Imaging of enzyme replacement therapy using PET

Article abstract of DOI:10.1073/pnas.1003247107

Direct enzyme replacement therapy (ERT) has been introduced as a means to treat a number of rare, complex genetic conditions associated with lysosomal dysfunction. Gaucher disease was the first for which this therapy was applied and remains the prototypical example. Although ERT using recombinant lysosomal enzymes has been shown to be effective in altering the clinical course of Gaucher disease, Fabry disease, Hurler syndrome, Hunter syndrome, Maroteaux-Lamy syndrome, and Pompe disease, the recalcitrance of certain disease manifestations underscores important unanswered questions related to dosing regimes, tissue half-life of the recombinant enzyme and the ability of intravenously administered enzyme to reach critical sites of known disease pathology.We have developed an innovative method for tagging acid β-glucocerebrosidase (GCase), the recombinant enzyme formulated in Cerezyme used to treat Gaucher disease, using an ¹⁸F-labeled substrate analogue that becomes trapped within the active site of the enzyme. Using micro-PET we show that the tissue distribution of injected enzyme can be imaged in a murine model and that the PET data correlate with tissue ¹⁸F counts. Further we show that PET imaging readily monitors pharmacokinetic changes effected by receptor blocking. The ability to ¹⁸F-label GCase to monitor the enzyme distribution and tissue half-life in vivo by PET provides a powerful research tool with an immediate clinical application to Gaucher disease and a clear path for application to other ERTs.

Full text of DOI:10.1073/pnas.1003247107

Imaging of enzyme replacement therapy using PET
Christopher P. Phenixa,b,2, Brian P. Rempelb,3, Karen Colobongc, Doris J. Doudetd, Michael J. Adama,
Lorne A. Clarkec, and Stephen G. Withersb,1
aTri-University Meson Facility (TRIUMF), 4004 Wesbrook Mall, Vancouver, BC, Canada V6T 2A3; bDepartments of Chemistry and Biochemistry, University of
British Columbia, Vancouver, BC, Canada V6T 1Z1; cChild and Family Research Institute, Department of Medical Genetics, University of British Columbia,
950 W28th Avenue, Vancouver, BC, Canada V5Z 4H4; and dDepartment of Medicine, Division of Neurology, 2221 Wesbrook Mall, Vancouver, BC, Canada
V6T 2B5
Edited* by Joanna S. Fowler, Brookhaven National Laboratory, Upton, NY, and approved April 29, 2010 (received for review March 14, 2010)
Direct enzyme replacement therapy (ERT) has been introduced as
a means to treat a number of rare, complex genetic conditions
associated with lysosomal dysfunction. Gaucher disease was the
first for which this therapy was applied and remains the prototypi-
cal example. Although ERT using recombinant lysosomal enzymes
has been shown to be effective in altering the clinical course of
Gaucherdisease, Fabry disease, Hurler syndrome, Hunter syndrome,
Maroteaux-Lamy syndrome, and Pompe disease, the recalcitrance
of certain disease manifestations underscores important unan-
swered questions related to dosing regimes, tissue half-life of the
recombinant enzyme and the ability of intravenously administered
enzyme to reach critical sites of known disease pathology. We have
developed an innovative method for tagging acid β-glucocerebrosi-
dase (GCase), the recombinant enzyme formulated in Cerezyme®
used to treat Gaucher disease, using an 18F-labeled substrate analo-
gue that becomes trapped within the active site of the enzyme.
Using micro-PET we show that the tissue distribution of injected
enzyme can be imaged in a murine model and that the PET data
correlate with tissue 18F counts. Further we show that PET imaging
readily monitors pharmacokinetic changes effected by receptor
blocking. The ability to 18F-label GCase to monitor the enzyme
distribution and tissue half-life in vivo by PET provides a powerful
research tool with an immediate clinical application to Gaucher
disease and a clear path for application to other ERTs.
the short half-life of 18F (109.8 min) renders the incorporation of
such labels during protein synthesis impossible. However, specific
integration of an 18F-label directly into the active site of an en-
zyme would avoid these issues, allowing imaging to be performed.
Given the difficulties associated with labeling the interior of the
protein, it is perhaps not surprising that no such studies have been
reported for monitoring administered enzyme therapeutics.
Gaucher Disease was the first lysosomal storage disease to be
treated by ERTand serves as the prototypical model for this ther-
apeutic approach. Gaucher Disease results from a deficiency in
GCase, a lysosomal β-glucosidase that catalyzes the hydrolysis of
β-glucosylceramide (GlcCer) to ceramide and β-D-glucose. This
enzyme plays a critical role in glycosphingolipid metabolism by its
influence on ceramide homeostasis (see Fig. 1A). Mutations in
the GCase gene (2) result in enzyme deficiency, which leads to
primary accumulation of GlcCer and, through complex and
poorly understood mechanisms, produces the symptoms of Gau-
cher Disease (3). Features common to all forms and thought to be
mediated by GlcCer storage within tissue macrophages include
hepatosplenomegaly, thrombocytopenia, and anemia.
Produced from Chinese hamster ovary cells, the first commer-
cially available ERT, Cerezyme®, is a recombinant GCase mod-
ified to ensure that N-linked glycans contain terminal mannose
residues, thereby conferring high affinity for the mannose recep-
tor resident on macrophages and other cell types (4) (See Fig. 2).
