Synthesis and activity of 1-aryl-1′-imidazolyl methyl ethers as non-thiol farnesyltransferase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2004.08.011

A series of imidazole-containing methyl ethers (4-5) have been designed and synthesized as potent and selective farnesyltransferase inhibitors (FTIs) by transposition of the D-ring to the methyl group on the imidazole of the previously reported FTIs 3. Several compounds such as 4h and 5b demonstrate superior enzymatic activity to the current benchmark compound tipifarnib (1) with IC₅₀ values in the lower subnanomolar range, while maintaining excellent cellular activity comparable to tipifarnib. The compounds are characterized as being simple, easier to make, and possess no chiral center involved.

Full text of DOI:10.1016/j.bmcl.2004.08.011

Bioorganic & Medicinal Chemistry Letters 14 (2004) 5371–5376
Synthesis and activity of 1-aryl-10-imidazolyl methyl ethers as
non-thiol farnesyltransferase inhibitors
Qun Li,* Gary T. Wang, Tongmei Li, Stephen L. Gwaltney, II, Keith W. Woods,
Akiyo Claiborne, Xilu Wang, Wendy Gu, Jerry Cohen, Vincent S. Stoll, Charles Hutchins,
David Frost, Saul H. Rosenberg and Hing L. Sham
Cancer Research, GPRD, Abbott Laboratories, Abbott Park, IL 60064-6101, USA
Received 29 June 2004; revised 6 August 2004; accepted 6 August 2004
Available online 3 September 2004
Abstract—A series of imidazole-containing methyl ethers (4–5) have been designed and synthesized as potent and selective farnesyl-
transferase inhibitors (FTIs) by transposition of the D-ring to the methyl group on the imidazole of the previously reported FTIs 3.
Several compounds such as 4h and 5b demonstrate superior enzymatic activity to the current benchmark compound tipifarnib
(1) with IC50 values in the lower subnanomolar range, while maintaining excellent cellular activity comparable to tipifarnib. The
compounds are characterized as being simple, easier to make, and possess no chiral center involved.
Ó 2004 Elsevier Ltd. All rights reserved.
X2
Cancer is an extremely complicated disease encompass-
ing hundreds of different disorders. Emerging science
within the past decade has created many opportunities
for fundamentally new approaches to tackle this dis-
ease.1,2 One such approach targets Ras, an oncogene
that is among the most frequently activating mutated
genes in tumors.3 Ras plays a major role in intracellular
signaling pathways that control cancer cell prolifera-
tion.4 Activation of Ras proteins requires posttrans-
lational farnesylation,a process of covalently attaching
a 15-carbon farnesyl moiety to conserved cysteine resi-
dues.5 Therefore,the search for inhibitors of farnesyl-
transferase (FTase) for the treatment of cancer has
generated considerable recent interest.6,7 Many FTase
inhibitors (FTIs) have demonstrated excellent antitumor
efficacy in preclinical human xenograft models and
several compounds are now in Phase II/III clinical
trials.8–10
Cl
L
Cl
H2N
Cl
C
C
D
D
N
1L2
Me
Me
A
N
Me
N
A
N
Me
B
O
O
N
N
2
1
R3
R1
R3
R1
D
C
C
D
L3
N
n
L3
A
Me
A
N
R2
N
R2
4: n=1
5: n=0
N
3
Figure 1. Modifications of 1 through 2 and 3 lead to achiral 1-aryl-10-
imidazolyl methyl ethers (4, 5).
structural refinement of those analogs resulted in the
identification of a series of biphenyl FTIs 3 (Fig. 1).15
Using a similar strategy as described in the preceding
paper—transposition of the D-ring to the methyl group
on the imidazole in 313—led to a series of methyl ethers
4–5 that are potent,and selective FTIs. 16 We report here
the design,synthesis,and biological activity of this
promising new series.
Tipifarnib(R115777, 1) is one of the most potent and
selective non-thiol containing FTI in clinical trials.11,12
In the preceding paper, we reported the discovery of
pyridones 2 and related analogs as potent FTIs obtained
through deletion of the B-ring of tipifarnib.13,14 Further
The required benzyl bromides 8 were prepared from 6
through Suzuki coupling then NBS bromination
(Scheme 1).13,15,17
Keywords: Farnesyltransferase inhibitors; Anticancer; Tipifarnib.
