Category Archives: Transhuman News

In the world of microbial warfare, sometimes you have to change the very fabric of who you are.

Viruses that infect bacteria fittingly called bacteriophages - and their prey have been at war for eons, each side evolving more devilish tactics to infect or destroy each other. Eventually, some bacteriophages took this arms race to a new level by changing the way they code their DNA.

At least, that's what we think happened.Once thought to be an outlier, new research published in three separate papers shows that there's a whole army of bacteriophages with non-standard DNA, which researchers call a Z-genome.

"Genomic DNA is composed of four standard nucleotides These nucleobases form the genetic alphabet, ATCG, which is conserved across all domains of life," biologists Michael Grome and Farren Isaacs write in a recent Scienceeditorialaccompanying the new research on bacteriophage genetics.

"However, in 1977, the DNA virus cyanophage S-2L was discovered with all instances of 'A' substituted with 2-aminoadenine (Z) throughout its genome forming the genetic alphabet ZTCG."

The reason appeared to be self-protection. Within the connecting 'rungs' of a DNA double helix, the 'Z' base forms a triple bond to the opposite 'T' base, one more than the two bonds of the regular A:T connection. This makes the viral genome hardier and more difficult for bacteria to prise apart with chemicals called nucleases.

Although scientists were fascinated, no other bacteriophages were found with the Z-genome, and with the difficulty of culturing S-2L in a lab, the Z-genome was set aside as a curiosity.

Now, research documented in three separate studies from researchers in France and China shows that this was not a one-off, whilst also characterizing how the Z-genome works and how it's assembled.

"Scientists have long dreamed of increasing the diversity of bases. Our work shows that nature has already come up with a way to do that," one of the teams, led by first author Yan Zhou from Tianjin University, wrote in their paper.

Zhou's team, along with another group led by Institut Pasteur microbiologist Dona Sleiman, found two major proteins which they called PurZ and PurB; these make up the 'Z' base.

A third group, led by Universit Paris-Saclay synthetic biologist Valerie Pezo, corroborated those findings and analysed an enzyme called DpoZ which is responsible for assembling the whole Z-genome together.

All three searched genetic sequence databases for the sequences relating to their proteins and enzymes, and found a wide variety of bacteriophages with similar genes.

"[The authors] have done an amazingly comprehensive job of showing that this is not one crazy outlier, but there's a whole group of bacteriophages that have this kind of genetic material," Jef Boeke, a molecular biologist at New York University who was not involved in the work, told The Scientist.

There are still plenty of questions to answer about the Z-genome.

For example, is a Z-genome compatible with regular cell machinery such as ours? And could it be used in the same way that artificial DNA is starting to be?

"The Z base has been unambiguously identified in a carbonaceous meteorite and proposed as a nucleobase that could have been available for the origin of life," the team lead by Zhou wrote in their paper.

"Considering that the Z base was discovered in a meteorite, our work may spark interest in interdisciplinary research on the origins of life and astrobiology."

The three papers have been published in Sciencehere, here, and here.

Read the original:
Some Viruses Have a Completely Different Genome to The Rest of Life on Earth - ScienceAlert

Significance

Medical visits and missed days of school and work caused by rhinoviruses cost tens of billions of US dollars annually. Currently, there are no antivirals against rhinoviruses, and the available treatments only treat the symptoms. Here, we present the molecular structure of human rhinovirus 14 in complex with its cellular receptor intercellular adhesion molecule 1. The binding of the virus to its receptor initiates the infection. Knowledge of the structure of the human rhinovirus 14intercellular adhesion molecule 1 interface and mechanism of interaction provides the basis for the design of compounds that may block the binding of rhinoviruses to receptors and thus prevent infection.

Most rhinoviruses, which are the leading cause of the common cold, utilize intercellular adhesion molecule-1 (ICAM-1) as a receptor to infect cells. To release their genomes, rhinoviruses convert to activated particles that contain pores in the capsid, lack minor capsid protein VP4, and have an altered genome organization. The binding of rhinoviruses to ICAM-1 promotes virus activation; however, the molecular details of the process remain unknown. Here, we present the structures of virion of rhinovirus 14 and its complex with ICAM-1 determined to resolutions of 2.6 and 2.4 , respectively. The cryo-electron microscopy reconstruction of rhinovirus 14 virions contains the resolved density of octanucleotide segments from the RNA genome that interact with VP2 subunits. We show that the binding of ICAM-1 to rhinovirus 14 is required to prime the virus for activation and genome release at acidic pH. Formation of the rhinovirus 14ICAM-1 complex induces conformational changes to the rhinovirus 14 capsid, including translocation of the C termini of VP4 subunits, which become poised for release through pores that open in the capsids of activated particles. VP4 subunits with altered conformation block the RNAVP2 interactions and expose patches of positively charged residues. The conformational changes to the capsid induce the redistribution of the virus genome by altering the capsidRNA interactions. The restructuring of the rhinovirus 14 capsid and genome prepares the virions for conversion to activated particles. The high-resolution structure of rhinovirus 14 in complex with ICAM-1 explains how the binding of uncoating receptors enables enterovirus genome release.

Human rhinoviruses are the cause of more than half of common colds (1). Medical visits and missed days of school and work cost tens of billions of US dollars annually (2, 3). There is currently no cure for rhinovirus infections, and the available treatments are only symptomatic. Rhinoviruses belong to the family Picornaviridae, genus Enterovirus, and are classified into species A, B, and C (4). Rhinoviruses A and B can belong to either major or minor groups, based on their utilization of intercellular adhesion molecule-1 (ICAM-1) or low-density lipoprotein receptor for cell entry (57). Type C rhinoviruses use CDHR3 as a receptor (8). Rhinovirus 14 belongs to the species rhinovirus B and uses ICAM-1 as a receptor. Receptors recognized by rhinoviruses and other enteroviruses can be divided into two groups based on their function in the infection process (9). Attachment receptors such as DAF, PSGL1, KREMEN1, CDHR3, and sialic acid enable the binding and endocytosis of virus particles into cells (1013). In contrast, uncoating receptors including ICAM-1, CD155, CAR, and SCARB2 enable virus cell entry but also promote genome release from virus particles (5, 1416).

Virions of rhinoviruses are nonenveloped and have icosahedral capsids (17). Genomes of rhinoviruses are 7,000 to 9,000 nucleotide-long single-stranded positive-sense RNA molecules (1, 17). The rhinovirus genome encodes a single polyprotein that is co- and posttranslationally cleaved into functional protein subunits. Capsid proteins VP1, VP3, and VP0, originating from one polyprotein, form a protomer, 60 of which assemble into a pseudo-T = 3 icosahedral capsid. To render the virions mature and infectious, VP0 subunits are cleaved into VP2 and VP4 (18, 19). VP1 subunits form pentamers around fivefold symmetry axes, whereas subunits VP2 and VP3 form heterohexamers centered on threefold symmetry axes. The major capsid proteins VP1 through 3 have a jelly roll -sandwich fold formed by two -sheets, each containing four antiparallel -strands, which are conventionally named B to I (2022). The two -sheets contain the strands BIDG and CHEF, respectively. The C termini of the capsid proteins are located at the virion surface, whereas the N termini mediate interactions between the capsid proteins and the RNA genome on the inner surface of the capsid. VP4 subunits are attached to the inner face of the capsid formed by the major capsid proteins. The surfaces of rhinovirus virions are characterized by circular depressions called canyons, which are centered around fivefold symmetry axes of the capsids (21).

The VP1 subunits of most rhinoviruses, but not those of rhinovirus 14, contain hydrophobic pockets, which are filled by molecules called pocket factors (17, 21, 23, 24). It has been speculated that pocket factors are fatty acids or lipids (25). The pockets are positioned immediately below the canyons. The exposure of rhinoviruses to acidic pH induces expulsion of the pocket factors, which leads to the formation of activated particles and genome release (17, 2632). The activated particles are characterized by capsid expansion, a reduction in interpentamer contacts, the release of VP4 subunits, externalization of N termini of VP1 subunits, and changes in the distribution of RNA genomes (17, 2629, 33, 34). Artificial hydrophobic compounds that bind to VP1 pockets with high affinity inhibit infection by rhinoviruses (35, 36).

ICAM-1 is an endothelial- and leukocyte-associated protein that stabilizes cellcell interactions and facilitates the movement of leukocytes through endothelia (37). ICAM-1 can be divided into an extracellular amino-terminal part composed of five immunoglobulin domains, a single transmembrane helix, and a 29-residuelong carboxyl-terminal cytoplasmic domain. The immunoglobulin domains are characterized by a specific fold that consists of seven to eight -strands, which form two antiparallel -sheets in a sandwich arrangement (3840). The immunoglobulin domains of ICAM-1 are stabilized by disulfide bonds and glycosylation (3841). The connections between the immunoglobulin domains are formed by flexible linkers that enable bending of the extracellular part of ICAM-1. For example, the angle between domains 1 and 2 differs by 8 between molecules in distinct crystal forms (38, 42). As a virus receptor, ICAM-1 enables the virus particles to sequester at the cell surface and mediates their endocytosis (5). The structures of complexes of rhinoviruses 3, 14, and 16, and CVA21 with ICAM-1 have been determined to resolutions of 9 to 28 (4246). It was shown that ICAM-1 molecules bind into the canyons at the rhinovirus surface, approximately between fivefold and twofold symmetry axes (4246). ICAM-1 interacts with residues from all three major capsid proteins. It has been speculated that the binding of ICAM-1 triggers the transition of virions of rhinovirus 14 to activated particles and initiates genome release (45, 47). However, the limited resolution of the previous studies prevented characterization of the corresponding molecular mechanism.

Here, we present the cryo-electron microscopy (cryo-EM) reconstruction of the rhinovirus 14 virion, which contains resolved density of octanucleotide segments of the RNA genome that interact with VP2 subunits. Furthermore, we show that the binding of ICAM-1 to rhinovirus 14 induces changes in its capsid and genome, which are required for subsequent virus activation and genome release at acidic pH.

The cryo-EM reconstruction of rhinovirus 14 virion was determined to a resolution of 2.6 (Fig. 1 A and B and SI Appendix, Fig. S1 and Table S1). The map enabled building of the structure of capsid proteins VP1 through 4 except for residues 1 to 17 and 290 to 293 of VP1, 1 to 6 of VP2, and 1 to 28 of VP4 (Fig. 1B). In addition to the capsid, the cryo-EM map contains resolved density corresponding to octanucleotide segments from the RNA genome (Fig. 1 AC). The quality of the map enabled building of the RNA structure; however, the nucleotide sequence could not be identified. The base of the second nucleotide from the 5 end of the RNA segment is flipped out from the RNA helix and forms a stacking interaction with the sidechain of Trp38 of VP2 (Fig. 1D and SI Appendix, Fig. S2). The residue Trp38 is conserved in the VP2 subunits of numerous picornaviruses, including polioviruses, rhinoviruses 2 and 16, coxsackievirus B3, coxsackievirus A9, and coxsackievirus A21 (SI Appendix, Fig. S3). Virion structures of these viruses contain disk-like densities that stack onto the tryptophane side chains, which were hypothesized to belong to a base of guanine nucleotide (23, 24, 4851). The structure of the RNA segment in the virion of rhinovirus 14 provides evidence that the previous speculations about the densities stacking onto Trp38 side chains were correct. To be consistent with the previous structures, we modeled the second nucleotide of the RNA segment in rhinovirus 14 as guanine (Fig. 1D and SI Appendix, Fig. S2). The stacking interaction between Trp38 and the base of the second nucleotide is the only direct contact between the RNA segment and the capsid (Fig. 1 A, C, and D and SI Appendix, Fig. S2).

Structure of virion of rhinovirus 14 contains resolved density corresponding to octanucleotides from its RNA genome. (A) Surface representation of cryo-EM of reconstruction of virion of rhinovirus 14 with front half of the particle removed to show internal structure. Density corresponding to VP1 is shown in blue, VP2 in green, VP3 in red, VP4 in yellow, and RNA segments in pink. Borders of a selected icosahedral asymmetric unit are indicated with a dashed triangle and positions of selected twofold, threefold, and fivefold symmetry axes are represented by an oval, triangle, and pentagon, respectively. (Scale bar, 5 nm.) (B) Cartoon representation of icosahedral asymmetric unit of rhinovirus 14 viewed from the inside of the capsid. The color coding of individual virus components is the same as in A. Positions of twofold, threefold, and fivefold symmetry axes are represented by an oval, triangle, and pentagon, respectively. (C) Two RNA octanucleotides that interact with each other and with VP2 subunits. Protein and RNA coloring is the same as in A. The cryo-EM density of the RNA segments is shown as a pink semitransparent surface. RNA bases and side chains of Trp38 of VP2 are shown in stick representation, in orange, and indicated with black arrowheads. The position of a twofold symmetry axis is indicated with an oval. (D) Detail of stacking interaction between Gua2 from RNA segment and Trp38 of VP2. The cryo-EM densities of RNA and protein are shown as semitransparent surfaces in pink and gray, respectively. (E) Interaction between N terminus of VP1 and genome. Capsid proteins are shown in cartoon representation with the same coloring as in A. Cryo-EM densities of individual proteins are shown as semitransparent surfaces colored according to the chain they belong to. The density of the RNA genome is shown in gray. The blue arrow indicates the contact between the N terminus of VP1 and the genome. The position of Thr17, the first modeled residue from the N terminus of VP1, is indicated.

Each RNA segment in the virion of rhinovirus 14 is associated with one protomer of capsid proteins VP1 through 4 (Fig. 1 A and B). The RNA is positioned next to a twofold axis, and two of the oligonucleotides interact with each other (Fig. 1 AC). To enable WatsonCrick base pairing between the two segments, the nucleotide at position seven was modeled as uracil and the nucleotide at position eight as adenine (Fig. 1C). Due to the constraints imposed by the interaction of the RNA with the capsid and of the interactions of two segments of the RNA with each other, the sequence of the oligonucleotide was built as 5 UGUUUUUA 3. Nevertheless, other sequences that fulfill the interaction conditions are equally possible.

The reconstruction of virion of rhinovirus 14 provides evidence that the N termini of VP1 subunits interact with the RNA genome (Fig. 1E). A similar function of the N terminus of VP1 was described previously in rhinovirus 2 (26). None of the interactions between the genomes and capsids of enteroviruses identified to date are sequence specific. Nevertheless, even the nonspecific interactions of the N termini of VP1 and Trp38 of VP2 with the RNA may enable the packaging of the enterovirus genome into a particle. Compounds that could prevent the RNAcapsid interaction by, for instance covering the side chain of Trp38, may interfere with the production of infectious virions.

The complex of rhinovirus 14 with ICAM-1 was prepared by mixing the components in phosphate-buffered saline (PBS) of pH 7.4 and incubating them at 34 C for 30 min (Fig. 2). The temperature was chosen to mimic that in the human upper respiratory tract (52). The binding of ICAM-1 did not induce the formation of activated particles or the genome release of rhinovirus 14 (Fig. 2 AC). This is in agreement with previous results showing that the ability of ICAM-1 to trigger genome release depends on temperature, receptor concentration, and buffer composition (47, 53). Experiments showing that ICAM-1 could induce genome release were performed in solutions with nonphysiological concentrations of salts, which may have destabilized the virus particles (42, 46, 47). Since enteroviruses have to deliver their genomes into the cytoplasm of a host cell to initiate infection, it would be detrimental if they released their genomes immediately upon binding to receptors at the cell surface. Our results show that rhinovirus 14 is stable when bound to ICAM-1 under native-like conditions (Fig. 2C), and the induction of genome release requires exposure to acidic pH in endosomes.

Binding of ICAM-1 to rhinovirus 14 is required for efficient genome release at acidic pH. (AD) Electron micrographs (Top), reference-free two-dimensional class averages (Center), and 3D reconstructions (Bottom) of rhinovirus 14 and rhinovirus 14ICAM-1 complex at neutral and acidic pH. The cryo-EM reconstructions are rainbow colored based on the distance of the particle surface from its center. Names above the reconstructions indicate the types of particles. Percentages below the reconstructions indicate the relative abundance of each type of particle in the sample. (A and B) Rhinovirus 14 at pH 7.4 (A) and 6.2 (B). (C and D) Rhinovirus 14ICAM-1 complex at pH 7.4 (C) and 6.2 (D). The top row of D contains two parts of a micrograph to show all types of particles present in the sample. Black arrowheads indicate virions, yellow indicate empty particles, cyan indicate rhinovirus 14ICAM-1 complex, green indicate activated particles, and red indicate open particles. (Scale bars, 30 nm.) (EH) Conformational changes to VP4 are induced by ICAM-1 binding but not by acidic pH. Surface representations of cryo-EM reconstructions showing the inner faces of capsids. The surfaces are color coded according to the capsid proteins with VP1 in blue, VP2 in green, VP3 in red, and VP4 in yellow. Only particles containing VP4 are shown. Positions of icosahedral symmetry axes are indicated with a black pentagon, triangle, and oval for fivefold, threefold, and twofold, respectively. Black arrows highlight C termini of VP4.

