Bloom syndrome DNA helicase deficiency is associated with oxidative stress and mitochondrial network changes

Bloom Syndrome (BS; OMIM #210900; ORPHA #125) is a rare genetic disorder that is associated with growth deficits, compromised immune system, insulin resistance, genome instability and extraordinary predisposition to cancer. Most efforts thus far have focused on understanding the role of the Bloom syndrome DNA helicase BLM as a recombination factor in maintaining genome stability and suppressing cancer. Here, we observed increased levels of reactive oxygen species (ROS) and DNA base damage in BLM-deficient cells, as well as oxidative-stress-dependent reduction in DNA replication speed. BLM-deficient cells exhibited increased mitochondrial mass, upregulation of mitochondrial transcription factor A (TFAM), higher ATP levels and increased respiratory reserve capacity. Cyclin B1, which acts in complex with cyclin-dependent kinase CDK1 to regulate mitotic entry and associated mitochondrial fission by phosphorylating mitochondrial fission protein Drp1, fails to be fully degraded in BLM-deficient cells and shows unscheduled expression in G1 phase cells. This failure to degrade cyclin B1 is accompanied by increased levels and persistent activation of Drp1 throughout mitosis and into G1 phase as well as mitochondrial fragmentation. This study identifies mitochondria-associated abnormalities in Bloom syndrome patient-derived and BLM-knockout cells and we discuss how these abnormalities may contribute to Bloom syndrome.


Results
Increased oxidative stress, DNA damage, and ROS-induced retardation of DNA replication speed in BLM-deficient fibroblasts. To evaluate oxidative stress in BLM-deficient cells we first assessed levels of reactive oxygen species (ROS) production. In addition to analyzing a Bloom-syndrome-patient-derived ). By flow cytometry, we observed an increase in cellular ROS levels in KSVS1452 (BLM KO ) using dihydroethidium (DHE) staining, which detects superoxide and hydrogen peroxide (Fig. 1A). In addition, we assessed cellular ROS using 2′,7′-dichlorofluorescin diacetate (DCFA), which is activated by most reactive oxygen species in cells. Both, the patient-derived GM08505 cells and the BLM-knockout KSVS1452 cells showed significantly increased ROS levels compared to the BLM-proficient GM00637 cells whereas BLM-complemented KSVS1454 cells showed normal ROS levels ( Fig. 1B). Flow cytometric analysis with MitoSOX-Red showed elevated levels of mitochondrial superoxide, a major intracellular source of ROS, in patient-derived BLM −/− cells and in BLM KO cells whereas reintroduction of wildtype BLM cDNA into KSVS1452 (KSVS1454, BLM KO/+ ) reversed mitochondrial superoxide ROS levels to normal (Fig. 1C). This was consistent with fluorescence microscopy using MitoSOX staining, which showed increased formation of fluorescent puncta in the BLM-deficient cell lines (Fig. 1D).
One of the well-characterized base lesions induced by ROS that is involved in DNA mutagenesis is oxidized guanine nucleoside (8-oxo-dG) 28 . Using immunofluorescence microscopy and flow cytometry, we found that GM08505 (BLM −/− ) and KSVS1452 (BLM KO ) cells had significantly higher levels of 8-oxo-dG, indicating that BLM-deficiency increases oxidative base damage (Fig. 1E,F).
BLM-deficient cells do not exhibit gross cell cycle defects (Supplemental Fig. S2A), but they grow more slowly (Supplemental Fig. S1F) and show a slight accumulation of G1-and S-phase cells (Supplemental Fig. S2B). As ROS was recently also shown to affect DNA replication fork velocity 9,29 we assessed whether ROS contributes to the decreased DNA replication speed observed in BLM-deficient cells 8,30 . Indeed, treatment of KSVS1452 (BLM KO ) cells with the antioxidant N-acetyl-cysteine (NAC) significantly reduced cellular ROS (Fig. 1G) and increased DNA replication speed to levels seen in BLM-proficient cells (Fig. 1H). In contrast, NAC had no effect on DNA replication speed in BLM-proficient cells (Fig. 1H). This suggests that elevated ROS contributes to the slowing of DNA replication in BLM-deficient cells.

