Synthesis and Antiproliferative Activity of Fluorinated -Acetylmannosamine Analogs.

Introduction
Fluorinated carbohydrates are valuable
probes and inhibitors of
carbohydrate-processing enzymes and carbohydrate-binding proteins. In addition, some fluorinated monosaccharides
can cross the cytoplasmic membrane, enter metabolic pathways, and
disrupt the biosynthesis of cell surface and extracellular glycans
in a controlled and targeted manner.
−

These fluorinated monosaccharide-based metabolic inhibitors of glycosylation
are typically acetylated prior to administration to facilitate diffusion
across the plasma membrane. The antiproliferative activity of some
acetylated fluorinated monosaccharides
,
limits their
use as metabolic inhibitors at higher concentrations, however, this
property can be exploited in the development of new antitumor therapeutics.
,
For example, fully acetylated 3-fluoro and 4-fluoro analogs of N-acetyl-d-glucosamine (GlcNAc), and 4-fluoro and
4,6-difluoro analogs of N-acetyl-d-galactosamine
(GalNAc) were cytotoxic against leukemia cells (IC
50 24–35 μM),
,
while fully
acetylated 6,6-difluoro-l-fucose 1 and 6,6,6-trifluoro-l-fucose 2 (Figure
) reduced the viability of the human colon cancer cells
HCT116 (IC
50 43 μM and 58 μM,
respectively).

In addition to fluorinated
monosaccharides, other
types of monosaccharide
glycomimetics may have antitumor activity. For example, 2-acetamido-3,4,6-tri-O-acetyl-d-mannopyranose (Ac3ManNAc)
and 2-acetamido-3,4,6-tri-O-butyryl-d-mannopyranose
(Bu3ManNAc, Figure
A) have demonstrated promising cytotoxic and antimigratory
activity against the cell line MDA-MB-231,
−

a human cell
line derived from highly aggressive triple-negative breast cancer
with limited treatment options. These
two compounds are structurally hemiacetals (also termed lactols) of
O-acylated N-acetylmannosamine (ManNAc), characterized
by an unprotected anomeric hydroxyl group and O-acyl groups at the
3-, 4-, and 6-positions. The GlcNAc hemiacetal Ac3GlcNAc
(a C2-epimer of Ac3ManNAc) also showed cytotoxicity, albeit
weak, but the GalNAc hemiacetal Ac3GalNAc (a C4-epimer
of Ac3GlcNAc) was virtually noncytotoxic (Figure
B). Consistent with the previously observed trend for ManNAc hemiacetals,
the cytotoxicity against MDA-MB-231 cells increased after the O-acetyl
groups were replaced with O-butyryl groups to give Bu3GlcNAc
and Bu3GalNAc (Figure
, Bu = butyryl).

Introducing fluorine at the 4-position of Ac3GlcNAc
and Ac3GalNActhereby combining the cytotoxicity
of hemiacetals with that of fluorosugarsproduced fluorinated
hemiacetals 4F-Ac2GlcNAc and 4F-Ac2GalNAc exhibiting
approximately 3-fold and 10-fold increased cytotoxicity to MDA-MB-231
cells, respectively (Figure
B). Hemiacetal 4F-Ac2GlcNAc also inhibited growth of human prostate cancer cells PC-3 in vitro. Introducing an additional
fluorine substituent at the 3- and 6-positions, or replacing the 2-acetamido
group with more lipophilic azido or phenyl-triazole groups, further
enhanced the antiproliferative effect against MDA-MB-231 cells.
,
In contrast to nonfluorinated hexosamine hemiacetals, however, the
substitution of butyryl esters for acetyl esters at the nonfluorinated
positions of these fluorinated hemiacetals mostly decreased cytotoxicity.

