On the mechanism of isomerization of all-trans-retinol esters to 11-cis-retinol in retinal pigment epithelial cells: 11-Fluoro-all-trans-retinol as substrate/inhibitor in the visual cycle

Article abstract of DOI:10.1016/j.bmc.2005.05.001

The synthesis of 11-fluoro-all-trans-retinol (11-F-tROL), which is shown to be an excellent substrate for processing by visual cycle enzymes, is described. It is isomerized to its 11-cis congener subsequent to its esterification by lecithin retinol acyl t

Full text of DOI:10.1016/j.bmc.2005.05.001

Bioorganic & Medicinal Chemistry 13 (2005) 5189–5194
On the mechanism of isomerization of all-trans-retinol
esters to 11-cis-retinol in retinal pigment epithelial cells:
11-Fluoro-all-trans-retinol as substrate/inhibitor in the visual cycle
Nathan Fishkin, Revital Yefidoff, Deviprasad R. Gollipalli and Robert R. Rando*
Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School,
45 Shattuck Street, Boston, MA 02115, USA
Received 30 March 2005; revised 3 May 2005; accepted 3 May 2005
Available online 25 May 2005
Submitted in honor of Professor Koji Nakanishi on the occasion of his 80th birthday and his being awarded the Tetrahedron Prize.
Abstract—The synthesis of 11-fluoro-all-trans-retinol (11-F-tROL), which is shown to be an excellent substrate for processing by
visual cycle enzymes, is described. It is isomerized to its 11-cis congener subsequent to its esterification by lecithin retinol acyl trans-
ferase (LRAT) approximately as well as is vitamin A itself. The enzymatic turnover of 11-F-tROL is unaccompanied by enzyme
inhibition. The previously reported lack of isomerization of this substrate had been suggested as evidence for a carbonium mech-
anism in the critical enzymatic isomerization pathway in vision. The mechanism of this process remains unknown.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
out 11-cis-retinoid biosynthesis were developed almost
20 years ago,4 the enzyme component(s) responsible
for the key 11-trans to cis conversion have not been
identified nor has a complete molecular mechanism been
established.
The primary event in vision is the photochemical cis !
trans isomerization of the 11-cis-retinal Schiff base chro-
mophore of rhodopsin. This isomerization reaction ini-
tiates the phototransduction cascade, which results in
the visual response.1 Continued vision in vertebrates de-
pends, of course, on the enzymatic reverse trans ! cis
isomerization to regenerate the 11-cis-retinal and there-
by complete the visual cycle. A diverse group of retinoid
binding proteins and enzymes, largely confined to the
retinal pigment epithelium (RPE), are required for the
operation of the overall visual cycle.2,3
It is known, however, that this process requires a mini-
mal three-component system in the RPE (Scheme 1)
comprised of lecithin retinol acyl transferase (LRAT),
mRPE65,
a chaperone for all-trans-retinyl esters
(tREs),5,6 and the isomerohydrolase (IMH), which pro-
cesses tREs into 11-cis-retinol.7 The latter reaction can
occur via a variety of mechanisms, which include SN20
and carbonium ion mechanisms.8 In either mechanism,
the driving force for reducing the bond order at
C11–C12 involves C–O bond cleavage, in the process
of which it expels the palmitate moiety (see Scheme 2).
The resulting formal positive charge could either be
spread out over the odd numbered carbons along the
retinoid (carbonium ion) or be neutralized by specific
enzymatic nucleophilic addition.
The endergonic trans-to-cis retinoid isomerization reac-
tion is of special interest. Despite the fact that the over-
all retinoid isomerization pathway is known and
conditions for an in vitro system capable of carrying
Abbreviations: tROL, all-trans-retinol; 11-F-tROL, 11-fluoro-all-trans-
retinol; tREs, all-trans-retinyl esters; LRAT, lecithin retinol acyl trans-
ferase; IMH, isomerohydrolase; RPE, retinal pigment epithelium;
CRALBP, cellular retinaldehyde binding protein; BSA, bovine serum
albumin.