*
Corresponding author. Tel.: +1 18479377125; fax: +1
18479361550; e-mail: qun.li@abbott.com
0960-894X/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2004.08.011

5372
Q. Li et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5371–5376
X
Ar1
Ar1
CH3
CH3
a
b
Br
R2
Y
6
R2
Y
R2
Y
8
NH2
NC
Br
a
7
X=Cl, B, I; Y=CH, N
NC
NC
8a
14
NC
Scheme 1. Reagents and conditions: (a) Ar1B(OH)2, Pd(OAc)2, Cy-
MAP1, CsF, dioxane, rt, overnight; (b) NBS, AIBN, CCl4, reflux, 12h.
O
H
b
N
N
N
N
N
Reduction of ester 918 with Ca(BH4)2 at room tempera-
ture provided alcohol 10 in 80% yield,which was quanti-
tatively converted to naphthylbenzyl bromide 8a when
treated with tribromophosphine (Scheme 2).
15
H
NC
16
Scheme 4. Reagents and conditions: (a) (i) NaN3, acetone, reflux, 4h;
(ii) PPh3, THF/H2O, reflux 1h, 28%; (b) NaBH(OAc)3, CH2Cl2, rt,
overnight, 50%.
The dimethyl ethers (4) were prepared as described in
Scheme 3. N-Tritylimidazole 11 underwent regio-selec-
tive alkylation with arylmethyl bromide using reaction
conditions developed by Anthony et al.19 to provide
the N-arylmethyl imidazole (12),which was hydrolyzed
to give alcohol 13. Coupling of alcohol 13 with the bro-
mides (8) in the presence of silver oxide furnished the
substituted dimethyl ethers (4).
Cl
Cl
Cl
I
OMe
b
OH
a
OMe
Cl
Cl
17
19
18
c
NC
Preparation of dimethylamine 16 is illustrated in Scheme
4. Accordingly, benzylbromide 8a is converted to benzyl-
amines 14 in a two-step sequence by reaction with
sodium azide and subsequent reduction with triphenyl-
phosphine (28% yield). Aldehyde 1513 underwent
reductive amination with amine 14 to afford the
dimethylamine (16) in 50% yield.
NC
Cl
O
N
N
Cl
Cl
N
N
5a
20
Scheme 5. Reagents and conditions: (a) 3-chorophenylboronic acid,
Pd(Ph3)4, NaHCO3, toluene/EtOH/H2O, reflux, 3h, 89%; (b) BBr3,
CH2Cl2, rt, 1h, 98%; (c) K2CO3, DMF, 55°C, 6h, 31%.
Synthesis of the compounds with a two-atom linker be-
tween the imidazole and aryl groups are illustrated in
Schemes 5 and 6. Suzuki coupling of 17 with 3-chloro-
benzeneboronic acid gave biphenyl 18 in 89% yield.
Compound 18 underwent demethylation (98% yield)
and the resulting phenol (19) was reacted with chloride
NC
NC
HO
Cl
F
N
13a
a
O
N
b
N
NC
NC
21
N
5b
CO2Et
Br
OH
a
b
NC
8a
NC
NC
NC
NC
9
10
X
+
Y
Scheme 2. Reagents and conditions: (a) Ca(BH4)2, THF/EtOH, rt,
overnight, 80%; (b) PBr3, DMF/CH2Cl2, 0°C, 2h, 100%.
c
NC
HO
O
22
N
N
N
Y
NC
X=Cl, Y=N
X=F, Y=CH
N
13b
5c:Y=CH; 5d: Y=N
Ar2
Ar2
N
N
AcO
Scheme 6. Reagents and conditions: (a) 1-naphthylboronic acid,
Pd(OAc)2, Cy-MAP-1, CsF, dioxane, rt, 2days, 92%; (b) NaH,
DMF, rt, overnight, 68%; (c) 13b, NaH, DMF, rt, overnight, 38–48%.
a
b
N
AcO
N
HO
Ph
Ph
N
Ph
11
N
13
12
2020 to furnish the desired product 5a in 31% yield. Sim-
ilarly, 5b was prepared in good yield from 21 through
Suzuki coupling and nucleophilic condensation with
13a (Scheme 6). Coupling of alcohol 13b (prepared in
Scheme 3 from the corresponding bromide)21 with the
para-cyano activated aryl halide (23),yielded targets
5c–d in 38–48% yield.
c
Ar1
R2
Ar2
Ar1
R2
Br
N
N
O
Y
Y
8
4
Scheme 3. Reagents and conditions: (a) Ar2CH2Br, AcOEt, 60°C,
20h; (b) LiOHÆH2O, THF/H2O, rt, 1h; (c) AgO, CH2Cl2, rt, 1day.