The confirmed positive role of the fluorine
substituent in enhancing
the cytotoxicity of acylated GlcNAc and GalNAc hemiacetals, suggests that replacing an acetoxy group with
fluorine could also significantly increase the cytotoxicity of acylated
ManNAc hemiacetals. This motivated us to synthesize a complete series
of acetylated mono-, di-, and trifluorinated ManNAc hemiacetal analogs 3, 6–11 (Figure
) and to determine their in vitro antitumor properties. The 3-fluoro ManNAc analog
was prepared with two different O-acyl groupsacetyl (compound 3) and propionyl (compound 4)to evaluate
the effect of alkyl chain length on cytotoxicity. The 2-azido-4-fluoro
analog 5 was included for comparison. During the synthesis
of 3,4-difluorinated and 3,4,6-trifluorinated ManNAc derivatives,
we unexpectedly obtained precursors of fluorinated N-acetyltalosamine with an amino group masked as an azide. Evaluation
of antitumor activity comprised the determination of the cytotoxic
activity by an MTT assay, a cell proliferation assay, and a colony-forming
assay. The antimigratory activity of selected compounds was evaluated
using a wound healing assay. To elucidate the mechanisms underlying
the cytotoxicity of the most cytotoxic fluoro analogs, we also conducted
cell cycle arrest analysis and Western blotting to detect proteins
characteristic for cell apoptosis and autophagy. The trifluorinated
mannosamine hemiacetal 11 displayed the strongest antiproliferative
and cytotoxic activity in all assays, placing it among the most cytotoxic
fluorinated monosaccharides reported.