Increased mitochondrial mass and TFAM upregulation in BLM-deficient cells. Quantitative
PCR showed that mtDNA content is significantly increased in BLM-deficient cells (Fig. 2C). NAO, a fluorescent dye that localizes to cardiolipin in the inner mitochondrial membrane in a membrane-potential-independent manner 31 , also showed increased staining of BLM-deficient cells by fluorescence microscopy ( Fig. 2A) and flow cytometry (Fig. 2B). Both, GM08505 (BLM −/− ) and KSVS1452 (BLM KO ) cells, also showed increased staining with MitoTracker CMXRos, which measures mitochondrial mass in a membrane-potential-dependent manner 31 (Fig. 2D). Together, these findings suggest increased mitochondrial mass with intact membrane potential in BLM-deficient cells.  Fig. S1G). Scale bars 10 µm. (B) Flow cytometry analysis of mitochondrial staining of GM00637 (BLM +/+ ) and BLM-deficient KSVS1452 (BLM KO ) cells with NAO. Analysis was performed in triplicate and mean ± SD is shown. (C) Relative mitochondrial DNA content measured by qPCR using primers against mitochondrial gene COX1 and nuclear gene 18S. Analysis was performed in triplicate and mean ± SD is shown. (D) Flow cytometry analysis of mitochondrial staining of GM00637 (BLM +/+ ) and BLM-deficient KSVS1452 (BLM KO ) and GM08505 (BLM −/− ) cells by membrane-potential-dependent Mitotracker Red CMXRos. Analysis was performed in triplicate and mean ± SD is shown. (E) TFAM mRNA levels in BLM-proficient (GM00637) cells and BLM-deficient (KSVS1452) cells were determined by qRT-PCR.  www.nature.com/scientificreports/ Mitochondrial transcription factor A (TFAM) binds to mtDNA, packaging it into nucleoids, and acts as a major factor in regulating mitochondrial transcription, replication and biogenesis 32,33 . TFAM was upregulated fourfold in KSVS1452 (BLM KO ) cells (Fig. 2F) and 2.5-fold in Bloom-syndrome-patient-derived GM08505 (BLM −/− ) cells (Fig. 2G), consistent with increased mitochondrial mass in these cells. Messenger RNA levels indicate transcriptional upregulation of TFAM in the absence of BLM (Fig. 2E). Immunofluorescence microscopy further demonstrated increased TFAM staining in BLM-deficient fibroblasts (Fig. 2H, Supplemental Fig. S3A) that was reduced to wildtype levels upon reintroduction of BLM cDNA (KSVS1454, Fig. 2H), verifying BLMdependent changes in TFAM levels. The increased amount of TFAM found in BLM-deficient cells localized normally to mitochondria (Fig. 2I).
Levels of Nrf1, a key activator of TFAM expression 34 , were slightly upregulated in BLM-deficient KSVS1452 cells whereas Nrf2, which also contributes to TFAM induction 34 , did not change significantly (Supplemental Fig. S3B,C), suggesting that oxidative stress regulators make some contribution to TFAM upregulation in BLMdeficient cells. Besides Nrf1/Nrf2, Myc also binds to the TFAM gene and upregulates its expression 35 . Interestingly, c-myc levels are increased in BLM-deficient cell lines and BLM negatively regulates c-myc by promoting c-myc www.nature.com/scientificreports/ degradation through the E3 ubiquitin ligase FBW7 [36][37][38][39] . Thus, increased myc levels may provide an additional mechanism for TFAM upregulation in BLM-deficient cells.
