Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Biochemical properties of naturally occurring human bloom helicase variants

  • Rachel R. Cueny,

    Roles Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing

    Affiliation Department of Biomolecular Chemistry, University of Wisconsin, Madison, WI, United States of America

  • Sameer Varma,

    Roles Conceptualization, Formal analysis, Investigation

    Affiliations Department of Cell Biology, Microbiology, and Molecular Biology, University of South Florida, Tampa, FL, United States of America, Department of Physics, University of South Florida, Tampa, FL, United States of America

  • Kristina H. Schmidt ,

    Roles Conceptualization, Investigation, Writing – original draft, Writing – review & editing

    kschmidt@usf.edu (KHS); jlkeck@wisc.edu (JLK)

    Affiliation Department of Cell Biology, Microbiology, and Molecular Biology, University of South Florida, Tampa, FL, United States of America

  • James L. Keck

    Roles Conceptualization, Funding acquisition, Investigation, Resources, Supervision, Writing – original draft, Writing – review & editing

    kschmidt@usf.edu (KHS); jlkeck@wisc.edu (JLK)

    Affiliation Department of Biomolecular Chemistry, University of Wisconsin, Madison, WI, United States of America

Abstract

Bloom syndrome helicase (BLM) is a RecQ-family helicase implicated in a variety of cellular processes, including DNA replication, DNA repair, and telomere maintenance. Mutations in human BLM cause Bloom syndrome (BS), an autosomal recessive disorder that leads to myriad negative health impacts including a predisposition to cancer. BS-causing mutations in BLM often negatively impact BLM ATPase and helicase activity. While BLM mutations that cause BS have been well characterized both in vitro and in vivo, there are other less studied BLM mutations that exist in the human population that do not lead to BS. Two of these non-BS mutations, encoding BLM P868L and BLM G1120R, when homozygous, increase sister chromatid exchanges in human cells. To characterize these naturally occurring BLM mutant proteins in vitro, we purified the BLM catalytic core (BLMcore, residues 636–1298) with either the P868L or G1120R substitution. We also purified a BLMcore K869A K870A mutant protein, which alters a lysine-rich loop proximal to the P868 residue. We found that BLMcore P868L and G1120R proteins were both able to hydrolyze ATP, bind diverse DNA substrates, and unwind G-quadruplex and duplex DNA structures. Molecular dynamics simulations suggest that the P868L substitution weakens the DNA interaction with the winged-helix domain of BLM and alters the orientation of one lobe of the ATPase domain. Because BLMcore P868L and G1120R retain helicase function in vitro, it is likely that the increased genome instability is caused by specific impacts of the mutant proteins in vivo. Interestingly, we found that BLMcore K869A K870A has diminished ATPase activity, weakened binding to duplex DNA structures, and less robust helicase activity compared to wild-type BLMcore. Thus, the lysine-rich loop may have an important role in ATPase activity and specific binding and DNA unwinding functions in BLM.

Introduction

Bloom syndrome (BS) is an autosomal recessive disorder that is characterized by shorter than average stature, sensitivity to sunlight, weakened immune response, and a predisposition to cancer [1]. Patients with BS are more likely to encounter age-associated complications early in life, including adult-onset diabetes, various cancers, and chronic obstructive lung disease [13]. BS can be diagnosed by assessing the level of damage and instability in chromosomes. BS is characterized by elevated chromosomal breakage, increased sister chromatid exchanges (SCEs), and quadriradial chromosomal structures [4]. Previous research has found that BS-causing mutations lead to a 10–12 fold increase in SCEs in human lymphocytes, and similar phenotypes have been demonstrated in DT40 lymphoma cells [5, 6].

BS is caused by mutations to BLM, which encodes the Bloom Syndrome DNA helicase (BLM) [4]. BLM is a member of the RecQ family of helicases and can unwind a variety of DNA secondary structures with 3′ to 5′ polarity, including double-stranded DNA (dsDNA), forked DNA substrates, G-quadruplex DNA (G4s), and Holliday junctions (HJs) [3, 7]. There are five human RecQ-family helicases, which include BLM, WRN, RECQL1, RECQL4, and RECQL5. Mutations in four of these RecQ helicases, including BLM, are associated with autosomal recessive disorders [8, 9]. As such, RecQ-family helicases are often defined as “caretakers” of the genome [10].

BLM is essential for promoting genome stability in the cell and has functions in DNA replication, DNA repair, transcription, and telomere maintenance [3]. In keeping with its role as a genome maintenance protein, BLM localizes to DNA replication and DNA damage sites, and to telomeres [3]. Further, BLM interacts with a large network of DNA replication and repair proteins and is thought to act in concert with a larger complex of proteins for many of its core functions [3, 11]. For example, BLM is part of the dissolvasome, which is composed of BLM, topoisomerase IIIα (Top3α), RecQ-mediated genome instability protein 1 (RMI1), and RMI2 [11]. The dissolvasome is involved in dissolution of double Holliday junctions (dHJs) generated during homologous recombination. This function of the dissolvasome prevents crossover during recombination, suppressing SCEs [12, 13]. Thus, defects in BLM cause many adverse effects on chromosome stability due to the multifaceted nature of BLM.

The BLM catalytic core includes a Helicase domain, a RecQ C-terminal domain (RQC), and a Helicase and RNase D C-terminal domain (HRDC) (Fig 1A) [14, 15]. The Helicase domain of BLM is composed of two RecA-like subdomains, which are important for ATP hydrolysis and helicase activity. The RQC domain is composed of Zinc-binding (ZN) and Winged Helix (WH) subdomains. The ZN subdomain includes four cysteine residues that coordinate a Zn2+ ion, and the WH is important for interactions with various DNA structures. The HRDC is likely involved in binding single-stranded DNA (ssDNA) and increasing BLM processivity [7]. The N-terminal and C-terminal regions of BLM are highly disordered and expected to be important for protein-protein interactions [7, 16].

thumbnail
Fig 1. Structure of BLM.

(A). Structure of BLM catalytic core (PDBID: 4O3M [14]), colored. according to domain map below structure. ADP is shown as sticks and magnesium and zinc ions are shown as green and grays spheres, respectively. Positions of substitutions P868L and G1120R are highlighted on domain map. (B). P868, K869, and K870 residues, shown as sticks. (C). G1120 residue, shown as sticks.

https://doi.org/10.1371/journal.pone.0281524.g001

Mutations in BLM that lead to BS include missense mutations, nonsense mutations, frameshifts, and splicing defects [4]. Many of the missense mutations interfere with helicase and ATPase activity of BLM [4, 17]. While the impact of homozygous mutations in BLM have been extensively characterized clinically, the health impacts of heterozygous BS-causing mutations is less well-studied. Limited studies have found that heterozygous BS mutations can lead to increased cancer risk in humans and in mice [18, 19]. Thus, although BS is recessive and manifests only when both copies of BLM are mutated, disrupting activity of one BLM allele may still have detrimental impacts on human health and longevity.

In addition to the lack of study on the impacts of heterozygous BLM mutations, there has been very little clinical or in vitro study of other naturally occurring BLM mutations that do not result in BS. There are over 1400 BLM mutations that exist in the human population, with 62 encoding for pathogenic mutations, 118 mutations considered benign, and the rest of uncertain or conflicting human health relevance [20]. In 2012, Mirzaei and Schmidt [21] investigated the effects of 41 BLM single nucleotide polymorphisms from the Short Genetic Variations database map by assessing the effects of these mutations in Saccharomyces cerevisiae. Interestingly, two substitutions, P868L and G1120R, led to a partial loss in function, demonstrated by increased sensitivity to hydroxyurea compared to wild-type (WT) BLM. These mutations also cause a 4–5 fold increase in SCEs, a moderate increase in sensitivity to hydroxyurea at low concentrations in human cells, and a decreased clearance of phosphorylated histone H2AX after treatment with camptothecin in human cells [22]. While these BLM mutations have less severe phenotypes than BS-causing mutations, P868L and G1120R substitutions still increase cellular genomic instability. Mutations in BLM that cause this intermediate phenotype could lead to increased risk of cancer or other adverse health impacts in humans. Interestingly, the BLM P868L allele has a 5.13% frequency in the human population, and the impacts of this heterozygous (or homozygous) BLM allele on human health and longevity are unknown [22]. Residue P868 resides in the helicase domain of BLM (Fig 1B) and is part of a lysine-rich loop (868-PKKPKK) that appears to be important for interacting with DNA at the ssDNA-dsDNA junction in BLM crystal structures [14, 15]. Residue G1120 is part of the BLM WH and terminates an alpha helix preceding a loop proximal to dsDNA in the BLM-dsDNA structure (Fig 1C).