It was postulated that if isomerization is active through
an SN20 mechanism then an isostere of retinol contain-
ing a fluorine atom at C11 instead of a hydrogen could
allow for irreversible covalent attachment of the retinoid
to the IMH (see Scheme 3), making it a potential suicide
inhibitor of isomerization. This is an appealing prospect,
Keywords: Visual cycle; 11-Fluoro-retinol; Isomerohydrolase
mechanism.
*
Correspondingauthor.Tel.:+6174321794;fax:+6174320471; e-mail:
robert_rando@hms.harvard.edu
0968-0896/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2005.05.001

5190
N. Fishkin et al. / Bioorg. Med. Chem. 13 (2005) 5189–5194
(11-F-tROL), after conversion to 11-F-tRE, is not
isomerized by RPE membranes.9 The electronegativity
of the 11-F was suggested to suppress carbonium ion
formation, thus implicating it as a mechanism in the
isomerization reaction. However, the more interesting
possibility that 11-F-tRE might inactivate IMH was
not suggested or experimentally tested. For this reason,
we reinvestigated this problem.
11-F-tROL 1 (Scheme 4) was synthesized using a new
and unambiguous route, and was tested for its ability
to irreversibly inhibit IMH activity. Compound 1 was
tested in vitro and was found to be a competitive
inhibitor of 11-cis-retinol formation with vitamin A
as the substrate, but, within experimental limits, did
not irreversibly inactivate IMH. But contrary to a
previous report,9 we found that 11-F-tROL is an
excellent substrate for the IMH, and conversion to
11-F-11-cis-ROL takes place with comparable kinetics
to vitamin A (tROL).
Scheme 1. Mammalian visual cycle.
2. Results
in that one retinol analogue might provide both mecha-
nistic information on the IMH mode of action and also
serve as the basis for a chemical affinity labeling agent for
isolation of the IMH. In possible support of this notion
is the reported observation that 11-F-all-trans-retinol
2.1. Synthesis of (7E,9E,11Z,13E)-11-fluoro-retinol
(all-trans)
A synthesis of 11-F-11-cis-ROL was reported sometime
ago in the literature;10 however, a direct route to the 11Z
Scheme 2. Carbonium ion mechanism (A) as well as an SN20 mechanism (B) are consistent with retinoid labeling experiments done on the
isomerohydrolase.

N. Fishkin et al. / Bioorg. Med. Chem. 13 (2005) 5189–5194
5191
Scheme 3. Potential SN20 mechanism by which 11-F-ROL could either act as a substrate (A) or irreversibly inactivate the isomerohydrolase (B).
Scheme 4. Reagents and conditions: (a) (fluorocarbethyoxymethyl)triphenylphosphonium bromide, nBuLi, CH2Cl2, ꢀ78 ꢁC, then 25 ꢁC, 2 h, 72%.
(b) DIBAL-H, THF, ꢀ78 ꢁC, then 0 ꢁC, 1 h. (c) BaMnO4, CH2Cl2, 25 ꢁC, 12 h, 80% two steps. (d) nBuLi, THF, ꢀ78 ꢁC, then aldehyde 4, then 25 ꢁC,
3 h, 62%. (e) nBu4NF, Et2O, 25 ꢁC, 30 min, 70%.
congener was never established. De Lera et al. report the
conversion of 11-F-11-cis-ROL to the corresponding all-
trans aldehyde using BaMnO4; however, the oxidation/
reduction procedure did not work reliably in our hands.
A route was therefore employed that sets the 11Z stereo-
chemistry much earlier in the synthesis. This route to
compound 1 is outlined in Scheme 4. Known protected
aldehyde 611 was reacted with (fluorocarbethoxy-
methyl)triphenylphosphonium bromide12 in a Wittig
coupling. Diene 5 was carried through to 11-F-tROL 1
using essentially the same procedure reported for the
11E isomer.10 The final product was obtained as a mix-
ture of 9E/Z isomers and was purified by semi-prepara-
tive HPLC (YMC-PACK, PVA-SIL, 250 · 10 mm, flow
4 ml/min, hexanes/dioxane 90:10) with the desired
9E,11Z isomer (all-trans) eluting first. Due to its lability
to oxygen, compound 1 was stored in aliquots at ꢀ80 ꢁC
in the presence of 10 mol% BHT. The UV spectra of
compound 1, as well as its 11E isomer, were measured
in hexane and are shown normalized in Figure 1.
