Structure-based discovery of first inhibitors targeting the helicase activity of human PIF1

Abstract

PIF1 is a conserved helicase and G4 DNA binding and unwinding enzyme, with roles in genome stability. Human PIF1 (hPIF1) is poorly understood, but its functions can become critical for tumour cell survival during oncogene-driven replication stress. Here we report the discovery, via an X-ray crystallographic fragment screen (XChem), of hPIF1 DNA binding and unwinding inhibitors. A structure was obtained with a 4-phenylthiazol-2-amine fragment bound in a pocket between helicase domains 2A and 2B, with additional contacts to Valine 258 from domain 1A. The compound makes specific interactions, notably through Leucine 548 and Alanine 551, that constrain conformational adjustments between domains 2A and 2B, previously linked to ATP hydrolysis and DNA unwinding. We next synthesized a range of related compounds and characterized their effects on hPIF1 DNA-binding and helicase activity in vitro, expanding the structure activity relationship (SAR) around the initial hit. A systematic analysis of clinical cancer databases is also presented here, supporting the notion that hPIF1 upregulation may represent a specific cancer cell vulnerability. The research demonstrates that hPIF1 is a tractable target through 4-phenylthiazol-2-amine derivatives as inhibitors of its helicase action, setting a foundation for creation of a novel class of anti-cancer therapeutics.

Graphical Abstract

Figure 1.

Overview of the 4-phenylthiazol-2-amine (compound 1) binding site in human PIF1. (A) A largely ‘cartoon’ overview of the structure of hPIF1HD complex with AMP-PNP and compound 1 bound (shown as ‘spheres’). The compound binding site is mostly between domains 2A and 2B, with Val 258 from domain 1A at one end of the binding site. Domain 2B, which is defined as starting in the middle of helix α11 is shown in green for the SH3-like part and orange for additional elements, including the ‘rail β-hairpin’ (β15–β16). (B) A similar view to A with a semi-transparent surface on the protein, showing the channel down which compound 1 likely entered the ‘binding pocket’ in the crystal. Note, hydrogen bonds to main-chain carbonyls of Leu 548 and Ala 551 and an indirect (via two waters, small red spheres) hydrogen bond with Asn 436, from the middle of helix α11. A probable sodium ion required for compound binding is also indicated. (C) The amino acid sequence of hPIF1HD coloured by domains. The ‘Domain’ line has domain names; amino acids with * underneath are in domain linkers. Note that domain 2B includes a region (in green) with an SH3-like domain fold (five β-strands). Amino acids within 3.8 Å of the compound are highlighted in yellow on the Hu_Pif1 sequence. Residues in α-helices or β-strands are underlined (some residues in 3/10 helices are also underlined). The ‘Sec str/10.’ line shows secondary structure names for α-helices, β-strands and 3/10 helices (η1 etc) and a. for every 10th residue in the sequence. Note, the hPIF1HD construct starts at residue Gln 206 and ends at Leu 620 and is preceded by four residues from the cloning tag (GSRM). The C-terminal residues in lower case and black are not in the structure or the construct.

Figure 2.

The compound 1 binding site is not present in a modelled hPIF1-ssDNA complex. (A, B) Focused and overview of the structure of the hPIF1HD/AMP-PNP/1 complex. The view is similar to that in Figure 1B. Note Leu 548 (light green), the last residue in domain 2B, is separated by two linker residues (449 and 550—grey) before Ala 551 (magenta—see sequence in Figure 1C). (C, D) Focused and overview of a model of a complex of hPIF1HD with ssDNA bound. Note the relatively large movement of Domain 2B which accompanies ssDNA binding and disrupts the compound 1 binding pocket. (E, F) Superposition of the hPIF1HD/AMP-PNP/1 structure and the modeled hPIF1HD-ssDNA structure, focusing on the compound 1 binding pocket in (E). Note the large relative shift in domain 2B. Note the change in orientation of the α11 helix at the N-terminus of domain 2B, as well as the change in relative positions of carbonyls of Leu 548 and Ala 551 at the C-terminus of domain 2B.

Figure 3.

Inhibitory activity of compound 1A and scheme for chemistry. (A) Activity of 1A determined by DNA strand displacement assay (10 nM FLhPIF1, 0.1 nM DNA substrate). An IC₅₀ 281 ± 336 μM determined, n = 3 repeat. Boil is the thermally denatured substrate and the ‘No PIF1’ lane shows native substrate. (B) An IC₅₀ of ∼1159 μM was determined for 1F, re-synthesized 1 (see also Supplementary Figure S7). 37 derivatives were synthesized in the phenyl ring (R¹ and R²) and 2-aminothiazole moiety (R³).

