Spoilt by choice – Which CYP-specific probe to use?

The use of in vitro metabolic surrogates (e.g. microsomes, recombinant CYP450s, cyro-preserved hepatocytes) is now widespread in drug discovery and evermore refined methods improving the utility of these model are in constant development. However, as any IVIV extrapolation is always subject to the reduced complexity of those model systems it is vital to understand their limitations (e.g. reduced expression of CYP3A4, PXR and CAR in CACO-2s; rapid loss of metabolic competence, canalicular and basolateral efflux transport in freshly isolated hepatocytes.) to avoid misinterpretation of data

In order to predict possible Drug-Drug Interactions (DDIs) it is necessary to understand the relative contribution of individual CYPs to the overall phase I metabolism of an NCE and to this end Relative Activity Factor (RAF), developed twenty years ago (1), has been used alongside inhibition approaches to elucidate the CYP reaction phenotype. Individual recombinant CYPs (rCYP), expressed in and isolated on bacterial membranes, can be used to measure the clearance (CLrCYP) of a CYP-selective probe.  The probe is then assayed in microsomes to obtain CLHLM and a correlation made of the relative levels of clearance in each system (RAF) for that particular CYP.  Once established the RAFs for each CYP can be used to assess the relative contribution of the individual CYPs to the metabolism of a NCE in microsomes.

Highly diverse RAFs are generated between various institutions due to the variability of microsome batches, rCYP expression levels and assay conditions but as long as these variables are maintained within any given laboratory the RAFs should generate internally consistent data. However, whilst it has been known for some time that the promiscuity of CYPs may be facilitated by multiple binding regions in the active site (2), until recently no one has directly assessed the effect of probe choice on whether the scaling from rP450 to HLMs is consistent between various P450-selective probe reactions and those of the test NCE by that P450 isoform.

To demonstrate this issue Sui et al (3) generated RAFs for 2C9 and 3A4 from three CYP-selective probes each.

CYP CYP-selective substrate
2C9 Diclofenac



3A4 Midazolam



Using the RAF generated by one probe the predicted microsome clearance (CLHLM (p)) was calculated for the other two probes then compared with the directly measured CLHLM for those probes.  This was performed in a crossover manner for each of the probes.

The CLrCYP and CLHLM were derived using standard Michaelis-Menton kinetics

snip 1

The RAF was then generated as a ratio between CLHLM and CLrCYP.snip 2

In each crossover the CLHLM (p) was then simply calculated as the measured clearance of the test probe with rCYP multiplied by the RAF.

snip 3

These predicted CLHLM(p) values were then subsequently compared with the actual measured CLHLM to give a value for the Intersystem Clearance Ratio (ICR)

snip 4

Fig 1. Crossover analysis of ICRs based on the RAFs derived from A = Diclofenac, B = tolbutamide and C = Warfarin for 2C9 and D = Midazolam, E = Nifedipine and F = Testosterone for 3A4


Whilst it is clear to see the effects of probe choice on ICR this then has a knock on effect when determining the relative CYP contribution to the metabolism of a test NCE. Fig 2. Shows the comparison of % relative contributions for CYP metabolism of substrates with RAFs generated from various combinations of 2C9 and 3A4 probes (RAFs generated 1A2, 2C19, and 2D6 from single probe throughout)

Fig 2. Variations in relative CYP contribution (fm, fraction of total metabolism attributed to specific CYPs) calculations subject to probe choice

marcus second

With Physiologically based pharmacokinetic (PBPK) modelling and simulation playing an increasingly large role in drug development the accuracy of the input data is therefore crucial to the predictive accuracy of a model. Here, the generation of fm is demonstrably affected by probe choice and if the RAFs for a given probe/CYP pair are not appropriate for the test NCE, deviations in fm from the true value may significantly impact, for instance, generation of risk assessments for drugs as potential DDI victims.

This study would suggest that for test NCEs, fm should be generated using various combinations of a limited number of CYP-specific probes which represent the full range of specific substrate binding sites for a given CYP.  Presently our understanding of CYP binding site multiplicity is limited although studies indicate that we may soon have probe/inhibitor pairs for discreet pharmacophores (4, 5) facilitating increased accuracy of fm prediction.

Blog written by Marcus Hanley


  1. Clarke S. E. (1998) In vitro assessment of human cytochrome P450. xenobiotica, 1998, vol. 28, no. 12, 1167-1202
  2. Korzekwa, K R (1998) Evaluation of atypical cytochrome P450 kinetics with two-substrate models: evidence that multiple substrates can simultaneously bind to cytochrome P450 active sites. Biochemistry, 24 March 1998, Vol.37(12), pp.4137-47
  3. Siu Y.A (2017) Impact of Probe Substrate Selection on Cytochrome P450 ReactionPhenotyping Using the Relative Activity Factor. Drug Metab Dispos 45:183–189
  4. Kumar V. (2006) CYP2C9 Inhibition: Impact of Probe Selection and Pharmacogenetics on in Vitro Inhibition Profiles. Drug Metab Dispos Vol. 34 (12):1966-1975
  5. Foti R.S. (2008) CYP2C19 Inhibition: The Impact of Substrate Probe Selection on in Vitro Inhibition Profiles. Drug Metab Dispos Vol. 36 (3): 523-528

Organs-on-a-chip: The future of drug discovery?