The structure of rhinovirus 14 in complex with the soluble ectodomain of ICAM-1 was determined to a resolution of 2.4 using cryo-EM and single-particle reconstruction (Fig. 3 andSI Appendix, Fig. S1 and Table S1). Domain 1 of ICAM-1 bound at the surface of rhinovirus 14 was resolved to a resolution of 2.6 (Fig. 3A and SI Appendix, Fig. S1). Levels of density in the map region corresponding to the domain 1 of ICAM-1 are similar to those in the capsid of the virus, indicating full occupancy of the receptors at the virus surface. Domains 2 and 3 of ICAM-1 are resolved to a resolution of 6 , and domains 4 and 5 are not visible in the cryo-EM reconstruction (Fig. 3A). The low resolution of the region of the map corresponding to domains 2 and 3 and the lack of density for domains 4 and 5 are probably caused by movements of those domains relative to domain 1, which is anchored at the virus surface (Fig. 3A) (38, 42, 43). In agreement with previous studies, domain 1 of ICAM-1 binds into the canyon of rhinovirus 14, approximately in the middle between fivefold and twofold symmetry axes (Fig. 3 B and C) (42, 43, 45). Previous studies of the interactions of rhinoviruses 3, 14, and 16 with ICAM-1 were limited to a resolution of 9.5 or lower (4246). The interpretation of the macromolecular interactions relied on the fitting of high-resolution structures, determined by X-ray crystallography, to the cryo-EM maps of the complex (4246). Therefore, the changes in the structures of the receptor and virus, induced upon their binding, could not be identified. Here, we show that ICAM-1 is wedged 3.4 deeper into the canyon and rotated 7.6 clockwise, when looking along the long axis of domain 1 toward the virus surface, relative to the structure reported previously (Fig. 4 and SI Appendix, Fig. S4A) (42, 43, 45). The interaction interface between ICAM-1 and rhinovirus 14 has a buried surface area of 1,500 2. The core of domain 1 of ICAM-1 is formed by -sheets ABED and GFC (Fig. 4 AC) (11, 54). Residues from the loops BC, DE, and FG and strands B, C, D, E, F, and G of ICAM-1 interact with rhinovirus 14 (SI Appendix, Fig. S4B). The mode of attachment of rhinovirus 14 to ICAM-1 is characteristic for uncoating receptors that bind to enterovirus canyons (5, 9, 14, 15). The uncoating receptors of enteroviruses, including ICAM-1, CD155, and CAR, have elongated molecules formed by domains with an immunoglobulin fold, which enables their insertion into the canyons of enterovirus particles (5, 14, 15).

Structure of rhinovirus 14 in complex with ICAM-1. (A) Surface representation of cryo-EM reconstruction of rhinovirus 14ICAM-1 complex color coded to distinguish individual proteins. Density corresponding to VP1 is shown in blue, VP2 in green, VP3 in red, and ICAM-1 in light magenta. Positions of selected icosahedral symmetry axes are indicated by a pentagon for fivefold, triangle for threefold, and an oval for twofold. The white triangle indicates the border of a selected icosahedral asymmetric unit. The yellow dashed rectangle indicates borders of the area shown in detail in B. (B) Roadmap projection showing residues forming the outer surface of rhinovirus 14 capsid (Left) and residues of domain 1 of ICAM-1 facing toward the virus (Right). Coloring is the same as in A. Residues involved in virusreceptor interactions are shown in bright colors. The polar angles and indicate positions at the capsid surface. (C) Schematic representation of ICAM-1. D1 to D5 indicate extracellular immunoglobulin domains; TM, transmembrane domain; cyt, cytoplasmic domain. Disulfide bridges (S-S) stabilizing the immunoglobulin domains are indicated. Red dashes highlight the binding site for rhinovirus 14. The ectodomain used in this study to determine the rhinovirus 14ICAM-1 interactions included residues 1 to 453.

Conformational changes associated with binding of rhinovirus 14 to ICAM-1. (A) Cartoon representation of icosahedral asymmetric unit of rhinovirus 14 in complex with ICAM-1. The VP1 subunit is shown in blue, VP2 in green, VP3 in red, VP4 in yellow, and domain 1 of ICAM-1 in magenta. The capsid proteins from virion of rhinovirus 14 are superimposed onto those of the rhinovirus 14ICAM-1 complex and are shown in white. The binding of ICAM-1 to rhinovirus 14 induces a 1.7 tilt of VP1 toward VP2 and VP3, which results in a narrowing of the canyon relative to the virion structure. Least-squares planes fitted to VP1 are shown to highlight the rotation of VP1. (B) Conformational change of FG loop of ICAM-1, shown in magenta, is required to prevent clashes with FG loop of VP3, shown in red. The native structure of ICAM-1 clashing with VP2 is shown in gray. (C) Cartoon representation of structure of ICAM-1 bound to rhinovirus 14 with side chains of cysteine residues shown in stick representation with red carbon atoms and yellow sulfur atoms. (D and E) Detail of disulfide bridge between Cys21 and Cys65 (D) and Cys25 and Cys69 (E). Cryo-EM density is shown as a semitransparent blue surface.

The binding of rhinovirus 14 to ICAM-1 is accompanied by the local restructuring of domain 1 of ICAM-1 and surface loops of capsid proteins, as well as by overall changes in the structure of the rhinovirus-14 capsid (Figs. 4 and 5 and SI Appendix, Figs. S5 and S6). VP1 subunits rotate 1.7 toward VP2 and VP3, which results in a contraction of the canyon (Fig. 4A). As a result, the capsid expands by 5 in diameter (Fig. 4A). The binding of ICAM-1 to rhinovirus 14 requires the bending of the FG loop of ICAM-1 8 toward the core of the immunoglobulin domain (Fig. 4B and SI Appendix, Fig. S5). This conformational change is necessary to prevent the clashing of the FG loop of ICAM-1 with residues 178 to 182 from the FG loop of VP3 of rhinovirus 14 (Fig. 4B). The conformational flexibility of the FG loop of ICAM-1 enables enlargement of its interaction interface with the capsid.

Changes of structure of C terminus of VP4 induced by ICAM-1 binding to rhinovirus 14. (A) Surface representation of cryo-EM reconstruction of capsid of rhinovirus 14 in complex with ICAM-1 viewed from inside the virion. Density corresponding to VP1 is shown in pale blue, VP2 in pale green, VP3 in pale red, and VP4 in semitransparent yellow. The structure of VP4 in the rhinovirus 14ICAM-1 complex is shown in cartoon representation in yellow, whereas the structure of VP4 in the virion of rhinovirus 14 is shown in magenta. The positions of selected icosahedral symmetry axes are indicated with a pentagon for fivefold, triangle for threefold, and oval for twofold. Borders of a selected icosahedral asymmetric unit are indicated with a dashed triangle. (B) Capsid structure of an empty particle of rhinovirus 14 containing pores around twofold symmetry axes and between twofold and fivefold symmetry axes through which VP4 may be released from the particle. (CF) Differences in structure of VP4 subunits in virion (C and E) and rhinovirus 14ICAM-1 complex (D and F). Capsid proteins are shown in cartoon representation. VP1 is shown in blue, VP2 in green, VP3 in red, VP4 in yellow, and RNA segments in pink. (C and E) Asn68 from C terminus of VP4 interacts with Asp11 and Arg12 of VP2 in virion of rhinovirus 14. The residues Asp11 and Arg12 are stabilized in position by the underlying loop of VP2 formed by residues 27 to 32 (highlighted in magenta). The side chain of Trp38 (highlighted in orange) forms a stacking interaction with Gua2 that is part of the resolved RNA segment positioned next to a twofold axis. (D and F) Binding of rhinovirus 14 to ICAM-1 induces conformational changes of virus capsid that include movement of residues 27 to 32 of VP2 toward particle center, which prevents interaction of C terminus of VP4 with residues Asp11 and Arg12 of VP2. The C terminus of VP4 acquires a new conformation, which covers the side chain of Trp38 of VP2 and blocks its interaction with RNA.

The structures of domain 1 of ICAM-1 determined to date contain the disulfide bonds Cys21 and Cys65 and Cys25 and Cys69 (3840). Residues Cys25 and Cys69 are located in the vicinity of the virus surface when ICAM-1 binds to rhinovirus 14 (Fig. 4 CE). Cys69 is part of the FG loop, whereas Cys25 is part of the BC loop (Fig. 4 C and E). The density connecting Cys25 and Cys69 of ICAM-1 in the complex with rhinovirus 14 is much weaker than that connecting Cys21 to Cys65 (Fig. 4 C and E). However, the positions of the two cysteines in the cryo-EM density map are consistent with the linkage of their side chains by a disulfide bond (Fig. 4 C and E) (55). Furthermore, mass spectrometry analysis of ICAM-1 molecules from the complex with rhinovirus 14 did not detect any peptides containing free Cys25 and Cys69 (SI Appendix, Fig. S7). However, peptides containing free cysteines were observed after the reduction of the disulfide bonds by dithiothreitol (DTT). This provides evidence that Cys25 and Cys69 of ICAM-1 in complex with rhinovirus 14 are linked by a disulfide bond. The low values of cryo-EM density may be caused by a higher flexibility of this part of ICAM-1, as indicated by the lower resolution than in the core of domain 1 (Fig. 3A). The binding of rhinovirus 14 to ICAM-1 also induces structural changes in the virus proteins. Residues 154 to 162 from the DE loop of VP1 shift 2 toward the core of the subunit (SI Appendix, Fig. S6). This movement helps to accommodate ICAM-1 in the canyon of rhinovirus 14.

The interaction between ICAM-1 and rhinovirus 14 is formed by 36 residues from domain 1 of ICAM-1 and 31, 8, and 13 residues of VP1, VP2, and VP3 of rhinovirus 14, respectively (Fig. 3B). The specificity of the interaction is controlled by a combination of the complementarity of the electrostatic interactions, a network of hydrogen bonds, and the positions of patches of hydrophobic interfaces. There are salt bridges between Lys77 of ICAM-1 and Glu210 from VP1 and Lys39 of ICAM-1 and C-terminal carboxyl group of Glu236 from VP3 (SI Appendix, Fig. S8 A and B). The interface includes a network of 18 hydrogen bonds. Furthermore, the resolution of the cryo-EM reconstruction of the complex was sufficient to enable the identification of water molecules, some of which form hydrogen bonds with both ICAM-1 and rhinovirus 14 and thus mediate interactions between the receptor and virus (SI Appendix, Fig. S8 C and D). For example, the amino group from the side chain of Lys29 of ICAM-1 interacts with two water molecules that form hydrogen bonds with side chains of Thr105 of VP3 and Asn68 of ICAM-1 (SI Appendix, Fig. S8D). It has been shown previously that the mutation of Thr75 of ICAM-1 to Ala reduces the efficiency of binding of rhinovirus 14 by more than 50% (54, 56). No ions were identified at the rhinovirus 14ICAM-1 interface (7, 57).

The interface between rhinovirus 14 and ICAM-1 contains complementary patches of hydrophobic interactions (SI Appendix, Fig. S8 E and F). Previous studies have shown that most mutations of Pro70 from the FG loop of ICAM-1 prevent the binding of rhinovirus 14 (54, 56). We show that Pro70 fits into a small hydrophobic pocket formed by Pro178, Phe86, and Thr180 of VP3 (SI Appendix, Fig. S8F). Fitting Pro70 of ICAM-1 into the hydrophobic cavity in VP3 requires movement and restructuring of the FG loop of ICAM-1 (Fig. 4 C and E and SI Appendix, Fig.S8F). Another residue of ICAM-1 that is critical for the binding of rhinovirus 14 is Leu30 (54, 56). In the complex, the side chain of Leu30 is situated between the side chains of Ile215 and Val217 from VP1, which form a hydrophobic pocket for the leucine side chain (SI Appendix, Fig. S8E). This explains why the mutation of Leu30 to Ser eliminates the binding of ICAM-1 to rhinovirus 14 (54, 56).

The structure of the C terminus of VP4 subunit in the rhinovirus 14ICAM-1 complex differs from that in the native virion (Figs. 2 EH and 5). The two structures of VP4 subunits are similar for residues 29 to 57, with rmsd of C atoms of the corresponding residues of 0.44 . However, residues 58 to 65 of VP4 extend toward a threefold symmetry axis of the capsid in the native virion, whereas the same residues point toward a twofold symmetry axis in the rhinovirus 14ICAM-1 complex (Fig. 5A). The movement of the C terminus of VP4 is induced by conformational changes of the major capsid proteins, which are triggered by ICAM-1 binding to the capsid. In the rhinovirus-14 virion, the C-terminal carboxyl group of Asn68 from VP4 forms a salt bridge with the side chain of Arg12 of VP2 (Fig. 5 C and E). Additionally, the side chain of Asn68 forms two hydrogen bonds with the side chain of Asp11 of VP2 (Fig. 5 C and E). As discussed above, the binding of ICAM-1 to rhinovirus 14 induces a rotation of VP1 toward VP2 and VP3 (Fig. 4A). These movements of capsid proteins bring residues 27 to 33 from the N terminus of VP2 into the space that is occupied by Arg12 of VP2 in the native virion (Fig. 5 D and F). This frees the C terminus of VP4 from the interaction with Arg12 of VP2 and probably enables its translocation toward a twofold axis (Fig. 5 A, D, and F). The restructuring of the C-terminal part of VP4 to point toward a twofold symmetry axis prepares the protein for release through either of the holes that form at and next to the twofold symmetry axes upon particle activation (Fig. 5 A and B) (29, 34).

The binding of ICAM-1 into the canyon of rhinovirus 14 induces the relocation of the C terminus of VP4 toward a twofold symmetry axis of the capsid (Fig. 5A). The movement of the C terminus of VP4 uncovers a patch of positively charged residues at the inner face of the capsid, adjacent to a threefold symmetry axis (Fig. 6 A and B). The positively charged surface attracts genomic RNA, which is represented in the cryo-EM reconstruction as a cylindrical appendage emanating from the spherical genome density filling the center of the virus particle (Fig. 6 CF). This indicates that parts of the RNA genome in various conformations interact nonspecifically with the positively charged regions of the capsid. Furthermore, the N termini of VP2 subunits probably interact with the RNA density positioned on a threefold axis (Fig. 6F). The C terminus of VP4 positioned next to a twofold axis of the capsid covers the side chain of Trp38 of VP2, which in the native virion forms a stacking interaction with a nucleotide from the RNA genome (Figs. 1D and 5 E and F and SI Appendix, Fig. S2). The loss of Trp38RNA contact relaxes the ordering of segments of the RNA genome that interact with the capsid around twofold symmetry axes in native virions (Fig. 6 G and H). A density corresponding to RNA at the twofold axis in the rhinovirus 14ICAM-1 complex does not have resolved features (Fig. 6 G and H). The interactions of the N termini of VP1 subunits with the RNA genome remain preserved even after the binding of rhinovirus 14 to ICAM-1 (Fig. 1E). The binding of rhinovirus 14 to ICAM-1 induces a rearrangement of the RNA genome, which may play a role in particle activation, as discussed below.

Conformational changes of capsid of rhinovirus 14 that are induced by binding to ICAM-1 trigger redistribution of RNA genome in the particle. (A and B) Detail of inner capsid surface around threefold symmetry axis of virion (A) and rhinovirus 14ICAM-1 complex (B). The surfaces are colored according to charge. (C and D) Electron densities of central slices of cryo-EM reconstructions of virion (C) and rhinovirus 14ICAM-1 complex (D) with a thickness of 1 . White represents high density values. Density representing ICAM-1 is highlighted in magenta. Positions of icosahedral symmetry axes are indicated with an oval, triangle, and pentagon for twofold, threefold, and fivefold axes, respectively. Black arrows in D point toward densities on threefold symmetry axes, which are not present in the virion. Red squares indicate positions of details shown at higher magnification in E and F. (Scale bar, 10 nm.) (E and F) Details of cryo-EM density distribution at the inner face of the capsid on a threefold symmetry axis. Capsid proteins are shown in cartoon representation with VP1 in blue, VP2 in green, VP3 in red, and VP4 in yellow. Cryo-EM density is shown as a semitransparent gray surface. Positions of the first resolved residues from the N termini of VP2 subunits are indicated. E and F show sections of particles with a thickness of 20 . (G and H) Comparison of structures of RNA genome interacting with VP2 subunits in virion (G) and rhinovirus 14ICAM-1 complex (H). The virion contains resolved cryo-EM density corresponding to octanucleotides (G). In contrast, there is a featureless density in the rhinovirus 14ICAM-1 complex (H). Capsid proteins are shown in cartoon representation, colored as in E and F.