Q. Li et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5371–5376
5373
Table 1. Activity of biaryl farnesyltransferase inhibitors
Table 1 (continued)
Compd Ar1
R2
R3
IC50 (nM)
EC50 (nM)
Rasc
processing
R3
FTa GGTb
Ar1
N
N
53%d
O
4r
4s
CN CN 0.92 3400
R2
CN CN 1.3
2300
43%d
1.6
Compd Ar1
R2
R3
IC50 (nM)
FTa GGTb
EC50 (nM)
Rasc
processing
1
Tipifarnibf
0.65 1100
Cl
4a
Cl
CN
0.62 8200
30%d
7.7
a Bovine farnesyltransferase.
b Bovine geranylgeranyltransferase.
c In H-ras NIH-3T3 cells.
d Inhibition at 100nM.
Cl
4b
CN CN
0.37 6800
e Not tested.
Cl
f Data from racemic mixtures.
4c
CN MeSO2 96
NTe
NTe
MeO
4d
CN Cl
CN Cl
2.2
>10,000 0%d
>10,000 34
Biological activities of the compounds were determined
against bovine FTase and cellular Ras processing in H-
ras transformed cells.22 Selectivity against geranylgera-
nyltransferase (GGTase), a closely related enzyme that
is responsible for prenylating the majority of prenylated
proteins, was also tested. These results are summarized
in Tables 1–4.
EtO
4e
1.4
1.2
OEt
4f
CN CN
2300
52%d
F3CO
4g
CN CN
CN CN
0.49 990
54%d
52
The substituted dimethyl ethers (4) demonstrate excel-
lent activity against FTase with IC50 values ranging
from 0.37 to 96nM (Table 1). In general, the dimethyl
ethers (4) are 2- to 5-fold more potent than the corre-
sponding compounds 3,15 while the activity against the
GGTase decreases in comparison with 3.15 A cyano
OCF3
4h
4i
4j
0.65 1300
AcNH
CN CN
CN CN
13
NTe
92%d
group,either at R , R3,or both,dramatically boosts
2
the activity,particularly in the Ras processing assay.
Cyano analog 4b,with an EC
COMe
of 7.7nM in the Ras
50
1.3
>10,000 77
processing assay,is much more potent than the corre-
sponding chloride (4a). Although the exact roles the
cyano group plays are not clear,the potency enhance-
ment may attribute to the tight fit of the cyano groups
into small pockets of the right size of a cyano group
and with proper vector off the aromatic ring. From
the X-ray structure,13 the D-ring cyano group (R3) fits
into a small pocket and accepts hydrogen bonds from
the main chain NH of both Tyr361 and Phe360 of
the b-subunit. Whereas the A-ring cyano group (R2)
occupies a small hydrophobic pocket formed by the
bound farnesylphosphate and Arg202 of the b-subunit
and accepts a hydrogen bond from the OH of Tyr166.
Bu-t
CF3
4k
4l
CN CN
CN CN
0.44 870
49%d
1.6
4.3
7.6
8600
OMe
>10,000 25%d
N
4m
4n
CN CN
CN CN
Cl
Cl
0.37 730
10%d
The effect of the position of the substituents at Ar1 on
activity is not clear. In general, the 3-substituted analogs
are superior. Although being moderate in enzymatic
activity (4.3nM), 4l is the most potent FTI in the cellular
assay as shown in Table 1, with an EC50 of 1.6nM in
inhibiting Ras processing.