Organs- on-a-chip were initially established at the Wyss Institute for Biologically Inspired Engineering located at Harvard University. The polymer chips containing microtubules are designed to recapitulate the structural, functional and mechanical attributes of human organs and they are only the size of a USB stick (Wyss Institute (2017)).

The chips are not just incredibly cool but also provide an innovative platform for drug discovery, which was recognised by the National Centre for the Replacement, Refinement and Reduction of animals in research (NC3Rs) back in 2012 (3Rs (2012)). The fact that the organs-on-a-chip mimic multiple aspects of the human body means that they could one day be used as an alternative approach to in vivo testing – a controversial but currently essential aspect of drug development.

The lung-on-a-chip, just one of the many organs developed, reconstructs the alveolar-capillary interface of the lung (Huh, D.et al (2010)). A central porous membrane is coated in extracellular matrix proteins that are present in the human lung. One side of the membrane is lined with alveolar epithelial cells isolated from a human lung and human pulmonary microvascular endothelial cells cover the other. The principal compartment allows for air to flow over the epithelial cells and for a blood-like solution containing nutrients to run beneath the endothelial cells. A vacuum on either side of the membrane causes the cells to stretch, in term conveying the movement experienced in the lung when we breathe.


olivia 1

Figure 1: Lung-on-a-chip diagram From Keane, J. (2013)

This model has proven to replicate a lung infection, where white blood cells migrate from the blood-like solution through onto the epithelial side and are observed engulfing bacteria present in the air space using time-lapse fluorescence microscopy (Huh, D.et al (2010)). Small airways-on-a-chip with its goblet and ciliated epithelial cells can be used to model diseases like chronic obstructive pulmonary disease (COPD) and asthma. Addition of interleukin 13 (IL-13) to the epithelium further replicates an asthmatic phenotype inducing goblet cell hyperplasia and cytokine hypersecretion. A paper in Nature has reported to be able to reverse this response with an inhibitor of the JAK-STAT pathway involved in the signalling of IL-13, showing the application of these models in the screening of new treatments (Benam, K. Villenave, R. et al. (2016)).

The next focus is the humans-on-a-chip, which utilize an automated device to connect multiple organs-on-a-chip via a shared vascular network (figure 2). This platform would enable scientist to investigate the pharmacokinetics and pharmacodynamics of a drug in a relevant system (Abaci, H.E. Shuler, M.L. (2015)). Whereby drugs could be administered via the lung-on-a-chip, absorbed by the gut-on-a-chip, metabolised by the liver-on-a-chip and excreted by the kidney-on-a-chip while accessing the efficacy and any toxicity throughout.

Olivia 2

Figure 2: Human-on-a-chip schematic From Mok, J. (2015)

An interesting application of this technology is in the scope of personalized drugs. A chip can be developed with an individual’s cell to determine personal response to a drug, providing a tailor made drug with optimal efficacy (Hamilton, G. (2016)). This notion can also be implemented to clinical trials, where potential new drugs could be tested on cells from a certain genetic populations or on cells from children for paediatric medicines. This technology still has a way to go but one day it may transform the drug discovery process.

Blog written by Olivia Simmonds


Abaci, H.E. Shuler, M.L. (2015). Human-on-a-chip design strategies and principles for physiologically based pharmocokinetics/pharmacodynamics modeling. Integr Biol (Camb). 7 (4), 383-391.

Benam, K. Villenave, R. et al. (2016) Small airway-on-a-chip enables analysis of human lung inflammation and drug responses in vitro Nature Methods. 13 (2), 151-157

Hamilton, G. (2016). Body parts on a chip. Available:  https://www.ted.com/talks/geraldine_hamilton_body_parts_on_a_chip. Last accessed 20th April 2017.

Huh, D.et al (2010) Reconstituting Organ-Level Lung Functions on a Chip Science. 328, 1662-1668

Keane, J. (2013) The End of Drug Testing on Animals, Lung-on-a-Chip Device. Available: http://www.industrytap.com/lung-on-a-chip-device-to-end-drug-testing-on-animals/2160. Last accessed 21st April 2017

Mok, J. (2015) Organs-on-Chips Emulates Human Organs for Better Biomedical Testing. Available: https://thenewstack.io/organs-on-chips-emulates-human-organs-for-better-biomedical-testing/. Last accessed 21st April 2017

Wyss Institute (2017)Human organs on a chip. Avalible: https://wyss.harvard.edu/technology/human-organs-on-chips/. Last accessed 21st April 2017.