Increased mitochondrial mass is not due to defective autophagy. An increase in mitochondrial mass could reflect changes in mitophagy 40 . To determine whether autophagic activity is impaired in cells lacking BLM, we measured changes in the levels of cytosolic LC3-I and autophagosomal marker LC3-II in the presence of the mitochondrial oxidative phosphorylation uncoupler CCCP and treatment with hydroxychloroquine (HCQ), which blocks autophagosome-lysosome fusion 41 . Treatment with CCCP induces autophagy and conversion of LC3-I to membrane-bound LC3-II whose accumulation can be visualized by blocking autophagic flux with HCQ. We observed that upon HCQ treatment LC3-II readily accumulated in both BLM-proficient GM00637 cells and BLM-deficient KSVS1452 cells, indicating normal autophagic flux (Fig. 3A). We further analyzed autophagic flux by immunofluorescence microscopy using double staining with antibodies against LC3 (autophagosome) and COXIV (mitochondria) (Fig. 3B). LC3 is recruited to the membranes of autophagosomes forming punctate structures. The number of LC3 puncta increased in BLM-proficient and BLM-deficient cells after HCQ exposure, indicating functional autophagy in both cell lines. LC3 foci colocalized with mitochondria in both cell lines; the slight increase in colocalization seen in the BLM-deficient KSVS1452 cells (Fig. 3C) could be due to the increased mitochondrial mass in these cells or indicate increased mitophagy. Finally, after treatment of BLM-proficient and BLM-deficient cell lines with the mitochondrial oxidative phosphorylation uncoupler CCCP, mitochondria localized to lysosomes (Fig. 3D). Overall, these findings indicate that mitophagy is not impaired in cells lacking BLM.
BLM deficiency is associated with mitochondrial fragmentation. Since BLM deficiency is associated with increased mitochondrial mass we evaluated the effect of BLM loss on the regulation of mitochondrial network dynamics. Throughout most of the cell cycle, mitochondria form a filamentous network, which is fragmented at the transition into mitosis to allow for the distribution of mitochondria into the daughter cells, linking mitochondrial fission to cell division 42 . On performing fluorescence microscopy with Mitotracker Red CMXRos, we observed that the mitochondrial network in BLM-deficient cells was fragmented (Fig. 4A). Specifically, mitochondria in BLM-deficient cells were significantly smaller, more circular, and less branched than mitochondria in BLM-proficient GM00637 cells (Fig. 4B). Electron microscopy showed no swelling or other gross abnormalities in the inner structure of mitochondria of BLM-deficient cells (Fig. 4C). Mitochondrial fission is mediated by the dynamin-like GTPase Drp1 whose phosphorylation at serine 616 leads to self-assembly into spirals that constrict the mitochondrial tubules, leading to fission 43 . BLM-deficient KSVS1452 cells exhibited a significant increase in Drp1 levels, which returned to normal upon reintroduction of BLM cDNA (KSVS1454) (Fig. 4D). However, fusion proteins Opa1 and Mfn1 were also upregulated, suggesting that mitochondrial fission and fusion protein levels are generally increased in BLM-deficient cells due to increased mitochondrial mass (Fig. 4D). Drp1 fission activity is regulated by phosphorylation, including at the transition to mitosis where it mediates mitochondrial fission prior to cell division 44 . To assess possible differences in Drp1 activity in BLM-deficient cells, we blocked cells at G2/M with a single thymidine arrest followed by nocodazole block and assessed Drp1-S616 phosphorylation after release into mitosis and the subsequent G1 phase over an 8-h time course (Fig. 4E). As expected, Drp1 was rapidly dephosphorylated after entry into mitosis in wildtype cells; however, in BLM KO cells Drp1 phosphorylation persisted throughout mitosis and into G1 phase. These findings suggest that persistent Drp1 activation could be responsible for the mis-timed mitochondrial fragmentation in KSVS1452 (BLM KO ) cells.

Unscheduled Cyclin B1 expression in BLM-deficient cells in late mitosis and G1.
One of the key phosphorylation events that activates Drp1 is mediated by cyclin B1/CDK1 during transition into mitosis 45 .