GCase belongs to a class of β-glycosidases that catalyze the
hydrolysis of glycosidic bonds via a two step, double displacement
mechanism utilizing a pair of glutamic acid residues (see Fig. 1B)
(5). In GCase, Glu340 acts as the catalytic nucleophile, displacing
the ceramide moiety from the anomeric carbon of glucose to form
a covalent glucosyl-enzyme intermediate (6). When presented
with substrate analogues in which the 2-hydroxyl has been
replaced by fluorine and a reactive leaving group installed at
the anomeric carbon (C-1), such enzymes form a long-lived cova-
lent glycosyl-enzyme intermediate (7) (see Fig. 1B). Importantly,
structures of several such trapped intermediates reveal that no
significant protein conformational changes accompany formation
of this intermediate, thus its biodistribution is not expected to
differ from that of free enzyme (8). If synthesized with a radio-
active 18F in place of the stable and common 19F isotope, these
inhibitors could serve as covalent, active-site directed, radioactive
labeling agents of GCase. Once radioactively labeled with the
mechanism-based enzyme inhibition ∣ PET Imaging ∣
Lysosomal Storage diseases ∣ Gaucher disease
he use of enzyme replacement therapies (ERT) to treat lyso-
T
somal storage diseases is increasing steadily as more recom-
binant enzymes become available. Central to the development of
ERT for lysosomal storage disorders was the elucidation of carbo-
hydrate-dependent receptor-mediated plasma membrane uptake
and endosomal targeting of lysosomal enzymes. Consequently the
enzymes employed must be administered in the appropriate glyco-
form. Although ERT has been demonstrated to be safe and
efficacious in a number of lysosomal storage diseases (1), there
are many unanswered questions related to appropriate dosing
regimens; the ability of recombinant proteins to target key tissues,
the in vivo half-lives of recombinant proteins in specific tissues as
well as long term efficacy. Advancements in these key areas would
significantly improve the design of recombinant proteins for clin-
ical use and ensure the best clinical outcomes for treated patients.
Many of these questions could be addressed by PET imaging.
Here, we demonstrate a unique method for labeling a specific
recombinant enzyme without modification of the exterior of the
protein and highlight the potential of PET imaging for monitoring
the biodistribution of therapeutic enzymes.
Author contributions: C.P.P., B.P.R., D.J.D., M.J.A., L.A.C., and S.G.W. designed research;
C.P.P., B.P.R., and K.C. performed research; M.J.A. contributed new reagents/analytic tools;
C.P.P., D.J.D., L.A.C., and S.G.W. analyzed data; and C.P.P., D.J.D., L.A.C., and S.G.W. wrote
the paper.
The authors declare no conflict of interest.
The satisfactory incorporation of 18F onto biomolecules such
as proteins, peptides, and antibodies is often very challenging as
structural modifications introduced by the surface attachment of
a radiolabel may alter the biological activity and/or physical and
metabolic stability of the final conjugate. The ideal situation
would involve the incorporation of a small radiolabel into the
interior of the protein without significant perturbation of protein
structure or of the behavior of the protein in vivo. Unfortunately
*This Direct Submission article had a prearranged editor.
1To whom correspondence should be addressed. E-mail: withers@chem.ubc.ca.
2Present address: Thunder Bay Regional Research Institute, 290 Munro Street, Thunder
Bay, ON, Canada P7A 7T1.
3Present address: Department of Science, The University of Alberta, Augustana Campus,
4901–46 Avenue, Camrose, AB, Canada T4V 2R3.
This article contains supporting information online at www.pnas.org/lookup/suppl/
doi:10.1073/pnas.1003247107/-/DCSupplemental.
10842–10847 ∣ PNAS ∣ June 15, 2010 ∣ vol. 107 ∣ no. 24
www.pnas.org/cgi/doi/10.1073/pnas.1003247107

A
B
Fig. 1. The biochemical reaction catalyzed by GCase and the mechanism of catalysis A) Reaction catalyzed by GCase during glycosphingolipid metabolism
B) Double displacement mechanism used by GCase when X ¼ OH and OR ¼ ceramide. When the 2-OH is replaced by F and the Ceramide replaced by 2,
4-dinitrophenol, these substrate analogues form a long-lived covalent glycosyl-enzyme intermediate.
positron-emitting 18F isotope, noninvasive biodistribution studies
of the therapeutic enzyme are possible using PET.
efficient, and stereoselective coupling process between FDG
and fluoro-2, 4-dinitrobenzene that is achievable at most radio-
chemical facilities that synthesize [18F]-FDG (see Materials and
Results and Discussion
Radiochemical Labeling of β-Glucosidases. The development of a
suitable 18F-labeled mechanism-based inhibitor for β-glucosidases
required the exploration of several different means of incorporat-
ing the isotopic label. While anattractive choicewas to incorporate
18F at the C-2 position, an alternative option was to presynthesize
the cold 2-fluoro sugar fluoride and incorporate the 18F isotope at
a remote position on the sugar structure. In our initial attempts we
radiochemically synthesized and tested two activated [18F]-fluor-
oglycoside derivatives: 2-deoxy-2-fluoro-β-mannosyl fluoride, and
2, 6-dideoxy-2, 6-difluoro-β-glucosyl fluoride (9, 10). The former
could be readily synthesized by 18F2 addition to a protected D-glu-
cal, while the latter could be generated by 18F− displacement of a
6-sulfonate derivative of 2-deoxy-2-fluoro-β-glucosyl fluoride.
However, each agent suffered from either a lengthy synthesis or
low bioconjugation rates, thus neither were useful.