Results

Synthesis
Initially,
we planned to prepare the 3-fluoro
ManNAc analog from the known d-altro-configured
2-azido intermediate 12
by
analogy with a recently described fluorination of its d-allo-configured counterpart 13, which upon
reaction with diethylaminosulfur trifluoride (DAST) yielded the desired
3-fluoro product 14 with an inversion of the configuration
at C3 in a usable yield of 41%, together with 5% of the elimination
products 15 and 16. However, fluorination of 12 predominantly yielded the d-altro-configured product 17 with
retention of the configuration at C3, containing traces of the desired
inseparable product 18 (Scheme
). The axial orientation of the fluorine
substituent in 17 was evidenced by a large vicinal coupling
between trans-diaxially disposed fluorine and H-4
(3
J
F,H‑4 = 29.6 Hz).
The addition of triethylamine trihydrofluoride improved the yield but 17 remained the main
reaction product. Conducting the reaction in toluene at higher temperatures
led predominantly to the elimination product 19, while the desired 18 was formed
in a disappointing 18% yield. The microwave-assisted reaction in dichloromethane
at 80 °C also gave the elimination product 19 along
with the low yield (19%) of an inseparable mixture of both C3-configurational
isomers 17 and 18.
An alternative route to the
required 3-fluoro
analogs was developed
starting from 3-fluorinated 1,6-anhydro-β-d-glucopyranose 20 (Scheme
), available in six steps from levoglucosan. The C2 hydroxyl group was activated as trifluoromethanesulfonate
and subsequent nucleophilic substitution by reaction with sodium azide
afforded 1,6-anhydro-2-azido-3-fluoro-mannopyranose 21. The equatorial position of the 2-azido group was indicated by the
large vicinal coupling constant (3
J
H‑2,F = 28.2 Hz). Oxidative de-O-benzylation with the
NaBrO3/Na2S2O4 system liberated the hydroxyl at the 4-position, giving
alcohol 22. Subsequent reaction with trimethylsilylthiophenol
(TMSSPh) effected cleavage of the internal acetal
,
and introduced a stable thiophenyl group at the anomeric position
to give compound 23 as a separable pair of anomers. Protection
of the anomeric position as a thioglycoside enabled us to manipulate
the remaining positions. Then, we could
regioselectively and chemoselectively introduce the unprotected anomeric
hydroxyl to obtain the desired hemiacetals (vide infra). In addition,
thioglycosides are highly useful as versatile glycosyl donors.
The two unprotected 4- and 6-hydroxyl groups
in 23 were either acetylated with acetic anhydride or
propionylated with
propionyl chloride in pyridine to give the acylated products 24 and 25. Hydrolysis of the thiophenyl glycoside
(NBS/acetone/water) yielded the hemiacetals 26 and 27, which were converted to the desired acylated 3-fluoro
ManNAc hemiacetals 3 and 4 on reaction with
thioacetic acid. The microwave-assisted
deoxyfluorination of diol 23 by reaction with DAST (1.3
equiv) proceeded regioselectively at the primary 6-hydroxyl group,
,
yielding 3,6-difluoro intermediate 28. To prevent migration
of the anomeric thiophenyl group to the 6- or 4-positions during the
reaction with DAST, α-thioglycoside α-23 was
used as the starting material.
,
Acetylation of the
4-OH group and subsequent hydrolysis of the thiophenyl group yielded
the 2-azido-hemiacetal 29, which was converted to the
desired acetylated 3,6-difluoro ManNAc hemiacetal 9 by
reaction with thioacetic acid (Scheme
).
To prepare the 4-fluoro and 4,6-difluoro analogs,
the 4-fluoro-mannosazide
intermediate 30, previously used in the chemoenzymatic
synthesis of 7-fluoro-neuraminic acid, was obtained in six steps from
methyl α-d-mannoside as reported (Scheme
). Zemplén deacylation liberated the hydroxyl group at the
6-position to give the alcohol 31. The sulfuric acid-catalyzed
acetolysis with Ac2O acetolysed both the benzyl ether and
the methyl glycoside group, producing
compound 5, which was included in the cytotoxicity assays
for comparison. The azide was converted to an acetamide by reaction
with AcSH to give the intermediate 32. The following
regioselective hydrolysis of the anomeric acetate by treatment with
hydrazine acetate yielded the target 4-fluoro hemiacetal 6.
The C6-deoxyfluorination of
alcohol 31 using a microwave-assisted
reaction with DAST afforded the 4,6-difluoro-mannosazide 33. Sulfuric acid-catalyzed acetolysis yielded the intermediate 34, which was isolated as an α-acetate, and the subsequent
azide-to-acetamide conversion followed by anomeric deprotection resulted
in the desired acetylated 4,6-difluoro ManNAc hemiacetal 10. The target acetylated 6-fluoro ManNAc hemiacetal 7 was prepared from diol 35 (Scheme
), which was synthesized from methyl α-d-mannopyranoside in five steps using the published route. Microwave-assisted deoxyfluorination using 1.2
equiv of DAST resulted in regioselective C6-deoxyfluorination to produce
6-fluoro-mannosazide 36. The intermediate 36 was then subjected to acetolysis followed by an azide-to-acetamide
conversion to give the acetate 37. Anomeric deprotection
afforded the target 6-fluoro ManNAc analog 7.
We
have originally intended to prepare the 3,4-difluoro ManNAc
analogs from the 3-fluoro analog 22 using Latrell-Dax
inversion at C4 followed by DAST-mediated
fluorination as the key reactions. Accordingly, the C4 hydroxyl in 22 was converted to trifluoromethanesulfonate ester and then
reacted with KNO2 to give the expected d-talosazide
derivative 38 (Scheme
A). However, the subsequent reaction with DAST proceeded
with complete retention of the configuration at the 4-position, giving
3,4-difluorotalosazide 39 as the sole product in good
yield of 85%. The equatorial position of the fluorine substituent
at the 4-position is evidenced from the large coupling between H-4
and F-3 (3
J
H‑3,F‑4 = 24.4 Hz). In addition, the magnitude of the geminal coupling constant 2
J
F‑4,C‑5 = 27.3
Hz indicates the anti-periplanar relationship between
bonds C4–F4 and C5–O5. Retentive
DAST-mediated fluorination at C4 of a 1,6-anhydrotalopyranose scaffold
has previously been reported for 1,6-anhydro-2,3-dideoxy-2,3-difluorotalose
40 and 1,6:2,3-dianhydrotalopyranose
42 to give the corresponding products 41 and 43, respectively (Scheme
B). On the other hand, DAST-mediated fluorination
of galactosan 44 was less stereoselective and afforded,
in addition to the major product 45 with inversion of
the configuration at C4, an inseparable minor product 46 with retention of the configuration. The available data suggest that deoxyfluorination of the C4 equatorial
hydroxyl in 1,6-anhydro-β-d-hexopyranoses is highly
stereoselective for talo-configured substrates, favoring
configurational retention. Retention of configuration at C4 in these
reactions is most likely due to the participation of the endocyclic
O5 oxygen of the pyranose ring and involves an epoxonium intermediate 47 (Scheme
C).
,