3Rs (2012). 3Rs Prize winners. Available: https://www.nc3rs.org.uk/3rsprizewinners. Last accessed 21st April 2017.


Towards a Heat-Stable Rapid-Acting Insulin


Insulin is a small globular protein hormone secreted by the pancreas to lower blood glucose levels. Subcutaneous injection of insulin has long been established as a treatment for diabetes mellitus in which either the pancreas ceases to produce insulin (type 1) or in which the tissues become insensitive to the action of insulin (type 2).

Structurally, the Insulin monomer consists of two chains, an A chain, containing 21 residues and a B-chain containing 30 residues (fig. 1a).

Raj 1




Raj 2



Figure 1. (a) The disposition of the two chains in the insulin molecule indicating the three disulphide bridges. (b) The hydrogen-bonds (dashed lines) of the antiparallel b-sheet formed between the two molecules of the insulin dimer and the cluster of non-polar residues at the interface [1].

Both chains pack to form a compact globular domain stabilized by three disulphide bridges (A6-A11, A7-B7, and A20-B19). PheB24 lies at the classical receptor-binding surface with its aromatic ring packed against the hydrophobic core (fig. 1b) and has been proposed to direct a change in conformation on receptor binding, with residues B24–B30 detaching from the core (a). It is this propensity for conformational change which makes the insulin monomer susceptible to fibrillation whereupon the structure changes from a predominantly α-helical state to a β-sheet rich conformation (fig. 2).

Raj 3

Figure 2. (a) Native structure of the insulin dimer. (b) View of insulin fibril model, looking down fibril axis [2]. (c) The protofilament structure of insulin amyloid fibrils [3]. (d) Insulin fibrils formed after incubation at 65°C for 2 days in the presence of 1 mM NaCl (pH 2.0)[4].

Possibly for this reason, in the pancreatic β-cell insulin has evolved to be stored as Zn2+-stabilized hexamers, arranged in crystalline arrays within mature storage granules (b).

Raj 4

Figure 3. The 2-zinc insulin hexamer of monomers (or a trimer of dimers). Each axial Zn2+ ion is coordinated by three B10 His side chains [1].

Upon secretion from the pancreas in response to rising blood glucose levels, the insulin Zn2+-hexamers dissociate into Zn2+-free dimers and monomers for immediate passage into blood capillaries. The rate of absorption of injected insulin is also limited by the time required for dissociation of the hexamers into the monomer. Recombinant DNA technology has made it possible to prepare rapid-acting insulin analogs with accelerated heaxmeric disassembly that remain dimeric or even monomeric at high concentration by introducing amino acid substitutions into the molecule.

Insulin KP (lispro – Humalog) is one such rapid-acting analog developed by Eli Lilly that can be injected just before meals. Lispro contains the substitutions ProB28→Lys and LysB29→Pro, which destabilise the classical dimer-forming C-terminal anti-parallel β-sheet, an inversion that mimics the sequence of the homologous insulin-Iike growth factor-1 (IGF-1)(Fig. 4).

Raj 5.png

Fig. 4. The hydrogen-bonds of the antiparallel b-sheet formed at the dimer interface of native insulin and insulin lispro[1].

However, despite destabilisation of the dimer interface insulin lispro still forms Zn2+ insulin hexamers in the presence of the phenolic excipients present in commercial pharmaceutical formulations (c). This provides the necessary stability against fibrillation during storage in the vial but absorption is still not as rapid as could be provided by a zinc-free formulation. Additionally lispro and other meal-time insulin analogs have reduced shelf life upon dilution by the patient or health-care provider.

There is a great need for an insulin analog which augments the stability of the insulin monomer while retaining the weakened dimer-related β-sheet of lispro.

Halogen stabilization in medicinal chemistry

Halogen atoms have long been used for compound optimization in medicinal chemistry (d). The utility of halogen substitutions in amino acids is also well established in medicinal chemistry. Of all the halogens, the effect of fluorine (with an atomic radius of 42 pm similar to that of hydrogen at 53 pm) incorporation on the physical and chemical properties of proteins is the most characterized in the scientific literature (e). Such observations have motivated the study of fluorinated amino acids for the structural stabilization of proteins, with the provisio that in the case of a biologically active polypeptide at least a significant proportion of the activity must also be maintained.

Thermalin Fluorolog

Fluorolog (f) is a rapid-acting, ultra-concentrated insulin that is in pre-clinical development by Thermalin Diabetes (Cleveland, Ohio). In this lispro analog B24 Phe has been substituted with ortho-monofluorophenylalanine (2F-Phe)(fig. 5).