2.2. 11-F-tROL does not inhibit IMH activity irreversibly
A 3 lM solution of either tROL or 1 was incubated with
RPE membranes at time points between 0 and 90 min at
room temperature. IMH activity was then measured at
each time point with all-trans-[3H]retinol (0.067 lM).
Figure 1. UV spectra of 11-F-tROL (black line) and 11-cis-11-F-ROL
(gray line) measured in hexane. The shoulder at 223 nm is diagnostic of
11-cis-11-F-ROL as the corresponding shoulder peak of native 11-cis-
ROL is red shifted to 233 nm.

5192
N. Fishkin et al. / Bioorg. Med. Chem. 13 (2005) 5189–5194
Membranes treated with 11-F-tROL were just as active
as those incubated with tROL at each time point tested
(data not shown). Had compound 1 been a potent irre-
versible inhibitor (through covalent attachment to
IMH), it would be expected that tracer conversion to
11-cis-retinol would drop off significantly with time
compared to the membranes incubated with native
tROL. This was the case with 11-cis-retinyl bromo-
acetate (cRBA), a specific and potent irreversible inhib-
itor of isomerohydrolase activity.13 Thus, 11-F-tROL is
not an irreversible inactivator of IMH. We then sought
to determine whether 11-F-tROL is a substrate for
IMH, after conversion into the corresponding tRE.
out the reaction showed no conversion to the 11-cis iso-
mer. This demonstrates that isomerization of compound
1 is enzymatic and not simply a background thermal
reaction. 11-F-retinyl esters (labeled in Fig. 2, panel B)
are also produced from compound 1 at both 37 ꢁC
and room temperature in the presence of membranes,
showing that 1 is a substrate for LRAT and, as ex-
pected, the isomerization also proceeds through a retinyl
ester intermediate.
Figure 3 shows the percentage of retinyl esters 11-cis and
13-cis, and all-trans isomers produced after incubation
with RPE membranes for 45 min IMH reaction using
either compound 1 or tROL as the substrate. LRAT
does not seem to distinguish between tROL and
11-F-tROL, and it appears that the 11-F isostere is as
effective an overall substrate for LRAT/IMH as native
tROL. This was further investigated by measuring com-
pound 1 conversion to the 11-cis isomer as a function of
time. The time course of the IMH activity is shown in
Figure 4. The initial rate of conversion of compound 1
to 11-F-11-cis-ROL was about 80% of that for the
IMH reaction with tROL. It also appears that satura-
tion occurs at only about 75% the activity of tROL. This
may be partly due to the fact that at longer reaction
times beyond the linear part of the curve, a significant
amount of 13-cis isomer is produced as a byproduct.
2.3. 11-F-tROL is a substrate for IMH
When RPE membranes are incubated with 1, several
new enzymatic products are generated. One of the major
compounds produced in this reaction is identified as
11-F-11-cis-ROL based on an identical HPLC retention
time and UV spectrum to an authentic standard.10
Figure 2 shows a control experiment in which com-
pound 1 is incubated without membranes in pH 8.0 buf-
fer for 45 min at 37 ꢁC. As can be seen in panel A, no
other 11-F retinol isomers are produced under these
conditions. However, when RPE membranes are added
under otherwise identical conditions, about 20% of com-
pound 1 is converted to 11-F-11-cis-ROL (Fig. 2, panel
B). Relatively small amounts of 11-F-13-cis-ROL as well
as 11-F-11,13-dicis-ROL are also formed in the reaction.
A control experiment in which the RPE membrane pro-
teins were denatured (100 ꢁC, 5 min) prior to carrying
Along these lines, we tested the recombinant cellular ret-
inaldehyde binding protein (CRALBP)14 for its ability
to stabilize and pull out of equilibrium the 11-F-11-cis-
ROL formed in the isomerohydrolase reaction. Carrying
Figure 2. Normal phase HPLC profile of (A) heat control with 11-F-tROL 1 (pH 8.0 buffer, 37 ꢁC, 45 min) without RPE membranes and
(B) reaction of 1 with buffered membranes under the same conditions. (i) 11,13-dicis-11-F-ROL, (ii) 11-cis-11-F-ROL, (iii) 13-cis-11-F-ROL, and
(iv) 11-F REs.