Recognition of
glycans by MR
Intravenous
GCase
Mannose
receptor (MR)
Endocytosis
Receptor
recycling
Normal lysosome
Early endosome
Based upon published kinetic data for inactivation of GCase,
we hypothesized that the ideal radiolabeled reagent might be
2-deoxy-2-[18F]-fluoro-β-D-glucosyl fluoride, which is known to
Lysosome
with excessive
GlcCer storage
react reasonably rapidly with GCase (ki∕Ki ¼ 20 M−1 min−1
)
(6). Unfortunately, synthesis of this compound would involve
three chemical steps after 18F incorporation, making its assembly
too time-consuming. An alternative reagent would be 2, 4-dinitro-
phenyl 2-deoxy-[18F]-2-fluoro-β-D-glucopyranoside (β-DNP-
[18F]-FDG) since the cold version of this compound is a known
inactivator of other β-glucosidases, but was unknown as an inacti-
vator of GCase (7). We therefore synthesized the cold version of
this compound and confirmed it to be an inactivator with a com-
parable inactivation rate constant to that of 2-deoxy-2-fluoro-β-D-
glucosyl fluoride for GCase (ki∕Ki ¼ 12 M−1 min−1). This result
opened up the attractive option of a one-step synthesis of β-DNP-
[18F]-FDG directly from pharmaceutical grade 2-deoxy-[18F]-2-
fluoro-D-glucose ([18F]-FDG), the most widely used PET radio-
pharmaceutical. If we could optimize reported procedures (11)
for the direct formation of dinitrophenyl glycosides from free
sugars plus fluorodinitrobenzene, an 18F glucosidase labeling
agent amenable to PET imaging would be produced. Indeed, by
iterative optimization ofreaction conditions, varying base, concen-
tration, temperature, and reaction time we developed a rapid,
GlcCer
hydrolysis
Late endosome
Fusion
:
:
GlcCer
N-linked
GlcNAc2Man3
Macrophage
Fig. 2. Macrophage uptake of intravenously administered recombinant
GCase (structure used in this figure taken from Dvir et al. (21)). The recombi-
nant GCase in Cerezyme, produced from Chinese hamster ovary cells, is mod-
ified to ensure that N-linked glycans contain terminal mannose residues thus
conferring high affinity for the mannose receptor (MR) resident on the macro-
phage and other cell types. Internalization of the enzyme occurs through
endocytosis of the MR receptor/GCase complex. Early endosomes containing
GCase mature into late endosomes while receptor sorting leads to the
recycling of the MR receptor back to the cellular membrane. Late endosomal
and lysosomal fusion delivers the therapeutic enzyme to the lysosome which
subsequently restores GlcCer to normal levels.
Phenix et al.
PNAS
∣
June 15, 2010
∣
vol. 107
∣
no. 24
∣
10843

we used a mutant Abg (Abg E358A) possessing an identical pri-
mary sequence to natural Abg, except that the active-site nucleo-
phile, Glutamic acid 358, was mutated to an Alanine (16). Such
enzymes have been shown to be catalytically incompetent,
although they still retain their secondary and tertiary structure
(17). RTLC analysis revealed that incubation of β-DNP-
[18F]-FDG with Abg E358A did not result in radioactive tagging
of the protein, clearly demonstrating that the fully intact catalytic
machinery in the active site of natural Abg is vital for the radio-
active labeling of the enzyme, and ruling out the incorporation of
18F in a nonspecific manner.
Fig. 3. Radiochemical synthesis of the glycosidase labeling agent β-DNP-
18F]-FDG and active-site labeling of GCase.
[
Methods and Fig. 3). HPLC and radio-TLC (RTLC) analysis using
authentic standards revealed that the major radioactive species
present after 10 min reaction was the desired β-DNP-
[18F]-FDG (∼80% radiochemical yield) with some [18F]-FDG
remaining and a third unidentified species that did not interfere
with enzymatic labeling and was easily removed by gel filtration
(see Fig. S1). Interestingly no 2, 4-dinitrophenyl 2-deoxy-[18F]-
2-fluoro-α-D-glucopyranoside was detected by HPLC. This find-
ing is consistent with earlier studies of this arylation reaction,
which revealed that the reaction was under kinetic control, with
predominant formation of the β-glycoside being due to the greater
reactivity of the β-hemiacetal in conjunction with rapid equilibra-
tion of the α- and β-hemiacetal anomers of the unreacted sugar.
(12, 13) Indeed the reaction yield and selectivity were sensitive
to temperature as no significant amounts of β-DNP-[18F]-FDG
were formed at shorter reaction times, neither at room tempera-
ture, nor at temperatures higher than 40 °C.
Reactivation of the 18F-fluoroglycosylated form of Abg. The F-Glc-
Abg intermediate is a stable but catalytically competent species
that breaks down very slowly by hydrolysis (t1∕2 ¼ 57;000 min)
but undergoes more rapid reactivation by transglycosylation onto
catalytically relevant sugars. Consistent with this, the addition of
20 mM thiophenyl β-D-glucopyranoside (TG) to the F-Glc-Abg
intermediate accelerates the transglycosylation reaction (t1∕2
¼
53 min) resulting in the formation of a new disaccharide and re-
generation of the catalytically active Abg enzyme (see Fig. S3)
(18). We employed this reactivating agent as an important control
to demonstrate that the radiolabeling of Abg was in fact mediated
through the 2-fluoroglucosyl-Abg intermediate in the active site of
the enzyme. Treatment of 18F-Glc-ABG with 60 mM TG resulted
in the relatively rapid formation of a new radioactive peak posses-
sing an Rf corresponding toaneutral phenyl glycoside, along with a
proportional decrease in the peak for 18F-Glc-ABG (see Fig. S4).