To better understand the effects of naturally occurring non-BS BLM mutations, we tested BLMcore constructs (residues 636–1298) with P868L or G1120R substitutions using a battery of biochemical assays. We also measured activity of a BLMcore K869A K870A mutant protein to assess the role of lysines within the BLM lysine-rich loop. The BLMcore retains helicase, DNA binding, and ATPase activity, making it the simplest model for characterizing the impacts of these naturally occurring mutant proteins in vitro [14, 15]. Previous in vitro studies investigating the impacts of BS-causing mutations utilized a nearly identical BLM catalytic core construct (residues 642–1290) and found that BS-causing mutations had severe impacts on the helicase and ATPase activity of BLM [17]. Additionally, defects detected with BLM P868L and BLM G1120R in yeast used a construct that included residues 648–1417 of BLM [21]. Therefore, we focused our experiments on BLMcore constructs to allow comparison with prior reports and to allow measurement of possible moderate impacts of the mutations on in vitro function of BLM. In addition to our biochemical experiments, we compared molecular dynamics (MD) simulations of the WT BLMcore and BLMcore P868L bound to DNA and ADP.

While we predicted that the P868L and G1120R substitutions would impact the in vitro function of BLMcore, these mutant proteins had similar DNA-dependent ATPase activity, DNA binding activity, and helicase activity to WT BLMcore. In contrast, BLMcore K869A K870A requires ~4-fold more DNA to stimulate ATPase activity and has less robust helicase activity compared to WT BLMcore. Additionally, DNA binding experiments suggest that BLMcore K869A K870A may be deficient in recognizing ssDNA-dsDNA junctions. Thermal stability measurements indicate that differences in the activity of BLMcore K869A K870A could not be explained by general destabilization associated with the mutations. Since the naturally occurring BLMcore mutant proteins maintain activity levels that are very similar to WT BLMcore, our results indicate that these mutations lead to moderate genome instability through impairment of other cellular functions of BLM. In alignment with these results, the MD simulations suggest that the P868L substitution could subtly weaken interaction of DNA with BLM’s winged-helix domain and alter the orientation of the N-terminal lobe of the ATPase domain, which could shed light on previously identified defects of the hypomorphic P868L mutant protein in vivo [22].

Materials and methods

Protein expression and purification

The BLMcore (residues 636–1298) and BLMcore mutant proteins were overexpressed as previously described [14]. Briefly, overexpression was carried out in Rosetta 2 (DE3) E. coli transformed with pLysS and BLMcore or BLMcore mutant protein overexpression plasmids. Cells were grown at 37°C in Terrific Broth supplemented with 50 μg/mL kanamycin and 50 μg/mL chloramphenicol until cultures reached an OD600 ~1.8. BLM overexpression was induced using 0.5 mM IPTG and cells were grown overnight at 18°C. Cells were then pelleted and stored at -80°C.

Cells pellets were resuspended in lysis buffer (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 10% (v/v) glycerol, 0.1% Triton X-100, 20 mM imidazole, 1 Pierce Protease inhibitor tablet/100 mL buffer (Thermo Fisher), and 0.1 mM phenylmethylsulfonyl fluoride), lysed using sonication, and cleared via centrifugation. Clarified lysate was loaded onto a 5 mL HisTrap FF column (Cytiva) that was equilibrated with Buffer A (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 10% (v/v) glycerol, and 20 mM imidazole). Protein was eluted with Buffer B (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 10% (v/v) glycerol, 250 mM imidazole). Fractions containing BLMcore were diluted to 0.25 M NaCl and loaded onto a HiPrep 16/10 Heparin FF Column (Cytiva). The column was washed with Buffer C (20 mM Tris-HCl, pH 8.0, 0.25 M NaCl, 10% (v/v) glycerol) and then BLMcore was eluted from the column using Buffer D (20 mM Tris-HCl, pH 8.0, 1 M NaCl, 10% (v/v) glycerol). Fractions containing BLMcore were concentrated in Vivaspin-20 10 kDa concentrator (Sartorius) to 2 mL and loaded onto HiPrep 16/60 Sephacryl S-300 HR column (Cytiva) using Sizing Buffer (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 5% (v/v) glycerol). Fractions containing BLMcore were concentrated to ~10 mg/mL then flash frozen in liquid nitrogen and stored at -80°C. BLMcore and mutant proteins were tested for nuclease activity and found to have no detectable nuclease activity (S1 Fig).

The BLMcore K869A K870A construct was purified with the following modifications to remove a contaminating nuclease. After collecting fractions from the HisTrap FF column, fractions were assessed for nuclease activity. Five μL of each fraction was incubated for 30 minutes at ambient temperature with 40 nM fluorescein labeled dT30 in 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM DTT, 0.1 mg/mL Bovine Serum Albumin (BSA), and 5 mM MgCl2. Five μL of stop buffer (2% SDS, 5 μg/mL proteinase K, 20% (v/v) glycerol, 0.1 ethylenediaminetetraacetic acid (EDTA)) was added to each sample, and 5 μL of each sample was loaded onto a 15% acrylamide 1.5-mm gel in Tris-Borate-EDTA (TBE) buffer supplemented with 100 mM KCl. Gels were run at 75 V for 1 hour at 4°C in 1x TBE running buffer with 100 mM KCl and imaged on an Azure c600 (Azure Biosystems). Fractions that contained significant nuclease activity were discarded, and fractions containing BLMcore K869A K870A without significant nuclease activity were diluted to 0.15 M NaCl and loaded onto the HiPrep 16/10 Heparin FF Column, washed with Buffer C2 (20 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 10% (v/v) glycerol), and eluted over a 20-column volume gradient up to 0.5 M NaCl. Fractions containing BLMcore were tested for nuclease activity. Fractions that contained BLMcore and no detectable nuclease activity were concentrated in Vivaspin-20 10 kDa concentrator (Sartorius) to 2 mL and loaded onto the HiPrep 16/60 Sephacryl S-300 HR column (Cytiva) using Sizing Buffer (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 5% (v/v) glycerol). Fractions were tested for nuclease activity, and fractions with no detectable nuclease activity were concentrated then flash frozen in liquid nitrogen and stored at -80°C.

Differential scanning fluorimetry

Differential scanning fluorimetry (DSF) was carried out as previously described with the following modifications [23]. WT BLMcore, BLMcore P868L, BLMcore G1120R, or BLMcore K869A K870A (5 μM) were incubated in DSF buffer (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM DTT, and 5 mM MgCl2) with 5X Sypro Orange (Millipore-Sigma) in 20 μL total volume for 10 minutes at room temperature. For reactions containing ADP or ATPγS, nucleotide was added to final concentration of 0.5 mM. Samples were then heated from 4°C to 95°C at a rate of 0.25°C s-1 with fluorescence measured every two seconds using the HEX filter in a C1000 Touch Thermal Cycler (BioRad). Each DSF trial was performed in triplicate.

DSF data were analyzed using DSFWorld (https://gestwickilab.shinyapps.io/dsfworld/) [24]. Raw fluorescence data were trimmed from 30°C to 60°C and the maximum of the first derivative (dRFU) of the denaturation curve is reported as the melting temperature (Tm). ΔTm values for ADP and ATPγS samples were determined by subtracting the average Tm for each protein alone from the Tm value measured for the same protein in the presence of ADP or ATPγS. Significance was determined using Welch’s two-tailed t-test using the default settings on Prism Version 9.3.1. Normalized fluorescence plots were produced using data trimmed from 30°C to 60°C in Prism Version 9.3.1.

DNA-dependent ATPase assay

WT BLMcore, BLMcore P868L, or BLMcore G1120R (5 nM) were incubated with either no DNA or dT20 serially diluted from 10 nM to 9.8 x 10−3 nM in ATPase buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 5% (v/v) glycerol, 0.1 mM DTT, 5 mM MgCl2, 0.1 mg/mL BSA, 2 mM 2-phosphoenolpyruvate, 3 U/mL Pyruvate Kinase/Lactate Dehydrogenase, 0.2 mM NADH). BLMcore K869A K870A at 5 nM was incubated with no DNA or with dT20 serially diluted from 100 nM to 0.98 nM in ATPase buffer. Reactions were initiated by addition of 1 mM ATP and A340 nm was monitored for 1 hour at 25°C. Data were analyzed as previously described [25] and plotted using Prism Version 9.3.1. ATPase assays were done in triplicate. Significance was determined using Welch’s two-tailed t-test using the default settings on Prism Version 9.3.1. For standard error of the mean (SEM) measurements, these values were calculated in Prism Version 9.3.1 by dividing the standard deviation by the square root of the number of replicates.