N. Fishkin et al. / Bioorg. Med. Chem. 13 (2005) 5189–5194
5193
is a substrate for the reaction. It should be noted that
there was no characterization offered of the highly
unstable 11-F-tROL in the previous study, so a direct
comparison of the two studies cannot be made.9
The fact that 11-F-tROL is an excellent substrate for
IMH after conversion to ester does not necessarily dis-
prove the carbonium ion mechanism. For example, fluo-
rine may compensate for its destabilization of a
neighboring positive charge through its inductive effect
with its ability to stabilize the charge through n–p back
donation.15 In addition, through resonance, any result-
ing positive charge on the retinoid polyene would likely
be spread over all of the odd numbered carbons and not
localized at C11. Indeed, some experimental justification
for this comes from the UV spectra of 11-fluoro-retinal
and its corresponding protonated Schiff base whose kmax
values are only slightly blueshifted (ꢁ10 nm) compared
to the non-fluorinated compounds.10 Thus, the question
of the mechanism of IMH action remains undecided.
Figure 3. Retinoid product distribution for 11-F-tROL and tROL
after a 45 min reaction with buffered membranes at 37 ꢁC. Substrate
concentrations in the reactions were 3 lM.
Under the conditions tested, 1 did not act as an irrevers-
ible inhibitor of IMH activity. This result does not,
though, rule out the possibility of an SN20 mechanism
for isomerization. The substrate could simply be turning
over many times before a fluoride ion gets released pref-
erentially to the enzymatic nucleophile, making covalent
attachment improbable.
In any event, the fact that compound 1 is shown here to
be a tROL surrogate suggests that 1 could be a useful
compound to further probe interactions between retinol
and such RPE proteins as LRAT and RPE65 using 19F
NMR perturbation experiments.16
3.1. Experimental
3.1.1. Ethyl (2Z,4E)-6[(tert-Butyldiphenylsilyl)oxy]-2-fluo-
ro-4-methylhexa-2,4-dienoate (5). (Fluorocarbethoxy-
methyl)triphenylphosphoniumbromide(4.56 g,0.01mol)
was dissolved in 100 ml dry CH2Cl2 and cooled to
ꢀ78 ꢁC. nBuLi (11.3 ml, 1.6 M in hexanes) was added
dropwise over 20 min. After 30 min, aldehyde 6 (4.0 g,
0.0118 mol) dissolved in dry CH2Cl2 was added drop-
wise over 45 min, and the reaction mixture was allowed
to warm to room temperature. After 2 h, water was
added to the orange slurry, followed by 200 ml hexanes.
The organic phase was washed with satd NaCl and dried
with MgSO4. The solvent was removed in vacuo and the
resulting orange oil was purified by flash column chro-
matography (SiO2, hexanes/EtOAc 10:1) to give 3.63 g
of 5 (72% yield) as a colorless oil.
Figure 4. Time course for the isomerization of 11-F-tROL and native
tROL to their corresponding 11-cis congeners. Activity is reported as
percent of maximal 11-cis retinoid formed.
out the reaction of compound 1 with membranes in the
presence of 60 lM CRALBP did have a significant effect
on the ratio of 11-cis:13-cis isomer produced; ꢁ3:1 in the
case of bovine serum albumin (BSA) versus 8:1 using
CRALBP. This result is consistent with the reported
ability of 11-F-11-cis-ROL to bind to CRALBP with
80% the affinity of native 11-cis-ROL.9
3. Discussion
1H NMR (200 MHz, CDCl3): d 1.08 (9H, s), 1.35 (3H, t,
J = 7 Hz), 1.79 (3H, s), 4.34 (2H, q, J = 7 Hz), 4.41
(2H, q, J = 6 Hz), 6.02 (1H, dt, JH–H = 6 Hz, JH–H =
1.2 Hz), 6.52 (1H, d, JH–F = 39 Hz), 7.4–7.5 (6H, m),
7.7–7.8 (4H, m).