Raj 6

Figure 5. (a) A vial of rapid-acting, ultraconcentrated U-500 Fluorolog. (b) Ortho-monofluoro-phenylalanine (2F-Phe). (c) The disposition of 2F-Phe against the body of the Fluorolog molecule [5].

The large inductive effects of the fluorine atom act to thermodynamically stabilize the entire molecule yielding an analog that no longer needs to form a hexamer to be stable. Fluorolog is not prone to clumping and can be formulated at high concentrations (500 units/mL) without the risk of fibrillation and targets three market niches [5]:

1) Fibrillation of insulin is enhanced by agitation and can result in blockage of insulin pumps used by some type 1 diabetics. In miniaturized insulin pumps, pump size is currently constrained by the size of the insulin reservoir which if shrunk by 80% could yield a pump that lasts a whole week. Insulin fibrillation is of even greater concern in implantable insulin pumps, where the insulin may be contained for up to 3 months at high concentration and at physiological temperature.

2) In developed countries some highly insulin resistant type 2 diabetics – often members of underprivileged minority communities- currently inject several hundred units of insulin each day (g). Injecting up to 0.5 mL of the standard lispro 100 units/mL (U-100) formulation is not only uncomfortable but also delays mealtime insulin absorption. Highly concentrated Fluorolog will allow these diabetics to inject smaller quantities at one site and closer to meals.

3) Because fibrillation is also enhanced at higher temperatures, insulin must optimally be kept refrigerated prior to use. For both type 1 and type 2 diabetics in underdeveloped parts of Africa and the Middle East there is a great need for an insulin that does not degrade without refrigeration. Fluorolog is stable for up to 3 months in warm climates without refrigeration.(h)

Phase 1 & 2 trials

Phase I studies have confirmed that the introduction of a single fluorine atom at the receptor-binding surface of Fluorolog stabilizes the monomer and protects it from degradation. Fluorolog exhibited the expected rapid-acting pharmacokinetic properties (even at U-500 formulation) in contrast to the impaired pharmacokinetic properties of native insulin. Furthermore, fluorination of B24Phe was able to mitigate the untoward effects of AspB10 in the DKP homolog on cellular proliferation in culture and on cross-binding to the IGF-1 receptor.

For its phase 2 trial Thermalin Diabetes is seeking to extend the pilot stability data to include individual aspects of chemical and physical degradation (such as disulphide cleavage, covalent polymer formation, and fibrillation) in preparation for an Investigational New Drug (IND) Application


a. In single-chain insulin (SCI: the recombinant precursor expressed in yeast) B29Lys and A1Gly are linked by a peptide bond. SCI crystallizes with the same fold as native insulin but is inactive.

b. In the early days insulin could only be crystallized if acid-ethanol extracted from pancreatic tissue in a galvanized bucket.

c.  DKP- insulin with the additional HisB10→Asp substitution is monomeric under a wide range of conditions. Although this insulin-like growth factor-1 equivalent substitution eliminates zinc binding it does raise the possibility of mitogenicity.

d. The activities of the widely prescribed statin atorvastatin (Liptor) and the SSRI antidepressant fluoxetine hydrochloride (Prozac) are both enhanced by the covalent incorporation of a fluorine atom.

e. Fluorine is distinguished from the normal C,H,O,N and S constituents of proteins by its atomic radius, electronegativity, stereo-electronic distribution of partial charges, and transmitted effects on the stereo-electronic properties of neighboring atoms.

f.  US Patent 20140128319 A1 (2008).

g.  As pre-filled insulin pens can only inject 60–80 units of insulin at a time, some highly insulin-resistant diabetics must inject up to 9 times a day or more.

h. Ideal for airmail.

Blog written by Raj Gill.


  1. Gill, R. & Wood, S. (2003) Structure and Phylogeny of Insulin. Chapter 12 in International Textbook of Diabetes Vol. 1, ed. Pickup, J.C. & Williams, G., Blackwell Scientific, Oxford.
  2.  Ivanova, M.I., Sievers, S.A., Sawaya, M.R., Wall, J.S. & Eisenberg, D. (2009) Molecular Basis for Insulin Fibril Assembly. Proc Natl Acad Sci USA 106, 18990–18995.
  3.  Landreh, M., Johansson, J., Rising, A., Presto, J. &Jörnvall, H. (2012) Control of Amyloid Assembly by Autoregulation. Biochemical J. 447, 185-192.
  4. Dyukov, M.I., Grudinin, M.P., Sirotkin, A.K., & Kiselev, O.I. (2008) Insulin Fibrillogenesis In vitro. Doklady Biochem. Biophys. 419, 79-81.
  5. http://www.thermalin.com/lead-product—fluorologtrade.hl