The binding of ICAM-1 to rhinovirus 14 triggers a cascade of structural changes that prepare the particle for activation and subsequent genome release (Fig. 7 A and B). The rotation of VP1 subunits results in a narrowing of the canyon and transmits the conformational changes to the inside of the capsid (Fig. 7 C and D). C termini of VP4 subunits reposition toward twofold symmetry axes, where they are optimally poised for externalization upon particle activation (Fig. 7 E and F). The conformational change to C termini of VP4 subunits uncovers patches of positively charged residues that attract genomic RNA toward threefold symmetry axes of the capsid (Fig. 7 G and H). The same conformational change prevents the interaction of Trp38 from VP2 with bases from the ordered RNA segments of the genome positioned next to twofold symmetry axes of the capsid (Fig. 7 G and H). Both of these effects result in reorganization of the RNA genome (Fig. 7 G and H). These changes in the capsid and genome structure of rhinovirus 14 induced by ICAM-1 binding are required for efficient genome release at acidic pH (Fig. 2). All in all, 90% of virions of rhinovirus 14 exposed to pH 6.2 remained in their native conformation, whereas the remaining particles were empty (Fig. 2 A, B, E, and F and SI Appendix, Table S1). The structures of rhinovirus 14 in their native conformation and empty capsids at acidic pH were determined to resolutions of 2.8 and 3.9 , respectively (Fig. 2 A and B and SI Appendix, Table S1). The exposure of rhinovirus 14 to acidic pH did not induce structural changes in VP4 subunits (Fig. 2 E and F). In contrast, the exposure of rhinovirus 14ICAM-1 complex to pH 6.2 resulted in activation and genome release from 95% of particles (Fig. 2 C, D, E, and H). The structure of the activated particle was determined to a resolution of 4.0 , empty particle to 3.9 , and open particle to 22 (Fig. 2 C and D and SI Appendix, Table S1). The changes in the capsid and genome of rhinovirus 14, which were induced by ICAM-1 binding, may lower the energy barrier of particle activation so that it can be overcome by random fluctuations in particle structure due to thermal motions termed capsid breathing (58, 59). This provides a putative explanation of how the reduction of capsid dynamics by antiviral compounds, which target VP1 pockets (59), blocks the activation and genome release of enteroviruses.

Overview of structural changes to rhinovirus 14 induced by binding of ICAM-1 that prepare the virus for activation and genome release. (A) Native virion diffuses toward cell membrane (green ribbon) decorated with ICAM-1 molecules (blue sticks with dark blue heads representing domain 1). The virus particle is represented by a central slice with the electron density of VP1 shown in light blue, VP2 in light green, VP3 in light red, VP4 in yellow, the genome in purple, and resolved RNA segments in pink. (B) Rhinovirus 14 is endocytosed by the cell after binding to ICAM-1. (CH) Sequence of structural changes in virion induced by binding to ICAM-1. C, E, and G represent native virion, whereas D, F, and H show rhinovirus 14ICAM-1 complex. (C and D) Binding of ICAM-1 induces rotation of VP1 subunit toward VP2 and VP3. Virus components are colored as in A. (E and F) ICAM-1 binding induces disruption of interactions between C terminus of VP4 and N terminus of VP2. C terminus of VP4 repositions from a threefold symmetry axis (indicated with a triangle) toward a twofold symmetry axis (oval). (G and H) Movements of C termini of VP4 subunits uncover positively charged residues around twofold symmetry axes, which attract negatively charged RNA genome. Furthermore, the C terminus of VP4 in the altered conformation covers Trp38 of VP2 and prevents its specific interaction with structured segments of the RNA genome, which relaxes the structure of RNA adjacent to the twofold symmetry axis.

The sample of complex of rhinovirus 14 with ICAM-1 exposed to acidic pH contained 7% empty particles missing a pentamer of capsid protein protomers (Fig. 2D). Open particles were previously speculated to enable enterovirus genome release (33). The expulsion of pentamers of capsid proteins results in the formation of a large hole in the capsid, which enables the diffusion of the RNA genome from the capsid within a microsecond (33, 60). The rapid release of a genome may be connected to its subsequent transport across the endosome membrane into the cytoplasm (61, 62).

Structural characterization of the rhinovirus 14ICAM-1 complex at atomic resolution provides detailed information about the conformational changes of both the receptor and virus that are required for its binding. Additionally, it provides insight into the structural changes of the virus that enable subsequent particle activation and genome release. In combination, these results provide the basis for the design of compounds that block enterovirus infection.

The extracellular part of ICAM-1 containing domains D1 to D5 was produced using the MutiBac system (Geneva Biotech). The full-length gene of ICAM-1 was a gift from Timothy Springer (Harvard Medical School, Boston, MA) (Addgene plasmid No. 8632; http://addgene.org/8632; RRID: Addgene 8632). The sequence encoding domains D1 to D5, the secretion signal peptide, and the C-terminal 10 His-tag were inserted into the pACEBac1 vector at the restriction site BamHI. The recombinant bacmid with the target sequence was prepared by recombination in DH10EMBacY Escherichia coli cells (Geneva Biotech). The recombinant baculovirus was prepared by transfecting SF9 cells with the recombinant bacmid. A total of 250 mL of the culture of SF9 cells were infected with the recombinant baculovirus and incubated for 96 h at 27 C with 120 rpm shaking. The produced protein was secreted into the medium. Cells and cell debris were pelleted by centrifugation at 20,000 g at 4 C for 15 min. The supernatant was filtered through a 0.2 m filter (Corning) and loaded into a HisTrap column (GE Healthcare) equilibrated in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4). Most of the impurities were removed by washing with PBS containing 70 mM imidazole. His-tagged ICAM-1 D1 to D5 was eluted using PBS with 500 mM imidazole. The eluted protein was buffer exchanged into PBS using 30 kDa cutoff centrifugal concentrators (Millipore, Merck). The target protein was further purified by size-exclusion chromatography, using a HiLoad 16/600 Superdex 200 pg column (GE Healthcare). Fractions containing ICAM-1 D1 to D5 were pooled and concentrated using centrifugal concentrators (Millipore, Merck) to a final concentration of 3.5 mg/mL.

Rhinovirus-14 strain 1059, obtained from ATCC, was propagated in HeLa-H1 (ATCC CRL195) cells cultivated in Dulbeccos modified Eagles medium enriched with 10% fetal bovine serum. HeLa cells grown to 100% confluence (100 tissue culture dishes, 150 mm diameter) were infected with a multiplicity of infection of 0.1. The infection was allowed to progress for 36 h until a complete cytopathic effect was observed. The media and cells were harvested and centrifuged at 15,000 g at 10 C for 30 min. The resulting pellet was subjected to three freezethaw cycles and resuspended in 5 mL PBS followed by homogenization in a Dounce tissue grinder. Cell debris was separated from the supernatant by centrifugation at 4,000 g for 30 min at 10 C. The virus-containing supernatant was combined with the media from infected cells. The virus particles were precipitated by the addition of PEG-8000 and NaCl to final concentrations of 12.5% (wt/vol) and 500 mM, respectively, and incubation overnight at 10 C. The precipitated virus was pelleted by centrifugation at 15,000 g at 10 C for 30 min. The pellet was resuspended in 20 mL PBS with 5 nm MgCl2. The sample was subjected to DNase (10 g/mL final concentration) and RNase (10 g/mL final concentration) treatment at room temperature for 30 min. Subsequently, trypsin was added to a final concentration of 0.5 g/mL, and the sample was incubated at 37 C for 15 min. EDTA (pH = 9.5) and Nonidet P-40 (Sigma-Aldrich) were added to final concentrations of 15 mM and 1% (vol/vol), respectively. The virus was pelleted through a 30% (wt/vol) sucrose cushion by centrifugation at 210,000 g in an Optima 80 ultracentrifuge (Beckman Coulter) using a Ti50.2 rotor. The pelleted virus particles were resuspended in 2 mL PBS, loaded on the top of a 60% (wt/wt) CsCl solution in PBS, and centrifuged at 160,000 g in an Optima 80 ultracentrifuge using a Beckman Coulter SW41Ti rotor at 10 C for 24 h. Opaque bands containing virus particles were harvested and subjected to buffer exchange in PBS using a Centricon Plus-70 centrifugal filter (Millipore) with a 100 kDa cutoff. The final concentration of rhinovirus 14 was 0.5 mg/mL. Purified virus was kept at 4 C.

The complex of rhinovirus 14 with ICAM-1 was prepared by mixing rhinovirus 14 with ICAM-1 at a molar ratio of 1:100 at pH 7.4 and incubating the mixture for 30 min at 34 C. Rhinovirus 14 and the rhinovirus 14ICAM-1 complex were transferred to acidic pH using DyeEx 2.0 (QIAGEN) spin columns containing PBS with pH 6.2. The samples were applied onto the columns and eluted by 1 min of centrifugation at 1,200 g. The samples were incubated at pH 6.2 at 34 C for 2.5 min, including the centrifugation step.

For vitrification, 3 L virus samples were applied onto a holey carbon-coated copper grid (R2/1, mesh 300, Quantifoil), blotted for 2 s, and vitrified by plunging into liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Grids for virion reconstruction were prepared by vitrifying a virus sample with a concentration of 0.5 mg/mL. The grids (except for those with rhinovirus 14 at pH 6.2) were then transferred to a Titan Krios electron microscope, operating at 300 kV at cryogenic conditions, equipped with a Falcon III direct electron detector (Thermo Fisher Scientific). The illuminating beam was aligned for parallel illumination in NanoProbe mode. Low-dose imaging was used with a total dose of 84.7 e/ 2. Nominal magnification was set to 75,000, resulting in a calibrated pixel size of 1.063 . The dataset was recorded automatically using EPU software (Thermo Fisher Scientific) in fast acquisition mode, using large image shifts. The samples of rhinovirus 14 and rhinovirus 14ICAM-1 complex were recorded using five acquisitions per hole, nine holes per stage shift. The sample of rhinovirus 14ICAM-1 complex at pH 6.2 was recorded using seven acquisitions per hole, nine holes per stage shift. The exposure time was set to 1 s, and each micrograph was recorded as a movie containing 40 frames. The target defocus range was 0.5 to 2.4 m.

Electron micrographs from the sample of rhinovirus 14 at pH 6.2 were collected using a Talos Arctica electron microscope (Thermo Fisher Scientific), operated at 200 kV under cryogenic conditions, equipped with a Falcon III direct electron detector (Thermo Fisher Scientific). The illuminating beam was aligned for parallel illumination in NanoProbe mode. Low-dose imaging was used with a total dose of 34.1 e/ 2. Nominal magnification was set to 120,000, resulting in a calibrated pixel size of 1.22 . The dataset was recorded automatically using EPU software (Thermo Fisher Scientific) in fast acquisition mode, using five acquisitions per hole, nine holes per stage shift. The exposure time was set to 1 s, and each micrograph was recorded as a movie containing 40 frames. The target defocus range was 0.5 to 2.4 m.

The beam-induced movements within one micrograph were corrected with the software MotionCorr2 using 5 5 patches (63). The motion-corrected micrographs were dose weighted, and defocus values were estimated using the program gCTF (64). Using crYOLO box manager (65), 200 particles were boxed manually and used as a template for ab initio model training. The resulting crYOLO model was used to pick particles. The particles were extracted using Relion3.1 (66) with a box size of 546 px. The particles were binned to a box size of 150 150 px and subjected to reference-free two-dimensional classifications in Relion3.1 (66). Particles from classes exhibiting high-resolution features were used for de novo model calculation with imposed icosahedral symmetry, using stochastic gradient descent as implemented in Relion3.1 (66). The resulting three-dimensional (3D) volume was used as a starting model for autorefinement in Relion3.1. After initial autorefinement, 3D classification in Relion3.1 was performed, omitting the alignment step. Particles belonging to the best class were reextracted and recentered box-size 512 512 px for rhinovirus 14 particles without ICAM-1 and 546 546 px for the rhinovirus 14ICAM-1 complex. Reextracted particles were subjected to another round of autorefinement in Relion3.1. Particles were then sorted into nine optic groups. The optics groups were determined by the position of the image shift used for the acquisition, whereas all the acquisition areas from the same foil hole were considered as one optics group. Therefore, only large (interhole) image shifts were considered as separate optics groups. Magnification correction was performed using Relion3.1, followed by beam-tilt correction, and subsequently by the estimation of third- and fourth-order Zernike polynomials. The aberration-corrected particles were further subjected to per-particle defocus and astigmatism correction and estimation of the CTF envelope function (CTF B-factor fitting). The particles were subjected to autorefinement with imposed icosahedral symmetry. Ewald sphere correction was performed as implemented in relion_reconstruct.py in Relion3.1 (67). The resulting map was used for Bayesian polishing of particles with default parameters. The polished particles were used for 3D autorefinement and CTF refinement followed by another 3D autorefinement. Finally, Ewald sphere correction was performed. The final map was threshold masked, divided by a modulation transfer function, and B-factor sharpened using Relion3.1. Local resolutions were estimated using the program MonoRes implemented in the Scipion software package (68, 69). Map B-factor sharpening based on local resolution estimation from MonoRes was performed using the program LocalDeblur (70). The dataset of rhinovirus 14 with ICAM-1 exposed to pH 6.2 contained empty capsids missing pentamers. These were identified by 3D classification with C5 symmetry, using the complete empty capsid as an initial model. Subsequent 3D autorefinement was performed with C5 symmetry. Neither CTF refinement nor Bayesian polishing were applied to this subset of particles.

The electrostatic potential map from cryo-EM reconstruction was oriented to the standard 222 icosahedral crystallographic orientation. The origin of the map was moved from the 0,0,0, coordinate to the center of the particle using the program mapman (71). The map was normalized and converted to crystallographic space group P23 using the CCP4i software suite (72). The higher-symmetry space group was used to reduce the computational demands of the model refinement. Crystal structures of rhinovirus 14 (Protein Data Bank [PDB]: 4RHV) and domain 1 of ICAM-1 (PDB: 1IC1) were manually fitted into the cryo-EM maps using the program Chimera and refined with the tool Fit in map (73). The cryo-EM structure of an empty particle of rhinovirus 14 in complex with a Fab fragment of antibody (PDB: 5W3O) was used as a starting model for the building of activated and empty particles (74). The fitted models were subjected to multiple rounds of real-space refinement in Phenix (version dev-3765), reciprocal-space refinement in REFMAC5, combined with manual corrections in Coot 0.9 and ISOLDE (7578). Hydrogen atoms were taken into account during the real-space refinement, whereas they were ignored in the reciprocal-space refinement. Waters were added automatically by the program find waters in Coot 0.9 and validated manually. Model validation parameters were calculated using MolProbity server and EMringer as implemented in phenix (79, 80). The RNA octanucleotide sequence in the native virion of rhinovirus 14 was initially built using the program Coot and refined with restraints using the program ISOLDE (78). Structural comparisons were performed in Chimera (73). Hydrogen bonds, salt bridges, and residues involved in the binding interface and buried surface areas were calculated using the program PDBePISA (https://www.ebi.ac.uk/pdbe/pisa/). Roadmaps were produced using the program Rivem (81).

Purified samples of ICAM-1 and the complex of rhinovirus 14 with ICAM-1 were digested with alpha-lytic protease (EC 3.4.21.12, Sigma-Aldrich catalog No. A6362) for 2 h at 37 C with shaking at 700 rpm. Half of the volume of each sample was then reduced using 10 mM DTT (for 45 min at 57 C with shaking at 700 rpm). After adding polyethylene glycol to a final concentration of 0.001%, the peptides were extracted from the vials using 25% formic acid/acetonitrile (1:1 vol/vol mixture) and vacuum concentrated. The peptide mixture was subjected to liquid-chromatography-mass spectrometry (LC-MS)/MS analysis using a RSLCnano system (ThermoFisher Scientific) coupled to an Impact II Qq-Time-Of-Flight mass spectrometer (Bruker). Prior to LC separation, peptides were online concentrated in a trap column (100 m 30 mm) filled with 3.5 m X-Bridge BEH 130 C18 sorbent (Waters). The peptides were separated using an Acclaim Pepmap100 C18 column (3 m particles, 75 m 500 mm; ThermoFisher Scientific) by the following LC gradient program (mobile phase A: 0.1% formic acid in water; mobile phase B: 0.1% formic acid in 80% acetonitrile; 300 nl/min): the gradient elution started at 1% of mobile phase B and increased to 56% over the first 50 min, then increased linearly to 80% of mobile phase B over the next 5 min and remained at this state for the final 10 min. Equilibration of the trapping column and the column was done prior to sample injection into the sample loop. The analytical column outlet was directly connected to a CaptiveSpray nanoBooster ion source (Bruker). The nanoBooster was filled with acetonitrile. MS and MS/MS spectra were acquired in a data-dependent strategy with a 3 s cycle time. The mass range was set to 150 to 2,200 m/z and precursors were selected from 300 to 2,000 m/z. The acquisition speed of MS and MS/MS scans was 2 Hz and 4 to 16 Hz, respectively. The speed of MS/MS spectra acquisition was based on precursor intensity. The preprocessing of the mass spectrometric data including recalibration, compound detection, and charge deconvolution was carried out using DataAnalysis software (version 4.2 SR1; Bruker).