Electrophoresis mobility shift assays

Electrophoresis mobility shift assays (EMSAs) were performed as previous described [26] with the following modifications. Partial duplex DNA (oRC32/FAM-oAV320, Table 1) or G4 DNA (FAM-oRC96, Table 1) constructs were folded by incubating 5 μM DNA in 10 mM Tris-HCl, pH 7.5 and 100 mM KCl at 95°C for 10 minutes and slowly cooling the sample to room temperature over several hours. DNA constructs were stored at 4°C. Serial dilutions of BLMcore or BLMcore mutant proteins were incubated with 40 nM partial duplex DNA, G4 DNA, or FAM-dT30 in 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM DTT, 0.1 mg/mL BSA, and 5 mM MgCl2 for 30 minutes at room temperature. 3.3% (v/v) glycerol was added to samples, and 5 μL of each sample was loaded onto a 5% acrylamide 1.5-mm gel in TBE buffer supplemented with 100 mM KCl. Gels were pre-run at 75 V for 20 minutes before loading protein/DNA complexes and running at 75 V for 1 hour at 4°C in 1xTBE running buffer supplemented with 100 mM KCl. Gels were imaged on the Azure c600 (Azure Biosystems). Experiments were done in triplicate.

Helicase assay

Helicase assays were performed as previous described [26] with the following modifications. Partial dsDNA or G4-dsDNA constructs were folded by incubating 5 μM DNA in 10 mM Tris-HCl, pH 7.5 and 100 mM KCl at 95°C for 10 minutes and slowly cooling the sample to room temperature over several hours, then stored at 4°C. Serial dilutions of BLMcore or BLMcore mutant proteins were incubated with 40 nM oRC32/FAM-oAV320 (partial duplex, Table 1) or oRC75/FAM-oAV320 (G4-dsDNA, Table 1) in helicase assay buffer (50 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM DTT, 0.1 mg/mL BSA, 5 mM MgCl2, 5 mM ATP, 4 μM oAV322) for 15 minutes at 37°C in 25 μL reactions. Folded DNA control was obtained by incubating reaction mixture without BLMcore, and melted control was obtained by omitting BLMcore and heating reaction to 95°C for 10 minutes. Five μL of stop buffer (2% SDS, 5 μg/mL proteinase K, 20% (v/v) glycerol, 0.1 mM EDTA) was added to each reaction and 5 μL of each sample was loaded onto a 15% acrylamide 1.5-mm gel in TBE buffer supplemented with 100 mM KCl. Gels were run at 75 V for 1 hour at 4°C in 1x TBE running buffer with 100 mM KCl. Gels were imaged on an Azure c600 (Azure Biosystems). BLMcore dsDNA, BLMcore P868L dsDNA and G4-dsDNA, and BLMcore K869A K870A dsDNA and G4-dsDNA were done in triplicate. BLMcore G4-dsDNA, BLMcore G1120R dsDNA and G4-dsDNA were done in quadruplicate. One data point was removed from the BLMcore G1120R G4-dsDNA dataset (at BLMcore G1120R concentration of 1.95 nM) after visual inspection of the gel which had additional fluorescence on the gel that obscured the results of the lane. DNA unfolding was quantified using ImageJ and analyzed using Prism Version 9.3.1 and fitting data to Eq (1). Significance was determined using Welch’s two-tailed t-test using the default settings on Prism Version 9.3.1.

(1)

Molecular dynamics simulations

The initial conformation for WT BLMcore was taken from one of the X-ray structures (PDB ID: 4CGZ) [15]. This structure contains a portion of the BLM protein (residues 637–1290) bound to DNA, an ADP molecule and a Zn2+ ion. Coordinates of the missing hydrogens were generated using the ‘pdb2gmx’ module of the GROMACS software [27], and the initial coordinates of a missing loop (residues 1194–1206) were constructed using MODELLER [28]. All titratable amino acids were assigned their default protonation states at pH of 7.4, except for the four cysteines (residues 1036,1055,1063,1066) that coordinate the bound Zn2+ ion, which are expected to be deprotonated [29]. Histidine protonation assignments (HD1 or HE2) were selected by carrying out a global optimization of the hydrogen bond network [30]. After building all missing atoms, the complex was energy minimized and then placed in a 12×12×12 nm cubic box containing 54,282 water molecules, 97 K+ ions and 79 Cl- ions, corresponding to an ionic strength of 100 mM. The difference in the numbers of K+ and Cl- ions serves to counter balance the net negative change of -18 eu on the complex. Following addition of solvent, the system was energy minimized for 500 steps, after which it was subjected to molecular dynamics (MD) simulation. The initial conformation of the P868L mutant protein was constructed from the energy minimized structure of the WT complex. BLM P868L was placed in a unit cell identical to the WT complex and it contains the same numbers of water molecules and salt ions. The P868L mutant protein unit cell was also energy minimized for 500 steps, and then subjected to MD.

All energy minimization and MD simulations were carried out using GROMACS version 2020 [27]. MD integration was carried out using the leap-frog algorithm and with a time step of 2 fs. All bonds were constrained [31]. MD simulations were carried out under isothermal-isobaric conditions. Pressure was regulated at 1.01325 bar using the Parrinello-Rahman extended ensemble approach [32], coupling time constant of 1.0 ps, and a compressibility of 4.5 × 10−5. Temperature was regulated at 310 K using the velocity-rescale thermostat [33] and a coupling constant of 0.1 ps, which uses a stochastic term for the generation of a proper canonical ensemble. Periodic boundary conditions were set in all directions, and electrostatic interactions were computed using a particle mesh Ewald scheme [34] with a short-range cutoff of 10 Å, Fourier grid spacing of 1 Å, and a fourth-order interpolation. van der Waals interactions were computed explicitly for interatomic distances smaller than 10 Å. Neighbor lists were constructed using grid search and were updated every 5 steps. Less frequent updates to neighbor lists, which are typically implemented to accelerate simulations on GPUs, resulted in DNA deformation. Water is described using the SPC/E model [35], and ion-water interactions are described using Joung and Cheatham parameters [36]. Protein, solvent ions, and DNA are described using the AMBER99SB-ILDN force field [37], along with NB-fix corrections for protein-DNA, ion-DNA and ion-ion interactions [38]. Note that the Joung and Cheatham ion parameters are not parameterized to reproduce reference data for ion-protein interactions [39], however, their use is justified as ions are not expected to bind directly to protein.

Results

BLMcore K869A K870A requires higher levels of ssDNA than WT BLMcore to stimulate ATPase activity

BLM requires ATP hydrolysis for translocation along DNA and helicase activity. Therefore, the human BLM mutant proteins BLM P868L and BLM G1120R could cause cellular defects due to a decrease in ATPase activity. To test this, we measured the DNA-dependent ATPase activity of mutant BLMcore proteins. We used a dT20 ssDNA substrate for this assay to examine only ssDNA-dependent activity, avoiding complications that could arise from ATPase activity associated with dsDNA unwinding. Interestingly, both of the BLMcore mutant proteins had similar maximum ATPase rates to WT BLMcore (Fig 2A, Table 2), indicating that BLMcore P868L and BLMcore G1120R are able to efficiently hydrolyze ATP when bound to ssDNA. While BLMcore P868L had an apparent increase in maximum ATPase rate compared to WT BLM (P value = 0.0116), there is a relatively small difference (35%) between the WT BLMcore and BLMcore P868L maximum ATPase rates (Table 2). Since the concentration of BLMcore directly impacts measured maximal ATPase rates, small inaccuracies in concentration determination can lead to modest apparent differences that are difficult to interpret. We therefore do not ascribe biochemical significance to this difference in maximum ATPase rate. The DNA concentration-dependence of the ATPase activity was also similar for WT BLMcore, BLMcore P868L, and BLMcore G1120R. The DNA concentrations required for half-maximum ATPase rate (KDNA) differed by less than 1-fold among all three proteins (Fig 2B, Table 2).

thumbnail
Fig 2. DNA-dependent ATPase assays.