Having access to 3,4-difluoro-talosazide 39, we attempted
to prepare 3,4-difluoro and 3,4,6-trifluoro N-acetyltalosamine,
however our efforts were thwarted by complications during azide to
acetamide conversion. The reaction of 39 with TMSSPh
afforded separable anomers of the thioglycoside 48 (Scheme
). The β-anomer
β-48 was acetylated to yield the 6-acetate 49, while DAST deoxyfluorination of the α-anomer α-48 afforded the trifluoro compound 50. Only the
α-anomer was used in the reaction with DAST to prevent thioaglycone
migration. Hydrolysis of the thiophenyl
group in compounds 49 and 50 yielded the
corresponding hemiacetals 51 and 52 in moderate
yields of around 40%. Hemiacetal 52 was obtained in about
90% purity due to partial decomposition during chromatography. Hemiacetals 51 and 52 showed sluggish and irreproducible
reaction under the conditions for azide-to-acetamide conversion that
were successful for hemiacetals 26, 27 and 29 (Scheme
) and previously for fluorinated GlcNAc and GalNAc hemiacetals. The addition of an excess of AcSH to 2-azidotaloses 51 and 52 led to mixtures of compounds and further
attempts at azide reduction were abandoned. However, a difluorinated
analog was surprisingly isolated from the product mixture after the
reaction of hemiacetal 52 with AcSH (Scheme
). Its HRMS spectrum indicated
the substitution of one fluorine with a thioacetyl group, and the
NMR spectrum suggested the galacto-configuration
(3
J
H‑2,H‑3 =
12.4 Hz, 3
J
H‑3,F‑4 = 35.3 Hz, 3
J
H‑3,H‑4 = 2.2 Hz, 3
J
H‑4,H‑5 < 1 Hz) and an unprotected α-configured anomeric hydroxyl.
There was also no fluorine or oxygen substituent at C3 (the chemical
shift of C-3 carbon δC‑3 = 43.4 ppm). The
product was assigned the structure of 3-thioester of 4,6-difluorinated
GalNAc analog 53 and was obtained as a crystalline α-anomer
(Scheme
). The structure
of the thioacetate 53 was eventually confirmed by an
X-ray diffraction analysis. Interestingly, the compound 53 crystallizes with two independent molecules, one of which has the
fluorinated exocyclic side chain C(5)–C(6)­H2F in
the gg conformation (Figure
). The other
independent molecule adopts the gt conformation,
which is typical with galactosides. The gg rotamer
is strongly disfavored in galactosides due to destabilizing 1,3-diaxial
interactions, here between the fluorine substituents F(4) and F(6). Although signal overlap in the 1H
NMR spectrum prevented a detailed conformational analysis, the low
magnitude of the vicinal coupling constant 3
J
(H5–F6) = 12.3 Hz indicates the minimal population
of the gg rotamer in a CDCl3 solution,
,,
as is typical for galactosides.
Because the planned route to the desired 3,4-difluoro
ManNAc analog
from the intermediate 22 failed (Scheme
), we decided to introduce the 3- and 4-fluorine
substituents prior to the installation of the 2-azido group. Advantageously,
we used the known 3,4-difluorinated 1,6-anhydroglucopyranose 54 available from levoglucosan in six steps as described by
Linclau and Giguère (Scheme
). Catalytic hydrogenation
of compound 54 liberated the hydroxyl group at the 2-position
to give the alcohol 55. The hydroxyl group was activated
as a trifluoromethanesulfonate ester and then substituted with azide,
producing the desired d-manno-configured
1,6-anhydro-2-azido-3,4-difluoropyranose intermediate 56. Reaction with TMSSPh afforded a mixture of α- and β-thioglycosides 57. Only the β-anomer β-57 was obtained
as a pure compound by column chromatography in 42% yield. Acetylation
of β-57 with acetic anhydride produced thioglycoside 58. NBS-promoted hydrolysis of the thiophenyl glycoside yielded
hemiacetal 59, and the final azide-to-acetamide conversion
afforded the target 3,4-difluoro ManNAc analog 8. Deoxyfluorination
of β-57 using a microwave-assisted reaction with
DAST yielded trifluorinated compound 60. No significant
migration of the thiophenyl group was observed on TLC despite the
β-configuration of the aglycone, which is favorable for migration. Subsequent NBS-promoted hydrolysis of the thiophenyl
glycoside produced hemiacetal 61. The azide-to-acetamide
conversion by reaction with AcSH resulted in the target 3,4,6-trifluoro-ManNAc
hemiacetal 11. All final acylated deoxyfluorinated hemiacetals
were obtained as a colorless gel-like mixture of both anomers, stable
in air or in alcoholic or DMSO-d
6 solution,
with only limited solubility in water, but sufficient for biological
assays.