The obtained data were searched with an in-house Mascot search engine (version 2.4.1; Matrixscience) against a custom database involving the ICAM-1 sequence and cRAP entries (downloaded from https://www.thegpm.org/crap/). The database searches were done without enzyme specificity and with oxidation (M) as a variable modification. The mass tolerances for peptides and MS/MS fragments were 10 ppm and 0.1 Da, respectively. Only peptides with a statistically significant peptide score (P < 0.05) were considered, and the obtained MS/MS data were validated manually.

Multiple sequence alignment of capsid proteins of selected viruses from the family Picornaviridae was performed in the Clustal Omega server (82). The multiple sequence alignment was visualized in the software Jalview 2.11.1.3 (83).

The cryo-EM maps and coordinates were deposited under the following accession codes: virion of rhinovirus 14 at neutral pH Electron Microscopy Data Bank EMD-12171 and PDB 7BG6; rhinovirus 14ICAM-1 complex at neutral pH EMD-12172 and PDB 7BG7; rhinovirus 14 in native conformation at acidic pH EMD-12599 and PDB 7NUQ; empty particle of rhinovirus 14 at acidic pH EMD-12597 and PDB 7NUO; rhinovirus 14 in native conformation at acidic pH originating from complex with ICAM-1 EMD-12596 and PDB 7NUN; activated particle originating from complex with ICAM-1 at acidic pH EMD-12594 and PDB 7NUL; empty particle originating from complex with ICAM-1 at acidic pH EMD-12595 and PDB 7NUM; and open particle originating from complex with ICAM-1 at acidic pH EMD-12598.

We gratefully acknowledge the Cryo-Electron Microscopy and Tomography Core Facility of Central European Institute of Technology (CEITEC) supported by Ministry of Education, Youth, and Sports of the Czech Republic (MEYS CR) (Grant LM2018127). This research was carried out under the project CEITEC 2020 (Grant LQ1601), with financial support from the MEYS of the Czech Republic under National Sustainability Program II. This work was supported by the IT4I project (Grant CZ.1.05/1.1.00/02.0070) and funded by the European Regional Development Fund and the national budget of the Czech Republic via the Research and Development for Innovations Operational Programme (RDI-OP) as well as the MEYS via Grant LM2011033. The research leading to these results received funding from Czech Science Foundation Grant GX19-25982X to P.P.

Author contributions: D.H. and P.P. designed research; D.H., L.., A.A., and O.. performed research; D.H., T.F., M.G., O.., Z.Z., and P.P. analyzed data; and D.H., T.F., and P.P. wrote the paper.

The authors declare no competing interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2024251118/-/DCSupplemental.

More:
ICAM-1 induced rearrangements of capsid and genome prime rhinovirus 14 for activation and uncoating - pnas.org

INTRODUCTION

Most breast cancer (BRCA)mutated cancers acquire resistance toward chemotherapeutic agents such as cisplatin and poly(adenosine 5-diphosphateribose) polymerase inhibitors (PARPi) (1). At present, besides the restoration of homologous recombination (HR), loss of poly(adenosine 5-diphosphateribose) glycohydrolase or acquired protection of stalled replication forks provides a mechanism that can promote drug resistance in a BRCA-deficient genetic background (14). However, identification of additional mechanisms underlying resistance to chemotherapeutics can provide a real opportunity to improve therapies in BRCA-deficient cancer patients.

BRCA proteins play a genetically separable role at the site of double-stranded breaks (DSBs) where they mediate an error-free HR repair and at replication forks where they facilitate protection of reversed forks from extensive nuclease-mediated degradation, to maintain genome stability (2, 3, 57). Similarly, the factors of nonhomologous end joining (NHEJ), an error-prone pathway, along with their role in repair of DSBs have been shown to associate with stalled forks either for their protection or to promote their restart (2, 8, 9). However, the factors involved in limiting fork stalling and subsequent restarting of forks upon endogenous or exogenously induced replication stress are poorly understood.

Proliferating cell nuclear antigen (PCNA) is a DNA clamp that associates with the active replication forks and functions as a processivity factor for DNA polymerases to carry out the DNA synthesis process but dissociates from stalled forks via an active unloading mechanism (8, 10, 11). During replication, PCNA rings are repeatedly loaded and unloaded by the replicating clamp loader replication factor C (RFC) complex and an alternative PCNA ring opener, ATAD5 (ELG1 in yeast)RFC-like complex (ATAD5-RLC). ATAD5-RLC unloads replication-coupled PCNA after ligation of Okazaki fragment and termination of DNA replication (1214). Maintenance of the delicate balance of PCNA levels onto DNA is crucial because PCNA levels can influence chromatin integrity (15, 16) and persistent PCNA retention on DNA causes genome instability (1719). However, mechanisms by which PCNA levels are regulated on replicating chromatin and the factors involved in this process still remain elusive.

Here, we uncover a novel function of human SMARCAD1 in regulating the fine control of PCNA levels at forks, which is required for the maintenance of replication stress tolerance and genome stability. SMARCAD1, a DEAD/H box helicase domain protein, belongs to a highly conserved adenosine 5-triphosphatedependent SWI/SNF (switch/sucrose non-fermentable) family of chromatin remodelers. Adenosine triphosphatase (ATPase) remodeling activity of SMARCAD1 is crucial for its function in HR repair and in maintenance of histone methyl marks for reestablishment of heterochromatin (20, 21).

In this study, we generated a separation-of-function mutant of human SMARCAD1, efficient in its HR function but defective in its interaction with the replication machinery. This strategy led to uncover a previously unrecognized role of SMARCAD1 in maintaining stability of active (unperturbed and restarted) replication forks, which is responsible for mediating resistance toward replication poisons. In the absence of SMARCAD1, replication fork progression requires BRCA1 to maintain the integrity of stalled forks to allow their restart. Furthermore, SMARCAD1 maintains replication fork stability and cellular viability in BRCA1-deficient nave or chemoresistant mouse breast tumor organoids, highlighting its essential role in the survival of tumor cells. Our results suggest a conserved role of SMARCAD1 and BRCA1 proteins at replication forks, SMARCAD1 at active forks while BRCA1 at stalled forks, to safeguard replication fork integrity and ensure genome stability.

Most factors associated with the active replisome are required to maintain the stability of the replication forks and could also be important for mediating efficient restart after stalling. To specifically identify novel factors involved in the stability of unperturbed forks, we performed isolation of proteins on nascent DNA (iPOND) coupled to stable isotope labeling with amino acids in cell culture (SILAC)based quantitative mass spectrometry (8, 9). Mouse embryonic stem cells (mESCs) were used to compare the proteins present at unperturbed active replication forks versus hydroxyurea (HU)induced stalled replication fork (fig. S1A). In total, 1443 common proteins were identified from two independent experiments (fig. S1, B and C). Consistent with previous reports, we observed a greater than twofold increase in replication stress response proteins, including RAD51 and BRCA1, at stalled forks (Fig. 1A) (8, 9). Levels of core components of the replicative helicase, such as mini-chromosome maintenance protein 6 (MCM6), remained largely unchanged during early replication stress (Fig. 1A). As shown previously (8), PCNA was enriched ~2-fold at the unperturbed forks when compared to the stalled forks, confirming that PCNA associates preferentially with active forks and showing proof of principle of this approach (Fig. 1A and fig. S1B). Among 66 proteins showing preferential enrichment at unperturbed replication forks (fig. S1C), we identified SMARCAD1, a conserved SWI/SNF chromatin remodeler (Fig. 1A and fig. S1B). KAP1/TRIM28 (KRAB-associated protein 1/tripartite motif-containing 28), a previously reported SMARCAD1 interacting partner, showed no preferential enrichment, a behavior that is similar to that of the MCM6 helicase, suggesting an additional and independent role of SMARCAD1 in replication fork dynamics (Fig. 1A) (8).

(A) Bar graph showing fold up-regulation of selected proteins under unperturbed (no HU)/HU-treated conditions based on SILAC H:L ratios. (B and C) Representative high-content microscopy images showing the colocalization of chromatin-bound (B, left) PCNA or (C, top) SMARCAD1 with 5-ethynyl-2-deoxyuridine (EdU) under given conditions. Box plots showing the mean intensity of (B, right) PCNA or (C, bottom) SMARCAD1 foci overlapping with EdU are shown (note that for HU condition, EdU labeling was performed before HU treatment). Numbers above each scatterplot indicate the mean intensity. n > 3000 mid-late S phase cells; ****P 0.0001, unpaired t test. a.u., arbitrary units. Scale bar, 5 m. (D) Schematic overview of protein domains in full-length and N-SMARCAD1. Asterisk represents the stop codon. (E) Immunoblot showing SMARCAD1 levels under indicated conditions. Tubulin is the loading control. Asterisk represents a nonspecific band. (F) Chromatin immunoprecipitation (IP) of SMARCAD1 in wild-type (WT) and N-SMARCAD1 cells, followed by immunoblot for SMARCAD1 and PCNA. Asterisk represents a nonspecific band. IgG, immunoglobulin G. (G) Representative images showing the expression of SMARCAD1 and EdU in WT, N-SMARCAD1, K528R-SMARCAD1, and SMARCAD1/ cells. Scale bar, 5 m. (H) Quantification of colony survival assay (means + SD; n = 3) in WT, N-SMARCAD1, and SMARCAD1/ cells is shown for given conditions. HU was given for 48 hours. ns, nonsignificant; unpaired t test.

To confirm our iPOND-SILAC mass spectrometry data and to assess whether the preferential enrichment of SMARCAD1 and PCNA at unperturbed replication forks is conserved across species, we performed immunofluorescence assays to measure the localization of these proteins with respect to the sites of replication in MRC5 human fibroblast cells. Sites of active DNA replication were labeled with 5-ethynyl-2-deoxyuridine (EdU), and localization of the chromatin-bound fraction of SMARCAD1, PCNA, and RAD51 within the sites of replication was measured in the presence or absence of HU using single cellbased, high-content microscopy. Consistent with the results of iPOND-SILAC mass spectrometry in mESCs, we observed that chromatin-bound SMARCAD1 and PCNA foci specifically colocalized with EdU. However, upon HU treatment, both these proteins showed a significant decrease in intensity at replication sites, suggesting that both SMARCAD1 and PCNA associate with unperturbed replication forks but dissociate from stalled forks (Fig. 1, B and C). As expected, RAD51 was found to be enriched significantly at replication sites upon HU treatment, suggesting a positive enrichment at stalled forks in contrast to PCNA and SMARCAD1 (fig. S1D) (8).

The N-terminal region of SMARCAD1 has been shown to be responsible for the PCNA-mediated localization of SMARCAD1 to replication forks (21, 22). To explore the role of this interaction at replication forks, we generated a SMARCAD1 mutant, using MRC5 cells, in which the canonical start site is disrupted and translation begins downstream at the next available start codon (Fig. 1D). Expression of this mutant gene results in a 137amino acid N-terminally truncated product, designated as N-SMARCAD1 that lacks the region responsible for its interaction with PCNA (22). The N-SMARCAD1 protein is approximately 100 kDa in size (Fig. 1E) and retains the downstream CUE1, CUE2, ATPase, and helicase domains (fig. S1E), crucial for chromatin remodeling and DNA repair functions (21, 23), intact. For comparative analysis, we also generated a complete SMARCAD1 knockout (SMARCAD1/) by replacing the SMARCAD1 gene with an mClover [a green fluorescent protein (GFP) variant] reporter gene (Fig. 1E). Both quantitative reverse transcription polymerase chain reaction (qRT-PCR) assays of the SMARCAD1 coding region and RNA sequencing (RNA-seq)based transcriptome analysis of cells containing the full length [wild-type (WT)] and those containing the truncated form (N-SMARCAD1) confirmed that expression levels of the two SMARCAD1 alleles were nearly identical (fig. S1, E and F). As expected, cells containing the knockout, SMARCAD1/, showed a lack of transcripts specific to the coding region of the gene.

To test the interaction between PCNA and the N-SMARCAD1 mutant, we generated a heterogeneously expressed GFP-tagged PCNA allele in both WT and N-SMARCAD1 genetic backgrounds (fig. S1G). Cross-linked chromatin immunoprecipitation (IP) of GFP-tagged PCNA confirmed that although N-SMARCAD1 associates with chromatin, it did not interact with GFP-PCNA, whereas the full-length WT SMARCAD1 protein retains this interaction (fig. S1H), as previously reported (21). Similarly, reverse chromatin IP of WT-SMARCAD1 and N-SMARCAD1 protein confirmed the lack of interaction between PCNA and N-SMARCAD1 protein (Fig. 1F). The interaction between SMARCAD1 and PCNA was also confirmed by immunoprecipitating PCNA under native conditions, again verifying the loss of interaction between N-SMARCAD1 and PCNA (fig. S1I). To determine whether a SMARCAD1 interaction with PCNA is required for its association with replication sites, we performed an immunofluorescence analysis to measure the localization of SMARCAD1 mutants at sites of DNA replication marked with EdU. Our data show that chromatin-bound foci of full-length SMARCAD1 colocalized with EdU+ sites, as previously reported (Fig. 1G) (21). As expected, no specific SMARCAD1 signal could be seen in SMARCAD1/ cells. Consistent with our IP data (Fig. 1F and fig. S1H), N-SMARCAD1 showed nuclear localization but no colocalization with EdU signals (Fig. 1G), suggesting that N-SMARCAD1 associates with chromatin but is not enriched at sites of replication.

Next, we sought to determine whether loss of SMARCAD1 association with replication forks affects cellular resistance to fork stalling agents such as HU, cisplatin, or the PARPi, olaparib. Both N-SMARCAD1 and SMARCAD1/ cells showed significant sensitivity to the replication poisons, suggesting that the presence of SMARCAD1 at replication forks is crucial for resistance to replication stress (Fig. 1H). To further explore the role of SMARCAD1 during DNA replication, we analyzed S phase progression by measuring EdU incorporation using high-content microscopy. We imaged >2000 cells and plotted for quantitative image-based cytometry analysis (QIBC) to obtain single cellbased cell cycle profile (24). Both N-SMARCAD1 and SMARCAD1/ cells displayed reduction in EdU intensities relative to WT cells, suggesting that loss of SMARCAD1 at forks causes DNA replication defects (fig. S1J).

Because the loss of SMARCAD1 causes defects in HR repair of DSBs due to inefficient DNA end resection (20, 23, 25), we next tested whether cells expressing N-SMARCAD1 also exhibited defects in HR repair. We measured HR efficiency using a direct repeatgreen fluorescent protein (DR-GFP) reporter assay (26). N-SMARCAD1 cells had an HR efficiency similar to that of WT (Fig. 2A). However, HR efficiency was significantly reduced in both WT and N-SMARCAD1 cells when SMARCAD1 was knocked down in these cells using small interfering RNA (siRNA), similar to that observed for BRCA1 knockdown (Fig. 2A). These data suggest that, although the complete loss of SMARCAD1 results in defective HR, expression of the truncated N-SMARCAD1 retains HR proficiency. In addition, chromatin fractionation and observation of RAD51 focus formation by immunofluorescence using high-content microscopy both showed a remarkable increase in chromatin-bound RAD51 upon olaparib treatment in both WT and N-SMARCAD1 but not in SMARCAD1-deficient cells (Fig. 2, B and C). These data further confirm that N-SMARCAD1 cells are proficient in the loading of RAD51 in response to DNA damage unlike SMARCAD1/. Furthermore, N-SMARCAD1 cells are also proficient in RAD51 foci formation similar to WT upon ionizing radiation (IR)induced DSB formation (fig. S2A). Consistently, N-SMARCAD1 does not show sensitivity to IR treatment in contrast to HR-defective BRCA1-depleted cells or NHEJ-defective p53-binding protein 1 (53BP1)depleted cells (fig. S2, A and B). Surprisingly, however, both N-SMARCAD1 and SMARCAD1 complete knockout showed similar sensitivity toward drugs, causing replication stress, olaparib, cisplatin, and HU (Figs. 1H and 2D and fig. S2C), arguing in favor of an uncoupling between HR repair function and resistance to replication stress in the N-SMARCAD1 cells, corroborating it to be a separation-of-function mutant.

(A) Quantification of HR efficiency using DR-GFP reporter assay. DR-GFP reporter and pcBASceI constructs were cotransfected into WT and N-SMARCAD1 MRC5 cells. Relative HR efficiency representing the percentage of GFP+ cells normalized to the transfection efficiency of the respective cell line is plotted. The means and SD are represented. n = 3; ***P 0.001, **P 0.01, and *P 0.05, unpaired t test. (B) Immunoblot showing the chromatin-bound fraction of RAD51 upon 7 M olaparib treatment for 24 hours in WT, N-SMARCAD1, and SMARCAD1/ cells. H1.2 is used as a loading control. The numbers below the blots show the fold change of RAD51 after normalization with H1.2, as compared to WT untreated samples, for the given blot (n = 3). (C) Top: Representative high-content microscopy images depicting RAD51 foci formation upon 7 M olaparib treatment for 24 hours in WT, N-SMARCAD1, and SMARCAD1/ cells. Scale bar, 50 m. Bottom: Quantification of RAD51 foci upon 7 M olaparib treatment for 24 hours using high-content microscopy. A total of 4700 cells were analyzed under each condition. ****P 0.001, one-way analysis of variance (ANOVA). Number above represents the fold change of RAD51 foci upon olaparib treatment compared to its own untreated samples. (D) Quantification of colony survival assay in WT, N-SMARCAD1, and SMARCAD1/ cells treated with different concentrations of olaparib. Error bars stand for SD (n = 3). ***P 0.001 and **P 0.01, unpaired t test.