An update on the progress of PARP inhibitors in the clinic

An article published at the beginning of 2016 titled ‘ PARP Inhibitors: The race is on’, describes the race to exploit single agent PARP inhibitors for the treatment of cancer by exploiting the concept of synthetic lethality to selectively target cancer cells deficient in the repair of DNA double strand breaks by the homologous recombination (HR) pathway1. The most studied defects in HR are associated with inactivating mutations in the proteins BRCA1 and BRCA2, which have essential roles in the pathway. Inhibition of PARP during the repair of damaged DNA bases by the base excision repair (BER) pathway causes single strand breaks in DNA that later become double strand breaks during DNA replication. The breaks are repaired by the error prone non-homologous end joining (NHEJ) pathway in cells deficient in homologous recombination (HR) repair mechanisms. The high burden of mutations in the DNA of rapidly proliferating cells caused by error prone NHEJ leads to cell death. Therefore, in accordance with the synthetic lethality concept, PARP inhibition in tumours that have deleterious mutations in BRAC1/2 should lead to cancer cell death and tumour shrinkage.2 Olaparib is the first example of a PARP inhibitor that is well tolerated in patients and clinically validates the synthetic lethality concept 3,4. The drug has been approved as a monotherapy for the treatment of ovarian cancer patients with germline BRCA mutations. Its success has encouraged the development of other PARP inhibitor programmes from Clovis, Tesaro, AbbVie and Medivation. So, as we end the first quarter of 2017, who is leading the race?

Clovis took an early lead with Rucaparib after submitting a new drug application to the FDA in June 2016 for patients with advanced ovarian cancer that have been treated with two or more chemotherapies and have either somatic or germline BRCA1/2 mutations. An efficacy study measuring the percentage of patients to experience complete or partial shrinkage of their tumors (progression free survival) was completed in a cohort of 106 patients with Rucaprib dosed twice per day (600mg). The company posted favorable comparative data showing a high response rate (54% vs 34% for Olaparib) with a marginally longer duration of response at 9.7 months compared to Olaparib’s 7.2 months. In December 2016 the FDA granted accelerated approval for Rucaparib. Furthermore, unlike Olaparib that is restricted to germline BRCA mutations, Rucaparib was approved for germline and/or somatic mutations in ovarian tumours.

In October 2016 Tesaro presented data on Niraparib at ESMO from a Phase III randomized clinical trial in patients with platinum responsive ovarian cancer. The study assessed progression free survival (PFS) in a cohort of 533 patients with or without BRCA mutations.5 Consistent with previous studies on PARP inhibitors, patients with germline BRCA1/2 mutations showed the greatest PFS at 21 months compared to placebo (5 months). Crucially, a statistically significant effect was demonstrated in non-germline BRCA mutant cancers with a PFS of 9.3 months vs 3 months for placebo. In December the FDA granted Niraparib priority review status and last week gave approval for the maintenance treatment for recurrent ovarian cancer in patients who are in complete or partial response to platinum based chemotherapy. Furthermore, the approval is not restricted to ovarian cancer patients with BRCA mutations, making the drug available to a larger population of ovarian cancer patients than either Rucaparib or Olaparib.

While PARP inhibitors have been tested in several types of solid tumour, the greatest response has been observed in ovarian cancer 6. AbbVie’s approach with Veliparib has focused on demonstrating sensitisation of classical chemotherapy drugs to combinations with PARP inhibitors in the more prevalent non-small cell lung cancer and breast cancer patient groups. The use of platinum based agents like carboplatin cause apoptosis of cancer cells by cross linking DNA, leading to tumour cell shrinkage 7. However, the DNA damage response can limit the effect of these agents. The resulting lesions are recognized by the DNA damage sensors and repaired by the base or nucleotide excision repair pathways. Inhibiting the role of PARP in these DNA repair pathways leads to the formation of single strand breaks, which are synthetically lethal in HR deficient cells. Studies in preclinical mouse models of breast cancer demonstrated a pronounced potentiation of cisplatin efficacy in combination with Veliparib 8. While PARP inhibitors are well tolerated, combination therapies risk narrowing the therapeutic window for both Veliparib and the chemotherapeutic agent. At the San Antonio Breast Cancer Symposium in December last year, AbbVie presented results from its phase II study of Veliparib in patients with locally recurrent or metastatic breast cancer with BRCA1 or BRCA2 mutations. While the addition of Veliparib to paclitaxel and carboplatin resulted in an improved overall response rate without an increase in adverse events, the study failed to show statistical significance in progression free survival. Despite the disappointing PFS data, the higher responder rate has encouraged continuation with the phase III trial (BROCADE) appropriately powered to detect improvements in PFS and overall survival. In addition to trails in Breast Cancer, AbbVie was given orphan drug status for Veliparib in NSCLC.

