Biochemical Properties of Naturally Occurring Human Bloom Helicase Variants

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.

. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. 228 Partial dsDNA or G4-dsDNA constructs were folded by incubating 5 µM DNA in 10 mM Tris-HCl, 229 pH 7.5 and 100 mM KCl at 95 °C for 10 minutes and slowly cooling the sample to room 230 temperature over several hours, then stored at 4 °C. Serial dilutions of BLM core or BLM core 231 mutant proteins were incubated with 40 nM oRC32/FAM-oAV320 (partial duplex, 237 (v/v) glycerol, 0.1 mM EDTA) was added to each reaction and 5 µL of each sample was loaded 238 onto a 15% acrylamide 1.5-mm gel in TBE buffer supplemented with 100 mM KCl. Gels were 239 run at 75 V for 1 hour at 4 °C in 1x TBE running buffer with 100 mM KCl. Gels were imaged on 240 an Azure c600 (Azure Biosystems). BLM core dsDNA, BLM core P868L dsDNA and G4-dsDNA, and 241 BLM core K869A K870A dsDNA and G4-dsDNA were done in triplicate. BLM core G4-dsDNA, . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint 243 quantified using ImageJ and analyzed using Prism Version 9.3.1 and fitting data to Equation (1).     290 of mutant BLM core proteins. We used a dT 20 ssDNA substrate for this assay to examine only . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint 292 associated with dsDNA unwinding. Interestingly, both of the BLM core mutant proteins had similar 293 maximum ATPase rates to WT BLM core (Fig 2A, Table 2), indicating that BLM core P868L and 294 BLM core G1120R are able to efficiently hydrolyze ATP when bound to ssDNA. While BLM core 295 P868L had an apparent increase in maximum ATPase rate compared to WT BLM (P value = 296 0.0116), there is a relatively small difference (35%) between the WT BLM core and BLM core P868L 297 maximum ATPase rates (Table 2). Since the concentration of BLM core directly impacts measured 298 maximal ATPase rates, small inaccuracies in concentration determination can lead to modest 299 apparent differences that are difficult to interpret. We therefore do not ascribe biochemical 300 significance to this difference. The DNA concentration-dependence of the ATPase activity was 301 also similar for WT BLM core , BLM core P868L, and BLM core G1120R. The DNA concentrations 302 required for half-maximum ATPase rate (K DNA ) differed by less than 1-fold among all three 303 proteins ( Fig 2B, Table 2). 304 305 BLM core K869A K870A was found to have a modestly different maximum ATPase rate compared 306 to WT BLM core (P value = 0.0348). More interestingly, this mutant protein had a ~4-fold higher 307 K DNA value (P value = 0.0183), indicating that higher concentrations of ssDNA were necessary 308 to stimulate its half-maximum ATPase activity (Fig 2,

330
We also assessed the ATPase rate for each mutant protein in the absence of DNA. WT BLM core , 331 BLM core P868L, and BLM core G1120R were capable of hydrolyzing ATP in the absence of DNA, 332 with rates of 4-10% of that observed with saturating levels of ssDNA. In contrast, BLM core K869A 333 K870A had no detectable DNA-independent ATPase activity ( 342 ssDNA overhang (partial dsDNA), each done in triplicate. WT BLM core bound to ssDNA and the 343 G4 substrate, as evidenced by slower migration of the labeled DNA on a gel (Fig 3). This shift 344 manifested as "smearing" for these substrates, which could indicate that these complexes are 345 dynamic, dissociating and re-associating in the assay. In contrast, when tested with partial . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint 347 stable complex with the substrate. The same trend was observed for BLM core P868L and BLM core 348 G1120R, which were both able to shift ssDNA and G4 DNA, indicated by smearing, and were 349 able to shift partial dsDNA, indicated by distinct band shifts (Fig 3). From estimating qualitative 350 binding affinities from the gel shift assays, WT BLM core , BLM core P868L, and BLM core G1120R all 351 appear to have similar affinities for each of these DNA substrates (  365 366 BLM core K869A K870A was also able to bind to ssDNA and G4 DNA (Fig 3) but required higher 367 protein concentrations to shift the DNA, consistent with a lower affinity for these substrates 368 (Table 2). This binding also resulted in a smear above the substrate migration site. Whereas WT 369 BLM core , BLM core P868L, and BLM core G1120R were able to bind the partial dsDNA causing a 370 distinct band shift, the BLM core K869A K870A mutant protein did not have this effect. Instead, 371 BLM core K869A K870A binding to partial duplex DNA resulted in a smear, similar to what was . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made 382 We next tested the ability of the BLM mutant proteins to unwind dsDNA and G4-dsDNA. This 383 experiment was done using a gel-based helicase assay with a substrate containing a 3' ssDNA 384 15 nucleotide overhang, either the human telomere G4-forming sequence or a control sequence 385 that cannot form a G4, and 18 base pairs of dsDNA (Fig 4A). The annealed strand contains a 386 fluorescent label on its 3' end and unwinding of this substrate can be observed by tracking 387 migration of fluorescent label in a gel (Fig 4B). The reaction is initiated by the addition of BLM core 388 or mutant protein, ATP, and an unlabeled trap oligo that has the same sequence as the 389 fluorescently labeled ssDNA. The total fraction of DNA unwound at various BLM core 390 concentrations is measured (Fig 4C), allowing a determination of the fraction of DNA that 391 BLM core or mutant protein can unwind and the concentration of BLM core required for half-392 maximum unwinding (K BLM ). (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint 449 distances are similar to those in the simulations of the WT form, the DNA-WH distance is larger 450 (Fig 5A), implying that the P868L substitution could weaken interaction between the WH domain 451 and DNA. This was not detected in our biochemical experiments, but this could be due to the 452 qualitative nature of the assay or less of a reliance on WH-DNA interactions than other domains 453 interactions with DNA for BLM core DNA binding.