Figure 4.

Inhibition of hPIF1HD ssDNA binding by compound 1 and derivatives. Binding reactions were assembled under helicase reaction conditions but with 0.25 nM radiolabelled oligo d(T)₅₅, 2 nM hPIF1HD, 5 mM AMP-PNP and 0.0078–2 mM SMI. Products were resolved by EMSA for quantification and IC₅₀ determination (n = 3 repeats). Bound ssDNA was calculated as the sum of the major and minor species, interpreted as one or two bound hPIF1HD molecules respectively. See also Supplementary Figure S10. IC₅₀ values are given along with those determined for hPIF1HD variants V258A and V258L (see Supplementary Figure S12).

Figure 5.

Pre-incubation time course of FLhPIF1 with SMIs before helicase assay. 100 nM FLhPIF1 was pre-incubated (0–3 h) with SMIs (100 μM) before addition to the helicase reaction. Top, gel image, products of unwinding assays resolved by electrophoresis. SMIs added, lanes 1–6, and 7, DMSO, are grouped according to time of pre-incubation. Below, summary of experimental scheme (left), key to lanes (centre) and quantified data (n = 4), right.

Figure 6.

SMI IC₅₀ for dsDNA unwinding determined following pre-incubation. SMIs (0.0078–2 mM SMI) were incubated for 2 h at 22°C (100 nM FLhPIF1 or hUPF1) before dilution (1:10) in the standard helicase assay (0.1 nM DNA substrate). For 1D and 6, structures, electrophoresis of FLhPIF1 reaction product and graphs for IC₅₀ value determination are shown. The graphed data include the activity of SMIs towards hUPF1 and IC₅₀ values obtained are given. See Supplementary Figure S14 for the complete data set.

Figure 7.

hPIF1 in clinical datasets.(A) Box plot of PIF1 mRNA expression (log₂ transformed expression data in transcripts per million (TPM)) across 33 cancer entities between tumour (T, red), metastatic (M, purple) and adjacent normal (N, blue) tissue, where available. The Wilcoxon rank sum test was used to compare the expression in tumour and adjacent healthy tissues when data was available, p values as indicated (**P < 0.01; ***P < 0.001). n indicates the number of samples available for each type. (B–H) Overall survival fraction over time (Kaplan–Meier analysis) comparing PIF1 low and high expression groups. (B) adrenocortical carcinoma (ACC, n = 64), (C) cholangiocarcinoma (CHOL, n = 36), (D) kidney renal clear cell carcinoma (KIRC, n = 450), (E) brain lower grade glioma (LGG, n = 463), (F) mesothelioma (MESO, n = 85), (G) primary skin cutaneous melanoma (SKCM, n = 94) and (H) uveal melanoma (UVM, n = 77). Log rank test was used to calculate hazard ratios (HR) between low and high PIF1 expression groups, P < 0.05 was considered significant. (I) Percentage of samples with PIF1 mutation in the different cancer types. Transcriptomic, genomic and clinical data were obtained from The Cancer Genome Atlas (TCGA) database. The abbreviation for the cancer types in A and I are: ACC, adrenocortical carcinoma; BLCA, bladder urothelial carcinoma; BRCA, breast invasive carcinoma; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; CHLO, cholangiocarcinoma; COAD, colon adenocarcinoma; DLBC, lymphoid neoplasm diffuse large B-cell lymphoma; ESCA, esophageal carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; KICH, kidney chromophobe; KIRC, kidney renal clear cell carcinoma; KIRP, kidney renal papillary cell carcinoma; LAML, acute myeloid leukaemia; LGG, brain lower grade glioma; LIHC, liver hepatocellular carcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; MESO, mesothelioma; OV, ovarian serous cystadenocarcinoma; PAAD, pancreatic adenocarcinoma; PCPG, pheochromocytoma and paraganglioma; PRAD, prostate adenocarcinoma; READ, rectum adenocarcinoma; SARC, sarcoma; SKCM, skin cutaneous melanoma; STAD, stomach adenocarcinoma; TGCT, testicular germ cell tumours; THAC, thyroid carcinoma; THTM, thymoma; UCEC, uterine corpus endometrial carcinoma; UCS, uterine carcinosarcoma; UVM, uveal melanoma.