To determine if persistent Drp1 phosphorylation in BLM-deficient cells is associated with changes in levels of cyclin B1/CDK1, we blocked GM00637 (BLM +/+ ) and KSVS1452 (BLM KO ) cells at G2/M with nocodazole, collected samples for 6 h after release and assessed degradation of cyclin B1 (Fig. 5A). Normally, cyclin B1 is highly expressed in late S and G2 phase to form the cyclin B1/CDK1 kinase complex where its function during the G2/M transition, along with Drp1 phosphorylation, is to target and phosphorylate all five-major multi-subunit respiratory complexes of the electron transport chain. This phosphorylation corresponds with a dramatic increase in mitochondrial respiration and ATP production, which allows cells to overcome the large energy expenditure of mitosis 46 . Cyclin B1 is then degraded as cells progress through mitosis following the inactivation of the spindle assembly checkpoint and entry into anaphase 47 . Here, we observed that cyclin B1 was degraded with the expected kinetics in the GM00637 (BLM +/+ ) cell line, whereas even after 6 h following the release from nocodazole block, cyclin B1 levels remained unchanged in BLM-deficient KSVS1452 (BLM KO ) cells (Fig. 5A). In contrast, the degradation pattern of cyclin A, which is also degraded during mitosis 47 , was similar in both cell lines, indicating proper re-entry into the cell cycle after release from nocodazole block (Fig. 5A). To verify that sustained cyclin B1 levels were associated with the loss of BLM, we assessed cyclin B1 degradation in the BLMcomplemented KSVS1454 (BLM KO/+ ) cell line. Indeed, cyclin B1 degradation in KSVS1454 cells returned to the normal pattern seen in GM00637 (BLM +/+ ) cells (Fig. 5A). We confirmed normal mitotic re-entry and progression into G1 after release from nocodazole by flow cytometry in all three cell lines (Supplemental Fig. S4A).
To eliminate the possibility that the prolonged presence of cyclin B1 in the BLM-deficient cells stemmed from inefficient release into G1 phase or that it was related to the synchronization procedure we sorted asynchronous GM00637 (BLM +/+ ), KSVS1452 (BLM KO ) and KSVS1454 (BLM KO/+ ) cultures into G1 and G2/M populations based on DNA content by flow cytometry (Supplemental Fig. S4B) and analyzed for cyclin B1 staining by immunofluorescence microscopy (Fig. 5B) 5D). We confirmed unscheduled cyclin B1 expression in G1 phase in KSVS1453 (BLM KO ) cells, a second independent clone of CRISPR/Cas9-mediated BLM disruption in GM00637 (Fig. 5C,D). Thus, cyclin B1 escapes mitotic degradation and is abnormally expressed during G1 phase in BLM-deficient cells. This abnormal cyclin B1 expression pattern in mitosis and in G1 is fully reversible by reintroducing BLM. Cyclin B1 localizes to mitochondria during the G2/M transition to upregulate ATP production for mitosis 46 . The vast majority of cyclin B1 untimely expressed during G1 in KSVS1452 (BLM KO ) cells also localized to mitochondria, as indicated by overlap with the mitochondria-specific protein COXIV, whereas some cyclin B1 formed non-mitochondrial foci (Fig. 5E).

Bioenergetic profile of BLM-deficient cells.
To evaluate the function of the mitochondria of BLMdeficient cells we performed respiratory assays using well-defined mitochondrial inhibitors (Fig. 6A). We found no significant difference in basal oxygen consumption rate (OCR) between the BLM-proficient GM00637 cells and the two isogenic, independent BLM-knockout cell lines KSVS1452 and KSVS1453 (Fig. 6B). This allowed us to normalize ATP turnover, reserve respiratory capacity and maximal respiratory capacity to basal respiration to decrease growth-rate-related variability. ATP turnover as assessed by adding oligomycin, an inhibitor of complex V (ATP synthase) of the electron transport chain, demonstrated equimolecular oxygen use for production of ATP by oxidative phosphorylation in the BLM-proficient and BLM-deficient cells (Fig. 6C). Next, we added FCCP, an uncoupler that disrupts the mitochondrial membrane potential, to obtain the maximum oxygen consumption rate and the respiratory reserve capacity (Fig. 6D,E). Reserve respiratory capacity is the theoretical extra capacity the electron transport chain has to drive oxidative phosphorylation under situations of ATP need. Interestingly, the two BLM-deficient cell lines had a 30% higher reserve respiratory capacity than the isogenic BLM-proficient cell line (Fig. 6E). Taken together with identical ATP turnover rates in the BLM-proficient and BLM-deficient cell lines, this indicates that compared to the BLM-proficient cells the energy expenditure of BLM-deficient cells is lower than their rate of ATP synthesis. This is also reflected in the significantly higher (2.5-fold) cellular ATP content of the BLM-deficient cell lines compared to the isogenic BLM-proficient cell line (Fig. 6G) and an ADP/ATP ratio significantly shifted towards ATP in the BLM-deficient cell lines compared to the BLM-proficient cell line (Fig. 6H). Finally, we observed a trend towards a higher glycolytic reserve in BLMdeficient cells, which is a measure of the compensatory increase in glycolysis following an inhibition of the ATP synthase (Fig. 6F). Together, these findings indicate that BLM-deficient cells have a higher ability to produce ATP by oxidative phosphorylation than BLM-proficient cells under energy-requiring conditions and that they may be able to compensate for a loss of oxidative phosphorylation better than BLM-proficient cells. Overall, our analysis of mitochondrial function suggests that the BLM-deficient cell lines are not energetically impaired under the tested conditions and have a larger potential to produce ATP than the BLM-proficient cell line.