We also performed transcriptome analysis to test whether the drug sensitivity observed in SMARCAD1 mutant cells could be a result of transcription deregulation of DNA damage response (DDR) genes in these cells, because transcription may be affected by its chromatin remodeling role. We observed a mild dysregulation in a subset of non-DDR genes (1.5-fold change in expression) in either N-SMARCAD1 or SMARCAD1/ cells, whereas almost no anomalous expression was observed in either mutant for a set of DDR genes (n = 179) (27), which included both HR and NHEJ DDR genes (fig. S2D). This suggests that the function of SMARCAD1 in promoting drug tolerance is unrelated to its role in heterochromatin maintenance or in transcriptional regulation. Furthermore, the efficient loading of RAD51 and the HR proficiency of cells expressing N-SMARCAD1, in contrast to those lacking SMARCAD1, is most likely not due to a differential transcriptome or cell cycle profile but due to the presence of intact CUE and ATPase/helicase domains in N-SMARCAD1 that are essential for its HR function (20, 25). The loss of PCNA interaction and association with the fork is the main cause for SMARCAD1-depleted cells to show sensitivity toward replication stressinducing drugs.

SMARCAD1 mutants displayed moderate but significant defects in progression through S phase (fig. S1J). To further monitor the dynamics of individual replication forks, we performed DNA fiber assay. We sequentially labeled WT and SMARCAD1 mutants (N-SMARCAD1 and SMARCAD1/) cells with CldU (red) and IdU (green), followed by track length analysis. N-SMARCAD1 cells exhibited a significant difference in the track lengths of both 5-chloro-2-deoxyuridine (CldU) and 5-iodo-2-deoxyuridine (IdU) in comparison to WT but similar to SMARCAD1/ cells (Fig. 3A). To test the possibility that accumulation of DNA damage over time in the mutant cells was causing the replication fork defect observed, we also analyzed fork progression in cells in which SMARCAD1 was depleted transiently with siRNA. The transient knockdown of SMARCAD1 resulted in similar fork progression defects than the one observed in N-SMARCAD1 and SMARCAD1/ (Fig. 3A). This suggests that SMARCAD1 directly facilitates the progression of replication forks.

(A) Top: Schematic of replication fork progression assay with CldU and IdU labeling in WT, N-SMARCAD1, and SMARCAD1/ (KO) cells. Representative DNA fibers for each condition are shown below the schematic. Scale bar, 5 m. Bottom: CldU (red) and IdU (green) track length (in micrometers) distribution for the indicated conditions. n = 3; ****P 0.0001, Kruskal-Wallis followed by Dunns multiple comparison test. (B) Top: Schematic of replication fork restart assay. Representative DNA fibers for each condition are shown below the schematic. Scale bar, 5 m. Bottom: CldU (red) and IdU (green) track length (in micrometers) distribution for the indicated conditions. ****P 0.0001, unpaired t test. All DNA fiber experiments presented here were repeated three times with similar outcomes. (C) Representative image of a normal (left) and a reversed replication fork (right) observed by electron microscopy (EM). D, daughter strand; P, parental strand; R, reversed arm. (D) Bar chart representing the percentage of fork reversal in WT and N-SMARCAD1 cells under untreated condition. n = 3; ****P 0.0001, unpaired t test. (E) Representative electron micrographs of single-stranded DNA (ssDNA) gaps. Green and blue arrows point toward ssDNA gaps at the fork and behind the fork, respectively. (F) Bar chart representing the distribution of ssDNA gaps behind the fork in WT and N-SMARCAD1 under untreated condition and 1 hour after release from 1 mM HU treatment. Chi-square test of trends was performed to assess the significance of internal ssDNA gaps between WT and N-SMARCAD1. n = 3; ****P < 0.0001. (G) Top: Pulsed-field gel electrophoresis (PFGE) analysis for DSBs shows WT and N-SMARCAD1 cells with and without 4 mM HU treatment for 3 hours and upon 16 hours of release after the HU treatment. Bottom: Quantification from the three independent experiments showing DSB levels.

Because SMARCAD1 deficiency displayed significant replication defects during unperturbed replication (Fig. 3A and fig. S1J), we wondered whether SMARCAD1 also plays a role in the progression after fork stalling. To assess the overall rate of DNA synthesis upon replication stress, we treated cells with 1 mM HU for an hour. The replication rate after stress was measured by allowing the EdU incorporation for various time points after release from HU and EdU intensities that were measured in >3000 cells using high-content microscopy. Upon 30 min of release from HU, we observed a mild reduction in EdU incorporation in N-SMARCAD1 cells. However, the reduction in EdU incorporation became more evident at later time points in N-SMARCAD1 cells (fig. S2E). To further verify this, we performed a fork restart assay using DNA fiber analysis. Cells were labeled with CldU, followed by a mild dose of HU (1 mM) treatment for an hour to stall the forks and subsequently released into IdU. Consistently, we observed significant defects in CldU track lengths, representing an internal control for unperturbed forks (Fig. 3B) similar to those observed in the fork progression assay performed in Fig. 3A. However, analysis of IdU track lengths representing stressed forks revealed an even higher shortening of the track lengths in N-SMARCAD1 cells, suggesting a more severe defect in the progression or restart of stalled forks (Fig. 3B). In addition, upon analysis of fork restart efficiency, we observed a significant difference between stalled and restarted forks in N-SMARCAD1 cells (25% restarted) when compared to WT cells (60% restarted) after 15 min of release from HU stress, whereas this difference significantly reduced after 30 min of release from HU (86% WT and 74% N-SMARCAD1) (fig. S2F, left), but the progression of restarted forks remained severely defective in N-SMARCAD1 cells (fig. S2F, right). These data suggest that forks restart in absence of SMARCAD1 with moderate delay but further show severe defects in progression of stressed forks. Thus, SMARCAD1 mediates both the efficient restart and progression of replication forks, which also supports the finding that cells lacking SMARCAD1 are sensitive to replication stressinducing agents.

To investigate whether the delayed restart and poor fork progression upon release from HU stress results in increased single-stranded DNA (ssDNA) levels in the N-SMARCAD1 cells, we analyzed RPA32, a surrogate for ssDNA, by chromatin fractionation. Upon HU treatment, the replication protein A, 32 kDa subunit (RPA32) signals were enhanced in WT cells (fig. S2G). Untreated N-SMARCAD1 cells showed a marked increase in chromatin-associated RPA32 compared to untreated WT cells, suggesting that the accumulation of under-replicated regions in the genome could be due to defects in normal fork progression (Fig. 3A and fig. S2G). However, a significant increase in RPA32 levels could be seen upon HU treatment and upon release from HU-mediated block in N-SMARCAD1 cells, suggesting that loss of SMARCAD1 at forks causes significant accumulation of under-replicated regions (fig. S2G). Furthermore, unperturbed N-SMARCAD1 cells showed significant phosphorylation of checkpoint kinase 1 (CHK1) but not ataxia-telangiectasia-mutated (ATM) protein, suggesting that absence of SMARCAD1 at the unperturbed forks specifically leads to activation of ataxia telangiectasia and Rad3-related protein (ATR) mediated checkpoint pathway, further corroborating replication stress in these cells (fig. S2H).

DNA replication stress, exogenous or endogenous, results in reversal of forks (2831), and we hypothesized that slower fork progression and accumulation of RPA in N-SMARCAD1 mutants under unperturbed conditions could be a result of frequent fork stalling that stabilizes into reversed forks. To test this hypothesis, we visualized replication intermediates formed in vivo using electron microscopy (EM) (9) in WT and N-SMARCAD1 mutant cells. We observed a higher frequency of reversed forks in N-SMARCAD1 than in WT cells, suggesting frequent stalling and remodeling of forks even under unperturbed conditions (Fig. 3, C and D). Moreover, we also observed an increase in the percentage of ssDNA gaps accumulated in daughter strands behind the fork of N-SMARCAD1 cells relative to WT, which further enhanced markedly upon release from HU-mediated stress (Fig. 3, E and F). We also quantified the length of ssDNA at the fork that determines nascent strand processing activity at the fork, which showed no significant difference in N-SMARCAD1 than compared to WT (fig. S2I). Together, these data further corroborate that the role of SMARCAD1 is critical in limiting fork stalling under unperturbed conditions and promoting efficient fork restart and fork progression globally upon replication stress.

We further investigated whether the increased accumulation of ssDNA upon replication stress leads to an increase in DSBs that would contribute to genome instability. To evaluate the accumulation of DNA damage, we performed pulsed-field gel electrophoresis (PFGE) to measure the physical presence of DSBs. There was no obvious increase in the level of DSBs upon the stalling of forks induced by HU treatment in either WT or N-SMARCAD1 cells, suggesting that forks stalled for 3 hours with HU treatment do not immediately collapse and convert into DSBs. These data were further supported by the efficient loading of RAD51 observed at stalled forks induced upon HU treatment in N-SMARCAD1 similar to WT (fig. S2J). However, after release from replication stress for 16 hours, a marked increase in the signal of broken DNA fragments can be observed in N-SMARCAD1 cells in comparison to WT cells (Fig. 3G). Together, these data suggest a role of SMARCAD1 at replication forks that is crucial to maintain genome integrity upon replication stress.

Because N-SMARCAD1 lacks interaction with PCNA (Fig. 1F and fig. S1H) and also display defects in fork progression (Fig. 3, A and B), we wondered whether the loss of SMARCAD1 at replication fork affects the PCNA clamp that acts as processivity factor for efficient DNA synthesis. We therefore measured the chromatin-bound PCNA levels in replicating cells labeled with EdU to observe the dynamics of PCNA localization during DNA synthesis. QIBC analysis showed significant reduction in chromatin-bound PCNA levels in replicating cells of N-SMARCAD1 in comparison to WT (Fig. 4A), whereas the total levels of PCNA protein were not affected (Fig. 4B). These data suggest that absence of SMARCAD1 at forks affect PCNA levels at the forks. A similar reduction in PCNA levels at replication sites was observed in SMARCAD1/ cells, suggesting that N-SMARCAD1 behaves similar to the complete loss of SMARCAD1 protein and that N-SMARCAD1 does not display a dominant negative phenotype (fig. S3A). We further monitored the impact of HU-mediated replication stress on PCNA recovery. Because PCNA dissociates from HU-mediated stalled forks (Fig. 1, A and B) (8), we hypothesized that aggravated defects in fork restart in N-SMARCAD1 were due to poor recovery of PCNA at the forks upon release from HU. Using QIBC analysis, we simultaneously assessed the EdU incorporation and PCNA recovery upon HU stress using an average of 3000 cells per condition (Fig. 4C). WT replicating cells showed significantly reduced PCNA levels upon 1 mM HU treatment for an hour and had recovered to their untreated levels by 45 min of release from HU stress (Fig. 4C and fig. S3B). Consistently, we observed reduced PCNA levels and reduced EdU incorporation in N-SMARCAD1 cells in comparison to WT cells under the untreated condition. N-SMARCAD1 cells showed severe defects in recovery of PCNA levels and reduced EdU incorporation upon release from HU-mediated replicative stress (Fig. 4C and fig. S3, B and C). The significantly reduced EdU incorporation is consistent with the results of the DNA fiber assay of fork restart upon HU stress, which revealed severe defects in the progression of restarted forks in N-SMARCAD1 cells (Fig. 3B). These data suggest that SMARCAD1 participates in the maintenance of PCNA levels at the unperturbed forks. Moreover, under stressed conditions, the absence of SMARCAD1 results in poor recovery of PCNA at restarting stalled forks, which subsequently causes inefficient fork restart and severe defects in fork progression upon replication stress.

(A) Left: Representative confocal images showing chromatin-bound PCNA (red) in EdU+(green) WT and N-SMARCAD1 MRC5 cells. Nucleus was stained with 4,6-diamidino-2-phenylindole (DAPI) (blue). Scale bar, 20 m. Right: QIBC analysis of the chromatin-bound PCNA in WT and N-SMARCAD1 cells. G01, S, and G2-M phase cells are labeled in red, blue, and green, respectively. Dashed lines represent the mean chromatin-bound PCNA intensity of S phase cells in WT cells. (B) Immunoblot showing the total level of PCNA in WT and N-SMARCAD1 cells. Tubulin is used as a loading control. Numbers below represent the quantification of PCNA level after normalized to the loading control. (C) QIBC analysis of PCNA versus EdU is shown in WT and N-SMARCAD1 cells in untreated, 1 mM 1-hour HU block, and 45-min release after HU conditions (note that for the HU block condition, EdU labeling was performed before HU treatment). A total of >1800 S phase cells were plotted under each condition. The color gradient represents the density of the cells. (D) Quantification of half-life of the GFP-PCNA fluorescence decay in GFP-tagged PCNA knock-in (KI) WT and N-SMARCAD1 clones, means S.D. ****P 0.0001, ***P 0.001, and **P 0.01, unpaired t test. (E) Immunoblot showing the whole-cell extract (WCE) and chromatin-bound fraction of RFC1, RFC4, and ATAD5 in WT and N-SMARCAD1 cells. H1.2 is used as a loading control. (F and G) Foci analysis of (F) chromatin-bound PCNA intensity and (G) EdU intensity in si-control and various concentrations of si-ATAD5treated WT and N-SMARCAD1 cells. ****P 0.0001 and **P 0.01, unpaired t test. (H) CldU (red) and IdU (green) track length (in micrometers) distribution for replication fork restart assay. ****P 0.0001, ***P 0.001, and **P 0.01, Kruskal-Wallis followed by Dunns multiple comparison test.

We further determined the dynamics of PCNA in replicating WT and N-SMARCAD1 cells using an inverse fluorescence recovery after photobleaching (iFRAP) live-cell imaging assay. iFRAP is an adapted FRAP approach optimized to analyze differences of dissociation rates (Koff) and involves continuous bleaching to quench the total nuclear fluorescence of a GFP-tagged protein with the exception of a small predefined area. Using this approach, we could determine the residence time of GFP-PCNA at the replication foci (unbleached area) as a direct readout of its turnover (fig. S3D). We performed iFRAP on GFP-tagged PCNA expressed from its endogenous allele in both WT and N-SMARCAD1 cell types (fig. S1G). We observed nearly twofold shorter residence times for GFP-tagged PCNA foci in N-SMARCAD1 cells compared to WT cells (Fig. 4D and fig. S3D). These data clearly suggest that the turnover of PCNA at replication forks is severely increased in the absence of SMARCAD1 at the forks, which may be caused by either a defect in the loading or unloading of PCNA in the absence of SMARCAD1 at the replication forks.

To further test this hypothesis, we performed chromatin fractionation to observe the chromatin-associated fraction of subunits of the PCNA loader, RFC (RFC1/RFC2-5), and of the unloader, RLC (ATAD5/RFC2-5) complex subunits (13, 32). We observed no obvious change in the level of RFC1, a major subunit of the RFC complex, in either cell type with or without HU treatment (Fig. 4E). The chromatin association of RFC4, a subunit shared between the RFC and RLC complexes, and that of ATAD5, a major subunit of the RLC complex, were found to be significantly enhanced in chromatin-bound fraction of N-SMARCAD1 cells, while the total level of these proteins as shown in whole-cell extracts remains similar to WT, which is also supported by the transcriptome analysis of these proteins (Fig. 4E and fig. S3E). This finding suggests that the increased chromatin binding of the PCNA-unloader ATAD5-RLC causes the increased release of PCNA in the absence of SMARCAD1. Next, we tested whether depleting ATAD5 levels might restore normal PCNA chromatin association and reduce replication defects in N-SMARCAD1 cells. Consistent with previous reports (33), we observed enhanced PCNA levels at replicating sites in WT cells and retention time of PCNA using iFRAP, upon strong ATAD5 knockdown (fig. S3, F to H). However, as previously reported (33), the strong reduction of ATAD5 significantly reduced the overall EdU incorporation even in WT cells, suggesting that the enhanced accumulation of PCNA at forks also affects overall DNA synthesis (fig. S3I). Therefore, we titrated the knockdown of ATAD5 in N-SMARCAD1 cells to bring PCNA levels equivalent to WT, using lower concentrations of si-ATAD5 (Fig. 4F and fig. S3J). We observed that 30- and 45-pmol concentrations of si-ATAD5 resulted in PCNA and EdU levels in N-SMARCAD1 similar to WT levels (Fig. 4, F and G). Further, using the controlled depletion of ATAD5 (45 pmol of si-ATAD5), we also observed the rescue in fork progression and fork restart efficiency (Fig. 4H). Similarly, the enhanced accumulation of PCNA and lower EdU incorporation with stronger depletion of ATAD5 could also be rescued by ectopically expressing Flag-tagged ATAD5 in WT and N-SMARCAD1 cells (fig. S3, K and L) (33). Together, these data suggest that fine regulation of ATAD5 levels at replication forks is required to maintain fine-controlled PCNA levels that maintain efficient DNA synthesis in cells.