Me
Me
O
4o
CN CN
0.60 >10,000 53%d
O
4p
4q
CN Cl
8.3
>10,000 11%d
Although the dimethyl ethers with disubstituted or bi-
cyclic Ar1 (4n–s) demonstrate good enzymatic activity,
they show poor cellular efficacy (Table 1). It appears
that a bicyclic aryl group such as naphthalene (4r–s) is
O
O
CN CN
0.81 NTe
54%

5374
Q. Li et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5371–5376
Table 2. Activity of biaryl farnesyltransferase inhibitors
Ar2
Ar1
Y
N
N
O
NC
Compd
Ar1
Cl
Y
Ar2
IC50 (nM)
EC50 (nM) Rasc processing
FTa
4.3
GGTb
NC
810
26%d
N
4t
NC
Cl
4u
CH
1.7
2200
155
N
NC
NC
NC
NC
NC
Cl
N
4v
4w
4x
4y
4z
CH
CH
CH
CH
CH
3.0
4.4
1.5
0.58
4.7
>10,000
>10,000
>10,000
>10,000
5500
43%d
49%d
35%d
17
MeO
MeO
EtO
N
N
N
O
O
O
O
29
N
O
O
NC
N
4aa
4bb
CH
CH
3.0
>10,000
NTe
16
NC
Br
O
N Me
>1000
NTe
a See Table 1.
b See Table 1.
c See Table 1.
d Inhibition at 100nM.
e Not tested.
less desirable as a C-ring component compared with
substituted phenyl groups.
sulted in a nearly four-fold drop in activity (4s vs 4gg).
In general,the compounds with these modifications
(Table 3) are inferior to the earlier series mainly because
of their sharply reduced cellular activity. Replacing the
oxygen in 4gg with a nitrogen resulted in sharply
increased Ras processing activity (16).
With the hope of improving cellular potency by lowering
the LogP, the A- and D-rings are substituted by pyri-
dines and pyridones (e.g., CLogP values of 4b and 4u
are 3.5 and 2.0, respectively). However, the compounds
(4t–bb) are generally less potent than the phenyl analogs,
with IC50 values ranging from 0.58 to 4.7nM and an
EC50 of 16nM for the best compound (4aa) (Table 2).
The pyridone (4bb) is completely inactive.
The activity of compound 5a (Table 4), which has a
two-atom linker, drops six-fold as compared with 4a,
suggesting that a three-atom linker is perhaps optimal
when the C-ring is attached at the 2-position (meta to
the cyano group of the A-ring or D-ring). However,
more potent compounds are obtained in the two-atom
linked series by attaching the C-ring ortho to the cyano
group on either the A-ring or the D-ring. Thus com-
pound 5b,with a 1-naphthyl group at the 3-position of
the A-ring,is significantly more potent than the corre-
sponding three-carbon-linked C-2 analog (4s) with an
IC50 of 0.38nM. Compound 5b is the most potent
compound of the current study in the cellular Ras
Other more drastic modifications including addition of
an extra aryl group at C-30 of the D-ring led to sharply
lowered activity, especially in the Ras processing assay
(4cc–ee, Table 3). Notably, transposition of the C-ring
aryl groups from the A-ring to C-20 of the D-ring has lit-
tle effect on the activity (4s vs 4ff),confirming the molec-
ular modeling result that the C-ring is actually very close
in space to the D-ring. However,further transposition
of the naphthylene from C-2 to C-3 on the A-ring re-
processing assay,with an EC
of 1.2nM,versus
50

Q. Li et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5371–5376
Table 3. Activity of biaryl farnesyltransferase inhibitors
5375
CN
3'
D
R4
2'
Ar1
3
2
N
W
A
N
NC
Compd Ar1
W
R4
IC50 (nM)
EC50
(nM) Rasc
processing
FTa
10
GGTb
Figure 2. Stereo view of an overlay of a model of compound 4b (in
green) over the X-ray crystal structure of tipifarnib( 1)13 (in purple) on
complex with FTase in the active site. Zn+2 is in shown in gray and
hydroxy farnesylpyrophosphate in blue.
Cl
Cl
O
4cc
O
O
87
0%d
NTe
2-
O
O
O
1.6nM for tipifarnib (1). Similarly,compounds 5c–d,
with a 1-naphthyl group at the 30-position of the D-ring
(ortho to the cyano group) are both subnanomolar
FTase inhibitors. Unfortunately,the improvement in
FTase inhibitory activity of 5b–d is accompanied by
the unfavorably increased activity against GGTase. On
average,selectivity for FTase of compounds 5b–d is
more than 10-fold lower than that of the corresponding
three-atom linked analogs in Table 1.
