A) have long been
hypothesized to form a multienzyme complex in cells (1–3). Our
investigation of this hypothesis in vivo successfully revealed that
the human de novo purine biosynthetic enzymes colocalize in the
cytoplasm of human cell lines upon purine depletion (Fig. 1 B
and C) (4, 5). Subsequently, we proposed a subcellular metabolic
organization for de novo purine biosynthesis, the “purinosome,”
in cells (4). More importantly, the association and dissociation of
the purinosome was regulated by changing the purine levels or by
manipulating the activity or expression levels of protein kinase
CK2 in live cells (4, 5).
Because cytoskeletal structures have been proposed to play an
important role in the organization of metabolic enzymes (3), we
explored, in this work, whether the purinosome is associated with
cellular structural elements. For example, glycolytic enzymes in-
cluding aldolase were identified as bound to actin cytoskeleton in
mammalian and yeast cells (6–8). Interestingly, dynamic alterna-
tion of actin structures during the cell cycle seems to be correlated
with the glycolysis-mediated production of ATP to satisfy an in-
creased demand for energy (9). Therefore, we sought the struc-
tural and functional relationships between the purinosome and
cellular cytoskeletal structures using human formylglycinamidine
ribonucleotide synthase (hFGAMS) fused with monomeric green/
orange fluorescent proteins (GFP/OFP) as a purinosome marker.
To visualize or manipulate the cytoskeleton in the presence of
purinosomes, we probed cellular actin networks using rhodamine-
conjugated phalloidineto stain F-actin structureswithin fixed cells
and alternatively inhibited actin polymerization by the addition of
cytochalasin D into live cells. In parallel, we stained microtubule
filaments in live cells using a TubulinTracker Green reagent
(Taxol conjugated with Oregon Green 488) and also treated live
cells with nocodazole, which directly binds to tubulin so as to in-
terfere with microtubule formation in cells. Moreover, we estab-
lished a cell-based assay monitoring the flux of de novo purine
biosynthesis to demonstrate the metabolic functionality of puri-
nosomes in the presence and the absence of small molecules.
Collectively, we propose that the spatial distribution of function-
ally active purinosomes is controlled by the network of micro-
tubules when cells demand purine production.
nzymes synthesizing inosine monophosphate through a de
Microtubule-Associated Purinosome Formation. We investigated
whether cytoskeletal structures are involved in the purinosome as-
sembly. First,we stained actin filaments with rhodamine-phalloidine
in fixed HeLa cells and also microtubules with a TubulinTracker
Green reagent in live HeLa cells in the presence of hFGAMS-GFP/
OFP as a purinosome marker. The cytosolic clusters did not coloc-
alize with the actin network (Fig. 2). However, purinosomes were
found associated with microtubule filaments in the cytoplasm (Fig.
3). In addition, we examined how purinosome assembly responds to
the small-molecule inhibitors cytochalasin D and nocodazole, which
interfere with actin and microtubule polymerization, respectively.
The addition of cytochalasin D in living HeLa cells did not have an
impact on the distribution of purinosomes (Fig. 4 A and B). How-
ever, it was apparent that the dissociation of purinosomes occurred
upon the addition of nocodazole (Fig. 4 C and D). These data with
small molecules are consistent with the cytoskeleton staining experi-
ments described above. Collectively, depolymerization of micro-
tubules by nocodazole appears to disfavor the cluster formation of
purinosomes even under purine-deficient conditions.
Suppression of Purinosome Activity by Nocodazole. We then estab-
lished a cell-based assay to monitor the flux of de novo purine
biosynthesis in the presence and the absence of inhibitors. Lawns
of HeLa cells were pulsed with [14C(U)]-glycine, a substrate of
glycinamide ribonucleotide synthetase at step 2 of de novo pu-
rine biosynthesis. Its incorporation into purines, the rate of
which represents the flux of de novo purine biosynthesis, was
determined via acid extraction and ion exchange resin column
chromatography. The14C incorporation into purines was nor-
malized to the total number of cells in the assay, plotted versus
the time after the pulse, and showed its linear incorporation as
a function of time.