3
4
The previously reported negative result of compound 1
not being an IMH substrate was used to support the no-
tion that isomerization takes place through a carbonium
ion mechanism, as shown in Scheme 2. It was argued
that an electron-withdrawing fluorine atom would
destabilize a putative carbocation intermediate and
therefore make the isomerization improbable. However,
the present findings show quite clearly that 11-F-tROL
3
Compound 5 was carried through to the desired 11-F-
retinol 1 using the same literature procedure10 used for
the corresponding E isomers. The 1H NMR spectra
are reported below for the new compounds.

5194
N. Fishkin et al. / Bioorg. Med. Chem. 13 (2005) 5189–5194
3.1.2. (2Z,4E)-6[(tert-Butyldiphenylsilyl)oxy]-2-fluoro-4-
methylhexa-2,4-dien-1-ol (4). 1H NMR (300 MHz,
CDCl3): d 1.06 (9H, s), 1.76 (3H, s), 4.09 (2H, d,
3JH–F = 13.2 Hz), 4.36 (2H, d, J = 6.3 Hz), 5.44 (1H, d,
3JH–F = 40.2 Hz, 5.73 (1H, t, J = 6.3 Hz), 7.4–7.5 (6H,
m), 7.7–7.8 (4H, m).
hexanes, concentrated under a stream of argon, and
then analyzed by HPLC. A UV profile of each eluted
peak was collected by an online PDA detector (Varian
Pro Star, Wakefield, RI). Isomers of native retinol were
quantified using known molar absorptivity coeffi-
cients.19 Compound 1 was quantified using kmax
=
321 nm (in hexane), e = 39,650 cmꢀ1 Mꢀ1 and the 11-
cis-11-fluoro-retinol using kmax = 317 nm (in hexane),
e = 28,900 cmꢀ1 Mꢀ1. All measurements were repeated
in triplicate and the average values were used for analy-
sis. Standard deviations are reflected in the error bars in
Figures 3 and 4. All manipulations were carried out un-
der dim red light.
3.1.3. (2Z,4E)-6[(tert-Butyldiphenylsilyl)oxy]-2-fluoro-4-
methylhexa-2,4-dienal.
1H
NMR
(200 MHz,
CD3COCD3): d 1.05 (9H, s), 1.78 (3H, s), 4.59 (2H, d,
J = Hz), 6.11 (1H, dt, JH–H = 6 Hz, JH–H = 1.2 Hz),
3
4
3
6.98 (1H, d, JH–F = 25 Hz), 7.4–7.5 (6H, m), 7.6–7.8
3
(4H, m), 9.86 (1H, d, JH–F = 21 Hz).
3.1.4. (7E,9E,11Z,13E)-tert-Butyldiphenylsilyl 11-fluoro
1
retinyl ether (2). H NMR (200 MHz, CDCl3): d 1.02
(6H, s), 1.05 (9H, s), 1.4–1.68 (4H, m), 1.70 (3H, s)
Acknowledgments
1.78 (3H, s), 1.94–2.05 (2H, m), 2.1 (3H, s), 4.41 (2H,
d, J = 6.4 Hz), 5.49 (1H, d, JH–F = 38.6 Hz), 5.79 (1H,
This work was supported by NIH Grant EY-04096 (to
R.R.). We are also grateful to an NIH Vision training
grant (to N.F.).
3
3
t, J = 6.4 Hz), 5.87 (1H, d, JH–F = 30 Hz), 6.12 (1H,
d, J = 16 Hz), 6.32 (1H, d, J = 16 Hz), 7.4–7.5 (6H,
m), 7.7–7.8 (4H, m). HRMS calculated 542.3380, found
543.3382 EI-MS.
References and notes
1. Fishkin, N.; Berova, N.; Nakanishi, K.. Chem. Rec. 2004,
4, 120.
2. Rando, R. R. Chem. Rev 2001, 101, 1881.
3. Saari, J. C. Invest. Ophthalmol. Vis. Sci. 2000, 41, 337.
4. Bernstein, P. S.; Law, W. C.; Rando, R. R. Proc. Natl.
Acad. Sci. U.S.A. 1987, 84, 1849.