O
4dd
340
NTe
2-
O
4ee
O
6.8
1400
0%d
461
2-
4ff
H
O
0.96 2100
Stereo view of an overlay of a model of 4b, which was
modeled based on the crystal structure of the close
chemical analog,15 and the X-ray crystal structure of
tipifarnib( 1)13 is shown in Figure 2. The model of 4b
superimposes very well with tipifarnib with a 2.3A dis-
tance between the active site Zn+2 and the imidazole
nitrogen. The A-ring extends out over the loop of resi-
dues Asp359-Phe360 forming a good van der Waals con-
tact with the loop. The C-ring is stacked against Trp106
and Trp102 and the D-ring stacks along the hydroxy
farnesyl pyrophosphate (HFP). The C- and D-rings also
stack together forming a strong p/p interaction.
3-
4gg
O
H
H
4.8
6.5
1300
220
5%d
5%d
3-
16
NH
a See Table 1.
b See Table 1.
c See Table 1.
d Inhibition at 100nM.
e Not tested.
We have previously reported the discovery of pyridones
(2) and close related analogs as potent FTIs based on
structural modifications of tipifarnib( 1).13,14 Further
structural refinements led to the identification of a series
of promising biphenyl FTIs as represented by 3.15 In the
current studies, a series of imidazole-containing methyl
ethers (4–5) have been designed and synthesized as po-
tent and selective farnesyltransferase inhibitors (FTIs)
by transposition of the D-ring to the methyl group on
the imidazole of 3. Several compounds such as 4l and
5b demonstrate potent in vitro enzymatic activity with
IC50 values in the subnanomolar range,while maintain-
ing excellent cellular activity comparable to tipifarnib.
These encouraging results warrant further efforts to
optimize the properties of the molecules in this series.
Table 4. Activity of biaryl farnesyltransferase inhibitors
CN
R4
3'
2'
D
Ar1
3
2
N
O
A
Y
N
X
Compd Ar1
X
Y
R4
IC50 (nM)
EC50 (nM)
Rasc pro-
cessing
FTa GGTb
Cl
2-
Cl
CH
H
H
4.0
1800
0%d
5a
3-
5b
CN CH
CN CH
0.38 240
0.32 565
1.2
5c
5d
H
H
49%d
References and notes
1. Traxler, P.; Bold, G.; Buchdunger, E.; Caravatti, G.;
Furet, P.; Manley, P.; OꢀReilly, T.; Wood, J.; Zimmer-
mann, J. Med. Res. Rev. 2001, 21, 499–512.
2. Baselga, J.; Averbuch, S. D. Drugs 2000, 60(suppl. 1), 33–
40.
CN
N
0.90 420
60
a See Table 1.
b See Table 1.
c See Table 1.
3. Bos, J. L. Cancer Res. 1998, 49, 4682–4689.
4. Downward, J. Nat. Rev. Cancer 2003, 3, 11–22.
d Inhibition at 100nM.

5376
Q. Li et al. / Bioorg. Med. Chem. Lett. 14 (2004) 5371–5376
5. Hurwitz, H. I.; Casey, P. J. Curr. Top. Membr. 2002, 52,
531–550.
6. Ayral-Kaloustian, S.; Salaski, E. J. Curr. Med. Chem.
2002, 9, 1003–1032.
7. Singh, S. B.; Lingham, R. B. Curr. Opin. Drug Dis. Dev.
2002, 5, 225–244.
8. Caponigro, F.; Casale, M.; Bryce, J. Expert Opin. Investig.
Drugs 2003, 12, 943–954.
Kohl, G. D.; Gibbs, J. B.; Oliff, A. I.; Graham, S. L.;
Hartman, G. D. Bioorg. Med. Chem. Lett. 1999, 9, 3301–
3306.
17. Tong, Y.; Lin, N.-H.; Wang, L.; Hasvold, L.; Wang, W.;
Leonard, N.; Li, T.; Li, Q.; Cohen, J.; Gu, W.-Z.; Zhang,
H.; Stoll, V.; Bauch, J.; Marsh, K.; Rosenberg, S. H.;
Sham, H. L. Bioorg. Med. Chem. Lett. 2003, 13, 1571–
1574.