Having established the role of SMARCAD1 at the replication forks, we further investigated the mechanism of how SMARCAD1 promotes replication fork progression. Earlier studies have shown a role for SMARCAD1 in displacing 53BP1 from the site of DSBs to promote HR repair (20). Moreover, SMARCAD1 and 53BP1 show contrasting enrichments at unperturbed versus stalled replication forks, shown by iPOND-SILAC mass spectrometry (Fig. 1A and table S1) (8). We further validated the enrichments of 53BP1 at stalled forks versus restarted forks using fluorescence microscopy in WT cells (Fig. 5A). The data clearly showed 53BP1 colocalization with EdU mainly upon HU treatment, suggesting its enrichment at stalled forks in WT cells, whereas upon release from HU stress, the EdU-labeled sites representing restarted forks show clear displacement between 53BP1 and EdU foci (Fig. 5A). We hypothesized that, similar to DSBs (20), SMARCAD1 might prevent 53BP1 to accumulate at active or restarted replication forks by promoting its displacement from the stalled forks. To test this hypothesis, we measured the levels of 53BP1 protein in replicating cells (EdU+) of N-SMARCAD1 compared to WT, in untreated and in cells released from HU stress. We observed a mild but significant increase in 53BP1 levels in replicating cells of N-SMARCAD1, and notably, a significantly higher accumulation of 53BP1 levels could be seen in cells released from HU stress (fig. S4A). We further measured the localization of 53BP1 protein relative to EdU-marked replication sites in N-SMARCAD1 compared to WT cells. Upon HU block, a significant percentage of replicating WT cells showed an overlap between EdU and 53BP1 foci, which significantly reduced upon release from HU stress (Fig. 5B). Significantly higher percentage of N-SMARCAD1 cells showed colocalization of EdU and 53BP1 foci in HU block cells, which remained remarkably higher even upon release from HU stress (Fig. 5B). Supporting this observation, the Pearsons overlap coefficient and Manders (M1/M2) overlap coefficients estimating the significance of overlap between EdU and 53BP1 foci were found to be significantly higher in N-SMARCAD1 than in WT (fig. S4B). Together, these data suggest that SMARCAD1 is required to displace 53BP1 from stalled replication forks possibly to allow their restart.

(A) Top: Representative image showing 53BP1 (green) and EdU (red) in WT cells treated with 4 mM HU for 3 hours (HU block) and 1-hour release after HU block (HU release). Bottom: The average distance between EdU and 53BP1 foci is measured. Error bars represent SD. (B) Top: Representative images showing 53BP1 foci in EdU-positive cells under indicated conditions. Bottom: Quantification of percentage of cells showing colocalization between EdU and 53BP1 is shown. ***P 0.001 and *P 0.05, unpaired t test. (C) Top: Schematic of fork restart assay. Bottom: CldU and IdU track length (in micrometers) distribution for the indicated conditions. n = 3; ****P 0.0001 and *P 0.05, Kruskal-Wallis with Dunns multiple comparison test. (D) Left: The frequency of reversed forks quantified using EM for indicated conditions. Right: Bar chart showing the distribution of ssDNA gaps behind the fork for the indicated conditions. n = 3; ***P 0.001, unpaired t test;****P 0.0001, chi-square test. (E) QIBC analysis of chromatin-bound PCNA and DAPI is shown under given conditions. Cells above dashed lines represent the S phase cells. The red arrows compare the PCNA level. (F) Quantification of colony survival assay under the indicated conditions. means + SD, n = 3; ***P 0.001 and **P 0.01, unpaired t test.

This observation led us to hypothesize that loss of 53BP1 may allow the normal progression of forks in N-SMARCAD1 cells, which shows frequent fork stalling even under unperturbed conditions (Fig. 3C). We, therefore, first investigated the progression rate of unperturbed forks using si-53BP1 in N-SMARCAD1 using a DNA fiber assay. Transient knockdown of 53BP1 completely rescued the fork progression defects of N-SMARCAD1 cells (fig. S4, C and D). In addition, we also performed fork restart assay and found that both IdU track lengths and CldU track lengths, representing stressed (after HU treatment) and nonstressed forks (before HU treatment), respectively, showed complete restoration of fork progression rates in N-SMARCAD1 (Fig. 5C). Consistently, we observed a rescue in accumulation of reversed forks and reduced accumulation of ssDNA gaps behind the fork in N-SMARCAD1 cells upon 53BP1 knockdown condition (Fig. 5D). As the severe defects in restart of replication forks in N-SMARCAD1 were correlated with the poor recovery of PCNA, we next sought to determine whether 53BP1 knockdown would also restore PCNA levels in N-SMARCAD1 cells. Consistently, QIBC analysis showed that upon HU-mediated block, PCNA levels were significantly reduced in replicating cells even upon 53BP1 knockdown. However, PCNA showed a significant recovery in N-SMARCAD1 similar to WT, when released from HU-mediated block (Fig. 5E and fig. S4E) under these conditions. Consistent with the restoration of PCNA levels, we also observed a marked reduction in chromatin-bound ATAD5 levels upon knockdown of 53BP1 in N-SMARCAD1 (fig. S4F), suggesting that 53BP1 further promotes PCNA unloading in absence of SMARCAD1 at forks through ATAD5 activity. The potential protein-protein interaction between 53BP1 and ATAD5 was further confirmed by yeast two-hybrid assay (fig. S4G) and by chromatin IP of 53BP1 (fig. S4H), showing positive interaction in WT cells that further enhances under either HU-induced replication stress conditions in WT cells or under unperturbed conditions of N-SMARCAD1 cells, both of which shows enhanced accumulation of stalled forks (Fig. 3C). We also noticed that the higher molecular weight band of ATAD5 was mainly immunoprecipitated with 53BP1 in chromatin IPs, which was further confirmed by notable reduction in signal of potentially phosphorylated ATAD5 band in cells targeted with si-ATAD5 (fig. S4H). The ATR-mediated phosphorylated form of ATAD5 has been reported to interact with RAD51 at stalled/regressed forks previously (34, 35). Together, these data suggest that 53BP1 interaction with ATAD5 regulates PCNA levels at stalled forks. Because loss of 53BP1 rescued genome instability, as monitored by the reduction of accumulated ssDNA gaps in N-SMARCAD1 (Fig. 5D), we next determined whether 53BP1 knockdown rescues the sensitivity of N-SMARCAD1 cells toward replication poisons. We observed a significant restoration of resistance toward cisplatin and olaparib treatment after the depletion of 53BP1 in N-SMARCAD1 cells (Fig. 5F). Together, these data imply that SMARCAD1 maintains fine PCNA levels by suppressing unscheduled 53BP1 accumulation at the active replication forks and thereby maintains genome stability and replication stress tolerance in the cells.

From these data, we further hypothesized that chromatin remodeling activity of SMARCAD1 is possibly required to displace 53BP1-associated nucleosomes to suppress the untimely accumulation of 53BP1-ATAD5 complex at replication forks. To investigate this, we generated knock-ins of complementary DNA (cDNA)SMARCAD1 that were either WT or contained an ATPase-disabling K528R mutation that can interact with replication forks but is defective in nucleosome remodeling activity, unlike N-SMARCAD1 that does not associate with replication forks at all (Fig. 1G) (20). As expected, we observed a rescue in fork progression defects in N-SMARCAD1 cells when corrected with fully functional SMARCAD1 but not with ATPase-dead K528R SMARCAD1 (fig. S4I). Moreover, K528R SMARCAD1 showed significant defects in fork progression and PCNA levels similar to N-SMARCAD1 (fig. S4, I and J). We further confirmed that defects of 53BP1 displacement at restarted forks observed in N-SMARCAD1 also existed in ATPase-dead SMARCAD1, detected by proximity ligation assay (PLA) approach between EdU and 53BP1 (fig. S4K). Furthermore, we also detected higher levels of 53BP1-associated ubiquitinated histone H2A lysine 15 (H2AK15Ub) nucleosomes at restarted forks in both ATPase-dead and N-SMARCAD1 cells (fig. S4L). These data strongly suggest that the chromatin remodeling activity of SMARCAD1 is required to evict 53BP1-associated nucleosomes to displace 53BP1-ATAD5 complex, preventing PCNA recovery at restarted forks, causing replication fork restart and progression defects.

Our data imply that SMARCAD1-mediated replication fork stability contributes to genome stability in a manner independent of its role in HR repair of DSBs. Similarly, HR-independent roles in the protection of stalled forks during replication stress have been uncovered for BRCA1 and BRCA2 (2, 3, 57). To further test whether SMARCAD1 also protects stalled forks, similar to BRCA1, we investigated fork degradation using DNA fiber assay. Loss of BRCA1 resulted in stalled fork degradation upon 3 hours of exposure to 4 mM HU, while N-SMARCAD1 showed no significant defects in fork protection similar to WT (Fig. 6A). Furthermore, as shown previously, longer exposure of cells to 4 mM HU (up to 8 hours) resulted in a moderate but significant processing of forks in WT cells (36), and we observed similar effects in N-SMARCAD1, while loss of BRCA1 led to severe fork degradation (Fig. 6A). Further, these data also suggest that SMARCAD1 is not defective in the processing of stalled forks, as proposed for its fission yeast homolog (37). Thus, these data along with fork progression data (Fig. 3, A and B), taken together, suggest that replication defects observed in absence of SMARCAD1 are due to defective active replication fork stability and not due to defective stalled fork protection or fork processing activities. Furthermore, in the absence of SMARCAD1, unperturbed cells showed frequent stalling of replication forks without subsequent accumulation of DSBs (Fig. 3, C and G), which could possibly be due to BRCA-mediated fork protection in SMARCAD1 mutant cells. To test this hypothesis, we knocked down BRCA1 transiently from MRC5 WT, N-SMARCAD1, and SMARCAD1/ cells to analyze replication fork dynamics. As previously reported, si-BRCA1 in WT cells showed no significant defects in the progression rate of unperturbed forks (2). However, in N-SMARCAD1 and SMARCAD1/ cells, loss of BRCA1 resulted in significantly shorter track length (fig. S5A), which could not be rescued by loss of 53BP1 (fig. S5B). These data suggest that upon loss of SMARCAD1, BRCA1 is required to maintain progression of forks, possibly by protecting stalled forks from DNA nucleasemediated degradation to allow their restart. To test whether indeed loss of BRCA1 in SMARCAD1 mutants leads to increased DNA damage, we performed QIBC analysis for the DNA-damage marker H2AX and observed significantly enhanced accumulation of DNA damage upon BRCA1 knockdown in both N-SMARCAD1 and SMARCAD1/ mutants compared to single mutants or WT cells (Fig. 6B), suggesting that BRCA1 could be required to protect stalled forks from degradation to prevent DNA damage accumulation.

(A) Top: Schematic of replication fork degradation assay with CldU and IdU labeling. Bottom: Ratio of IdU to CldU tract length was plotted for the indicated conditions. ****P 0.0001 and *P 0.05, Kruskal-Wallis with Dunns multiple comparison test. (B) QIBC analysis of H2AX versus EdU is shown in WT, N-SMARCAD1, and SMARCAD1/cells under si-control and si-BRCA1 conditions. A total of >1000 cells were plotted under each condition. The color gradient represents the H2AX levels in each cell. (C) Top: Schematic of replication fork progression assay with CldU and IdU labeling. Bottom: CldU (red) and IdU (green) track length (in micrometers) distribution for the indicated conditions. n = 3; ****P 0.0001 and ***P 0.001, Kruskal-Wallis followed by Dunns multiple comparison test. (D) Left: Representative images of KB1P (Brca1/;p53/) mouse tumor cells imaged at day 10, after transduction of scramble control shRNA and shSMARCAD1 #1 and #3. Right: Quantification of cell viability using crystal violet staining assay. Error bars stand for +SD. n = 3; ****P 0.0001, ***P 0.001, and **P 0.01, unpaired t test. (E) Top: Representative images of KB1P mouse tumor organoid. Image was taken 5 days after the transduction of scramble control shRNA and shSMARCAD1 #1 and #3. Scale bars, 1000 m. Bottom: Quantification of cell viability using cell titer blue assay. Error bars stand for +SD. n = 3; ***P 0.001, **P 0.01, and *P 0.05, unpaired t test. (F) Top: Schematic of replication fork progression assay. Bottom: CldU (red) and IdU (green) track length (in micrometers) distribution in KB1P mouse tumor cells treated with si-control or si-SMARCAD1. n = 3; ****P 0.0001 and *P 0.05, Kruskal-Wallis followed by Dunns multiple comparison test.

As previously reported, BRCA1 protects stalled forks from degradation mediated by DNA nuclease Mre11 (7). Therefore, to test this hypothesis, we treated cells with Mirin, an inhibitor of DNA nuclease Mre11, and monitored fork progression using a DNA fiber assay. Notably, Mirin treatment completely rescues the severe fork progression defects observed upon loss of BRCA1 in the SMARCAD1 mutant (Fig. 6C). These data suggest that, in the absence of SMARCAD1 stalled forks indeed require BRCA1 protection to allow fork progression and maintain genome integrity.

Previously, SMARCAD1 was reported to play a critical role in the metastasis of triple-negative breast cancer (38, 39). To test whether differential levels of SMARCAD1 expression could be an indicator of patient responses to replication stressinducing platinum chemotherapy, we analyzed patients with high-grade serous ovarian cancer (HGSOC) for their correlation between BRCA1 and SMARCAD1 expression levels to their response to chemotherapy. Survival analysis demonstrated that platinum-treated BRCA1-low patients, but not BRCA1-high patients, with low SMARCAD1 expression were correlated with a longer progression-free survival (PFS), while higher expression of SMARCAD1 correlated was with poor response to chemotherapy (fig. S5C). These data suggest that SMARCAD1 levels could be a biomarker for acquired resistance to platinum-based chemotherapy in BRCA1-low/deficient ovarian cancers.

To further verify this experimentally, we queried whether SMARCAD1 is required for fork progression in BRCA1-deficient tumor cells and whether its loss could hypersensitize HR-deficient BRCA1/ mouse breast tumor cells generated using K14Cre;Brca1F/F;p53F/F (KB1P) mouse mammary tumor models (40). We generated short hairpin RNA (shRNA)mediated knockdowns of SMARCAD1 in Brca1/;p53/ defective mouse breast tumorderived cell lines (fig. S5D). Unexpectedly, the loss of SMARCAD1 resulted in a significant reduction in colony formation in HR-defective BRCA1/ (KB1P-G3; PARPi nave) (41) tumor cells but not in KB1P-G3 tumor cells that were reconstituted with human BRCA1 (KB1P-G3B1) and proficient in HR (42), suggesting that loss of SMARCAD1 causes synthetic lethality in BRCA1-deficient tumor cells (Fig. 6D). These data indicate a potential role of SMARCAD1 in maintaining active fork stability, which may be the reason for the survival of BRCA1-deficient HR-defective tumor cells. Furthermore, we also tested whether BRCA1 and 53BP1 double-knockout tumor cells, which are proficient for HR and resistant to PARPi treatments (KB1P-177.a5; PARPi resistant) (41), require SMARCAD1 for proliferation. A SMARCAD1 knockdown, again, resulted in lethality in these cells, suggesting that SMARCAD1s role is essential for the proliferation of BRCA-defective tumor cells, irrespective of their HR status (Fig. 6D). Furthermore, 53BP1 deficiency in BRCA1-defective genetic background could not rescue defects of SMARCAD1 knockdown, which suggests that fork protection mediated by BRCA1 becomes critical for cellular survival in the absence of SMARCAD1, similar to what we observed in human fibroblast cells (fig. S5, A and B). In addition, we tested the effect of SMARCAD1 knockdown on KB1P-derived, PARPi-nave (KB1P4.N), and PARPi-resistant (KB1P4.R) tumor organoids grown in ex vivo cultures (43). Consistent with our results in KB1P tumor cell lines, we observed a synthetic lethality in the three-dimensional (3D) tumor organoids, suggesting that SMARCAD1 is essential for the survival of BRCA1-mutated tumors (Fig. 6E). These data strongly suggest a conserved and nonepistatic role of SMARCAD1 and BRCA1 at replication forks.