MTT Cytotoxicity Assay
The in vitro cytotoxicity of the prepared fluorinated
hemiacetals was determined
using the MTT cell viability assay after 72 h of treatment (Table
) and expressed as IC
50 values. We used two cell lines, the MDA-MB-231
cell line derived from human triple-negative breast adenocarcinoma,
and the human mammary gland epithelial cell line MCF-10A, which is
a nontumorigenic epithelial cell line. The MDA-MB-231 cell line was
selected because it has previously shown sensitivity to nonfluorinated
acylated mannosamines
−

and fluorinated acylated GlcNAc and GalNAc.
,
The MCF-10A cell line was selected as a control to evaluate the
effects of the compounds on noncancerous cells. Selected cytotoxicity
values against MDA-MB-231 cells for acetylated difluorinated and trifluorinated
GlcNAc and GalNAc hemiacetals available in the literature are shown in Table
for comparison.
Nonfluorinated Ac

3

ManNAc and Bu

3

ManNAc showed
moderate cytotoxicity with the butyryl analog showing slightly increased
activity. Mono- and difluorinated analogs of ManNAc that contained
a 6-fluoro substituent exhibited low or no antiproliferative activity.
Thus, the 6-fluoro-ManNAc analog 7, and the 3,6-difluoro-ManNAc
analog 9 were completely ineffective (IC
50 > 200 μM, Table
, entries 7 and 9), while the 4,6-difluoro-mannosamine 10 showed weak cytotoxicity only against the MDA-MB-231 cells
(entry 10). This contrasts with the corresponding GlcNAc and GalNAc
hemiacetals, which all exhibited measurable cytotoxicity (see entries
12–15 for difluorinated GlcNAc). The azido analog 5 was cytotoxic only to the noncancerous
MCF-10A cells (entry 5).
The remaining fluorinated ManNAc hemiacetals 3, 4, 6, 8, and 11 exhibited
cytotoxic activity to both tested cell lines with the IC
50 values ranging from 21 μM to 57 μM. The
most cytotoxic compound to both cell lines was the 3,4,6-trifluoro-ManNAc
analog 11 (entry 11), which performed slightly better
than the corresponding GlcNAc and GalNAc analogs (entries 16 and 17).
Its cytotoxicity toward MCF-10A cells was comparable to cisplatin.
The 3,4-difluoro ManNAc analog 8 was only slightly less
cytotoxic than trifluoro ManNAc 11. No significant selectivity
for the MDA-MB-231 cancer cell line was observed.

Cell Proliferation
Assay
In addition to assessing the
metabolic activity and viability by the MTT assay, the cytotoxic compounds 4, 6, 8, and 11 were
tested in a cell proliferation assay at different concentrations based
on the IC
50 values obtained (10 μM,
25 μM, and 50 μM for compounds 4 and 6; 10 μM, 20 μM, and 30 μM for compounds 8 and 11). This assay served as a complementary
experiment to the MTT assay that would also allow us to assess potential
morphological changes. As no detectable morphological changes were
observed and the results were consistent with the MTT assay, the experiment
was performed in a single biological replicate. The obtained data
(Figure
) largely
aligns with the IC
50 values from the MTT
assay, with one exception: the propionylated 3-fluoro ManNAc analog 4 was the least effective of all compounds tested in the cell
proliferation assay, despite its relatively high cytotoxicity in the
MTT assay (IC
50 = 27 ± 1 μM).
Conversely, the trifluorinated analog 11 exhibited the
strongest antiproliferative effect among the tested compounds, consistent
with the results of the MTT assay.