The time scale of this reaction was consistent with the previously
measured half-life, thereby providing solid evidence that the
radioactive tagging of Abg is through a covalent 18F-fluoroglyco-
syl-enzyme intermediate. Due to the known, very slow turnover of
Radiolabeling of Abg and GCase. Two different enzyme systems were
employed in the evaluation of approaches and reagents.
Initial studies were performed on the well characterized β-gluco-
sidase from Agrobacterium sp. (Abg). Abg is a stable, easily
handled enzyme that has been extensively studied by our group
(14, 15). Having worked out ideal conditions and proven the
active-site location of the label, studies were then performed on
the therapeutically relevant enzyme GCase.
the 2-fluoroglucosyl enzyme intermediate on GCase (t 21
>
1;300 min) it was not possible to perform the equivalent controls
on this enzyme with 18F-labeled reagents (t 21 ¼ 109.8 min) (6).
Covalent inactivation of Abg by β-DNP-FDG is known to occur
rapidly, with a second-order rate constant for formation of the
covalent fluoroglucosylated-ABG (F-Glc-Abg) intermediate
(ki∕Ki) of 500 mM−1 min−1. As expected, incubation of freshly
synthesized β-DNP-[18F]-FDG with Abg resulted in the formation
of radioactive enzyme within several minutes. By contrast, GCase
is inactivated at a much lower rate (ki∕Ki of 0.012 mM−1 min−1),
largely due to a much higher Ki value, and consequently required
longer incubation times for radioactive labeling. To minimize
this problem, β-DNP-[18F]-FDG recovered from the HPLC was
evaporated to dryness and dissolved in minimal amounts of assay
buffer (<200 μL) prior to addition of concentrated GCase to keep
its concentration as high as possible. Reaction times required for
the radioactive tagging of GCase were stillsignificantly longer than
for Abg, typically about 45 min in total, but now consistent with
the requirements for isotopic labeling with 18F required for
PET imaging. As seen in the RTLC analysis shown in Fig. S2, effi-
cient labeling was observed.
Inhibition of GCase labeling by 6-nonylisofagomine. In order to
demonstrate that GCase labeling was an active-site-dependent
process, radioactive tagging was attempted in the presence of
6-nonylisofagomine, a potential active-site-specific chaperone
known to be a potent inhibitor (IC50 ¼ 0.6 nM) of the enzyme
(see Fig. S5) (19). Indeed, incubation of GCase with β-DNP-
[18F]-FDG and 6-nonylisofagomine under identical conditions
to those used previously resulted in no incorporation of radiolabel
into the enzyme.
Isolation of Radiochemically Pure 18F-Glc-GCase. Performance of
biodistribution studies of 18F-Glc-GCase in mice, required access
to radiochemically pure 18F-Glc-GCase. Initial attempts to re-
move excess β-DNP-[18F]-FDG and other radioactive molecules
by HPLC using several different gel filtration columns failed as
GCase was retained in the matrix of the gel, a behavior reported
previously (20). Fortunately, it was found that GCase remains
soluble at high concentrations in 50 mM MES buffer containing
150 mM NaCl without the need for stabilizing detergents (21). By
sequential repetitive dilution of the taurocholate and Triton-
X100 with MES buffer and concentration by ultrafiltration using
Centricon 50 kDa centrifugal devices, the radiolabeled enzyme
was concentrated, while simultaneously removing excess
β-DNP-[18F]-FDG and other contaminating small molecules. A
final clean up step and buffer exchange was accomplished using
a desalting spin column previously saturated with buffer. Follow-
ing this procedure, we were able to routinely isolate about
200 μCi of radiochemically pure 18F-Glc-GCase (see Fig. S6)
for PET studies in about 2.5 h, including all radiochemical steps.
This synthesis should be scalable to permit clinical imaging of
Gaucher patients. However, to limit radiation exposure to the
Demonstration That 18F Labeling Occurs in the Enzyme’s Active Site.
While the 18F tagging of both enzymes was demonstrated by
RTLC and size exclusion chromatography, we required further
evidence that the labeling chemistry was occurring in the active
sites of each enzyme. A variety of control experiments on GCase
and Abg, including protection against inactivation and acceptor-
dependent reactivation, demonstrated that radioactive tagging
was active-site-specific, as follows.
Mutant Abg lacking the catalytic nucleophile is not radioactively
tagged. To demonstrate that the radioactive labeling of Abg
was dependent on the intact catalytic machinery of the enzyme,
10844
∣
www.pnas.org/cgi/doi/10.1073/pnas.1003247107
Phenix et al.

radiochemist, automation of the chromatography steps would be
required but is entirely feasible.
An advantage of using [18F]-FDG as the radioactive tag is that,
in the event of enzymatic turnover or proteolytic degradation of
18F-Glc-GCase, the resulting radioactivity released as [18F]-
FDG would be rapidly phosphorylated by hexokinase and unlikely
to enter back into circulation within the time frame of the scan.
However, even this scenario is unlikely since 18F-Glc-GCase
has a half-life of hydrolysis of 1,300 min in vitro, with a similar rate
being observed in mice (27). Significant release of [18F]-FDG from
the active site through spontaneous hydrolysis or metabolic break-
down of 18F-Glc-GCase is therefore highly unlikely to compromise
the biodistribution data over the 2 h micro-PETacquisition time.
