CD2: Modern Approaches to Drug Discovery William Hodds, wh738@york.ac.uk.

Structure-Based Discovery of the Selective BCL-2 Inhibitor ABT-199: A Review

Leukaemia and The Role of BCL-2 in Lymphocytic Leukaemia
Leukaemia is a large group of similar diseases, all of which share a common characteristic: they
involve the unregulated proliferation of red and white blood cells with atypical metabolism and
protein expression, which entirely ignore Hayflicks normal division limit of ~50 divisions. The
ability to ignore internal apoptotic signals is an essential hallmark of cancer[1] and leads to
aggressive proliferation of the cell, resulting in the formation of large clusters of abnormal cells,
known as tumours. Tumours disrupt tissue homeostasis and cause local necrosis of surrounding
tissue[2], often leading to patient fatality.
In normal physiology, one such apoptotic signal is the release of Cytochrome C (Cyt C) from the
mitochondrial periplasm[3] through specialised channels in the outer membrane, normally
suppressed by the BCL-2 and BCL-XL proteins[4]. Release of Cyt C leads to the activation of an
executioner caspase protease cascade, which ultimately leads to cell death. BCL-2 prevents
release of Cyt C from the mitochondria by binding the 2-helix (BH3 domain) of pro-apoptotic
proteins BAX and BAK[5] (which serve to open the channels[6]), thereby inhibiting them.
In lymphocytic leukaemia, however, BCL-2 becomes highly, constitutively expressed, reducing
the intracellular [pro-apoptosis]:[anti-apoptosis] protein ratio, and reducing B-cell propensity to
apoptosis significantly[7], ultimately leading to tumour formation. Therefore, modulation of BCL2 is highly desirable as a viable means of treating this type of leukaemia.
Structural Characteristics of BCL-2
BCL-2 is a 239 amino acid protein consisting of 7 -helices, 4 of which participate in binding, 2,
3, 4, and 6 (see figure 1). The binding groove, formed by these 4 active -helices, is mostly
hydrophobic and large at ~20 in length, reflecting the role of BCL-2 to bind the 2 helix of
other, pro-apoptotic, BCL-2 family proteins, such as BAX or BAK[8]. The hydrophobic groove is
made up of 2 major pockets, P2 and P4, which are the major mediators of strong ligand binding,
though minor electrostatic interactions from a conserved Arg10 (or Arg15) does contribute
binding to the 2 of pro-apoptotic proteins, the natural ligand of BCL-2.

a)

b)

P4

3
2

P2

Figure 1. a) Ribbon model of BCL-2,

showing -helices 1-6, with the 4 -helices
which constitute the binding groove
shown in yellow. b) Space-filling model of
BCL-2 showing the large 20 hydrophobic
binding groove, made up of two pockets,
P4 and P2. Negative and positive charges
are shown in red and blue, respectively.

CD2: Modern Approaches to Drug Discovery William Hodds, wh738@york.ac.uk.

The development of a BCL-2 selective (BCL-XL sparing) inhibitor

As noted before, BCL-2 exerts cellular effect by the inhibition of pro-apoptotic proteins such as
BAX and BAK, thereby prolonging the life of the cancerous cell, by avoiding apoptosis.
Therefore, inhibition of BCL-2 itself will prevent this prolongation, and induce cell death. In
previous work, it is shown that ABT-263 (navitoclax, 1) effectively inhibits BCL-2[9]. 1 binds BCL-2
by the hydrophobic pockets P2 and P4, but does not interact with the BCL-2 family conserved
Arg10 that is exploited in natural BH3 domain binding. To compensate, minor electrostatic
interactions between backbone NH-CO of Gly142 and the sulfonamide of 1, and NH-CO
backbone of Tyr199 and the morpholine of 1 are engaged in 1-BCL-2 binding.
In addition, intramolecular -stacking is observed in pocket P4, when 1 is bound to BCL-2,
between the sulfuryl substituted aryl ring and the thioaryl ring.
However, due to the high sequence homology of BCL-2 to another anti-apoptotic protein, BCLXL, 1 also inhibits BCL-XL , resulting in severe thrombocytopenia (reduction in blood platelet
concentration), as BCL-XL is also the primary survival factor of blood platelets[10].
Therefore, it is desirable to modify 1 to obtain a BCL-2 inhibitor with high affinity for BCL-2 (Ki =
0-10 nM) whilst having a low affinity for BCL-XL (Ki = 102-3 Ki (BCL-2)), all the while retaining the
oral bioavailability that 1 possesses.
In order to do this, Souers et al[11] determined the binding affinities (Ki) of a range of analogues
of 1, and revealed that analogues without the thioaryl ring unit began to become selective
toward BCL-2 whilst sparing BCL-XL. The crystal structure of the new analogue, 2, bound to BCL2 was determined, and showed that 2 adopted a similar overall binding mode to 1, but
occupied a smaller cavity volume.

P2

P4

P2

P4

a)

b)

Figure 2. a) The binding mode of 1 to BCL-2. b) The binding of 2 to BCL-2, showing the vacancy
in the P4 pocket which, in the crystal structure, is occupied by Trp30 of another BCL-2 protein,
engaging in hydrogen bonding with Asp103.
The additional vacant volume allows for the intercalation of the indole of Trp30 from another
BCL-2 protein into the P4 pocket, which is engaged in -stacking with the nitro-aryl unit of 2 in a
manner reminiscent of the intramolecular -stacking seen in 1. The homodimerisation of two
BCL-2 proteins, to fulfil the -stacking interaction in the P4 pocket, highlights the importance of
this hydrophobic interaction in the ligand binding to BCL-2.