Of all the PARP inhibitors in late stage clinical trials, the most potent and selective is Talazaporib, recently acquired by Pfizer after its take-over of Medivation. While PARP inhibitors have been developed to potently inhibit the enzymatic activity of PARP, recent evidence suggests that the ability of the compounds to trap PARP at the DNA single strand break closely correlates with efficacy in preclinical models 9. Talazaporib has superior trapping ability than any other PARP inhibitor on the market, and with the promise of superior efficacy at a lower dose with fewer side effects, there is significant interest in the ongoing clinical studies. The drug is in a phase III clinical study for advanced and/or metastatic breast cancer with a BRCA mutation. Results from the study are keenly awaited.

PARP inhibitors are widely accepted as a clinically proven target class that validates the synthetic lethality concept. The future growth of PARP inhibitors will depend on the ability to combine it with other targeted approaches that can demonstrate synthetic lethality in tumour cell backgrounds while sparing normal cells. The good tolerability of PARP inhibitors in combination with other chemotherapy bodes well. Studies are ongoing for PARP inhibitors in different cancer populations in combination with radiation, Wee1 inhibitors, ATR inhibitors, Hsp90 inhibitors and drugs targeting the PI3K/Akt pathway. Studies may complete in 2017/18 and the results are eagerly awaited.

Blog written by Darren Le Grand


  1. Brown, J. S., Kaye, S. B. & Yap, T. A. PARP inhibitors: the race is on. Br. J. Cancer 114, 713–715 (2016).
  2. Farmer, H. et al. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434, 917–21 (2005).
  3. Ledermann, J. et al. Olaparib Maintenance Therapy in Platinum-Sensitive Relapsed Ovarian Cancer. N. Engl. J. Med. 366, 1382–1392 (2012).
  4. Ledermann, J. A. et al. Articles Overall survival in patients with platinum-sensitive recurrent serous ovarian cancer receiving olaparib maintenance monotherapy: an updated analysis from a randomised, placebo-controlled, double-blind, phase 2 trial. Lancet Oncol (2016). doi:10.1016/S1470-2045(16)30376-X
  5. Mirza, M. R. et al. Niraparib Maintenance Therapy in Platinum-Sensitive, Recurrent Ovarian Cancer. N. Engl. J. Med. 375, 2154–2164 (2016).
  6. Ledermann, J. A. PARP inhibitors in ovarian cancer. Ann. Oncol. 27, i40–i44 (2016).
  7. Kelland, L. The resurgence of platinum-based cancer chemotherapy. Nat. Rev. Cancer 7, 573–584 (2007).
  8. Donawho, C. K. et al. ABT-888, an Orally Active Poly(ADP-Ribose) Polymerase Inhibitor that Potentiates DNA-Damaging Agents in Preclinical Tumor Models. Clin. Cancer Res. 13, (2007).
  9. Murai, J. et al. Trapping of PARP1 and PARP2 by Clinical PARP Inhibitors. Cancer Res. 72, 5588–99 (2012).


Crystallisation protein construct for Protein-Ligand Crystal complexes

The functional clues about various cellular processes in living organisms relies on the interaction of biological macromolecules, especially proteins with small molecules (ligands). A deeper understanding of these proteins – ligand interactions at molecular level is essential for answering the functional biological questions and key concepts required for structure based drug design.  The high-resolution structural data for protein-ligand complexes are generally obtained by either X-ray crystallography or nuclear magnetic resonance spectroscopy. These structural knowledge elucidate the molecular architecture for the protein of interest as well as convey experimental evidence of binding mode how the small molecule binds, which is of extremely beneficial during the design of lead molecules for the treatment of diseases.

The availability of the suitable crystallization construct for the protein of interest is the basis for generation of protein-ligand complex crystals for X-ray structural studies. In this blog, few of the key points focusing this area is mentioned. A recent review by (Muller, 2017) illustrates a flowchart (Fig.1) to achieve this key step.

Mohan 1


Fig 1: Adapted from Muller, 2017 Acta Crystallogr D Struct Biol D73, 79-92

The first step of protein construct design is of identifying a well-ordered structural domain for the protein of interest, avoiding the unstructured regions by carrying out detailed structure based sequence alignments using various bioinformatics domain prediction tools such as Pfam (Finn et al., 2016), pDomTHREADER (Lewis et al., 2013), DISOPRED (Ward et al., 2004) RONN (Yang et al., 2005) etc. An insilico-based validation of the initially designed protein domain for its ability to crystallise could also be analysed by XtalPRed-RF tool (Slabinski et al., 2007) is also a good choice.

Once the core domains are identified, the regions of disorder either between the domains or within the domains could be replaced with equivalent less flexible residue of the homologue proteins or by introduction of short flexible linkers which could also aid in the crystal formation. Few successful stories from the literature are listed here (Clifton et al., 2015, Ocasio et al., 2016, Blair et al., 2010, Muller, 2017).

It is also worth looking at the chemical modifications occurring for the protein of interest, which could also hinder the crystallisation process. For kinases production studies, several groups (Cowan-Jacob et al., 2007, Mace et al., 2013) were successful in obtaining higher yields of proteins of interest by altering the phosphorylation profiles with kinase inhibitors or removing the phosphorylation sites or phospho-mimetics.