(A). Maximum DNA-dependent ATPase rates for WT BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A. These data represent the mean of three replicates with error bars representing the standard error of the mean. P-values are indicated above the bars. (B). ssDNA concentrations required for half-maximum ATPase rate (KDNA) for WT BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A. These data represent the mean of three replicates with error bars representing the standard error of the mean. P-values are indicated above the bars.

https://doi.org/10.1371/journal.pone.0281524.g002

thumbnail
Table 2. Summary of ATPase, DNA binding, and helicase activity.

https://doi.org/10.1371/journal.pone.0281524.t002

BLMcore K869A K870A was found to have a modestly different maximum ATPase rate compared to WT BLMcore (P value = 0.0348). More interestingly, this mutant protein had a ~4-fold higher KDNA value (P value = 0.0183), indicating that higher concentrations of ssDNA were necessary to stimulate its half-maximum ATPase activity (Fig 2, Table 2). Unlike maximal ATPase measurements, KDNA values are independent of protein concentration, indicating that this difference is biochemically significant. These data show that disrupting the charge of the lysine-rich loop impacted the concentration of ssDNA required to stimulate DNA-dependent ATP hydrolysis in BLMcore whereas the P868L and G1120R substitutions did not. This result is consistent with a reduced DNA binding affinity for BLMcore K869A K870A relative to the other BLMcore constructs tested.

We also assessed the ATPase rate for each mutant protein in the absence of DNA. WT BLMcore, BLMcore P868L, and BLMcore G1120R were capable of hydrolyzing ATP in the absence of DNA, with rates of 4–10% of that observed with saturating levels of ssDNA. In contrast, BLMcore K869A K870A had no detectable DNA-independent ATPase activity (Table 2). This finding aligns with the higher DNA concentration requirement for this mutant protein and may point to a defect in its overall ATPase functions.

BLMcore K869A K870A has reduced specificity for binding ssDNA-dsDNA junctions

BLM is able to bind and unwind a variety of DNA structures [3]. To test the ability of the BLMcore proteins to bind to different DNA structures, we used Electrophoretic Mobility Shift Assays (EMSAs) to assess binding to ssDNA, G4 DNA with a 3’ ssDNA overhang, and dsDNA with a 3’ ssDNA overhang (partial dsDNA), each done in triplicate. WT BLMcore bound to ssDNA and the G4 substrate, as evidenced by slower migration of the labeled DNA on a gel (Fig 3). This shift manifested as “smearing” for these substrates, which could indicate that these complexes are dynamic, dissociating and re-associating in the assay. In contrast, when tested with partial dsDNA, WT BLMcore was able to form a distinct band shift, consistent with WT BLMcore forming a stable complex with the substrate. The same trend was observed for BLMcore P868L and BLMcore G1120R, which were both able to shift ssDNA and G4 DNA, indicated by smearing, and were able to shift partial dsDNA, indicated by distinct band shifts (Fig 3). From estimating qualitative binding affinities from the gel shift assays, WT BLMcore, BLMcore P868L, and BLMcore G1120R all appear to have similar affinities for each of these DNA substrates (Table 2).

thumbnail
Fig 3. DNA binding assays for WT BLM and mutant proteins.

Electrophoretic mobility shift assays for WT BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A (left to right) for binding ssDNA (top), G4 DNA (middle), and partial-duplex DNA (bottom). Lowest and highest concentrations are indicated below each gel, and the estimated apparent KD is indicated with an arrow. Concentrations for WT BLMcore with G4 and with ssDNA, BLMcore P868L with G4, ssDNA, and partial-dsDNA, and BLMcore G1120R with G4, ssDNA, and partial-dsDNA are 1.00 μM, 500 nM, 250 nM, 125 nM, 62.5 nM, 31.2 nM, 15.6 nM, 7.81 nM, 3.91 nM, and 1.95 nM. WT BLM with partial-dsDNA concentrations are 600 nM, 400 nM, 267 nM, 178 nM, 118 nM, 79.0 nM, 52.7 nM, 35.1 nM, 23.4 nM, and 15.6 nM. Concentrations for BLMcore K869A K870A with G4, ssDNA, and partial-dsDNA are 2.5 μM, 1.25 μM, 625 nM, 312 nM, 156 nM, 78.1 nM, 39.1 nM, 19.5 nM, 9.77 nM, and 4.88 nM. DNA concentration for each assay is 40 nM. EMSAs were run in triplicate with the gel shown here as a representative example.

https://doi.org/10.1371/journal.pone.0281524.g003

BLMcore K869A K870A was also able to bind to ssDNA and G4 DNA (Fig 3) but required higher protein concentrations to shift the DNA, consistent with a lower affinity for these substrates (Table 2). This binding also resulted in a smear above the substrate migration site. Whereas WT BLMcore, BLMcore P868L, and BLMcore G1120R were able to bind the partial dsDNA causing a distinct band shift, the BLMcore K869A K870A mutant protein did not have this effect. Instead, BLMcore K869A K870A binding to partial duplex DNA resulted in a smear, similar to what was observed for binding to ssDNA and G4 DNA. Additionally, a band corresponding to the labeled ssDNA from the substrate was consistently observed with the highest concentration of BLMcore K869A K870A (1 μM). This could be due to modest ATP-independent unwinding at high concentrations of this mutant protein. As described in the Materials and Methods, this effect is not due to a contaminating nuclease in the BLMcore K869A K870A purification. Thus, BLMcore K869A K870A appears to have reduced overall DNA binding stability and reduced ability to discriminate among different DNA structures.

BLMcore K869A K870A is a less efficient helicase than WT BLMcore, BLMcore P868L, and BLMcore G1120R

We next tested the ability of the BLM mutant proteins to unwind dsDNA and G4-dsDNA. This experiment was done using a gel-based helicase assay with a substrate containing a 3’ ssDNA 15 nucleotide overhang, either the human telomere G4-forming sequence or a control sequence that cannot form a G4, and 18 base pairs of dsDNA (Fig 4A). The annealed strand contains a fluorescent label on its 3’ end and unwinding of this substrate can be observed by tracking migration of fluorescent label in a gel (Fig 4B). The reaction is initiated by the addition of BLMcore or mutant protein, ATP, and an unlabeled trap oligo that has the same sequence as the fluorescently labeled ssDNA. The total fraction of DNA unwound at various BLMcore concentrations is measured (Fig 4C), allowing a determination of the fraction of DNA that BLMcore or mutant protein can unwind and the concentration of BLMcore required for half-maximum unwinding (KBLM).

thumbnail
Fig 4. WT BLMcore and BLMcore mutant protein helicase assays unwinding duplex DNA and G4 DNA.

(A) Schematic of experimental system. (B). (Top to bottom): WT BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A gel-based helicase assays assessing duplex DNA unwinding (left) and G4-dsDNA unwinding (right). For WT BLMcore and BLMcore G1120R, protein concentrations are 1.00 μM, 500 nM, 250 nM, 125 nM, 62.5 nM, 31.2 nM, 15.6 nM, 7.81 nM, 3.91 nM, and 1.95 nM. For BLMcore P868L, protein concentrations are 100 nM, 50 nM, 25 nM, 12.5 nM, 6.25 nM, 3.12 nM, 1.56 nM, 0.781 nM, 0.391 nM, and 0.195 nM. For BLMcore K869A K870A, protein concentrations are 2.5 μM, 1.25 μM, 625 nM, 312 nM, 156 nM, 78.1 nM, 39.1 nM, 19.5 nM, 9.77 nM, and 4.88 nM. DNA concentrations are 40 nM for each assay. Gels are representative examples of helicase assays run in triplicate or quadruplicate. (C). Quantification of fraction of substrate unwound for dsDNA unwinding (black) and G4-dsDNA unwinding (blue). Data points are the mean values of helicase assays with error bars representing the standard error of the mean.

https://doi.org/10.1371/journal.pone.0281524.g004

WT BLMcore was able to efficiently unwind both the dsDNA and the G4-dsDNA substrates (Fig 4B and 4C), with over 90% maximum unwinding for both. The KBLM value for WT BLM was 15 ±

2.4 nM and 17 ± 2.4 nM for dsDNA and G4-dsDNA, respectively (Table 2). Both BLMcore G1120R and BLMcore P868L were able to unwind over 90% of the dsDNA substrate and G4-dsDNA substrate as well, indicating that these mutant proteins have similar maximum activity levels to WT BLMcore. BLMcore P868L had KBLM values of 12 ± 2.4 nM and 6.2 ±1.4 nM for dsDNA and G4-dsDNA, respectively (Table 2), indicating that BLMcore P868L requires similar amounts of protein to unwind dsDNA as WT BLMcore but that it was somewhat more efficient at unwinding G4-dsDNA substrates (P values = 0.36 and 0.012). BLMcore G1120R had KBLM values of 13 ± 2.4 nM and 23 ± 3.1 nM for dsDNA and G4-dsDNA unwinding, indicating that BLMcore G1120R unwound dsDNA and G4-dsDNA with a similar efficiency to that of WT BLMcore (P values = 0.56 and 0.22).