Without an inhibitor, the rate of14C incorporation of glycine
into the pool of cellular purines in HeLa cells grown under purine-
depleted conditions was ∼42% greater than the rate observed for
cells grown under purine-rich conditions (Fig. 5A). We then
treated cells with the microtubule disrupting agent nocodazole to
assess its effect on the functionality of purinosomes. After a 1-h
incubation with nocodazole, [14C(U)]-glycine was similarly pulsed
for 3 h (Fig. S1). Although purine-rich HeLa cells barely respon-
ded to nocodazole (Fig. 5B), the flux of de novo purine bio-
synthesis for purine-depleted HeLa cells was suppressed by ∼36%
at the 3-h time point in the presence of nocodazole relative to the
DMSO control (Fig. 5B). This experiment clearly showed that
purine-deficient cells had diminished de novo purine biosynthesis
in the absence of microtubules. Thus, HeLa cells maintained in
Author contributions: S.A. and S.J.B. designed research; S.A., Y.D., and J.W.T. performed
research; M.K. contributed new reagents/analytic tools; S.A., Y.D., J.W.T., M.K., and S.J.B.
analyzed data; and S.A. and S.J.B. wrote the paper.
The authors declare no conflict of interest.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| July 20, 2010
| vol. 107
| no. 29www.pnas.org/cgi/doi/10.1073/pnas.1008451107
purine-depleted conditions rely on the functional clustering of
purinosomes that is spatially organized by microtubule networks.
We conducted cellular imaging of cytoskeletal structures in the
presence of purinosomes. Purinosomes were clearly embedded
within a network of microtubules, but actin filaments were not
associated or colocalized with purinosomes. In addition, dis-
ruption of the microtubule network by the addition of nocoda-
zole was sufficient to dissociate purinosomes in live HeLa cells.
The spatial distribution of purinosomes results from their being
embedded within microtubule networks.
We then demonstrated that the association and dissociation of
purinosomes were correlated with the rate of de novo purine bio-
synthesis in HeLa cells. By monitoring14C-glycine incorporation
into the pool of cellular purines, we were able to observe increased
purine biosynthesis in cells with purinosomes. More importantly,
nocodazole attenuated the metabolic flux of de novo purine bio-
synthesis in purine-depleted HeLa cells. Therefore, this cell-based
purinosome activity assay indeed supports the role of microtubules
for functional purinosome formation in live cells.
The observed localization phenotype between purinosomes
and microtubules (Fig. 3) is also supported by our previous
observations. When a region of interest containing a purinosome
cluster was photobleached, fluorescent intensities were recovered
in the same location as the photobleached area (4). In addition, in
experiments in which sequential enrichment and depletion of
purine levels triggered the dissociation and the association of
purinosomes (4), reclustering of purinosomes did not occur in the
same location. Therefore, we propose that newly forming puri-
nosomes would stochastically nucleate at any location in the cy-
toplasm guided by microtubules but would remain at that location
until they functionally dissociate.
As mentioned earlier, glycolytic enzymes are reversibly bound
to actin structures, and glycolysis-driven ATP production was
associated with actin cytoskeleton dynamics (9). In parallel, the
anticipated association of de novo purine biosynthesis with de
novo ATP synthesis suggests an advantage for the subcellular
localization of purinosomes to microtubules owing to their need
for a ready energy source to perform microtubule-mediated cel-
lular dynamics. Alternatively, because purinosomes are distrib-
uted across the entire cytoplasm, microtubule-assisted functional
pyrophosphate (PRPP) to inosine monophosphate (IMP) in 10 steps. AIRS, aminoimidazole ribonucleotide synthetase; AICAR Tfase, aminoimidazole carbox-
amide ribonucleotide transformylase; ASL, adenylosuccinate lyase; CAIRS, carboxyaminoimidazole ribonucleotide synthase; FGAMS, formylglycinamidine ri-
bonucleotide synthase; GAR, glycinamide ribonucleotide synthetase; GARS, GAR synthetase; GAR Tfase, GAR transformylase; IMPCH, IMP cyclohydrolase; PPAT,
PRPP amidotransferase; and SICARS, succinylaminoimidazole carboxamide ribonucleotide synthetase. Steps 2, 3, and 5 are catalyzed by a trifunctional enzyme,
TrifGART; steps 6and7 are catalyzed byabifunctional enzyme, PAICS;and steps 9and 10are catalyzed bya bifunctionalenzyme, ATIC. (B andC)Distribution of
hFGAMS-GFP transiently expressed in HeLa cells grown in purine-rich (B) and purine-depleted (C) media. (Scale bar, 10 μm.)