The exploration of various affinity and solubility tags (Pina et al., 2014, Costa et al., 2014) to enhance the folding and solubility for the protein of interest to aid in protein purification could also speed up the whole crystallisation process

Not in all cases, but the presence of large flexible amino acids such as Lys, Glu, Gln on the protein surface could sometimes hinder the protein crystallisation. Surface entropy reduction (SER) provides solution to this by designing various mutant protein constructs replacing these flexible amino acids to smaller amino acids such as alanine found to be effective. On in-silico basis, SERp server (Goldschmidt et al., 2007) helps in identifying the amino acids for surface mutation to enhance the crystallisation probability.

Once the protein construct for crystallisation is designed based on the above parameters, the next step is to successfully prepare a homogenous preparation of pure protein for structural studies. There are many expression systems such as E. coli (Rosano & Ceccarelli, 2014), Yeast, Baculovirus (Unger & Peleg, 2012) or Mammalian for the production of protein of interest (Structural Genomics et al., 2008, Muller, 2017). The selection of resource efficient and suitable expression system for each protein of interest varies and it involved various experimental optimizations to be carried out.

In some cases, the addition of ligands during protein expression and/or purification (Muller, 2017) as well as thermal stability assays were also worth trying out during the production of proteins on large scale.

When the pure protein is available, the next task of generating protein-ligand crystal complexes can be achieved by soaking or co-crystallisation techniques.

The main requisite for soaking is the availability of good quality apo crystals for the protein of interest. As a general guide, 10-50 mM for fragments and 0.1 – 1mM for high molecular weight compounds could be used if Kd is not known (Muller, 2017), but these parameters varies a lot. The treatment of soaking crystals with stabilisation buffer, cross-linking with glutaraldehyde (Lusty, 1999), step-wise increase of ligand concentration during soaking, soaking incubation timings are some of the optimisations normally carried out for an efficient soaking (Hassell et al., 2007). A high-throughput automated technique for carrying out soaking experiments is also getting popular (Collins et al., 2017)

The process of co-crystallisation involves the pre-incubation of protein with the ligand of interest to make the protein-ligand complex before the crystallisation set up (Hassell et al., 2007, Muller, 2017). As a general rule of thumb, the compound concentration should be three times the Kd value (Muller, 2017). Optimisations involving varying the drop ratios (1:1,1:2, 2:1 for protein:well solution) (Ng et al., 2016), ‘dry’ co-crystallisation (Gelin et al., 2015) in which crystallisation wells are pre-coated with ligands before the crystallisation set up could speed up the entire process.

In general, the whole process of generating the protein–ligand complex crystals consists of various stages such as designing protein construct, protein production on large scale followed by crystallisation experiments with co-crystallisation and/or soaking and each stage involves trial and error methods for the better outcome.

Blog by Mohan Rajasekaran


Blair, W. S., Pickford, C., Irving, S. L., Brown, D. G., Anderson, M., Bazin, R., Cao, J., Ciaramella, G., Isaacson, J., Jackson, L., Hunt, R., Kjerrstrom, A., Nieman, J. A., Patick, A. K., Perros, M., Scott, A. D., Whitby, K., Wu, H. & Butler, S. L. (2010). PLoS Pathog 6, e1001220.

Clifton, M. C., Dranow, D. M., Leed, A., Fulroth, B., Fairman, J. W., Abendroth, J., Atkins, K. A., Wallace, E., Fan, D., Xu, G., Ni, Z. J., Daniels, D., Van Drie, J., Wei, G., Burgin, A. B., Golub, T. R., Hubbard, B. K. & Serrano-Wu, M. H. (2015). PLoS One 10, e0125010.

Collins, P. M., Ng, J. T., Talon, R., Nekrosiute, K., Krojer, T., Douangamath, A., Brandao-Neto, J., Wright, N., Pearce, N. M. & von Delft, F. (2017). Acta Crystallogr D Struct Biol 73, 246-255.

Costa, S., Almeida, A., Castro, A. & Domingues, L. (2014). Front Microbiol 5, 63.

Cowan-Jacob, S. W., Fendrich, G., Floersheimer, A., Furet, P., Liebetanz, J., Rummel, G., Rheinberger, P., Centeleghe, M., Fabbro, D. & Manley, P. W. (2007). Acta Crystallogr D Biol Crystallogr 63, 80-93.

Finn, R. D., Coggill, P., Eberhardt, R. Y., Eddy, S. R., Mistry, J., Mitchell, A. L., Potter, S. C., Punta, M., Qureshi, M., Sangrador-Vegas, A., Salazar, G. A., Tate, J. & Bateman, A. (2016). Nucleic Acids Res 44, D279-285.