BLMcore K869A K870A was unable to unwind the same percentage of substrate, with lower than 80% total unwinding for both dsDNA and G4-dsDNA (Table 2). At high concentrations of BLMcore K869A K870A, we observed inhibition of unwinding (Fig 4C), which is also observed at high concentrations of WT BLMcore. The KBLM values obtained from BLMcore K869A K870A unwinding dsDNA and G4-dsDNA were 22 ± 6.6 nM and 46 ± 8.8 nM, respectively, however standard error of the mean (SEM) was much higher for this mutant protein (Table 2). These values were determined to not be statistically significant when compared to the WT BLMcore dsDNA and G4-dsDNA KBLM values (P values = 0.42 and 0.072), but this is likely due to the large SEM for BLMcore K869A K870A.

All BLMcore mutant proteins are stably folded under assay temperatures

One possible explanation for the impact of the substitutions in the BLMcore K869A K870A protein was that the substitutions destabilize the protein. To test whether sequence changes in the BLMcore mutant proteins altered proper protein folding, the melting temperature (Tm) of each BLMcore mutant protein was measured using differential scanning fluorimetry (DSF). WT BLMcore and BLMcore P868L have similar Tm values (46.3 ± 0.12°C and 45.7 ± 0.23°C, respectively), whereas BLMcore G1120R and BLMcore K869A K870A have Tm values of 42.6 ± 0.23°C and 42.1 ± 0.30°C, respectively (S2 Fig, Fig 5A). The Tm differences among the proteins are modest, and all of the proteins are stably folded at temperatures used for all the ATPase assays, DNA binding assays, and helicase assays of this study.

thumbnail
Fig 5. BLMcore mutant protein melting temperatures.

(A) Melting temperature (Tm) of BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A. Error bars indicate standard error of the mean and p-values are indicated above bars. (B) Change in Tm (ΔTm, ADP) for each protein with the addition of 0.5 mM ADP compared to the Tm of each mutant protein without nucleotide present. ΔTm, ADP values were obtained by subtracting the mean Tm in (A) of each mutant protein from each Tm obtained in the presence of ADP. Error bars indicate the standard error of the mean and p-values are indicated above bars. (C) The ΔTm, ATPγS for each protein with the addition of 0.5 mM ATPγS compared to the Tm of each mutant protein without nucleotide present. ΔTm, ATPγS values were obtained as in (B). Error bars indicate the standard error of the mean and p-values are indicated above bars.

https://doi.org/10.1371/journal.pone.0281524.g005

We also tested the impact of ADP and ATPγS on the Tm of BLMcore and BLMcore mutant proteins. All proteins were stabilized by the presence of ADP or ATPγS (Fig 5B, 5C), indicating that each BLMcore protein could bind these nucleotides. Interestingly, BLMcore G1120R is stabilized to a greater extent by ADP, with a ΔTm, ADP of 3.7°C compared to the ΔTm, ADP of WT BLMcore of 2.9°C. BLMcore K869A K870A was also stabilized to a greater extent by ADP, with a ΔTm, ADP of 4.8°C. BLMcore K869A K870A was also stabilized to a greater extent by ATPγS, with a ΔTm, ATPγS of 3.6°C compared to the WT BLMcore ΔTm, ATPγS of 2.1°C. Overall, these data shows that all BLMcore mutant proteins are stabilized by nucleotide binding, with BLMcore K869A K870A having the greatest increase in ΔTm for both ADP and ATPγS.

Molecular dynamics simulations of blmcore P868L

The BLM P868L allele has a 5.13% frequency in the human population and causes an increased frequency of sister chromatid exchanges in cell culture, making it essential to better understand how this substitution impacts BLM function. To examine effects of the P868L substitution on the structure and dynamics of the BLMcore complexed with partial dsDNA and ADP, we carried out MD simulations. In MD simulations, macromolecules are placed in bulk solvent and all atoms belonging to macromolecules and solvent are set free to move in three-dimensional space under the influence of each other’s forces and under applied boundary conditions of pressure (density) and temperature. During simulations, the structure of the macromolecule is free to change, and the solvent is free to move around. These simulations are run for a fixed period of time, which provides atomic-level insights into the structural evolution and dynamics of macromolecules. In this work, we carried out six 1-μs long MD simulations, three of the BLMcore P868L mutant protein and three of the WT BLMcore (control). Each of these simulations required about two months of wall clock time on a single-GPU, 20-core workstation.

First, we analyzed the simulation data to examine the association of DNA with BLM. In the X-ray structure of WT BLMcore (PDB ID: 4CGZ [15]), the dsDNA (12 bp) interacts directly with the RecA-like lobe 2 of the ATPase domain (RecA-D2) and with the ZN and WH subdomains of the RQC domain of BLM. The single-stranded portion of the DNA (5 nucleotides) interacts directly only with RecA-D2. We find that this general pattern of association between BLM and DNA is maintained in the MD simulations of the WT BLMcore. The one difference that we note between the X-ray structure and MD simulation is that the distance between DNA and ZN domain is larger in MD simulations (Fig 6A). This is unlikely to be an artifact of the employed force field because we note such increased DNA-ZN distances even when the complex is simulated with an older version of the DNA-protein force field (amber-99sb-ILDN without DNA-protein NB-fix corrections [37]) in which protein-DNA binding is stronger.

thumbnail
Fig 6. Effects of the P868L substitution on the structure and dynamics of the BLMcore protein complexed with DNA and ADP.

(A) Effect of the P868L substitution on the structure and dynamics of bound DNA. Top panels: Fluctuations in distances between centers of mass. (CoMs) of DNA and BLM domains (ZN-zinc-binding domain, WH-winged helix domain, D2- RecA-like lobe 2 of the ATPase domain). The dashed vertical lines indicate CoM distances in the X-ray structure. Bottom panels: MD snapshots superimposed over theBLMcore X-ray crystal structure. The purpose of choosing a coarse-grained representation to illustrate the finding is that in MD, the DNA moves away from the BLM protein. The D2, ZN and WH domains are shown using their CoMs connected by solid lines: the CoM of D2 is connected to the CoM of ZN and the CoM of ZN is connected to the CoM of WH. DNA is shown using two line segments, one connecting the CoMs of the two halves of the double stranded region, and the other connecting the CoM of its single stranded region with the CoM of one half of its double stranded region. Sixty snapshots are shown, twenty taken from each of the three MD simulations at regular time intervals of 50 ns. The coordinates of the CoMs and coarse-grained representation of the DNA are determined after the backbone atoms of D2 in that snapshot are fitted onto the backbone atoms D2 in the X-ray structure. Consequently, the CoM of D2 appears stationary in time. (B) Effect of P868L substitution on the angle between the CoMs of ZN, D2, and D1 domains. The dashed vertical line on the figure on the left is the angle observed in the X-ray structure. The figure on the right compares the distribution of the domain CoMs (spheres) in 60 equally spaced snapshots extracted from MD simulations. The grey spheres are CoMs of domains in the WT BLM and those in other colors are CoMs of domains in the P868L mutant protein. (C) P868L induced shift (η) in conformational ensemble of the D2 domain. The X-ray structure of the D2 domain is shown as cartoon putty where the width of the putty as well as color of the putty denote the magnitude of η. Higher values of η imply larger effects of P868L induced shifts in conformational ensembles.

https://doi.org/10.1371/journal.pone.0281524.g006

Comparison of WT and P868L simulations reveals one key effect of the P868L substitution on DNA-BLM binding: while DNA-ZN and DNA-D2 distances are similar to those in the simulations of the WT form, the DNA-WH distance is larger (Fig 6A). Therefore, effectively, the P868L mutation pushes the DNA further away from the BLM core. This is also visually discernible in the structural coarse-grained representations of BLM-DNA complex in the bottom panel of Fig 6A compared to WT as there are distinctly more conformations of DNA that are further away from BLM in P868L. From an energetics standpoint, this implies that the P868L substitution weakens the interaction between the BLM core and DNA. We note that our current in vitro biochemical experiments do not show any discernible effect of P868L on BLM activity, which appears to be in contrast with both MD simulations and our previous cellular assays (21, 22). This may perhaps be due to the low sensitivity and qualitative nature of the biochemical assays employed in this study.