Because BRCA1-deficient cells show reduced fork protection and high levels of endogenous stress (7, 44), we hypothesized that the loss of SMARCAD1 further enhances replication stress due to the defective progression of forks, causing proliferation defects. To test this hypothesis, we used siRNA to transiently deplete SMARCAD1 protein (45) in KB1P 2D tumor-derived cell lines (fig. S5E) to monitor individual fork progression using DNA fiber assay. We sequentially labeled human BRCA1-reconstituted, KB1P-G3B1 cells as control, KB1P-G3 (HR deficient), and KB1P-177.a5 (chemoresistant; HR proficient) with CldU (red) and IdU (green), followed by track length analysis. In support to the survival assays, although sublethal SMARCAD1 knockdown affects only mildly the cell cycle of all three cell lines (fig. S5F), it led to a significantly shorter track lengths of both CldU and IdU in both KB1P-G3 and KB1P-177 cells in comparison to BRCA1-reconstituted KB1P-G3B1 cells, suggesting an essential role of SMARCAD1 in mediating fork progression in the absence of BRCA1 (Fig. 6F). Consistently, we also observed the reduction in PCNA levels and enhanced 53BP1 enrichments at the fork, using the PLA approach with EdU, upon loss of SMARCAD1 in BRCA1/ mouse tumor cells, similar to human cells (fig. S5, G and H). Together, these results strongly suggest that the SMARCAD1-mediated stability of active replication forks is a physiologically important process for cellular proliferation of BRCA1-deficient tumors, irrespective of their HR status (fig. S6).

Our study has revealed a novel mechanism of active fork stability that has important implications in the survival of tumor cells.

As opposed to the commonly attributed role of DNA repair factors in replication fork protection (6, 7, 9, 46), here, we identify a new function of SMARCAD1 in maintaining the stability of active (unperturbed and restarted) replication forks, while its absence does not disturb stalled fork protection and fork processing activities (Figs. 3, A and B, and 6A and fig. S2I). Using a separation-of-function SMARCAD1 mutant (N-SMARCAD1), we show that SMARCAD1s role in stabilization of active replication forks is genetically separable from its role in HR repair and is critical in maintaining genome stability especially upon replication stress. The physical interaction between SMARCAD1 and PCNA, established using in vitro and in vivo assays (21), was suggested to be responsible for SMARCAD1s association with replication machinery (21, 22). Our biochemical and immunofluorescence assays further confirm that the N-SMARCAD1 protein, lacking initial 137 amino acids, can bind to chromatin but lacks the ability to interact with PCNA. This finding is consistent with the lack of association between N-SMARCAD1 and replication forks, as previously suggested (22). However, other components may also be involved in promoting SMARCAD1s association with replication machinery, such as phosphorylation of SMARCAD1 by cyclin-dependent kinase (CDK). A CDK phosphorylation site at the N terminus of SMARCAD1 is among the 137 amino acids that are missing in the N-SMARCAD1 protein (47). Nonetheless, the CUE-dependent protein-protein interactions and ATPase-dependent chromatin remodeling activity, in the context of HR repair and nuclear association, seem to remain functional in the N-SMARCAD1 protein. Notably, cells with a transient depletion of SMARCAD1, SMARCAD1-null (SMARCAD1/) genotype, and those expressing the N-SMARCAD1 allele show similar defects in fork progression, suggesting that it is the direct effect of loss of protein at the replication forks and not the secondary effects of mutants accumulating damages that result in slower fork progression. Furthermore, the similar sensitivity toward replication poisons of HR-proficient N-SMARCAD1 and HR-deficient SMARCAD1/ cells argues that the role of SMARCAD1 at replication forks is, in fact, crucial in mediating resistance to replication stressinducing drugs rather than its role in HR.

Furthermore, our data showed evidence of frequent accumulation of stalled forks and ssDNA gaps behind the replication forks in N-SMARCAD1 cells. The accumulation of ssDNA and stalled forks could be indicative of a hindered replication fork progression through certain difficult-to-replicate regions, such as highly transcribing regions or repetitive regions of the genome (48). Alternatively, ssDNA accumulation could also be a resultant of the repriming events by PRIMPOL at stalled forks that in the process of reinitiating, the DNA synthesis leads to the accumulation of ssDNA gaps (49, 50). However, in BRCA1-challenged cells, PRIMPOL activity was shown to be responsible for DNA synthesis upon replication stress condition. Here, our study shows a unique pathway of active fork stabilization mediated by SMARCAD1, which is critical for fork progression in BRCA1-deficient cells even under unperturbed conditions. This implies that SMARCAD1-mediated active replication fork stability is a central and a separate pathway for stabilization of replication forks than from recently described PRIMPOL-mediated fork repriming or well-established BRCA1-mediated fork protection pathway (51).

Our findings suggest a hitherto unrecognized role for SMARCAD1 in maintaining the fine control of PCNA levels at the forks. In this study, along with previously published study (21, 22), we have strong evidence of a positive interaction between SMARCAD1 and PCNA, which is also responsible for SMARCAD1 association with replication machinery. A global reduction in chromatin-bound PCNA levels at the fork and a faster dissociation rate of PCNA foci in N-SMARCAD1 cells further suggest a mutualistic interaction between SMARCAD1 and PCNA at the replication forks (Fig. 4, C and D). Consistently, an increase in PCNA unloading by the ATAD5-RLC complex was observed in N-SMARCAD1 cells. A recent report demonstrated a critical role of ATAD5 in the removal of PCNA from stalled forks to promote the recruitment of fork protection factors (34). Consistent with this report, we observed reduced PCNA levels at replication forks, accompanied by an increased accumulation of ATAD5-RLC complex and increased frequency of reversed forks (protected stalled forks) in unperturbed N-SMARCAD1 cells. Furthermore, a significant number of peptides arising from RFC2-5 protein subunits that are shared between PCNA loading (RFC) and unloading (ATAD5-RLC) complexes were obtained from SMARCAD1 coimmunopurification (21). These data may indicate the direct involvement of SMARCAD1 in regulating loading/unloading activity of PCNA at replication forks. However, an interesting finding from our study is that loss of 53BP1 results in a significant restoration of PCNA levels in N-SMARCAD1 cells accompanied with a significant reduction in ATAD5 levels at replication forks. Furthermore, the direct interaction observed between 53BP1 and ATAD5 in WT cells is enhanced in N-SMARCAD1 cells or HU-treated WT cells possibly due to ATR-mediated posttranslationally modified ATAD5. Whether the posttranslational modification of ATAD5 is solely ATR mediated or additional mechanisms play a role in its regulation (34) could distinguish between the physiological roles of ATAD5 in regulating PCNA dynamics that involves continuous loading/unloading events during normal fork progression versus the persistent unloading of PCNA from stalled forks.

Our study shows an unforeseen role of SMARCAD1 in preventing 53BP1 accumulation at active restarted replication forks. Previously, 53BP1 has been shown to bind to H2AK15Ub nucleosomes at DSBs (52), while SMARCAD1 was proposed to displace 53BP1-associated nucleosomes at DSBs to promote HR repair (20). This observation is consistent with the finding that SMARCAD1 and its homologs in yeast can slide, evict, and exchange H2A-H2B dimer, also regulating histone turnover in replicating cells of fission yeast cells (48, 5355). Consistent with these observations, it has been shown that the loss of SMARCAD1 results in a prolonged enrichment of 53BP1 at DSBs (20, 25). SMARCAD1 and 53BP1 also show contrasting enrichments at stalled versus unperturbed forks, suggesting that their coexistence is possibly also prohibited by SMARCAD1 at replication forks in a manner similar to that of their interaction at DSBs (Figs. 1, A and C, and 5A) (8). Notably, we found increased 53BP1 and the histone epigentic mark that it associates with, at restarted forks in N-SMARCAD1 and ATPase-dead SMARCAD1. These data imply that both the ability of SMARCAD1 to localize to forks and its chromatin remodeling activity are required to evict the 53BP1-associated nucleosomes to prevent untimely 53BP1-ATAD5 accumulation on active forks. As shown previously, the ATR-mediated phosphorylation of ATAD5, upon HU-induced stalled fork accumulation, interacts with proteins at reversed forks proteins (34). We suggest that in the absence of SMARCAD1, enhanced ATAD5-RLC levels causing PCNA dissociation from forks lead to frequent fork stalling and, consequently, accumulation of reversed forks resulting in activation of ATR checkpoint. The chromatin remodeling activity of SMARCAD1 is required to evict 53BP1-bound H2AK15Ub nucleosomes at reversed arm of stalled forks for their restart. However, in the absence of SMARCAD1, enhanced accumulation of 53BP1 possibly further stabilizes ATR-mediated phosphorylated ATAD5 at the reversed forks, which leads to continuous PCNA unloading, causing severe defects in restart/progression of forks. In addition to loss of 53BP1, the controlled depletion of ATAD5 could also restore normal PCNA levels at the fork that rescued the overall DNA synthesis and replication fork restart efficiency (Fig. 4, F to H). This suggests that active replication fork stability is indeed regulated by maintaining fine-controlled PCNA levels at the forks.

Furthermore, it was previously suggested that the loss of 53BP1 restores HR in SMARCAD1-depleted cells, which is responsible for developing resistance to replication stress-inducing drugs (20). However, this study using a separation-of-function SMARCAD1 mutant, which is HR proficient but defective for fork stability, shows that the extent of damage generated upon replication stress is rather responsible for the cellular sensitivity and not unrepaired DSBs due to lack of HR. This further suggests that the role of SMARCAD1 at forks is crucial for tolerance to replication stressinducing agents. We have, therefore, revealed a moonlighting function of SMARCAD1 at the replication forks in displacing 53BP1 to maintain replication fork progression and genome stability. Other NHEJ factors such as mammalian Rap1-interacting factor 1 (RIF1), Pax transactivation domain-interacting protein (PTIP), and others have also been found in association with replication forks. Therefore, it would be interesting to investigate whether 53BP1 works in complex with NHEJ machinery or have a separate role in association with ATAD5-RLC complex to regulate PCNA homeostasis and thereby fork dynamics.

BRCA1/2 factors, independent of their role in HR, protect replication forks and prevent their collapse into genome-destabilizing DSBs (6, 7). Although SMARCAD1and BRCA1 have been shown to act epistatically during HR repair (20, 25), here, we show contrasting differences in role of SMARCAD1 and BRCA1 at replication forks that can be observed by (i) differential enrichment of SMARCAD1 and BRCA1 at the replication forks, where SMARCAD1 preferentially associates with active forks, while BRCA1 associates with stalled forks (Fig. 1A) (8); (ii) stalled forks in the absence of SMARCAD1 remain protected and do not degrade unlike in absence of BRCA1; (iii) loss of SMARCAD1 but not BRCA1 causes defects in unperturbed replication fork progression (Fig. 3A and fig. S5A) (2); and last, (iv) loss of 53BP1 in BRCA1-deficient cells that restores HR repair capacity does not rescue sensitivity of BRCA1 mutants to cisplatin treatment (Fig. 5F) (56). However, loss of 53BP1 in SMARCAD1 mutant rescues cisplatin sensitivity, suggesting that replication stress sensitivity is uncoupled from HR repair and that SMARCAD1s role at active replication forks is distinct from that of BRCA1s role at stalled replication forks to maintain tolerance toward replication stressinducing agents. Thus, loss of SMARCAD1 results in enhanced accumulation of replication forkassociated DNA damage and, ultimately, synthetic lethality in mouse BRCA1-defective tumors irrespective of their HR status. Loss of 53BP1 could not rescue severe replication fork progression defects observed under SMARCAD1 and BRCA1 double-mutant condition. This suggests that frequently accumulated stalled forks in the absence of SMARCAD1 essentially require BRCA1-mediated fork protection, which could only be rescued by Mre11 inhibition. Together, these data suggest a distinct role of SMARCAD1 and BRCA1 at replication forks, acting in two independent pathways, where SMARCAD1 mediates active fork stability, while BRCA1 mediates stalled fork protection. However, both the pathways are interdependent for maintaining replication fork integrity, which is also conserved across species, from mouse to human.

In summary, we have found a distinct pathway of active fork stabilization mediated by SMARCAD1 and have shown a conserved interplay between SMARCAD1 and BRCA1 in stabilization of replication forks to maintain genome integrity (fig. S6). Notably, SMARCAD1-mediated stabilization of unperturbed forks promotes cellular proliferation in BRCA1-deficient mouse breast tumor, cells, and organoids, independently of their HR and PARPi resistance status. Similarly, the correlation of reduced chances of survival after chemotherapy in cancer patients with enhanced expression of SMARCAD1 along with reduced expression of BRCA1 suggests that stabilization of active forks promotes tolerance toward chemotherapy in BRCA1-defective tumors. Last, the observation that SMARCAD1 becomes essential for genome stability and cellular survival in the absence of BRCA1 suggests that targeting the stability of active replication forks has the potential to be a clinically effective remedy for BRCA-deficient tumors, nave or chemoresistant. It also suggests that SMARCAD1 could be a strong candidate for development of novel therapeutic treatment for BRCA1-deficient cancer patients.

Plasmid transfections were performed using X-tremeGENE 9 DNA transfection agent (Roche) according to the manufacturers protocol. To generate MRC5 N-SMARCAD1 cells, MRC5 WT cells were transfected with pLentiCRISPR-V2 plasmid (Addgene #52961) containing a guide RNA (gRNA) sequence targeting exon 2 of SMARCAD1, followed by puromycin selection (1 g/ml). To generate MRC5 SMARCAD1/, two gRNA sequences targeting exon 2 and exon 24 of SMARCAD1 were selected and cotransfected with a homolog repair template containing an mClover gene.

To express mClover-SMARCAD1 full-length/SMARCAD1 K528Rmutant cDNA, gRNAs targeting SMARCAD1 exon 2 and exon 24 were cotransfected with mClover-SMARCAD1 full-length/ SMARCAD1 K528Rmutant cDNA, respectively, in MRC5 WT and N-SMARCAD1 cells. The K528R mutant was generated by site-directed mutagenesis of full-length SMARCAD1 cDNA.

To generate GFP-tagged PCNA knock-in MRC5 cells, a gRNA sequence targeting exon 2 of PCNA was selected and inserted into lentiCRISPR V2 (Addgene #52961). MRC5 WT and N-SMARCAD1 cells were transfected with the gRNA and the FLAG-GFP-PCNA repair template and sorted by fluorescence-activated cell sorting (FACS). Sequences of gRNAs and mutagenesis primers are listed in table S3.

All MRC5 human fibroblasts were cultured in a 1:1 ratio of Dulbeccos modified Eagles medium (DMEM) and Hams F10 (Invitrogen) supplemented with 10% fetal calf serum (FCS; Biowest) and 1% penicillin-streptomycin (PS; Sigma-Aldrich) at 37C and 5% CO2 in a humidified incubator. KB1P-G3, KB1P-177.a5 (41, 42), and KB1P-G3B1 (42) have been described previously. All KB1P mouse tumor cell lines were cultured in DMEM/F12 and GlutaMAX (Gibco) containing insulin (5 g/ml; Sigma-Aldrich), cholera toxin (5 ng/ml; Sigma-Aldrich), murine epidermal growth factor (EGF; 5 ng/ml; Sigma-Aldrich), 10% FCS, and 1% PS under low oxygen conditions (3% O2 and 5% CO2 at 37C).

All tumor-derived organoid lines have been described before (43). KB1P4.N1 and KB1P4.R1 tumor organoids were derived from a mammary KB1P PARPi-nave and PARPi-resistant tumor, respectively (female donor). Cultures were embedded in Cultrex Reduced Growth Factor Basement Membrane Extract Type 2 (BME; Trevigen; 40 ml of BME:growth medium 1:1 drop in a single well of 24-well plate) and grown in Advanced DMEM/F12 (Gibco) supplemented with 1 M Hepes (Gibco), GlutaMAX (Gibco), PS (50 U/ml), B27 (Gibco), 125 mM N-acetyl-l-cysteine (Sigma-Aldrich), and EGF (50 ng/ml). Organoids were cultured under standard conditions (37C and 5% CO2).

mESCs were maintained in 2i medium deficient in lysine, arginine, and l-glutamine (PAA) at 37C and 5% CO2 in a humidified incubator. For SILAC labeling, cells were grown in medium containing light [12C6]-lysine (73 g/ml) and [12C6, 14N4]-arginine (42 g/ml) (Sigma-Aldrich) or similar concentrations of heavy [13C6]-lysine and [13C6, 15N4]-arginine (Cambridge Isotope Laboratories).

siRNA transfection, shRNA transduction, and cell titer assay. siRNA transfection was performed with Lipofectamine RNAiMAX (Thermo Fisher Scientific) according to the manufacturers protocol. Details of siRNA oligomers and shRNAs used in this study are given in table S3.

Transductions were performed in duplicate in KB1P mouse tumor cells. After 3 days of selection, KB1P mouse tumor cells were expanded to 10-cm dishes. Five days after passage, samples were fixed with 4% formaldehyde and stained with 0.1% crystal violet, and quantification was carried out by determining the absorbance at 590 nm after extraction with 10% acetic acid.