Cellular localization of hFGAMS-GFP participating in de novo purine biosynthesis. (A) De novo purine biosynthetic pathway transforms phosphoribosyl
fixed hFGAMS-GFP–forming clusters in the cytoplasm (green channel). (B) Actin networks stained by rhodamine-phalloidine in the cytoplasm (red channel).
(C) Merged image of hFGAMS-GFP (A, green in C) and actin cytoskeleton (B, red in C). (Scale bar, 10 μm.)
Localization of purinosomes and actin filaments in fixed HeLa cells grown in purine-depleted medium. (A) Transiently expressed and subsequently
An et al.PNAS
| July 20, 2010
| vol. 107
| no. 29
purinosome formation might be an alternative means of main-
taining cellular energy homeostasis throughout the cytoplasm.
Although it is possible that specific microtubule-associated pro-
teins facilitate purinosome formation, we conclude that the net-
work of microtubules minimally provides nucleation sites for
functionally active purinosome formation in the cytoplasm upon
Materials and Methods
Materials. The hFGAMS-GFP and hFGAMS-OFP constructs were prepared as
described before (4). Rhodamine-phalloidine and TubulinTracker Green
were purchased from Molecular Probes. Cytochalasin D and nocodazole
were obtained from Sigma. [14C(U)]-Glycine was from DuPont/New En-
Transfection of Mammalian Cells. A human cervical cancer cell line, HeLa
(ATCC), was maintained and transfected for this study as described before (4).
Briefly, HeLa cells were subjected to the following: “purine-depleted me-
dium,” RPMI 1640 (Mediatech) supplemented with dialyzed 5% FBS (Atlanta
Biological) and 50 μg/mL gentamicin sulfate (Sigma); and “purine-rich me-
dium,” MEM (Mediatech) with 10% FBS and 50 μg/mL gentamicin sulfate. FBS
was dialyzed against 0.9% NaCl at 4 °C for ∼2 d. Lipofectamine 2000 (Invi-
trogen) as a transfection reagent was used by following the manufacturer’s
protocol as previously described (4). Of note, an alternative purine-depleted
medium (i.e., MEM, dialyzed 10% FBS and 50 μg/mL gentamicin sulfate) was
stained by a TubulinTracker Green reagent in the cytoplasm (green channel). (B) Transiently expressed hFGAMS-OFP–forming clusters in the cytoplasm,
representing formation of purinosomes (red channel). (C) Merged image of microtubules (A, green in C) and hFGAMS-OFP (B, red in C). (D) Representative
region of interest highlighted in the white box in panel (C). Of note, the enlarged image of panel (D) was enhanced for clarification by adjustments of
brightness, contrast and/or color balance. (Scale bar, 10 μm.)
Subcellular localization of purinosomes harbored by microtubule filaments in HeLa cells grown in purine-depleted medium. (A) Microtubule networks
(C and D) was supplied to HeLa cells displaying purinosomes formed by hFGAMS-GFP. Individual images were taken before addition of the inhibitors (un-
treated; A and C) and after the cells had been incubated with the inhibitors for a given time (B, 90 min; D, 60 min). (Scale bar, 10 μm.)
Effects of small molecules on purinosome assembly formed in HeLa cells grown in purine-depleted medium. Cytochalasin D (A and B) or nocodazole
| www.pnas.org/cgi/doi/10.1073/pnas.1008451107An et al.
evaluated with respect to purinosome formation in HeLa cells by transiently
Fluorescence Microscopy of Live and Fixed Cells. All samples were imaged at
ambient temperature (∼25 °C) with a 60× objective (1.49 numeric aperture;
Nikon Apo TIRF) using a Photometrics CoolSnap ES2CCD detector mounted
onto a Nikon TE-2000E inverted microscope as described before (5). Oregon
Green 488 and GFP detection was accomplished using a S484/15x excitation
filter (Chroma Technology), S517/30m emission filter (Chroma Technology),
and Q505LP/HQ510LP dichroic (Chroma Technology). Rhodamine and OFP
detection was carried out using a S555/25x excitation filter (Chroma Tech-
dichroic (Chroma Technology).