Other than BLM-DNA interactions, the P868L substitution also changes the orientation of the RecA-D1 relative to the rest of BLM in MD simulations. The P868L substitution reduces the angle between the centers of mass (CoMs) of the D1, D2 and ZN domains (Fig 6B). Since the sites for ATP-binding and hydrolysis are at the interface between RecA-D1 and RecA-D2, this result suggests potential effects of P868L on ATP hydrolysis and, thus, BLM activity. Changes were not observed in our biochemical assays, but this may be due to use of ssDNA instead of partial dsDNA. However, interpretation of ATPase rate data with partial duplex DNA in the biochemical assay is not simple, since ATPase activity would be coupled to DNA unwinding that would change the structure of the stimulating DNA from partial duplex to ssDNA during steady-state measurements.

To understand how P868L substitution alters RecA-D1 orientation, we examined the effect of P868L substitution on the conformational ensemble of RecA-D2. We used Direct Comparison of Ensembles (DiCE) [4042], which computes the physical overlap between two conformational ensembles in high-dimensional space. It yields the difference in ensembles in terms of a quantity ‘η’ that is normalized to vary between 0 and 1. The larger the difference in ensembles, the closer the value of η is to 1. η is normalized with respect to amino acid size, which allows comparison of η between different amino acids. We found that for most amino acids η < 0.5 (Fig 6C). If conformational ensemble shifts were purely in mean positions, and thermal fluctuations were of the order of 1.0 Å, then η = 0.5 would correspond roughly to a shift in mean position by 0.7 Å. Therefore, amino acids with η < 0.5 were considered to have small induced shifts. We found that amino acids with η > 0.5 (the most affected with η ~ 0.7) are close to RecA-D1 (Fig 6C). This suggests that the signal that originates at the P868 substitution site at the periphery of RecA-D2 [21] could allosterically reach the catalytic cleft and the linker between the two lobes, explaining the potential effect of P868L on RecA-D1 orientation.

Discussion

BLM P868L and BLM G1120R both exist in the human population and are not associated with BS. However, both of these mutations cause increased SCEs and decreased recovery from hydroxyurea at low concentrations and decreased clearance of phosphorylated histone H2AX after treatment with camptothecin in human cells [22]. While these phenotypes are not as severe as what is observed in BS-causing mutations, these mutations still lead to genome instability and could cause an increased risk of cancer or other negative health outcomes. Many BS-causing BLM mutant proteins have been investigated in vitro and have been found to have decreased helicase activity and ATPase activity [17]. We hypothesized that these human BLM mutant proteins would lead to a moderate decrease in ATPase activity, DNA binding activity, and/or helicase activity.

Interestingly, BLMcore P868L and BLMcore G1120R are both active ATPases that have similar maximum ATPase rates to WT BLMcore and require similar concentrations of ssDNA for ATPase activity stimulation (Fig 2, Table 2). Additionally, both mutant proteins have ATPase activity even in the absence of DNA, similar to WT BLMcore. When assessing the ability of these mutant proteins to bind to ssDNA, G4 DNA, and partial dsDNA, both were able to bind substrates with similar affinities to WT BLM BLMcore While binding to ssDNA and G4 DNA was indicated by a smeared complex, binding to the partial duplex DNA induced a distinct band shift. The smearing observed with binding to ssDNA and G4 DNA substrates could indicate weaker overall binding and be caused by dissociation of BLMcore from the DNA while running in the gel. The distinct band shift observed from the partial duplex is consistent with WT BLMcore, BLMcore P868L, and BLMcore G1120R binding more stably to the partial duplex than the other DNA constructs.

We also assessed the ability of BLMcore P868L and BLMcore G1120R to unwind dsDNA and G4-dsDNA substrates. WT BLMcore, BLMcore P868L, and BLMcore G1120R were able to unwind dsDNA efficiently and required similar amounts of BLM to reach half-maximum unwinding (Table 2). Interestingly, BLMcore P868L was able to unwind G4-dsDNA more efficiently than WT BLMcore, indicating that this mutant protein is slightly better at unwinding G4s than WT BLMcore. This could indicate that BLMcore P868L has a tighter binding affinity for G4s than WT BLMcore, causing more proficient unwinding. However, direct binding studies did not demonstrate a higher affinity of BLMcore P868L for G4s than WT BLMcore (Table 2), so other properties such as processivity may be the source for this modest difference. BLMcore G1120R is less efficient at unwinding the G4-dsDNA substrate than the dsDNA substrate alone, requiring more BLMcore G1120R for half-maximum unwinding. This could be because this mutant protein is less efficient at unwinding G4s than dsDNA or because the BLMcore G1120R mutant protein is less processive on longer substrates.

Overall, the in vitro activities of BLMcore P868L and BLMcore G1120R are strikingly similar to that of WT BLMcore, despite these mutant proteins causing moderate genome instability in cells. While it has been observed that BS-inducing BLM mutant proteins have decreased helicase and ATPase activity in vitro, this is not the case for these non-BS human BLMcore mutant proteins [17]. However, there are several other ways that these mutations could cause genome instability in cells. Many RecQ helicases, including BLM, are regulated by posttranslational modifications (PTMs), such as phosphorylation, acetylation, and SUMOylation [3]. For example, BLM is phosphorylated in response to replication associated DNA damage or dsDNA breaks. Mutations that impact phosphorylation of BLM can decrease the ability of cells to recover from hydroxyurea exposure and other DNA damage [1]. Residues 868 and 1120 are not known sites of PTMs, making it unlikely that this is the reason that these mutant proteins cause increased SCEs [43]. However, residue K873, which is part of the lysine rich loop proximal to P868 is a reported site of SUMOylation and ubiquitination [43]. Because P868 terminates a β-strand at the start of the lysine-rich loop, it is possible that the mutation to leucine changes the structure of the loop and disrupts recognition of K873 by cellular PTM enzymes. This could result in decreased BLM activity in cells but more limited impact on in vitro activity of BLM.

Another possibility is that these mutations lead to increased SCEs in cells by producing BLM mutant proteins with disrupted dHJ dissolution activity and/or altered interaction with other proteins. BLM interacts with over 20 different proteins, including other dissolvasome proteins that are involved in dissolving dHJs [1, 3, 12]. If the BLM dissolvasome function is impaired due to weaker interactions between BLM and Top3α, RMI1, or RMI2 because of these mutations, this could lead to increased genome instability. From our studies, we cannot determine if P868L or G1120R human mutations decrease protein-protein interactions of BLM, but this will be an interesting target for future studies. From the MD simulations, a weakened interaction between DNA and the WH domain was observed for BLM P868L. This difference could contribute to the partial loss of cellular function of BLM P868L by disrupting BLM P868L dHJ resolution. Additionally, the change in orientation of the RecA-D1 lobe, which affects the catalytic cleft of the BLM helicase core, as well as subtle conformational changes in conserved helicase motifs V and VI (S3 Fig), could contribute to the functional deficits of BLM P868L in yeast and in human cells [21, 22]. However, the effects of these changes may be too small to be detected by the in vitro biochemical assays used in this study and would be more likely to have observable impact on other in vivo functions of BLM.