5. Gollapalli, D. R.; Maiti, P.; Rando, R. R. Biochemistry
2003, 42, 11824.
6. Xue, L.; Gollapalli, D. R.; Maiti, P.; Jahng, W. J.; Rando,
R. R. Cell 2004, 117, 761.
3.1.5. (7E,9E,11Z,13E)-11-Fluoro-all-trans-retinol (1).
1H NMR (200 MHz, CDCl3): d 1.07 (6H, s, H16 and
H17), 1.4–1.65 (4H, m, H2 and H3), 1.68 (3H, s,
H18) 1.90 (3H, s, H20), 1.94–2.05 (2H, m, H4), 2.08
(3H, s, H19), 4.18 (2H, d, J = 6.2 Hz, H15), 5.46
3
(1H, d, JH–F = 39 Hz, H12), 5.70 (1H, t, J = 6.2 Hz,
H14), 5.87 (1H, d, JH–F = 30 Hz, H10), 6.10 (1H, d,
J = 16 Hz, H8), 6.28 (1H, d, J = 16 Hz, H7). Calculated
3
for C20H29FONa = 327.2095, ESI [M + Na]+ = 327.2107.
UV kmax = 321 nm (in hexane), e = 39,650 cmꢀ1 Mꢀ1
.
7. Deigner, P. S.; Law, W. C.; Canada, F. J.; Rando, R. R.
Science 1989, 244, 968.
8. Rando, R. R. Angew. Chem., Int. Ed. Engl. 1990, 102, 507.
9. McBee, J. K.; Kuksa, V.; Alvarez, R.; de Lera, A. R.;
Prezhdo, O.; Haeseleer, F.; Sokal, I.; Palczewski, K.
Biochemistry 2000, 39, 11370.
10. Francesch, A.; Alvarez, R.; Lopez, S.; de Lera, A. R. J.
Org. Chem 1997, 62, 310.
11. Torrado, A.; Iglesias, B.; Lopez, S.; de Lera, A. R.
Tetrahedron 1995, 51, 2435.
12. Thenappan, A.; Burton, D. J. J. Org. Chem. 1990, 55,
2311.
13. Gollapalli, D. R.; Rando, R. R. Biochim. Biophys. Acta
2003, 1651, 93.
14. Saari, J. C.; Bredberg, D. L. Exp. Eye Res. 1988, 46, 569.
15. Gassman, P. G.; Tidwell, T. T. Acc. Chem. Res. 1983, 16,
279.
3.1.6. IMH inhibition assay. Bovine RPE membranes
were prepared from whole eye cups as described.17
A
buffered membrane suspension (100 mM Tris/HCl, pH
8.0 and 200 lg protein) was incubated with either 1 or
native all-trans-retinol (3 lM) (DMSO vehicle) at room
temperature. Aliquots were removed at 0, 5, 15, 25, 45,
60, and 90 min, and all-trans-retinol [11,12-3H2],
(0.067 lM), DPPC (100 lM), and BSA (2% by wt) were
added. Each reaction was allowed to proceed at 37 ꢁC
for 45 min and then quenched with MeOH. The reti-
noids were extracted in hexanes and were analyzed as
previously reported.18 The amount of radioactive 11-
cis-retinol formed was used as a measure of isomerohy-
drolase activity.
16. Liu, R. S. H.; Colmenares, L. U. Proc. Natl. Acad. Sci.
U.S.A. 2003, 100, 14639.
3.1.7. Substrate conversion assay. IMH reactions were
carried out on a 250 ll scale with 20 lg buffered RPE
membranes (pH 8.0), 100 lM DPPC, 2% BSA or
60 lM CRALBP, and 3 lM 1 or native all-trans-retinol
as the substrate. Reactions were quenched with MeOH
at the indicated time points in Figure 4, extracted with
17. Barry, R. J.; Canada, F. J.; Rando, R. R. J. Biol. Chem.
1989, 264, 9231.
18. Winston, A.; Rando, R. R. Biochemistry 1998, 37, 2044.
19. Sporn, M. B., Roberts, A. B., Goodman, D. S., Eds.; The
Retinoids: Biology, Chemistry, and Medicine; Raven Press:
New York, 1994.