Discussion
In this study, we have identified abnormalities in Bloom-syndrome-patient-derived cells and BLM-knockout cells that include increased ROS that causes DNA base damage and reduced DNA replication speed, increased mitochondrial mass and TFAM expression, and mitochondrial fragmentation. BLM-deficient cells also had higher ATP content and increased reserve respiratory capacity, but an otherwise unremarkable bioenergetic profile that was similar to the isogenic BLM-proficient parental cell line. Although the cause of mitochondrial network fragmentation in BLM-deficient cells is unknown, it is associated with untimely cyclin B1 expression into G1 phase and persistent activation of Drp1 at S616, which promotes mitochondrial fission at mitosis entry. Upregulation of TFAM is a characteristic marker of many different cancer tissues and is associated with malignant progression and poor prognosis [48][49][50][51][52] , raising the possibility that it also contributes to cancer risk in  (Fig. S1G). Whole cell extracts were prepared from three independent cultures and quantification was performed using ImageJ (version 1.53a; freely available at http://image J.nih.gov/ij). Ran    www.nature.com/scientificreports/ Bloom syndrome. In addition to its role in cell proliferation and mitochondrial function, TFAM has been shown to regulate apoptosis. In certain types of cancer, TFAM binds to the promoters of apoptosis-related genes and regulates their expression 53 . TFAM depletion induces p21-dependent G1 arrest and p21-deficient cells exhibit elevated TFAM expression levels 53,54 . Thus, TFAM upregulation in BLM-deficient cells might also provide a mechanism to stimulate cell proliferation and prevent apoptosis to survive the defects caused by the absence of BLM activity. In addition to increased TFAM expression and mitochondrial mass, we observed increased production of mitochondrial ROS in BLM-deficient cells. At low non-toxic concentrations, ROS play a role in signal transduction, but at higher levels they cause oxidative stress characterized by DNA base oxidation, lipid peroxidation and disruption of cellular redox status 38 . Unlike the human RecQ-helicase family member RecQ4, BLM does not localize to mitochondria 55 . Instead, increased ROS in BLM-deficient cells could be the result of the increased mitochondrial mass, but could also indicate changes in the arrangement of electron transport chain complexes in the mitochondrial membrane as a result of increased mitochondrial fragmentation. Indeed, increased ROS production has been shown to coincide with increased mitochondrial fragmentation [56][57][58][59] . Local cellular concentrations of ROS can lead to translocation of mitochondria to distinct cellular regions for signaling functions 60 . For example, the superoxide generating chemical MEN causes mitochondrial redistribution from the cell periphery to the perinuclear region 61 . Similarly, hypoxia causes perinuclear mitochondrial clustering 60 . Thus, high levels of ROS in BLM-deficient cells might contribute to the increased mitochondrial fragmentation and perinuclear clustering that we observed. Interestingly, such clustering of mitochondria around the nuclear periphery increases ROS levels in the nucleus 60 and may contribute to the oxidative DNA damage we observed in BLM-deficient cells here as well as to the reduced DNA replication speed in Bloom syndrome patient cells previously reported 8,30 .