Biodistribution and PET Imaging of 18F-Glc-GCase. To study the in vivo
biodistribution of GCase, radiochemically pure 18F-Glc-GCase
was administered as a bolus tail vein injection into healthy mice,
and the tissue distribution analyzed both by measuring the
amounts of radioactivity in postmortem tissue samples and non-
invasively using micro-PET. Radioactive uptake was observed in
various tissues and organs (see Fig. 4A), and data compared favor-
ably to the limited GCase biodistributions previously reported
based on enzymatic assays of tissue homogenates (22–24). Macro-
phage-rich organs such as the liver and spleen demonstrated the
highest overall uptake. Interestingly, uptake was also observed in
the gall bladder, bones (femur, vertebrae, pelvis, and shoulders),
bone marrow, kidneys, intestines, thymus, and transiently in the
heart. Insignificant amounts of radioactivity were detected in
the brain, and skeletal muscle. Blood sampling by heart puncture
at the end of some studies revealed that, by 1 h postinjection, very
little radioactivity remained in the circulation (<0.5% injected
dose), thereby demonstrating rapid enzyme uptake by some
tissues. Tissue activity in micro-PET studies correlated well with
the data obtained from tissue activity counts from killed animals
(see Fig. 4B) and images were reconstructed to illustrate organ
uptake (see Fig. 5 A and B). Higher resolution analysis such as
cellular distribution is precluded by the inherent spatial limitations
of the PET technique. Elimination of 18F-Glc-GCase was ob-
served through renal and hepatobiliary routes as radioactivity
was detected in the urinary bladder, gall bladder, and large intes-
tine in later time points. Renal elimination is not surprising given
that the molecular mass of the recombinant GCase is ∼60 kDa
(25), and the molecular weight cut off value for glomerular filtra-
tion is estimated to be about 60 kDa (26). Different biodistribu-
tions may be observed if the labeled enzyme is administered by
infusion rather than as a bolus injection.
Receptor Blocking Experiments. In order to test the ability of this
imaging approach to readily and noninvasively monitor how the
enzyme biodistribution can be controlled through coadministra-
tion of external agents, or possibly by modifying surface glycan
structures on the enzyme, receptor blocking experiments were per-
formed. The macrophage uptake of glycoproteins such as recom-
binant GCase from the circulatory system relies on surface glycans
terminating in mannose and N-acetylglucosamine, and primarily
occurs through the mannose/N-acetylglucosamine (MR) receptor
system and other receptor types (28) (Fig. 2). The mannose recep-
tor-ligand complex is rapidly internalized and recycling of new
receptors is thought to occur (29). To demonstrate that radioactive
uptake in the liver was mannose receptor-dependent and to deter-
mine the ability of the PET-imaging approach to monitor changes
in uptake, we performed a receptor blocking experiment by the
intravenous administration of mannan prior to injection of
18F-Glc-GCase. As can be seen in Fig. 4 C and D and Fig. 5 C
and D, a significant decrease in the rate of uptake by the liver
was observed for mice pretreated with mannan when compared
to control mice receiving only 18F-Glc-GCase. After 2 h the total
amounts of 18F-Glc-GCase accumulating in the liver of the man-
nan-treated mice approached that of the control mice, consistent
with decreasing tissue mannan concentrations and receptor recy-
cling. These results demonstrate how this PET-imaging approach
A
B
C
D
Fig. 4. Uptake of 18F-Glc-GCase in various organs. A) Biodistribution of 18F-Glc-GCase in various tissues at 2 h postinjection in mice. Mice were killed and tissue
radioactivity was measured by γ counting (n ¼ 4 for all organs except the gall bladder where n ¼ 2) Bars represent the average tissue uptake plus/minus
standard deviation (expressed as % injected dose∕g of tissue). B) Comparison of the biodistribution of 18F-Glc-GCase as determined by PET (blue bars expressed
% injected dose∕mL) and by postmortem tissue γ counting (purple bars expressed % injected dose∕g) 2 h postinjection. C) Liver uptake of 18F-Glc-GCase
determined by PET in an untreated (green triangles) and a mannan-treated mouse (blue diamonds). The red square represents the postmortem tissue activity
counts of the mannan-treated mouse 2.5 h postinjection D) Bar graphs comparing the rate of liver uptake for untreated mice vs mice receiving yeast mannan
(15 mg∕kg) prior to injection of 18F-Glc-GCase (n ¼ 2 in each instance). The rates of uptake were determined by calculating the slope of the linear portion of the
liver uptake curves over the first 15 min postinjection.
Phenix et al.
PNAS
∣
June 15, 2010
∣
vol. 107
∣
no. 24
∣
10845

(preequilibrated with 1 mL of H2O) and a further 1 mL of H2O was passed
through the Sep Pak, collecting four fractions of approximately 0.33 mL.
The second fraction containing the bulk of the radioactivity (generally
between 3–5 mCi) was purified on a Synergy semiprep reverse phase HPLC
column (Phenomenex) using a gradient of 60% CH3OH/40% H2O to 100%
CH3OH over 10 min. The solvent composition was maintained at 100%
CH3OH for a further 10 min. The DNP-[18F]-FDG was collected in a 2.0 mL
v-vial at approximately 14 min in mobile phase which consisted of a high
percentage of CH3OH, which was evaporated at 60 °C.
Kinetic Analysis of the Inactivation of GCase by β-DNP-FDG. β-DNP-FDG was
chemically synthesized as described previously (7). To measure inactivation
rates, varying concentrations of β-DNP-FDG in assay buffer (180 μL total
volume, 20 mM citrate, 60 mM phosphate, pH 5.5, containing 1 mM EDTA,
0,25% (v∕v) Triton X-100, and 0.25% taurocholate) were warmed to 37 °C.