Colony Forming Assay
We also evaluated the ability
of MDA-MB-231 cells to survive and form colonies after treatment with
the cytotoxic fluorinated hemiacetals 3, 4, 6, 8, and 11 at a concentration
of 2.5 μM. The average number of colonies formed from at least
ten separate wells for each treatment is displayed in the box plot
graph in Figure
.
Control cells treated with DMSO formed an average of 34 colonies per
well. Treatment with all compounds, except for 3-fluoro ManNAc analog 4, resulted in a significant reduction in the number of colonies
formed. The 3,4,6-trifluoro-ManNAc analog 11 was the
most effective in inhibiting colony formation, reducing the average
number of colonies to 18. This was followed by 3,4-difluoromannosamine 8, which reduced the average number of colonies to 22.

Cell Cycle Analysis
Flow cytometry was performed to
analyze the cell cycle of MDA-MB-231 cells treated with compounds 3, 4, 6, 8, and 11; DMSO was used as the control. The data obtained (Figure S1) were analyzed using two-tailed t tests to determine if there were significant differences
in the percentage of cells in each phase between the control and treated
cells. None of the ManNAc analogs tested significantly disrupted the
cell cycle.

Western Blotting Analysis
To elucidate
the mechanisms
underlying the cytotoxic effects of the tested compounds, immunodetection
using Western blotting (Figure
) was performed to analyze selected proteins involved in apoptosis
(PARP), DNA damage (γH2AX), cell cycle arrest (p21, cyclin B1,
and cyclin D), monitoring of cellular energy balance and response
to metabolic stress (AMPK and p-AMPK), and autophagy (p62 and LC3B).
Treatment with the most cytotoxic 3,4,6-trifluoro-ManNAc analog 11 caused a distinct increase in γH2AX phosphorylation,
indicating DNA damage, and a slight induction of PARP cleavage, suggesting
the initiation of apoptotic processes. Compound 11 also
showed upregulation of the autophagy markers p62 and LC3B.

Wound Healing Assay
Compounds 5, 7, and 9 that were inactive against MDA-MB-231
cells in the MTT assay were tested for their antimigratory properties
using a wound healing assay, as nonfluorinated ManNAc hemiacetals
exhibited antimigratory properties in earlier studies. None of the tested compounds showed reduced
cell migration. On the contrary, cells treated with these hexosamine
analogs appeared to migrate even more than control cells treated with
DMSO (Figure S2).

Discussion
Motivated by the previous finding that fluorine
introduction enhances
the cytotoxicity of acylated GlcNAc and GalNAc hemiacetals, in this
study, we synthesized fluorinated acetylated ManNAc hemiacetals and
evaluated their in vitro antitumor activity. Previously,
only two monofluorinated nonacetylated ManNAc analogs were reported:
4-fluoro-ManNAc and 6-fluoro-ManNAc. The synthesis of the fluorinated ManNAc hemiacetals
was more complicated than we initially expected. Our experience, as
well as that of the other groups, emphasizes
the importance of gathering empirical results from which useful trends
and patterns can emerge to guide future synthetic attempts. For example,
a trans-diaxially disposed vicinal azido group appears
to increase the likelihood of retentive deoxyfluorination on reaction
with DAST as shown here for compound 12 (Scheme
) and previously for methyl
3-azido-4,6-O-benzylidene-3-deoxy-α-d-altropyranoside and 1,6-anhydro-3-azido-3-deoxy-2-O-tosyl-β-d-glucopyranose. A possible explanation is neighboring group participation
of the azido substituent
−

in spite of the fact that an
azide behaves as an inert and nonparticipating group in glycosylation
with 2-azido-donors.
−

Similarly, the deoxyfluorination of compound 38 corroborates
the propensity of 1,6-anhydrotaloses to undergo retentive deoxyfluorination
at the 4-position on reaction with DAST (Scheme
A,B). If the proposed O5 participation is
the correct reaction mechanism (Scheme
C), this tendency of 1,6-anhydrotaloses is surprising
because an antiperiplanar electronegative substituent at C2 (with
respect to the endocyclic pyranose O5 oxygen)for example an
azide in 38 or fluorine in 40should
reduce the electron-donating capacity of the pyranose O5 oxygen and
its ability to form a cationic intermediate such as 47 (Scheme
C), as previously
noted by Linclau.