Cytochalasin D and nocodazole were added to cells after three washes
with buffered saline solution (20 mM Hepes, pH 7.4, 135 mM NaCl, 5 mM KCl,
1 mM MgCl2, 1.8 mM CaCl2and 5.6 mM glucose). Cells transiently expressing
hFGAMS-GFP were imaged before and after the addition of either 2 μL cy-
tochalasin D (1 mg/mL in DMSO) or 4 μL nocodazole (4.2 mg/mL in DMSO) to
give final concentrations of 1 μg/mL cytochalasin D and 8 μg/mL nocodazole,
respectively. Control experiments were also performed by the addition of
4 μL DMSO.
To stain cellular microtubule filaments, live HeLa cells transiently ex-
pressing hFGAMS-OFP were washed with buffered saline solution, followed
by incubation with a TubulinTracker Green reagent (1 mM in DMSO; final
concentration, 250 nM) at 37 °C for 30 min. In addition, to investigate the
cellular distribution of actin filaments in fixed HeLa cells, cells were prepared
similarly to those used for live cell imaging; however, the cells transfected
with hFGAMS-GFP were fixed with freshly prepared 3% formaldehyde,
permeabilized with 0.2% Triton X-100, and blocked with 10% normal goat
serum (Jackson ImmunoResearch Laboratory) for 30 min at RT as described
before (4). The cells were then incubated for 20 min at RT with rhodamine-
phalloidine (8.3 μM in DMSO; 5 μL/sample) in PBS (PBS: 10 mM Na2HPO4,
pH 7.4, 2 mM KH2PO4, 137 mM NaCl, and 2.7 mM KCl).
Determination of de Novo Purine Biosynthetic Rates. The rate of de novo
purine synthesis was determined by the incorporation of [14C(U)]-glycine
(DuPont/New England Nuclear, NEC-276E, 111.70 mCi/mmol) into cellular
purines using the method of Boss and Erbe (10). HeLa cells were maintained
in purine-rich and purine-depleted media (MEM supplemented with 10%
FBS or 10% dialyzed FBS, respectively, with 50 μg/mL gentamicin sulfate) for
at least three passages. These cells were then seeded into T75 flasks con-
taining the appropriate, gentamicin-free media at 2 × 106and 3 × 106cells/
flask for the purine-rich and purine-depleted conditions, respectively. After
allowing ≈36 h for the cells to achieve midlog phase growth, the cells were
placed in 2 mL fresh media. After reequilibration, the cells were pulsed with
[14C(U)]-glycine (125 μM, 20 mCi/mmol, 5 μCi/flask) for the desired time. The
media was aspirated, and the cells were washed three times with 10 mL ice-
cold Dulbeccos’s PBS (Cellgro). Cells were harvested by treatment with
0.25% trypsin–EDTA solution.
To each cell pellet, 1 mL perchloric acid (0.4 M) was added, followed by
vigorous vortexing to suspend cells. Incubation of cell suspensions in a boiling
water bath for 1 h completely lysed the cells and extracted all purines. Im-
mediately after the acid extraction, the tubes were cooled on ice. Cellular
debris was pelleted by centrifugation and the supernatant was loaded onto
0.8 × 3 cm AG50W-X8 (100–200 mesh, Bio-Rad) columns that had been
preequilibrated with 0.1 M HCl. The columns were washed with 5 mL HCl
(1 M), and purines were then eluted with 5 mL of HCl (6 M). Quantitation
was achieved by mixing 1 mL of the eluant with 10 mL Ecoscint (National
Diagnostics) followed by liquid scintillation counting using a Beckman
Coulter LS6500 instrument. To obtain comparable rates between purine-rich
and purine-depleted conditions, de novo purine biosynthesis was normal-
ized to the total number of cells.
Effects of Nocodazole on de Novo Purine Biosynthetic Rate. HeLa cells cultured
in purine-rich and purine-depleted media were inoculated into T75 flasks
following the protocol for measuring the de novo purine biosynthetic rate.
On the day of assay, cells were rinsed and equilibrated in buffered saline
solution for 1 h before adding nocodazole (final concentration, 8 μg/mL) or
DMSO as control. After an additional 1 h incubation with nocodazole or
DMSO, the cells were pulsed with [14C(U)]-glycine (125 μM, 20 mCi/mmol,
5 μCi/flask) and incubated at 37 °C for 1, 2, and 3 h before harvesting the
cells to measure14C incorporation into purines as described above.14C in-
corporation into newly synthesized purines was normalized to the total
number of cells. We also performed an unpaired one-tailed Student t test
using Microsoft Excel to determine whether the effect of nocodazole on
purine-depleted HeLa cells was statistically significant.