The inactivation was initiated by the addition of GCase (20 μL, 0.5 μg∕mL)
obtained from the unused portions of patient vials and diluted into assay
buffer. At appropriate time points, 20 μL aliquots were withdrawn and added
to a solution containing 2–10× the Michaelis constant, Km of 2, 4-dinitrophe-
nyl β-D-glucopyranoside, along with a large volume of buffer (600 μL)
warmed to 37 °C. This halts the inactivation reaction both by dilution of
the β-DNP-FDG and by competition with a large excess of substrate. The re-
sidual enzyme activity was then assayed using a continuous enzymatic assay
monitoring the rate of release of 2, 4-dinitrophenolate (2, 4-DNP) at a
wavelength of 400 nm. The rate of 2, 4-DNP release is directly proportional
to the concentration of catalytically active enzyme remaining in the solution.
Pseudo-first order rate constants for the loss of enzymatic activity for each
concentration of inactivator were determined by nonlinear fitting of the de-
cay curve to a single exponential decay equation using GraFit. The individual
kobs values were then replotted as a function of inactivator concentration.
Individual values for ki and Ki were obtained by fitting to the equation:
Fig. 5. PET images obtained by injection of 18F-Glc-GCase into mice. Images
are as follows: A) Coronal slice highlighting the accumulated uptake of tracer
in spleen, kidneys, and liver over 2 h acquisition time B) Sagittal slice of the
image shown in A) but highlighting the liver, bladder, and spine C) For com-
parison purposes the accumulated tracer uptake into organs of an untreated
mouse summed over the first 20 min postinjection D) Accumulated tracer
uptake into the organs of a mouse treated with mannan summed over
the first 20 min postinjection
will allow direct analysis of the consequences of modification of
administration protocols as well as administered reagents without
bias arising from the modification of the exterior of the protein.
kobs ¼ ki½Iꢀ∕Ki þ ½Iꢀ:
Conclusion
In summary, we have developed a unique mechanism-based meth-
od for radioactively labeling β-glycosidases using β-DNP-[18F]-
FDG derived from the widely available [18F]-FDG, and used this
method to monitor the biodistribution of the therapeutic enzyme
GCase in mice. Our preliminary results indicate that PET-imaging
may be useful in detailed studies of ERT including pharmacoki-
netic analysis of enzyme uptake in various tissues in vivo, as well
as in the development of methods to alter tissue distribution and
refine the targeting of recombinant proteins. This noninvasive
technology could be applicable to Gaucher Disease patients
and as such should provide valuable insights into the pharmaco-
kinetic properties of recombinant enzyme and the role of perso-
nalized dosing regimens based on these in vivo studies. In addition,
this PET-imaging approach will enable direct in vivo comparison
of the recently introduced preparations of recombinant GCase
derived from plant and human sources (30, 31). Currently,
experiments are being designed to analyze the distribution of
18F-Glc-GCase in mice with a conditional knock out of the GCase
gene (32).
Radiolabeling of Abg and Control Experiments. To a v-vial containing an eva-
porated DNP-[18F]-FDG reaction mixture, Abg in phosphate buffer (50 mM,
pH 6.5) was added and the solution incubated for 5 min at 37 °C. RTLC and
size exclusion chromatography revealed the rapid formation of 18F-Glc-Abg.
To demonstrate that radioactive labeling was occurring in the active site
of Abg, a mutant Abg (Abg E358A) was incubated with DNP-[18F]-FDG as
described for natural Abg. For the reactivation experiments, Abg was radi-
olabeled as described above followed by addition of thiophenyl glucoside
(TG) to a final concentration of 60 mM. Samples were taken at various time
points and analyzed by RTLC.
Radiolabeling and Isolation of GCase. To the v-vial containing the freshly
purified DNP-[18F]-FDG, recombinant GCase (∼60 units obtained from the
unused portions of patient vials of Cerezyme) in ∼250 μL of reaction buffer
(20 mM phosphate, 40 mM citrate, 0.25% sodium taurocholate, 0.1% Triton
X-100, pH 5.5) was added and the solution incubated for 45 min at 37 °C. The
solution was then diluted to a volume of ∼1;000 μL by MES buffer (50 mM
MES, 100 mM NaCl, pH 6.5) and divided between two Microcon centrifugal
filter devices (50 kDa MW cut off), spun for 8 min at 14,000 x g in a Beckman
centrifuge. Another portion of MES buffer ( 500 μL) was added and centri-
fuged for an additional 8 min. Final concentrate volumes were ∼100 μL,
which were then subjected to two rounds of gel filtration using Pierce
Desalting Spin Columns following the manufacturer’s instructions. Desalting
columns were preequilibrated with Cerezyme buffer (60 mM Citrate, 20 mM
Phosphate, 0.1% Tween 80 polysorbate80, pH 6.6). Approximately 200 μCi of
radiochemically pure (>97% by radio-TLC) 18F-Glc-GCase possessing a specific
activity of ∼50 Ci∕mmol was routinely synthesized following this method.
Radioactive labeling of GCase can be blocked by the inclusion of 6-nonyliso-
fagomine (10 mM) in the radiolabeling buffer mixture.