We also investigated the effects of substitutions to the lysine-rich loop proximal to the P868 residue in the BLMcore K869A K870A construct. BLMcore K869A K870A is moderately less stable than WT BLMcore, but still has a Tm well above the temperatures used for our in vitro assays, indicating that this protein was stably folded in our studies (Fig 5). This BLMcore K869A K870A mutant protein had a 4-fold increase in ssDNA concentration required for half-maximum stimulation of ATPase activity and a loss of ATPase activity in the absence of DNA. While DNA is known to stimulate ATPase activity of RecQ helicases, it was surprising that BLMcore K869A K870A had no ATPase activity in the absence of DNA. It is important for helicases to couple ATP hydrolysis to their functions of DNA translocation and DNA unwinding, and it has been found in the bacterial RecQ that the aromatic-rich loop (ARL) (residues 798–808 in BLM) is important for coupling ATP hydrolysis to the unwinding activity of RecQ. Substitutions to the RecQ ARL produce mutant proteins that either require more DNA for maximum ATP hydrolysis or with increased DNA-independent ATPase activity [44]. While the BLMcore K869A K870A substitutions are not part of the ARL of BLM, this loop might have a similar effect on regulating ATPase activity and ATP-coupled functions of BLM. The MD simulations suggest that P868L has allosteric effects that alter the RecA-D1/D2 interface that forms the catalytic cleft of the BLM helicase domain. There could be multiple allosteric signaling pathways that connect the P868 site, and likely the lysine-rich loop, to the RecA-D1 lobe [45]. Identifying these signaling pathways and hotspots will require additional simulations [42], and expanding the simulations to other functionally impaired BLM mutants, including G1120R and mutants of the lysine-rich loop, will require additional simulations, and we expect this to be a target area for future studies.

In addition to the diminished ATPase activity, BLMcore K869A K870A has a decreased apparent affinity for ssDNA and G4 substrates. Further, BLMcore K869A K870A is deficient in stably and selectively binding to ssDNA-dsDNA junctions. While WT BLMcore band shifts partial dsDNA, BLMcore K869A K870A only causes a smear above the substrate. This smear is consistent with an overall reduced DNA binding stability for BLMcore K869A K870A. Additionally, at high concentrations of BLMcore K869A K870A, there is the appearance of a band at the size of the unannealed fluorescent-labeled ssDNA. This band appears to be the result of some amount of ATP-independent unwinding of the substrate from BLMcore K869A K870A. If the lysine-rich loop of BLM is important for linking ATPase activity of BLM to its other functions, such as helicase activity, it is possible that substitutions to this loop are able to decouple these activities and generate a mutant protein that has weak ATP-independent helicase activity. Despite this, BLMcore K869A K870A is unable to unwind the same amount of dsDNA or G4-dsDNA as WT BLMcore. BLMcore K869A K870A does not reach over 80% unwinding of either substrate and requires higher concentrations of protein to unwind both dsDNA and G4-dsDNA. This may be due to enhanced strand annealing in the mutant protein, as has been observed with BLM previously [46]. BLMcore K869A K870A requires approximately 2-fold more protein to unwind the G4-dsDNA compared to the dsDNA substrate. This could be because this mutant protein has reduced G4 binding or unwinding or that this mutant protein is less processive than WT BLMcore.

From this work, we conclude that BLMcore P868L and BLMcore G1120R are competent helicases in the in vitro activities tested herein. However, since both of these mutant proteins lead to genome instability phenotypes in human cells, these mutant proteins are clearly defective in some activities that are essential for genome stability. Future studies should focus on assessing the protein interaction network, dHJ dissolution activity, and cellular localization of these mutant proteins. We have also shown that the lysine-rich loop of BLM is an important regulator of both ATPase activity and helicase activity of BLM. This linker could play a role in coupling ATP hydrolysis with the unwinding action of BLM and can provide a rich area for future BLM studies.

Supporting information

S1 Fig. Purity and nuclease test of WT BLMcore and BLMcore variants.

(A). Assessment of the purity of WT BLMcore and BLMcore variants. Left to right: Blue Prestained Broad Range Protein Standard (New England Biolabs), WT BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A loaded onto a 4–15% Mini-PROTEAN TGX Precast Protein gel (BioRad). Each protein was at a concentration of ~2.5 μM. (B). Assessment of nuclease contamination of WT BLMcore and BLMcore variants at 1 μM concentration. Proteins were incubated with 40 nM fluorescein labeled dT30 in 50 mM Tris-HCl, pH 7.5, 50 mM KCl, 1 mM DTT, 0.1 mg/mL Bovine Serum Albumin (BSA), and 5 mM MgCl2 for 30 minutes at room temperature. Five μL of stop buffer (2% SDS, 5 μg/mL proteinase K, 20% (v/v) glycerol, 0.1 ethylenediaminetetraacetic acid (EDTA)) was added to each sample, and 5 μL of each sample was loaded onto a 15% acrylamide 1.5-mm gel in Tris-Borate-EDTA (TBE) buffer supplemented with 100 mM KCl. Gels were run at 75 V for 1 hour at 4°C in 1xTBE running buffer with 100 mM KCl and imaged on an Azure c600 (Azure Biosystems). Samples in gel (from left to right) are DNA alone, WT BLM, BLM P868L, BLM G1120R, and BLM K869A K870A.

https://doi.org/10.1371/journal.pone.0281524.s001

(TIF)

S2 Fig. Normalized melting curves for each BLMcore construct.

The normalized melt curves between 30°C and 60°C is shown for BLMcore, BLMcore P868L, BLMcore G1120R, and BLMcore K869A K870A in the absence (blue circles) or presence of 0.5 mM ADP (red squares) or ATPγS (green triangle). Points on graph indicate the mean of three replicates and error bars indicate standard deviation calculated by Prism (version 9.3.1).

https://doi.org/10.1371/journal.pone.0281524.s002

(TIF)

S3 Fig. P868L-induced shift (η) in conformational ensemble of lobe 2 (D2 domain) of the ATPase domain of human BLM.

The X-ray structure of the D2 domain is shown as a cartoon putty where the width of the putty denotes the magnitude of η. The three conserved helicase motifs present in D2 and the linker connecting D2 to D1 are highlighted in color. The red sphere indicates the position of the P868 residue at the opposite side of D2.

https://doi.org/10.1371/journal.pone.0281524.s003

(TIF)

Acknowledgments

The authors thank members of the Keck laboratory for critical reading and evaluation of the manuscript.