ACKNOWLEDGMENTS. This work was funded by National Institutes of
Health Grant GM24129 (to S.J.B.).
[14C(U)]-glycine incorporated into cellular purines in HeLa cells cultured in purine-rich (■) and purine-depleted (□) media. Incorporation was found to be
linear with time up to 4 h, and ratio of de novo purine biosynthesis rates in purine-depleted to purine-rich media was 1.42 by fitting data with the least-
squares line method (10, 11). However, data could alternatively be fit with a single exponential function, resulting in larger difference between the two data
sets (i.e., 1.60). Error bar indicates SD of three independent assays. Of note, the data points at t = 0 from the two cell culture conditions overlap. (B) Effects of
nocodazole on de novo purine biosynthesis was evaluated in a similar way by measuring [14C(U)]-glycine incorporation for 3 h (Fig. S1). For each type of cells,
de novo purine biosynthesis was compared in the absence (i.e., DMSO) and presence of nocodazole. Purine biosynthesis in purine-depleted HeLa cells was
decreased by ∼36% in the presence of nocodazole at 3 h. Bar height is the14C incorporation into purines per million cells. Error bar indicates SD of three
independent assays. *Unpaired one-tailed Student t test revealed that the effect of nocodazole on purine-depleted HeLa cells was statistically significant (P <
0.001). It should be noted that the cells for Fig. 5A were maintained in the preferred growth medium until harvesting, whereas the cells for Fig. 5B were rinsed
with buffered saline solution to be treated with nocodazole and then maintained in buffered saline solution until harvesting, to be consistent with cellular
Metabolic flux measurement of de novo purine biosynthesis for HeLa cells. (A) De novo purine biosynthesis is measured by determining amount of
An et al.PNAS
| July 20, 2010
| vol. 107
| no. 29
1. Rowe PB, Wyngaarden JB (1968) Glutamine phosphoribosylpyrophosphate ami-
dotransferase. Purification, substructure, amino acid composition, and absorption
spectra. J Biol Chem 243:6373–6383.
2. Smith GK, Mueller WT, Wasserman GF, Taylor WD, Benkovic SJ (1980) Charac-
terization of the enzyme complex involving the folate-requiring enzymes of de novo
purine biosynthesis. Biochemistry 19:4313–4321.
3. Srere PA (1987) Complexes of sequential metabolic enzymes. Annu Rev Biochem 56:
4. An S, Kumar R, Sheets ED, Benkovic SJ (2008) Reversible compartmentalization of de
novo purine biosynthetic complexes in living cells. Science 320:103–106.
5. An S, Kyoung M, Allen JJ, Shokat KM, Benkovic SJ (2010) Dynamic regulation of
a metabolic multi-enzyme complex by protein kinase CK2. J Biol Chem 285:
6. Minaschek G, Gröschel-Stewart U, Blum S, Bereiter-Hahn J (1992) Microcom-
partmentation of glycolytic enzymes in cultured cells. Eur J Cell Biol 58:418–428.
7. Pagliaro L, Taylor DL (1988) Aldolase exists in both the fluid and solid phases of
cytoplasm. J Cell Biol 107:981–991.
8. Götz R, Schlüter E, Shoham G, Zimmermann FK (1999) A potential role of the
cytoskeleton of Saccharomyces cerevisiae in a functional organization of glycolytic
enzymes. Yeast 15:1619–1629.
9. Bereiter-Hahn J, Stübig C, Heymann V (1995) Cell cycle-related changes in F-actin
distribution are correlated with glycolytic activity. Exp Cell Res 218:551–560.
10. Boss GR, Erbe RW (1982) Decreased purine synthesis during amino acid starvation of
human lymphoblasts. J Biol Chem 257:4242–4247.
11. Yamaoka T, et al. (2001) Feedback inhibition of amidophosphoribosyltransferase
regulates the rate of cell growth via purine nucleotide, DNA, and protein syntheses.
J Biol Chem 276:21285–21291.
| www.pnas.org/cgi/doi/10.1073/pnas.1008451107An et al.