Gelin, M., Delfosse, V., Allemand, F., Hoh, F., Sallaz-Damaz, Y., Pirocchi, M., Bourguet, W., Ferrer, J. L., Labesse, G. & Guichou, J. F. (2015). Acta Crystallogr D Biol Crystallogr 71, 1777-1787.

Goldschmidt, L., Cooper, D. R., Derewenda, Z. S. & Eisenberg, D. (2007). Protein Sci 16, 1569-1576.

Hassell, A. M., An, G., Bledsoe, R. K., Bynum, J. M., Carter, H. L., 3rd, Deng, S. J., Gampe, R. T., Grisard, T. E., Madauss, K. P., Nolte, R. T., Rocque, W. J., Wang, L., Weaver, K. L., Williams, S. P., Wisely, G. B., Xu, R. & Shewchuk, L. M. (2007). Acta Crystallogr D Biol Crystallogr 63, 72-79.

Lewis, T. E., Sillitoe, I., Andreeva, A., Blundell, T. L., Buchan, D. W., Chothia, C., Cuff, A., Dana, J. M., Filippis, I., Gough, J., Hunter, S., Jones, D. T., Kelley, L. A., Kleywegt, G. J., Minneci, F., Mitchell, A., Murzin, A. G., Ochoa-Montano, B., Rackham, O. J., Smith, J., Sternberg, M. J., Velankar, S., Yeats, C. & Orengo, C. (2013). Nucleic Acids Res 41, D499-507.

Lusty, C. J. (1999). Journal of Applied Crystallography 32, 106-112.

Mace, P. D., Wallez, Y., Egger, M. F., Dobaczewska, M. K., Robinson, H., Pasquale, E. B. & Riedl, S. J. (2013). Nat Commun 4, 1681.

Muller, I. (2017). Acta Crystallogr D Struct Biol 73, 79-92.

Ng, J. T., Dekker, C., Reardon, P. & von Delft, F. (2016). Acta Crystallogr D Struct Biol 72, 224-235.

Ocasio, C. A., Rajasekaran, M. B., Walker, S., Le Grand, D., Spencer, J., Pearl, F. M., Ward, S. E., Savic, V., Pearl, L. H., Hochegger, H. & Oliver, A. W. (2016). Oncotarget 7, 71182-71197.

Pina, A. S., Lowe, C. R. & Roque, A. C. (2014). Biotechnol Adv 32, 366-381.

Rosano, G. L. & Ceccarelli, E. A. (2014). Front Microbiol 5, 172.

Slabinski, L., Jaroszewski, L., Rychlewski, L., Wilson, I. A., Lesley, S. A. & Godzik, A. (2007). Bioinformatics 23, 3403-3405.

Structural Genomics, C., China Structural Genomics, C., Northeast Structural Genomics, C., Graslund, S., Nordlund, P., Weigelt, J., Hallberg, B. M., Bray, J., Gileadi, O., Knapp, S., Oppermann, U., Arrowsmith, C., Hui, R., Ming, J., dhe-Paganon, S., Park, H. W., Savchenko, A., Yee, A., Edwards, A., Vincentelli, R., Cambillau, C., Kim, R., Kim, S. H., Rao, Z., Shi, Y., Terwilliger, T. C., Kim, C. Y., Hung, L. W., Waldo, G. S., Peleg, Y., Albeck, S., Unger, T., Dym, O., Prilusky, J., Sussman, J. L., Stevens, R. C., Lesley, S. A., Wilson, I. A., Joachimiak, A., Collart, F., Dementieva, I., Donnelly, M. I., Eschenfeldt, W. H., Kim, Y., Stols, L., Wu, R., Zhou, M., Burley, S. K., Emtage, J. S., Sauder, J. M., Thompson, D., Bain, K., Luz, J., Gheyi, T., Zhang, F., Atwell, S., Almo, S. C., Bonanno, J. B., Fiser, A., Swaminathan, S., Studier, F. W., Chance, M. R., Sali, A., Acton, T. B., Xiao, R., Zhao, L., Ma, L. C., Hunt, J. F., Tong, L., Cunningham, K., Inouye, M., Anderson, S., Janjua, H., Shastry, R., Ho, C. K., Wang, D., Wang, H., Jiang, M., Montelione, G. T., Stuart, D. I., Owens, R. J., Daenke, S., Schutz, A., Heinemann, U., Yokoyama, S., Bussow, K. & Gunsalus, K. C. (2008). Nat Methods 5, 135-146.

Unger, T. & Peleg, Y. (2012). Methods Mol Biol 800, 187-199.

Ward, J. J., McGuffin, L. J., Bryson, K., Buxton, B. F. & Jones, D. T. (2004). Bioinformatics 20, 2138-2139.

Yang, Z. R., Thomson, R., McNeil, P. & Esnouf, R. M. (2005). Bioinformatics 21, 3369-3376.