Materials and Methods
Radiosynthesis of DNP-[18F]-FDG. Pharmaceutical grade 2-deoxy-2-[18F]-fluoro-
D-glucose (2-[18F]-FDG) was a gift from the BC Cancer Agency, Vancouver
Canada. The 2-[18F]-FDG was synthesized using a commercially available auto-
mated radiochemistry system TRACERlab FXFDG following the manufacturer’s
instructions. Briefly, the system facilitates the automated synthesis of 2-[18F]-
FDG via nucleophilic displacement of the peracetylated mannose triflate with
[
18F]-fluoride prepared by cyclotron bombardment of 1 mL of H218O with
12.5 MeV protons in a niobium target. Following deprotection with HCl a
series of solid phase extraction (SPE) columns yield radiochemically pure
2-[18F]-FDG in an isotonic solution (10% NaCl). After quality control analysis
had ruled out the presence of [18F]-fluoride, 10–15 mCi of 2-[18F]-FDG was
placed into a 2 mL v-vial containing a small stir bar and fitted with a rubber
septum. The H2O was evaporated by heating to 100 °C while simultaneously
passing helium gas through the v-vial vented into a sodium lime trap. The vial
was cooled to 37 °C after which 150 μL of aqueous NaHCO3 (∼0.93 M) and
150 μL of 1-fluoro 2, 4-dinitrobenzene in ethanol (∼13 mg∕mL) was added.
The reaction was stirred for 10 min at 37 °C, quenched with acetic acid
(25 μL) and evaporated at 60 °C as described above. The yellow residue was
dissolved in 300 μL of distilled water, and loaded onto a C18 Sep Pak column
Micro-PET Imaging. Mice (C57BL/6 and 20 w of age n ¼ 4) were subjected to
inhalational isofluorane anesthesia (5% gas initial dose and then maintained
at 1–2% gas with a mask). Once the animal was anesthetized, it was placed
flat on its belly and taped to the scanner bed. Approximately 25–50 μCi of
18F-Glc-GCase (∼80 Units∕kg) in approximately 50–100 μL of Cerezyme buffer
was intravenously administered into the tail vein as a bolus injection. All stu-
dies were performed on a Siemens/Concorde Focus 120 micro-PET scanner.
Following a 6 min transmission scan with a 57Co source, a 2 h emission scan
was performed starting at tracer injection. Data were histogrammed into
4 · 30, 2 · 60, 11 · 300, 6 · 450, and 2 · 480 s frames and reconstructed using
Fourier rebinning and filtered backprojection after applying normalization,
10846
∣
www.pnas.org/cgi/doi/10.1073/pnas.1003247107
Phenix et al.

scatter, attenuation, and sensitivity corrections. Two mice were similarly
injected with the tracer but were used only for blood sampling and tissue
counting and were killed at 1 hr postinjection. All the animals were eutha-
nized prior to recovery from anesthesia by cervical dislocation and tissues
were harvested for gamma counting within the PET unit. Regions of interest
were chosen guided by the use of a mouse organ atlas. For the mannan
blocking experiments, 15 mg∕kg of mannan were injected into the tail vein
5–10 min prior to 18F-Glc-GCase (n ¼ 2 mice). All biodistribution data were
decay corrected using a half-life of 109.8 min for the 18F isotope. All proce-
dures involving mice were approved by the University of British Columbia
Animal Care Committee.
ACKNOWLEDGMENTS. We thank Siobhan McCormick and Katie Dinelle for
performing the micro-PET scans and data acquisition, Harmony Hemmelgarn
and Rajwinder Dhami for assistance with animal studies. We are grateful to
Milan Vuckovic and Yulia Rozen of the BC Cancer agency for providing us
with FDG for our synthesis. Many thanks to the people in the Nuclear
Medicine division at TRIUMF for their technical assistance. A thank you also
goes out to Dr. Ethan Goddard-Borger for covering due to illness and Dr. Tom
Wennekes for producing Fig. 2. We acknowledge the financial support of
the Canadian Institutes of Health Research, the Natural Sciences and
Engineering Research Council of Canada, the Children's Gaucher Research
Fund and the National Gaucher Foundation (US).
1. Rohrbach M, Clarke JTR (2007) Treatment of lysosomal storage disorders—progress
with enzyme replacement therapy. Drugs 67:2697–2716.
18. Street IP, Kempton JB, Withers SG (1992) Inactivation of a beta-glucosidase through
the accumulation of
a stable 2-Deoxy-2-Fluoro-Alpha-D-Glucopyranosyl enzyme
2. Liou B, et al. (2006) Analyses of variant acid beta-glucosidases—effects of Gaucher
Disease mutations. J Biol Chem 281:4242–4253.
3. Fuller M, et al. (2008) Glucosylceramide accumulation is not confined to the lysosome
in fibroblasts from patients with Gaucher Disease. Mol Genet Metab 93:437–443.
intermediate—a detailed investigation. Biochemistry 31:9970–9978.
19. Zhu XX, Sheth KA, Li SH, Chang HH, Fan JQ (2005) Rational design and synthesis of
highly potent beta-glucocerebrosidase inhibitors. Angew Chem International Edition
44:7450–7453.
4. Sato Y, Beutler
E (1993) Binding, internalization, and degradation of mannose-
20. Choy FYM (1986) Purification of human placental glucocerebrosidase using a 2-step
high-performance hydrophobic and gel-permeation column chromatography meth-
od. Anal Biochem 156:515–520.
21. Dvir H, et al. (2003) X-ray structure of human acid-beta-glucosidase, the defective
enzyme in Gaucher Disease. Embo Rep 4:704–709.
22. Xu Y, et al. (1996) Turnover and distribution of intravenously administered mannose-
terminated human acid beta-glucosidase in murine and human tissues. Pediatr Res
39:313–322.