Unexpected formation
of product 53 from hemiacetal 52 in reaction
with AcSH involves epimerization at the 2-position,
suggesting that 53 is formed through an elimination-addition
mechanism involving an unsaturated intermediate 62 (Scheme
). A 1,4-conjugate
addition of thioacetic acid and concomitant reduction of the azido
group would then yield the GalNAc analog 53. The formation
of the 3-thiolated products 64 and 65, which
was previously reported for the reaction of the acetylated 3-fluoro-GlcNAc
hemiacetal 63 with 2-phenylethanethiol, followed by acetylation,
probably proceeds through the same mechanism with the elimination
of hydrogen fluoride and addition of the thiol. Noteworthy, it was demonstrated that O-acetylated (nonfluorinated)
GlcNAc hemiacetal can eliminate acetic acid and add the thiol group
of cysteine at the 3-position via an elimination-addition mechanism.
This reaction with cysteine residues has been identified as the cause
of undesired S-glyco-modification of proteins.

Interestingly, no product resulting from migration
of the β-linked
thiophenyl aglycone from the anomeric to the 6-position was isolated
after the microwave-assisted DAST deoxyfluorination of the thioglycoside
β-57, indicating that this side reaction occurred
to a negligible degree with this starting compound (Scheme
). We have previously observed
a significant migration with phenyl 2-azido-6-hydroxy-1-thio-β-galactoside
β-66, which predominantly produced migration product 68, in addition to desired 6-fluorothiogalactoside 67, in microwave-assisted deoxyfluorination with DAST (Scheme
). However, thioglucoside β-69 was much less prone
to migration, giving predominantly 6-fluorothioglucoside 70 together with a low quantity of the inseparable migration product 71. Taken together, thioaglycone
migration does not appear to be a general reaction mechanism with
2-azido-β-thioglycosides, and the extent of migration depends
considerably on the configuration and substitution pattern of the
starting thioglycoside.
In contrast to acylated GlcNAc
and GalNAc hemiacetals,
where deoxyfluorination
consistently increased cytotoxicity, the effect of deoxyfluorination
on acetylated ManNAc hemiacetal was more variable and very sensitive
to the fluorination pattern. Introducing fluorine at the 6-position
mostly reduced the cytotoxicity compared to the nonfluorinated Ac

3

ManNAc hemiacetal.
For example, 6F–Ac2ManNAc 7 was not
cytotoxic, however, 6-fluorination increased cytotoxicity of propionylated
GlcNAc and GalNAc hemiacetals, albeit to a lesser extent than 3- and
4-fluorination. Strikingly, 3,6- and
4,6-diF-AcManNAc 9 and 10 were virtually
noncytotoxic, while the corresponding acetylated GlcNAc and GalNAc
counterparts had potent antiproliferative effects on MDA-MB-231 cells. The common trend for all fluorinated N-acetylhexosamine hemiacetals is that the trifluorinated
analogs are among the most cytotoxic fluoro analogs. This applies
to both GlcNAc and GalNAc analogs and
is consistent with the finding that 3,4,6-triF-ManNAc 11 was the most cytotoxic of the compounds tested in this study. Together
with 4,6-diF-PrGalNAc (Table
, entry 15), it is one of the most cytotoxic fluorinated monosaccharides
reported.
Direct molecular targets in cells responsible for
the observed
antiproliferative effects of fluorinated hemiacetals have not yet
been identified. It has been hypothesized that nonspecific S-glyco-modification
of cellular proteins contributes to the cytotoxic effects of hexosamine
hemiacetals. It is possible that deoxyfluorination
increases the levels of damaged S-glyco-modified proteins in cells
due to higher tendency of fluorinated hexosamine hemiacetals to react
by elimination-addition mechanism, but confirming this hypothesis
is beyond the scope of this study. Additionally, certain fluorinated
monosaccharides can act as metabolic inhibitors of specific cell-surface
glycan structures.
,
According to a previous report, however, this metabolic perturbation of the
cell-surface glycome did not correlate with the antiproliferative
activity of the hemiacetal 4F-Ac2GlcNAc in human prostate
cancer PC-3 cells, suggesting that the glycome remodeling alone is
not responsible for the antiproliferative activity of this fluoro
analog.
Western blot analysis of selected stress and death-related
proteins
revealed that compound 11, which exhibited the highest
cytotoxicity, antiproliferative effect, and inhibition of the formation
of colonies induced pronounced molecular changes consistent with multifactorial
cell stress responses. Notably, γH2AX, a marker of DNA double-strand
breaks, was strongly upregulated in response to compound 11, with a ∼7.4-fold increase compared to control, indicating
extensive DNA damage. This was accompanied by an increase in LC3B
and p62 levels, which is consistent with the activation of autophagy,
as well as PARP cleavage, a hallmark of apoptosis. Although changes
in AMPK and p-AMPK expression were also observed, suggesting metabolic
stress modulation, it does not seem to be a dominant mechanism.
Furthermore, Western blot analysis revealed a downregulation of
cyclin B1 and cyclin D1. However, these changes were not manifested
by a detectable shift in cell cycle distribution in flow cytometric
analysis of DNA content (Figure S1), indicating
that compound 11 does not induce a significant cell cycle
arrest under the tested conditions. This indicates that the observed
antiproliferative activity is likely mediated through mechanisms other
than cell cycle blockade, such as DNA damage, apoptosis, and autophagy,
as supported by γH2AX accumulation, PARP cleavage, and LC3B
upregulation.
Taken together, these findings suggest that the
antiproliferative
activity of compound 11 is mediated through a multimodal
cellular response, involving DNA damage, apoptosis, and autophagy,
rather than through direct cell cycle arrest or modulation of the
surface glycome. While the precise molecular targets remain unidentified,
the consistent activation of stress and death-related pathways supports
the notion that this trifluorinated hexosamine analog elicits a complex
and multifactorial mechanism of action.