512
513 Interestingly, BLM core P868L and BLM core G1120R are both active ATPases that have similar 514 maximum ATPase rates to WT BLM core and require similar concentrations of ssDNA for ATPase 515 activity stimulation (Fig 2, Table 2). Additionally, both mutant proteins have ATPase activity 516 even in the absence of DNA, similar to WT BLM core . When assessing the ability of these mutant 517 proteins to bind to ssDNA, G4 DNA, and partial dsDNA, both were able to bind substrates with 518 similar affinities to WT BLM BLM core While binding to ssDNA and G4 DNA was indicated by a 519 smeared complex, binding to the partial duplex DNA induced a distinct band shift. The smearing 520 observed with binding to ssDNA and G4 DNA substrates could indicate weaker overall binding 521 and be caused by dissociation of BLM core from the DNA while running in the gel. The distinct 522 band shift observed from the partial duplex is consistent with WT BLM core , BLM core P868L, and 523 BLM core G1120R binding more stably to the partial duplex than the other DNA constructs.
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint 525 We also assessed the ability of BLM core P868L and BLM core G1120R to unwind dsDNA and G4-526 dsDNA substrates. WT BLM core , BLM core P868L, and BLM core G1120R were able to unwind 527 dsDNA efficiently and required similar amounts of BLM to reach half-maximum unwinding 528 (Table 2). Interestingly, BLM core P868L was able to unwind G4-dsDNA more efficiently than WT 529 BLM core , indicating that this mutant protein is slightly better at unwinding G4s than WT BLM core . 530 This could indicate that BLM core P868L has a tighter binding affinity for G4s than WT BLM core , 531 causing more proficient unwinding. However, direct binding studies did not demonstrate a 532 higher affinity of BLM core P868L for G4s than WT BLM core (Table 2), so other properties such as 533 processivity may be the source for this modest difference. BLM core G1120R is less efficient at 534 unwinding the G4-dsDNA substrate than the dsDNA substrate alone, requiring more BLM core 535 G1120R for half-maximum unwinding. This could be because this mutant protein is less efficient 536 at unwinding G4s than dsDNA or because the BLM core G1120R mutant protein is less 537 processive on longer substrates.

538
539 Overall, the in vitro activities of BLM core P868L and BLM core G1120R are strikingly similar to that 540 of WT BLM core , despite these mutant proteins causing moderate genome instability in cells. 547 Mutations that impact phosphorylation of BLM can decrease the ability of cells to recover from 548 hydroxyurea exposure and other DNA damage (1). Residues 868 and 1120 are not known sites 549 of PTMs, making it unlikely that this is the reason that these mutant proteins cause increased . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint 551 which is part of the lysine rich loop proximal to P868 is a reported site of SUMOylation and 552 ubiquitination (40). Because P868 terminates a β-strand at the start of the lysine-rich loop, it is 553 possible that the mutation to leucine changes the structure of the loop and disrupts recognition 554 of K873 by cellular PTM enzymes. This could result in decreased BLM activity in cells but more 555 limited impact on in vitro activity of BLM. 556 557 Another possibility is that these mutations lead to increased SCEs in cells by producing BLM 558 mutant proteins with disrupted dHJ dissolution activity and/or altered interaction with other 559 proteins. BLM interacts with over 20 different proteins, including other dissolvasome proteins 560 that are involved in dissolving dHJs (1,3,12). If the BLM dissolvasome function is impaired due 561 to weaker interactions between BLM and Top3α, RMI1, or RMI2 because of these mutations, 562 this could lead to increased genome instability. Additionally, if BLM mutations lead to a 563 decreased affinity for HJs, this could be detrimental for BLM function in vivo. The G1120 residue 564 is important for maintaining the WH α2-α3 loop and is conserved among many RecQ family 565 helicases (21). Interestingly, substitutions in the α2-α3 loop, S1121A and K1125A, have been 566 shown to induce a modest decrease in binding affinity for HJs in vitro, indicating that this loop 567 could be important for HJ binding and recognition (41). While BLM G1120R functions similarly to 568 WT BLM in our assays, the G1120R substitution could cause a decrease in BLM binding affinity 569 for HJs. Thus, BLM G1120R might lead to decreased dissolvasome activity, and subsequently, 570 increased SCEs in cells. The weakened interaction between DNA and the WH domain observed 571 in the MD simulations could contribute in a similar manner to the partial loss of cellular function 572 of BLM P868L. Additionally, the change in orientation of the RecA-D1 lobe, which affects the 573 catalytic cleft of the BLM helicase core, as well as subtle conformational changes in conserved 574 helicase motifs V and VI (Supplemental Fig S2), could contribute to the functional deficits of . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted January 26, 2023. ; https://doi.org/10.1101/2023.01.26.525669 doi: bioRxiv preprint