9. Dancey, J. E. Curr. Pharm. Des. 2002, 8, 2259–2267.
10. Purcell, W. T.; Donehower, R. C. Curr. Oncol. Rep. 2002,
4, 29–36.
11. Venet, M.; End, D.; Angibaud, P. Curr. Top. Med. Chem.
2003, 3, 1095–1102.
18. Claiborne, A. K.; Gwaltney, S. L., II; Hasvold, L. A.; Li,
Q.; Li, T.; Lin, N.-H.; Mantei, R. A.; Rockway, T. W.;
Sham, H. L.; Sullivan, G. M.; Tong, Y.; Wang, G.; Wang,
L.; Wang, X.; Wang, W. WO02074747, 2002.
19. Anthony, N. J.; Gomez, R. P.; Schaber, M. D.; Mosser, S.
D.; Hamilton, K. A.; OꢀNeil, T. J.; Koblan, K. S.;
Graham, S. L.; Hartman, G. D.; Shah, D.; Rands, E.;
Kohl, N. E.; Gibbs, J. B.; Oliff, A. I. J. Med. Chem. 1999,
42, 3356–3368.
20. Lee, H.; Lee, J.; Lee, S.; Shin, Y.; Jung, W.; Kim, J. H.;
Park, K.; Kim, K.; Cho, H. S.; Ro, S.; Lee, S.; Jeong, S.
W.; Choi, T.; Chung, H. H.; Koh, J. S. Bioorg. Med.
Chem. Lett. 2001, 11, 3069–3072.
21. Gwaltney, S. L., II; OꢀConnor, S. J.; Nelson, L. T. J.;
Sullivan, G. M.; Imade, H.; Wang, W.; Hasvold, L.; Li,
Q.; Cohen, J.; Gu, W.-Z.; Tahir, S. K.; Bauch, J.; Marsh,
K.; Ng, S.-C.; Frost, D. J.; Zhang, H.; Muchmore, S.;
Jacob, C. G.; Stoll, V.; Hutchins, C.; Rosenberg, S. H.;
Sham, H. L. Bioorg. Med. Chem. Lett. 2003, 16, 1363–
1366.
22. Assay methods described in: Vogt, A.; Qian, Y.; Blaskov-
ich, M. A.; Fossum, R. D.; Hamilton, A. D.; Sebti, S. M.
J. Biol. Chem. 1995, 270, 660–664.
12. Norman, P. Curr. Opin. Invest. Drugs 2002, 3, 313–319.
13. Li, Q.; Claiborne, A.; Li, T.; Hasvold, L.; Stoll, S. V.;
Muchmore, S.; Jakob, C. G.; Gu, W.; Cohen, J.; Hutchins,
C.; Frost, D.; Rosenberg, S. H.; Sham, H. L. Bioorg. Med.
Chem. Lett. 2004, 14, preceding paper. doi:10.1016/
j.bmcl.2004.08.012.
14. Hasvold, L. A.; Wang, W.; Gwaltney, S. L.; Rockway, T.
W., II; Nelson, L. T. J.; Mantei, R.; Fakhoury, S.; Sullivan,
G.; Li, Q.; Lin, N.-H.; Wang, L.; Zhang, H.; Cohen, J.; Gu,
W.-Z.; Marsh, K.; Bauch, J.; Rosenberg, S.; Sham, H.
Bioorg. Med. Chem. Lett. 2003, 13, 4001–4005.
15. Wang, L.; Wang, G. T.; Wang, X.; Tong, Y.; Sullivan, G.;
Park, D.; Leonard, N.; Li, Q.; Cohen, J.; Gu, W.-Z.;
Zhang, H.; Bauch, J.; Jacob, C. G.; Hutchins, C. W.; Stoll,
S. V.; Marsh, K.; Rosenberg, S. H.; Sham, H.; Lin, N.-H.
J. Med. Chem. 2004, 47, 612–626.
16. Dinsmore, C. J.; Williams, T. M.; OꢀNeill, T. J.; Liu, D.;
Rands, E.; Culberson, J. C.; Lobell, R. B.; Koblan, K. S.;