Remarkably, we found that normal DNA replication speed can be restored in Bloom syndrome cells by treating cells with an antioxidant, whereas antioxidant treatment had no effect on DNA replication speed in BLM-proficient cells. ROS could decrease DNA replication speed by generating DNA lesions that block DNA polymerases 62 , by oxidizing proteins involved in replication [63][64][65][66] , or by reducing the nucleotide pool 9 . Notably, replication fork velocity was recently shown to be coupled to ROS signaling through the fork accelerator TIME-LESS-TIPIN 29 . As shown here for endogenous ROS-induced replication slowdown in BLM-deficient cells, NAC reverses slow replication speed induced by exposure to exogenous H 2 O 2 whereas fork slowdown due to other causes, such as treatment with aphidicolin, is not affected by NAC 29 . Besides changing replication dynamics, increased nuclear ROS also induces transcriptional changes 67 , which could contribute to the risk of malignant transformation of Bloom syndrome cells.
The maintenance and networking of mitochondria is mediated by mitochondrial fusion and fission events and disruption of this balance is associated with developmental defects, neurodegeneration, metabolic diseases and aging 44,[68][69][70][71][72] . In BLM-deficient cells, increased mitochondrial fragmentation was associated with persistent phosphorylation of the fission protein Drp1 whereas mitophagy appeared normal. The major phosphorylation event that activates Drp1 during the G2/M transition occurs at serine 616 by cyclin B1/CDK1 and links mitochondrial fission to cell division 42,45 . BLM-deficient cells failed to fully degrade cyclin B1 at anaphase and inappropriately accumulate cyclin B1 during G1, which is mostly localized to mitochondria. It is unclear how much of the constitutive mitochondrial fragmentation phenotype exhibited by BLM-deficient cells can be attributed to unscheduled cyclin B1 expression and oxidative-stress-induced fragmentation by Drp1 42,45 . However, any aberration in mitochondrial fission kinetics is likely to be problematic for proliferating cells. Mitochondrial fragmentation is observed in tumor cells and is associated with increased invasiveness and metastatic potential 73,74 . Drp1-induced mitochondrial fragmentation, such as seen here in BLM-deficient cells, also promotes tumor growth 75 . Mitochondrial dysfunction and mitochondrial fragmentation are also associated with insulin resistance, pancreatic β-cell dysfunction and development of Type 2 diabetes 70,76,77 . While we currently have no evidence that the mitochondrial abnormalities in BLM-deficient cells are associated with a loss of mitochondrial function, it is worth noting that ~ 17% of persons with Bloom syndrome develop Type 2 diabetes at a young age 78 .
An important function of the cyclin B1/CDK1 kinase complex during the G2/M transition is to target and phosphorylate all five major multi-subunit respiratory complexes of the electron transport chain to increase www.nature.com/scientificreports/ ATP production for mitosis 46,79 . It is plausible that the unscheduled mitochondrial cyclin B1 expression in G1 performs the same function and contributes to the increase in total ATP content in BLM-deficient cells. Since the G1/S and G2/M transition checkpoints are energy sensitive, it will be interesting to determine in future studies whether BLM-deficient cells require these increased levels of ATP for survival and if they compromise cell cycle processes and genomic stability. Unscheduled cyclin B1 expression during G1 is observed in a growing number of cancer cell lines as well as in patient-derived primary cells from a broad range of tumor types 80,81 . The mechanisms underlying unscheduled appearance of cyclin B1 in G1 or its ability to escape normal degradation during the metaphase to anaphase transition have yet to be identified. However, cyclin B1 mRNA and translation appear to persist after mitosis, suggesting that rapid proteolysis is the more important mechanism for clearing cyclin B1 during G1 phase 82 . Similarly, in yeast, cyclin B1 degradation continues into G1 83 . It is unclear why this degradation process is impaired in the absence of BLM. One possibility is that the mitochondrial location of unscheduled cyclin B1 in G1 phase makes it less accessible to the G1 phase degradation system. This also raises the possibility that unscheduled cyclin B1 expression in late mitosis and in G1 phase is linked to the mitochondrial abnormalities in BLM-deficient cells. Indeed, not only is abnormal cyclin B1 expression associated with tumorigenesis, evidence for a link between the mitochondrial fragmentation phenotype and promotion of tumor growth is also increasing 75,[84][85][86] .