Conclusion
Although
the synthesis of fluorinated analogs
of N-acetylmannosamine posed a greater challenge
than the synthesis of N-acetylglucosamine and galactosamine
analogs reported previously, a complete
set of deoxyfluorinated O-acetylated
ManNAc hemiacetals has been prepared. The synthesis involves the preparation
of advanced polyfluorinated intermediates, including polyfluorinated
phenyl 2-azido-thiomannosides and -talosides. These thioglycosides
are fluorinated glycosyl donors for challenging 1,2-cis-glycosylation. Retentive deoxyfluorination of 2-azido-altropyranoside 12 suggests that the trans-diaxially positioned
azido group provides anchimeric assistance in the nucleophilic deoxyfluorination
with DAST. Retentive deoxyfluorination with DAST at the equatorial
4-position in 1,6-anhydro-talopyranose derivatives likely proceeds
via participation of the pyranose ring oxygen and is a valuable route
to 4-fluoro-talopyranoses. Introducing fluorine at the 3-position
makes hexosamine-derived hemiacetals susceptible to the elimination
of hydrogen fluoride and subsequent conjugate addition, as demonstrated
by compound 53. Our results also indicate that the migration
of thioaglycones from β-thioglycosides considerably depends
on the substitution pattern and configuration. It can occur to an
insignificant extent with some 2-azidohexosamine β-thioglycosides,
not compromising the yield of the desired fluorinated sugars.
A common trend for ManNAc, GalNAc and GlcNAc hemiacetals is that
their 3,4,6-trideoxyfluorinated analogs showed significant cytotoxic
effects. In contrast, monodeoxy- and dideoxyfluorinated analogs displayed
variable activity likely influenced by the position and relative configuration
of the fluorine substituents. The potent but unselective antiproliferative
activity exhibited by the trifluoro ManNAc analog 11 was
confirmed by three independent cell-based assays. This cytotoxicity
was consistently associated with DNA damage, apoptosis, and autophagy,
supporting a multifactorial mechanism of action. These findings highlight
the promise of trifluorinated amino sugar hemiacetals as cytotoxic
scaffolds for further biomedical research.

Safety Statement
Reactions with DAST are potentially
hazardous. Appropriate personal
protective equipment, including gloves and eye protection, should
be worn, and reaction should be conducted with caution. According
to the empirical safety rule for organic azides, a compound is considered
safe to handle if the number of nitrogen atoms does not exceed the
number of carbon atoms, and if (N
C + N
O)/N
N ≥ 3.
In this formula, N is the number of atoms. All synthesized
azide-containing compounds fulfill these criteria and are therefore
not expected to exhibit explosive or impact-sensitive behavior.

Supplementary Material