References

  1. 1. Manthei KA, Keck JL. The BLM dissolvasome in DNA replication and repair. Cell Mol Life Sci. 2013;70(21):4067–84. pmid:23543275
  2. 2. Bloom D. Congenital Telangiectatic Erythema Resembling Lupus Erythematosus in Dwarfs. Am Med Assoc Am J Dis Child. 1954;754–8.
  3. 3. Croteau DL, Popuri V, Opresko PL, Bohr VA. Human RecQ Helicases in DNA Repair, Recombination, and Replication. Annu Rev Biochem. 2014;83:519–52. pmid:24606147
  4. 4. German J, Sanz MM, Ciocci S, Ye TZ, Ellis NA. Syndrome-Causing Mutations of the BLM Gene in Persons in the Bloom’s Syndrome Registry. Hum Mutat. 2007;28(8):743–53. pmid:17407155
  5. 5. Chaganti RSK, Schonberg S, German J. A Manyfold Increase in Sister Chromatid Exchanges in Bloom ‘ s Syndrome Lymphocytes. Proc Natl Acad Sci. 1974;71(11):4508–12. pmid:4140506
  6. 6. Wang W, Seki M, Narita Y, Sonoda E, Takeda S, Yamada K, et al. Possible association of BLM in decreasing DNA double strand breaks during DNA replication. EMBO. 2000;19(13):3428–35. pmid:10880455
  7. 7. Bennett RJ, Keck JL. Structure and Function of RecQ DNA Helicases. Crit Rev Biochem Mol Biol. 2004;39(2):79–97. pmid:15217989
  8. 8. Wang W, Seki M, Narita Y, Nakagawa T, Yoshimura A, Otsuki M, et al. Functional Relation among RecQ Family Helicases RecQL1, RecQL5, and BLM in Cell Growth and Sister Chromatid Exchange Formation. Mol Cell Biol. 2003;23(10):3527–35. pmid:12724411
  9. 9. Abu-libdeh B, Jr RMB, Stewart GS, Abu-libdeh B, Jhujh SS, Dhar S, et al. RECON syndrome is a genome instability disorder caused by mutations in the DNA helicase RECQL1. J Clin Invest. 2022;132(5):1–17. pmid:35025765
  10. 10. Hickson ID. RecQ Helicases: Caretakers of the Genome. Nat Rev Cancer. 2003;3:169–78. pmid:12612652
  11. 11. Hoadley KA, Xu D, Xue Y, Satyshur KA, Wang W, Keck JL. Structure and Cellular Roles of the RMI Core Complex from the Bloom Syndrome Dissolvasome. Structure [Internet]. 2010;18(9):1149–58. Available from: pmid:20826341
  12. 12. Wu L, Hickson ID. The Bloom’s syndrome helicase suppresses crossing over during homologous recombination. Nature. 2003;426(18):870–4. pmid:14685245
  13. 13. Seki M, Nakagawa T, Seki T, Kato G, Tada S, Takahashi Y, et al. Bloom Helicase and DNA Topoisomerase IIIa Are Involved in the Dissolution of Sister Chromatids. Mol Cell Biol. 2006;26(16):6299–307.
  14. 14. Swan MK, Legris V, Tanner A, Reaper PM, Vial S, Bordas R, et al. Structure of human Bloom’s syndrome helicase in complex with ADP and duplex DNA research papers. Acta Crystallogr. 2014;1370:1465–75.
  15. 15. Newman JA, Savitsky P, Allerston CK, Bizard AH, Sarl K. Crystal structure of the Bloom ‘ s syndrome helicase indicates a role for the HRDC domain in conformational changes. Nucleic Acids Res. 2015;43(10):5221–35. pmid:25901030
  16. 16. Mirzaei H, Syed S, Kennedy J, Schmidt KH. Sgs1 Truncations Induce Genome Rearrangements but Suppress Detrimental Effects of BLM Overexpression in Saccharomyces cerevisiae. J Mol Biol [Internet]. 2011;405(4):877–91. Available from: pmid:21111748
  17. 17. bing Guo R, Rigolet P, Ren H, Zhang B, dong Zhang X, xing Dou S, et al. Structural and functional analyses of disease-causing missense mutations in Bloom syndrome protein. Nucleic Acids Res. 2007;35(18):6297–310. pmid:17878217
  18. 18. Gruber SB, Ellis NA, Scott KK, Almog R, Kolachana P, Bonner JD, et al. BLM Heterozygosity and the Risk of Colorectal Cancer. Science (80-). 2013;297.
  19. 19. Goss KH, Risinger MA, Kordich JJ, Sanz MM, Straughen JE, Slovek LE, et al. Enhanced Tumor Formation in Mice Heterozygous for Blm Mutation. Science (80-). 2002;297:2051–4. pmid:12242442
  20. 20. Karczewski KJ, Francioli LC, Tiao G, Cummings BB, Alföldi J, Wang Q, et al. The mutational constraint spectrum quantified from variation in 141,456 humans. Nature. 2020;581.
  21. 21. Mirzaei H, Schmidt KH. Non-Bloom syndrome-associated partial and total loss-of-function variants of BLM helicase. Proc Natl Acad Sci. 2012;109(47):19357–62. pmid:23129629
  22. 22. Shastri VM, Schmidt KH. Cellular defects caused by hypomorphic variants of the Bloom syndrome helicase gene BLM. Mol Genet Genomic Med. 2016;4(1):106–19. pmid:26788541
  23. 23. Voter AF, Killoran MP, Ananiev GE, Wildman SA, Hoffmann FM, Keck JL. A High-Throughput Screening Strategy to Identify Inhibitors of SSB Protein–Protein Interactions in an Academic Screening Facility. SLAS Discov [Internet]. 2017;23(1):94–101. Available from: pmid:28570838
  24. 24. Wu T, Yu J, Gale-day Z, Woo A, Suresh A, Hornsby M, et al. Three Essential Resources to Improve Differential Scanning Fluorimetry (DSF) Experiments. Biorxiv. 2020;1–23.
  25. 25. Manthei KA, Hill MC, Burke JE, Butcher SE, Keck JL. Structural mechanisms of DNA binding and unwinding in bacterial RecQ helicases. Proc Natl Acad Sci. 2015;112(14):4292–7. pmid:25831501
  26. 26. Foster HR, Lin X, Srikant S, Cueny RR, Falbel TG, Burton BM. Natural Transformation Protein ComFA Exhibits Single-Stranded DNA Translocase Activity. J Bacteriol. 2022;204(3):1–14. pmid:35041498
  27. 27. James M, Murtola T, Schulz R, Smith JC, Hess B, Lindahl E. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX. 2015;2(1):19–25.
  28. 28. Webb B, Sali A. Comparative Protein Structure Modeling Using MODELLER. Curr Protoc Bioinforma. 2016;54:116–22.
  29. 29. Dudev T, Lim C. Factors Governing the Protonation State of Cysteines in Proteins: An Ab Initio/CDM Study. J Am Chem Soc. 2002;124:6759–66. pmid:12047197
  30. 30. Dolinsky TJ, Nielsen JE, Mccammon JA, Baker NA. PDB2PQR: an automated pipeline for the setup of Poisson–Boltzmann electrostatics calculations. Nucleic Acids Res. 2004;32(Web Server issue):665–7. pmid:15215472
  31. 31. Hess B. P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation. J Chem Theory Comput. 2008;4:116–22.
  32. 32. Mencel K, Starynowicz P, Siczek M, Jakubas R, Medycki W. Symmetry breaking structural phase transitions, dielectric properties and molecular motions of formamidinium cations in 1D and 2D hybrid compounds: (NH2CHNH2)3[Bi2Cl9] and (NH2CHNH2)3[Bi2Br9]. Dalt Trans. 2019;48:14829–38. pmid:31552956
  33. 33. Bussi G, Donadio D, Parrinello M, Bussi G, Donadio D, Parrinello M. Canonical sampling through velocity rescaling. J Chem Phys. 2007;126(014101):1–7. pmid:17212484
  34. 34. Darden T, York D, Pedersen L. Particle mesh Ewald: An N⋅log (N) method for Ewald sums in large systems in large systems. J Chem Phys. 1993;98(10089):1–5.
  35. 35. Berendsen HJC, Grigera JR, Straatsma TP. The Missing Term in Effective Pair Potentials. J Phys Chem. 1987;91(24):6269–71.
  36. 36. Joung IS, Cheatham TE. Determination of Alkali and Halide Monovalent Ion Parameters for Use in Explicitly Solvated Biomolecular Simulations. J Phys Chem. 2008;112:9020–41. pmid:18593145
  37. 37. Lindorff-Larsen K, Piana S, Palmo K, Maragakis P, Klepeis JL, Dror RO, et al. Improved side-chain torsion potentials for the Amber ff99SB protein force field. Proteins. 2010;78(8):1950–8. pmid:20408171
  38. 38. Yoo J, Aksimentiev A. Improved Parameterization of Amine−Carboxylate and Amine−Phosphate Interactions for Molecular Dynamics Simulations Using the CHARMM and AMBER Force Fields. J Chem Theory Comput. 2016;12(1):430–43. pmid:26632962
  39. 39. Delgado JA, Wineman-Fisher V, Pandit S, Varma S. Inclusion of High-Field Target Data in AMOEBA’s Calibration Improves Predictions of Protein−Ion Interactions. J Chem Inf Model. 2022;62(19):4713–26. pmid:36173398
  40. 40. Leighty RE, Varma S. Quantifying Changes in Intrinsic Molecular Motion Using Support Vector Machines. J Chem Theory Comput. 2013;12(9):868–75. pmid:26588730
  41. 41. Dutta P, Siddiqui A, Botlani M, Varma S. Stimulation of Nipah Fusion: Small Intradomain Changes Trigger Extensive Interdomain Rearrangements. Biophys J [Internet]. 2016;111(8):1621–30. Available from: pmid:27760350
  42. 42. Duro N, Varma S. Role of Structural Fluctuations in Allosteric Stimulation of Paramyxovirus Hemagglutinin-Neuraminidase. Structure [Internet]. 2019;27(10):1601–11. Available from: pmid:31402221
  43. 43. Hornbeck P V, Zhang B, Murray B, Kornhauser JM, Latham V, Psp PR. PhosphoSitePlus, 2014: mutations, PTMs and recalibrations. Nucleic Acids Res. 2015;43:D512–20. pmid:25514926
  44. 44. Zittel MC, Keck JL. Coupling DNA-binding and ATP hydrolysis in Escherichia coli RecQ: role of a highly conserved aromatic-rich sequence. Nucleic Acids Res. 2005;33(22):6982–91. pmid:16340008
  45. 45. Botlani M, Siddiqui A, Varma S. Machine learning approaches to evaluate correlation patterns in allosteric signaling: A case study of the PDZ2 domain. J Chem Phys [Internet]. 2018;148(241726):1–9. Available from: pmid:29960337
  46. 46. Cheok CF, Wu L, Garcia PL, Janscak P, Hickson ID. The Bloom’s syndrome helicase promotes the annealing of complementary single-stranded DNA. Nucleic Acids Res. 2005;33(12):3932–41. pmid:16024743