Determining the bioenergetic profile of BLM-deficient cells revealed that their mitochondrial function was not impaired. On the contrary, their ATP levels were increased 2.5-fold and respiratory reserve capacity was increased by 30% compared to the isogenic BLM-proficient cell line. These differences suggest that BLM-deficient cells use less ATP than they produce, possibly due to their slower growth rate (Supplemental Fig. S1F), and have a higher capacity to respond to events of higher energy demand. Although we do not yet know the source of the increased respiratory reserve capacity in BLM-deficient cells, it is generally associated with increased cell survival under challenging conditions and may present a mechanism by which BLM-deficient cells cope with high levels of genome instability.
Even though it has been more than 60 years since Bloom syndrome was first described, the molecular mechanisms underlying some of its features have remained unclear 1 . As genome instability, cellular aging, degeneration and cancer predisposition have been linked to ROS, we propose that endogenous ROS overproduction, mitochondrial network abnormalities, especially as they relate to fission and fusion, and unscheduled cyclin B1 expression may contribute to the development of some symptoms of Bloom syndrome. Further studies will be needed to assess the potential pathogenicity of the mitochondrial abnormalities and cyclin B1 dysregulation and to determine if mitochondrially targeted antioxidants and free radical scavengers could serve as a prophylactic strategy to delay the onset of clinical symptoms in persons with Bloom syndrome. Inhibitors of mitochondrial fission are also being investigated as therapeutics for diseases with a mitochondrial fission phenotype and oxidative stress 87 . In addition to the well-understood role of BLM in homologous recombination and the DNA damage response, this study suggests that BLM may have additional roles in cell cycle processes and mitochondrial biogenesis through which it contributes to normal cell function.

Materials and methods
Cell culture and stable cell lines. SV40-transformed normal skin fibroblast cell line GM00637 (BLM +/+ ) and Bloom syndrome patient fibroblast cell line GM08505 (BLM −/− ) were obtained from Coriell Cell Repository. The biallelic BLM-knockout cell line KSVS1452 (BLM KO ) was derived from the GM00637 using a CRISPR/Cas9 mediated genome engineering protocol 88 . Guide RNA sequences targeting exon 8 of BLM gene were designed using a CRISPR guide design tool (http://tools .genom e-engin eerin g.org), cloned into pSpCas9 (BB)-2A-Puro and used for transfection of GM00637. BLM-knockout clones were screened by immunoblotting and biallelic gene disruptions verified by sequencing of both BLM alleles (KSVS1452 BLM allele 1: BLM c.del2040_2046; g.del90,763,123_90,763,130; BLM p.Leu681LysTer686; BLM allele 2: BLM c.delins2036_2041AG; g.delins90,763 ,119_90,763,124AG; BLM p.Leu678AlaTer686). The cellular phenotype of KSVS1452 was confirmed to resemble that of Bloom-syndrome-patient-derived cells as we have described before 89 (Supplemental Fig. S1). KSVS1453 is a second, independent clone obtained by CRISPR/Cas9-mediated disruption of BLM exon 8. KSVS1452 was transfected with a plasmid expressing wildtype BLM cDNA (pcDNA3-BLM) and a stable clone selected to yield KSVS1454 (BLM KO/+ ). BLM expression levels in KSVS1454 corresponding to those of GM00637 cells were verified by Western blot (Supplemental Fig. S1G). Cells were grown in Modified Eagle's Medium (MEM) supplemented with 10% Fetal Bovine Serum (FBS), 2 mM l-glutamine and 1% penicillin-streptomycin-glutamine (PSG) at 37 °C in the presence of 5% CO 2 . Immunofluorescence microscopy. Cells cultured on glass coverslips (Corning) were washed with PBS and fixed using 4% PFA and permeabilized using 0.25% Triton X-100. For imaging after sorting, cells were  Fluorescence activated cell sorting. To determine DNA content cells were washed twice with sample buffer (0.001% glucose in PBS, filtered through a 0.22 μm filter), resuspended at a concentration of 10 6 cells/ml, and fixed in 70% ethanol at 4 °C overnight. Fixed cells were stained with propidium iodide (50 µg/ml in sample buffer with 100 Kunitz units/ml RNase A) for 1 h at room temperature. Fluorescence was measured by a BD FACSCantoII flow cytometer and analyzed using BD FACSDiva and FlowJo v. 10.7 software (BD Life Sciences, https ://www.flowj o.com/solut ions/flowj o/downl oads). For cell sorting, asynchronous cells were washed and processed as described above. Fixed cells were sorted into G1 and G2/M populations based on DNA diploidy using a BD FACSMelody cell sorter.