23. Bijsterbosch MK, et al. (1996) Quantitative analysis of the targeting of mannose-term-
inal glucocerebrosidase. Predominant uptake by liver endothelial cells. Eur J Biochem
237:344–349.
24. Friedman BA, et al. (1999) A comparison of the pharmacological properties of carbo-
hydrate remodeled recombinant and placental-derived beta-glucocerebrosidase: im-
plications for clinical efficacy in treatment of Gaucher Disease. Blood 93:2807–2816.
25. Kacher Y, et al. (2008) Acid beta-glucosidase: insights from structural analysis and
relevance to Gaucher Disease therapy. Biol Chem 389:1361–1369.
26. Trejtnar F, Laznicek M (2002) Analysis of renal handling of radiopharmaceuticals.
Q J Nucl Med 46:181–194.
27. McCarter JD, Adam MJ, Hartman NG, Withers SG (1994) In-Vivo inhibition of beta-glu-
cosidase and beta-mannosidase activity in rats by 2-Deoxy-2-Fluoro-Beta-Glycosyl
fluorides and recovery of activity in-Vivo and in-Vitro. Biochem J 301:343–348.
28. Du H, et al. (2005) The role of mannosylated enzyme and the mannose receptor in
enzyme replacement therapy. Am J Hum Genet 77:1061–1074.
terminated glucocerebrosidase by macrophages. J Clin Invest 91:1909–1917.
5. Kempton JB, Withers SG (1992) Mechanism of Agrobacterium beta-glucosidase—
kinetic studies. Biochemistry 31:9961–9969.
6. Miao SC, et al. (1994) Identification of Glu(340) as the active-site nucleophile in human
glucocerebrosidase by use of electrospray tandem mass-spectrometry. J Biol Chem
269:10975–10978.
7. Withers SG, Street IP, Bird P, Dolphin DH (1987) 2-Deoxy-2-Fluoroglucosides—a novel
class of mechanism-based glucosidase inhibitors. J Am Chem Soc 109:7530–7531.
8. Zechel DL, Withers SG (2000) Glycosidase mechanisms: Anatomy of a finely tuned
catalyst. Accounts Chem Res 33:11–18.
9. McCarter JD, Adam MJ, Withers SG (1992) Synthesis of 2-Deoxy-2-[F-18]-Fluoro-
Beta-Mannosyl [F-18] fluoride as a potential imaging probe for glycosidases. J Labelled
Compd 31:1005–1009.
10. Wong AW, Adam MJ, Withers SG (2001) Synthesis of 2, 6-dideoxy-2-fluoro-6-[F-18]-
fluoro-beta-D-glucopyranosyl fluoride (2, 6FGF) as a potential imaging probe for
glucocerebrosidase. J Labelled Compd 44:385–394.
11. Sharma SK, Corrales G, Penades
unprotected 2, 4-Dinitrophenyl glycosides. Tetrahedron Lett 36:5627–5630.
12. Berven LA, Dolphin D, Withers SG (1990) The base-catalyzed anomerization of
S (1995) Single-step stereoselective synthesis of
dinitrophenyl glycosides—evidence for
68:1859–1866.
a novel reaction-mechanism. Can J Chem
13. Koeners HJ, Dekok AJ, Romers C, Vanboom JH (1980) The use of the 2, 4-Dinitrophenyl
group in sugar chemistry reexamined. Recl Trav Chim Pay-B 99:355–362.
14. Namchuk MN, Withers SG (1995) Mechanism of Agrobacterium beta-glucosidase:
Kinetic analysis of the role of noncovalent enzyme/substrate interactions. Biochemis-
try 34:16194–16202.
29. Stahl P, Schlesinger PH, Sigardson E, Rodman JS, Lee YC (1980) Receptor-mediated
pinocytosis of mannose glycoconjugates by macrophages—characterization and
evidence for receptor recycling. Cell 19:207–215.
30. Aviezer D, et al. (2009) Novel enzyme replacement therapy for Gaucher Disease:
ongoing phase III clinical trial with recombinant human glucocerebrosidase expressed
in plant cells. Mol Genet Metab 96:S13–S14.
15. Wang Q, Trimbur D, Graham R, Warren RAJ, Withers SG (1995) Identification of the
acid/base catalyst in Agrobacterium-faecalis beta-glucosidase by kinetic-analysis of
mutants. Biochemistry 34:14554–14562.
31. Shaaltiel Y, et al. (2007) Production of glucocerebrosidase with terminal mannose
glycans for enzyme replacement therapy of Gaucher’s Disease using a plant cell
system. Plant Biotechnol J 5:579–590.
16. Wang QP, Graham RW, Trimbur D, Warren RAJ, Withers SG (1994) Changing enzy-
matic-reaction mechanisms by mutagenesis—conversion of a retaining glucosidase
to an inverting enzyme. J Am Chem Soc 116:11594–11595.
32. Sinclair GB, Colobong KE, Choy FYM, Clarke LA (2007) Generation of a conditional
knockout of murine glucocerebrosidase: utility for the study of Gaucher disease.
Mol Genet Metab 90:148–156.
17. Trimbur DE, Warren RAJ, Withers SG (1992) Region-directed mutagenesis of residues
surrounding the active-site nucleophile in beta-glucosidase from Agrobacterium-
faecalis. J Biol Chem 267:10248–10251.
Phenix et al.
PNAS
∣
June 15, 2010
∣
vol. 107
∣
no. 24
∣
10847