CD2: Modern Approaches to Drug Discovery William Hodds, wh738@york.ac.uk.

In addition to this finding, it was found that the indole group of the intercalating Trp30 was
oriented so to form a hydrogen bond with the Asp103 of the other BCL-2 of the dimer. This
Asp103 hydrogen bond was used to discriminate between BCL-2 and BCL-XL, as in BCL-XL Glu96
takes the place of Asp103, and cannot participate in hydrogen bonding in the same manner at
physiological pH, as a hydrogen bond acceptor is necessary for interaction with the N of the
Trp30 indole.
To replicate the stabilising interactions that 2 revealed to favour BCL-2 but spare BCL-XL,
analogue 3 was made (ABT-199), and was shown to effectively capture the Asp103 hydrogen
bond with the azaindole N atom in the crystal structure (not available on PDB), as was seen in
the 2-BCL-2 complex with the indole of Trp30, and due to this, a preferential binding toward
BCL-2 was obtained, whilst sparing BCL-XL and other proteins of the BCL-2 family.
Table 4. Data taken from the work of Souers et al [11], showing the high affinity of ABT-199 for
BCL-2, but significantly lower affinity for other anti-apoptotic proteins within the BCL-2 family,
most notably BCL-XL.
Protein
BCL-2
BCL-XL
BCL-W
MCL-1

function
Anti-apoptotic
Anti-apoptotic
Anti-apoptotic [15]
Anti/Pro-apoptotic [16]

Ki (to ABT-199)
< 0.010 nM
48 nM
245 nM
> 444 nM

ABT-199

Figure 5. Summary of the structure-guided chemical development, from 1 to 2 to ABT-199,

showing the use of an azaindole group in ABT-199 to gain an additional hydrogen bonding
interaction with Asp103 from BCL-2, thus discriminating against binding to BCL-XL.
Binding in the P2 pocket was found to be highly preferential to hydrophobic interactions,
therefore the change of the moiety in 1 to a 6 membered allyl-ether was reverted back to the
3,3-dimethyl cyclohexenyl moiety. Both 2 and ABT-199 also have no chirality, thereby
simplifying the chemical synthesis of 2 and ABT-199, relative to 1, which would need
enantiomeric purification, and would give a reduced yield of the desired enantiomer.

CD2: Modern Approaches to Drug Discovery William Hodds, wh738@york.ac.uk.

The clinical efficacy of ABT-199

Two different Interleukin-3 (IL-3) dependent mouse FL5.12 cell lines engineered to rely on BCL2 or BCL-XL for survival in the absence of IL-3 were tested in vitro to determine the pre-clinical
efficacy of ABT-199 at selectively binding BCL-2 and sparing BCL-XL [12]. The EC50 (concentration
of ABT-199 at which 50% cell death is achieved) was determined for both these cell lines, and
was found to be 4 nM for the BCL-2 dependent cells, and 261 nM for the BCL-XL cells [18],
showing selectivity for BCL-2.
Further to this finding, monitoring the release of Cyt C from the mitochondria upon
administering ABT-199 in a similar cell line (using southern blotting) reveals that Cyt C
concentration in the cytosol increases in an ABT-199 concentration-dependent manner,
proportional to a decrease in the mitochondrial concentration of Cyt C. This confirms that the
observed cellular effect is obtained through the release of Cyt C from the mitochondria, and not
another effect, such as Cyt C upregulation in the cell cytosol, for example.
ABT-199 conc. (M):

0.001

0.003

0.01

0.03

0.1

0.3

Cyt C
(mitochondrial)
Cyt C
(cytosolic)

Figure 6. Schematic southern blotting assay showing the ABT-199 concentration-dependent

increase of cytosolic Cyt C, and decrease of mitochondrial Cyt C.
This suggests that, clinically, thrombocytopenia should be greatly reduced in leukaemia patients
dosed with ABT-199 relative to leukaemia patients dosed with navitoclax, whilst ABT-199
patients should still benefit from tumour regression. This was confirmed by a small-scale clinical
trial in 3 patients with refractory lymphocytic leukaemia, where, upon dosing of 100 mg or 200
mg of ABT-199, mass tumour lysis was detected but no clinically-significant blood platelet
decrease was observed, within 6 hours[13], nor was any major organ dysfunction observed. One
patient did experience mild intravascular coagulation, but this was attributed to the
overwhelming of the excretion system due to tumour lysis metabolites. Therefore, ABT-199 was
shown to have high efficacy at treating lymphocytic leukaemia in vivo, whilst avoiding platelet
reduction side effects, verifying the pre-clinical predictions.
References
1. Jerry W. Shay & Woodring E. Wright, Nature Reviews Molecular Cell Biology 1, 72-76 (October 2000).
2. R. Weinberg, Trends in Cancer, 2015, 1, 4-5.
3. 5. G. Kroemer and J. Reed, Nature Medicine, 2000, 6, 513-519.
4. 6. 7. R. Weinberg, The biology of cancer, Garland Science, New York., 2007. Pg 336 341.
8. G. Kroemer and J. Reed, Nature Medicine, 2000, 6, 513-519.
9. A. Petros, Biochimica et Biophysica Acta (BBA) - Molecular Cell Research, 2004, 1644, 83-94.
10. M. Wendt, Expert Opinion on Drug Discovery, 2008, 3, 1123-1143.
11.12.13. A. Souers, et al., Nature Medicine, 2013, 19, 202-208. (main source of information).