Immunoblotting.
Measurements of mitochondrial respiration. Mitochondrial respiration was determined in realtime using an XF-96 Extracellular Flux Analyzer (Seahorse Bioscience, Agilent). Cell lines GM00637 (BLM +/+ ), KSVS1452 (BLM KO ) and KSVS1453 (BLM KO ) were seeded overnight in Seahorse assay media (Seahorse Bioscience, Agilent), supplemented with 1 mM pyruvate and adjusted to pH 7.4. Oxygen consumption rate (OCR) was measured to establish a baseline of respiratory rate. Subsequently, wells were injected with either 1 µM oligomycin or 0.5 µM carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone (FCCP) to determine respectively ATP turnover (drop in OCR after addition of oligomycin) and reserve respiratory (increase in OCR after addition of FCCP). Finally, 2 µM antimycin A was added to all wells as a control to measure level of non-mitochondrial oxygen consumption. Extracellular acidification rates (ECAR) was determined to establish a measure of glycolytic reserve calculated as change in ECAR due to addition of 1 µM oligomycin. As the different cell lines had different growth rates (Supplemental Fig. S1F) all measures of OCR and ECAR were normalized to basal respiration rate (third measurement, at 22 min).
Real-time PCR. Total RNA was extracted from cells using the Quick-RNA Miniprep kit (Zymoresearch).
Following this, 1 µg of RNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription kit (Thermo Fisher). Real-time PCR was performed using SYBR-Green PCR master mix (Bio-Rad) and an ABI prism 7500 sequence detection system (Applied Biosystems). TFAM and human β-actin primers were designed www.nature.com/scientificreports/ with Primer3 92 . Values were normalized against β-actin and relative expression was calculated using the deltadelta Ct method.
DNA fiber assay. DNA replication rates were measured by labeling DNA fibers based on a protocol described earlier 93,94 . Briefly, cells at ~ 50 to 70% confluency were sequentially labeled with 20 μM CldU (Sigma) for 30 min and 100 μM IdU (Sigma) for 30 min. Cells were then resuspended in PBS at a concentration of 1 × 10 6 cells/ml. Labeled cells diluted with unlabeled cells were allowed to air-dry on a microscopy slide and incubated with 10 μl of lysis buffer (0.5% SDS, 200 mM Tris-HCl, pH 7.4, 50 mM EDTA). Slides were inclined at 15° to stretch the fibers. DNA spreads were fixed by incubation with 3:1 methanol:acetic acid followed by denaturation with 2.5 N HCl for 70 min. Following this, the slides were blocked with 10% goat-serum/PBS-T (PBS+ 0.1% Triton X-100) for 1 h. This was followed by incubation with rat anti-BrdU (anti-CldU; Abcam) and mouse anti-BrdU(anti-IdU; Becton Dickinson) at a dilution of 1:100 and 1:50 respectively in 10% goat-serum/ PBS-T(PBS+ 0.1% Triton X-100) for 2 h. Slides were washed with PBS and incubated with AlexaFluor 594 goat anti-rat IgG (H+L) (Invitrogen) and goat anti-mouse (H+L) AlexaFluor 488 Plus (Invitrogen) in the dark for 1 h. Images were acquired using a Keyence BZ-X Fluorescence microscope with a CFI Achromat 60×/0.8 objective.
Electron microscopy. For transmission electron microscopy (TEM) imaging, BLM-proficient GM00637 and BLM-deficient KSVS1452 human fibroblast cell lines were prepared using high-pressure freezing/freeze substitution fixation (HPF/FS) method. The cells were fixed in 4% paraformaldehyde and 0.5% glutaraldehyde in 0.