3D tumor-derived organoids were transduced according to a previously established protocol (43). Puromycin selection (3 g/ml) was carried out for three consecutive days after transduction. Pictures were taken at day 5. For quantification, cells were incubated with CellTiter-Blue (Promega) reagent at day 5.

Cells were lysed in lysis buffer [30 mM Hepes (pH 7.6), 1 mM MgCl2, 130 mM NaCl, 0.5% Triton X-100, 0.5 mM dithiothreitol, and EDTA-free protease inhibitor], at 4C for 30 min. Chromatin-containing pellet was spinned down by centrifugation at 16,000g for 10 min and resuspended in lysis buffer supplemented with Benzonase (250 U/l; Merck Millipore) and incubated for 15 min at 4C.

Live-cell confocal laser scanning microscopy was carried out as described before (57), with minor adjustments. All live-cell imaging experiments were performed using a Leica TCS SP5 microscope equipped with HCX PL APO CS 63 oil immersion objective, at 37C and 5% CO2. For iFRAP, GFP-PCNAexpressing WT and N-SMARCAD1 MRC5 cells were continuously bleached at high 488-nm laser outside the selected GFP-PCNA foci, and the fluorescence decrease in the selected foci was determined over time. The resulting dissociation curves were background corrected and normalized to prebleach values, set at 1.

The procedure for DR-GFP reporter was described previously (26) and applied with minor alterations. After being seeded in a six-well plate overnight, cells were cotransfected with DR-GFP reporter plasmid (Addgene #26475) and I-Scel expression vector (Addgene #26477) or empty vector using X-tremeGENE 9 DNA transfection agent (Roche) according to the manufacturers protocol for two consecutive days. p-MAX-GFP plasmid (Addgene #16007) was transfected in parallel to assess transfection efficiency. On day 3, GFP expression was analyzed by flow cytometer.

Cells were sequentially pulse labeled with 30 M CldU (MP Biomedicals) and 250 M IdU (Sigma-Aldrich) according to the schematic in each figure. For Mirin treatment, 100 M Mirin was added to the medium for 2 hours before the experiment. DNA fiber analysis was carried out according to the standard protocol as mentioned previously (30). Fibers were visualized and imaged by Axio Imager D2 microscope (Carl Zeiss). ImageJ software was used for the quantification. The Kruskal-Wallis test followed by Dunns multiple comparison test was applied for statistical analysis using the GraphPad Prism software. The combined summary of DNA fiber spread data analysis is given in table S2.

After lysis with radioimmunoprecipitation assay (RIPA) buffer (whole-cell lysate) or resuspended in chromatin fractionation lysis buffer (chromatin-bound proteins), samples were mixed with 2 Laemmli sample buffer, boiled for 5 min, loaded on Bis-Tris Gel, and transferred to a polyvinylidene difluoride membrane. Membranes were blocked with 5% nonfat milk in tris-buffered saline (TBS) for 1 hour and incubated with primary antibody diluted in 5% bovine serum albumin (BSA) in TBS overnight at 4C. Membranes were then washed in 0.1% Tween 20 in TBS, incubated with a secondary antibody coupled to near-IR dyes CF 680/770, and visualized using Odyssey CLx infrared scanner (LI-COR). ImageJ software was used for quantification. Primary and corresponding secondary antibodies are listed in table S4.

Cells were labeled with EdU (10 M) for 30 min, unless otherwise mention. For HU-treated samples, EdU was labeled before the treatment. In analysis of chromatin-bound protein, cells were first preextracted with 0.1% Triton X-100 in cytoskeletal (CSK) buffer on ice and then fixed in 4% formaldehyde in phosphate-buffered saline (PBS) for 15 min at room temperature for SMARCAD1, 53BP1, RAD51, and H2AX or 100% 20C methanol for 10 min for PCNA. Subsequently, samples were permeabilized in 0.1% Triton X-100 in PBS for 10 min, blocked with 5% BSA in PBS, and stained with a primary antibody diluted in blocking buffer, followed by incubation in fluorescence-conjugated secondary antibody. EdU was visualized with a Click-IT reaction using Alexa Fluor 488 azide or Alexa Fluor 594 azide (Invitrogen) according to the manufacturers protocol. Samples were stained with 4,6-diamidino-2-phenylindole (DAPI) and mounted on slides using ProLong Gold (Invitrogen).

Cells were washed once with 1 PBS, treated with 0.1% Triton X -100 in CSK buffer on ice, and fixed with ice-cold methanol for 10 min (PCNA) or with 4% formaldehyde (FA) in PBS for 15 min (53BP1 and H2AK15ub). Subsequently, cells were permeabilized with 0.1% Triton X-100 in PBS for 10 min and blocked with 5% BSA in PBS at room temperature for 1 hour. Afterward, cells were treated with Click-iT reaction according to the manufacturers protocol for 1 hour and were incubated with PCNA (PC10), 53BP1, H2AK15ub, and biotin at 4C overnight in a humid chamber. After washes with PBS with 0.1%Tween-20 (PBST), cells were incubated with anti-mouse minus and anti-rabbit plus PLA probes (Sigma-Aldrich) at 37C for 1 hour. Following the manufacturers instructions, the PLA reaction was performed with the Duolink In Situ Detection Reagents. Cells were stained with DAPI and mounted on slides using ProLong Gold. Images were captured using Metafer5 and quantified using MetaSystem.

Coverslip images were obtained using a LSM700 microscope equipped with a Plan-Apochromat 63/1.4 oil objective (Carl Zeiss), MetaSystems5 equipped with an EC Plan-Neofluar 40/0.75 objective (Carl Zeiss), or SP5 microscope equipped with HCX PL APO CS 63 oil objective (Leica). Detection of EdU-positive cells was performed in combination with the DAPI channel applying a cross entropybased thresholding and binary watershed segmentation. The brightness and contrast adjustment was applied differently because of differential backgrounds in the indicated cell lines of Fig. 1G for the qualitative representation. To compute the Pearson and Manders overlap coefficients in fig. S4B, the 53BP1 foci in 488- and 568-nm channels for EdU-positive cells were segmented using an trous wavelet transform with three scales, and the wavelet coefficients were thresholded at the level of 3-sigma (58). To measure the distance between 53BP1 and EdU foci in Fig. 5A, a line of 3 m was drawn across the proximal foci, and the intensity of the two channels were measured using Multi Plot in ImageJ. Further analysis was performed using Microsoft Excel. For high-content imaging given in Figs. 1 (B and C), 2C, 4 (A and C), and 5E and figs. S1D and S2J, all the data were obtained using the Opera Phenix High-Content Screening System (PerkinElmer) with a 40 water objective (numerical aperture, 1.1) and analyzed with the Harmony v4.9 high-content imaging and analysis software (PerkinElmer). At least 75 fields were imaged as a Z-stack of eight planes (step size, 1 m). In the maximum projection, nuclei were detected using DAPI. Selection of S phase cells was based on EdU signal under untreated (UT) and HU block condition. Under HU release conditions, S phase cells were determined by intensity of PCNA median. The pixel intensities (sum) were determined in DAPI, 488- and 568-nm channel for each nucleus. PCNA sum normalized to DAPI sum was shown in the bar chart. For quantification of EdU-positive foci in Fig. 1 (B and C) and fig. S1D, an additional mask was generated on the basis of the detection of local intensity maxima (region to spot intensity) in the EdU channel and was used for quantification of spot intensities together with spot contrast in the 488- and 568-nm channels. For quantification of foci in Fig. 2C and figs. S2A, S4 (K and L), and S5 (G and H), a mask was generated using the detection of spot contrast and intensity, with threshold for spot radius. The quantified values for each foci/cell were exported to the TIBCO Spotfire software.

Total RNA was extracted using the ReliaPrep RNA Miniprep Systems (Promega). One thousand nanograms of total RNA was used to synthetize cDNA using Moloney Murine Leukemia Virus Reverse Transcriptase, Ribonuclease H Minus, Point Mutant (Promega). qPCR was performed using the GoTaq qPCR Master Mix (Promega), and -actin was used for normalization. Primers used for qPCR are listed in table S3.

Next-generation sequencing short reads were trimmed using fastp and processed using Kalliso, an RNA-seq quantification program that uses a pseudo-alignment method of assigning reads to genomic locations in lieu of a more costly traditional alignment (59). The human transcriptome, version GRCh38.p12, was indexed, the paired, trimmed reads were assigned to transcripts, and read counts were converted to transcripts per million (TPMs) by Kallisto. TPMs from transcripts originating from the same gene were aggregated, and relative expression levels were computed as the log2 fold change relative to the matched WT using an in-house script (available as a separate file in the Supplementary Materials). RPKM (reads per million kilobases) values were computed from TPMs using the median transcript length per gene.

Pseudo-alignments, output by Kallisto in a standard BAM format, were used to assess transcript structure such as the assignment of the transcription start for N-SMARCAD1. Box plots and bar plots were produced using ggpubr and ggplot2, respectively, in the R program (the R Foundation).

Light lysine and arginine labeled mESCs were incubated with 10 M EdU for 10 min and treated with 4 mM HU for 3 hours. Heavy lysine and arginine labeled mESCs were incubated with 10 M EdU for 10 min. iPOND mass spectrometry was performed essentially as described. At least two peptides were required for protein identification. Quantitation is reported as the log2 of the normalized heavy/light ratios. SILAC data were analyzed using MaxQuant. The resulting output tables of two independent experiments were merged and used as the input for calculating the average fold change to identify significantly up-regulated proteins in unperturbed forks and stalled forks based on the ratio of heavy and light peptides (H/L ratio) in the SILAC experiment in MaxQuant software (9).

Cells were cross-linked in 1% formaldehyde in serum-free medium for 10 min at room temperature. Cross-linking reaction was quenched with 0.125 M glycine, and cells were washed with PBS. Cross-linked cells were scrapped, and chromatin was purified as described (57). Chromatin was sheared using a Bioruptor sonicator (Diagenode) using cycles of 20-s on, 60-s off during 15 min, after which samples were centrifuged. The supernatant containing cross-linked chromatin was used for IP. For native IP, cells were collected by trypsinization and lysed with lysis buffer [1 mM MgCl2, 150 mM NaCl, 1 mM EDTA, 0.5% NP-40, and 30 mM Hepes buffer (pH 7.6)] for 20 min at 4C. Chromatin fraction was collected by spinning. DNA was fragmented by passing the lysed suspension 10 times through a needle attached to a 1-ml syringe, after which samples were centrifuged. The supernatant containing the chromatin fraction was used for IP.

For IP, extracts were incubated with GFP-Trap beads (ChromoTek), 53BP1 (1.8 g), PCNA (1.8 g), or SMARCAD1 (1.8 g) antibody overnight at 4C. For IP with PCNA, 53BP1, and SMARCAD1 antibodies, protein A agarose/salmon sperm DNA slurry (Millipore) was added for 4 hours at 4C. Subsequently, beads were washed five times in RIPA buffer, and elution of the precipitated proteins was performed by extended boiling in 2 Laemmli sample buffer for immunoblotting analysis.

Cells were seeded in triplicate in 10-cm culturing dish and treated with olaparib (Selleckchem), cisplatin (Sigma-Aldrich), or HU (Sigma-Aldrich) 1 day after seeding. HU was given at the indicated concentration for 24 or 48 hours, as indicated in the figure legend. Olaparib treatment was given throughout the whole experimental process. Different concentrations of cisplatin were given for 4 hours before being replaced with new medium, except the 1 M cisplatin group in Fig. 5F, which were given throughout the whole experimental process.

After 1 week, colonies were fixed and stained in a mixture of 43% water, 50% methanol, 7% acetic acid, and 0.1% Brilliant Blue R (Sigma-Aldrich) and subsequently counted with GelCount (Oxford Optronix). The survival was plotted as the mean percentage of colonies detected following the treatment normalized to the mean number of colonies from the untreated samples.

Cells were grown to 70 to 80% confluency, labeled with EdU for 30 min, and fixed for 10 min in 4% formaldehyde in PBS at room temperature. Cells were then washed with 1% BSA/PBS and permeabilized in 0.5% saponin buffer in 1% BSA/PBS. EdU was labeled with the Click-iT reaction using Alexa Fluor 594 azide according to the manufacturers protocol (Invitrogen). DAPI was used to stain the DNA.

EM analysis was performed according to the standard protocol (9). For DNA extraction, cells were lysed in lysis buffer and digested at 50C in the presence of proteinase K for 2 hours. The DNA was purified using chloroform/isoamyl alcohol, precipitated in isopropanol, given 70% ethanol wash, and resuspended in elution buffer. Isolated genomic DNA was digested with Pvu II high-fidelity restriction enzyme for 4 to 5 hours. After digestion, the DNA solution was transferred to a Microcon DNA fast flow centrifugal filter. The filter was washed with tris-EDTA (TE) buffer after spinning for 7 min. The benzyldimethylalkylammonium chloride method was used to spread the DNA on the water surface and then loaded on carbon-coated nickel grids, and last, DNA was coated with platinum using high-vacuum evaporator MED 010 (Bal-Tec). Microscopy was performed with a transmission electron microscope FEI Talos, with 4K by 4K complementary metal-oxide semiconductor camera. For each experimental condition, at least 200 replication fork intermediates were analyzed from three independent experiments, and MAPS software (Thermo Fisher Scientific) was used to analyze the images.

For HU-treated samples, cells were treated with 4 mM HU for 3 hours, following or not with a 16-hour release, before harvesting for PFGE assay. DSB detection by PFGE was performed as reported previously (9). The gel was stained with ethidium bromide and imaged on a Uvidoc-HD2 imager. ImageJ software was used for the quantification of broken DNA normalized to unbroken DNA for each lane.

N-SMARCAD1 protein was purified from whole-cell lysate using MRC5 N-SMARCAD1 cell line. Cells were resuspended in the IP buffer, sheared 10 times as 15-s on and then 45-s off at mode high using a Bioruptor sonicator (Diagenode) at 4C, and incubated with 500 U of Benzonase (Merck Millipore) for 60 min, after which samples were centrifuged. The supernatant was used for IP. For IP, extracts were incubated with SMARCAD1 (1.8 g) antibody overnight at 4C. Protein A agarose/salmon sperm DNA slurry (Millipore) was added for 2 hours at 4C. Subsequently, beads were washed five times in IP buffer, and elution of the protein was performed by extensive boiling in 2 Laemmli sample buffer. Eluted protein was run on bis-tris gel, gel slices were trypsinized, and peptides were analyzed by mass spectrometry to determine the protein sequence as described previously (57).

Disease-free survival curves of The Cancer Genome Atlas (TCGA) patients with HGSOC were generated by the Kaplan-Meier method, and differences between survival curves were assessed for statistical significance with the log-rank test. We divided the TCGA patients with ovarian carcinoma expressing replication stress markers (CCNE1 overexpression, CDKN2A-low expression, and/or RB1 deletion) into cohorts according to their BRCA1 mRNA expression levels: BRCA1 low (below median) and BRCA1 high (above median) (60). In each of these cohorts, we analyzed the correlation between SMARCAD1 expression and outcome. Normalization of expression values was performed using z score transformation, such that SMARCAD1-low expression with z score < 0.75 and SMARCAD1-high expression with z score > 0.75 (fig. S5C). Cohort with BRCA1-high and SMARCAD1-low expression, n = 66; BRCA1-low and SMARCAD1-high expression, n = 10. Cohort with BRCA1-low and SMARCAD1-low expression, n = 87; BRCA1-low and SMARCAD1-high expression, n = 10.

Human 53BP1 full length was fused to the LexA protein in pBTM116 and was coexpressed with human ATAD5 full length fused to the GAL4 activation domain in pGAD-HA in the yeast strain L40. Interactions were assayed using the (LexAop)4-HIS3 reporter system.

For all data, the means, SD, and SEM were calculated using either Microsoft Excel or GraphPad Prism 8.

Acknowledgments: We thank R. Kanaar, W. Vermeulen, and C. Wyman for stimulating discussions and sharing important reagents used in the manuscript; K. Myung and K. Lee for ATAD5 antibody and sharing technical information; D. Chowdhury for advice on PFS analysis; P. Zegerman for Y2H reagents; and E. Goggola for help with the initial phase of mouse tumor cells culture. We acknowledge infrastructural support from the Josephine Nefkens Precision Cancer Treatment Program. Funding: This work was supported by grant NWO-Vidi (193.131) and the Dutch Cancer Society funded grant (11008/2017-1) to A.R.C., the Oncode Institute partly financed by the Dutch Cancer Society funded grant (KWF grant 10506) to J.A.M. and J.J., the European Unions Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant (agreement no. 722729) to J.J., Daniel den Hoed Stitching Young Investigator Award grant (DDHS no. 10834) to N.T., and start-up funds from the Erasmus MC to N.T. Author contributions: C.S.Y.L. conducted all the QIBC, FACS, and PFGE experiments. M.v.T. performed iFRAP and chromatin fractionation experiments and, with help from Y.Z., performed chromatin IP experiments. V.G. performed all the DNAfiber and immunofluorescence experiments related to ATAD5. M.P.D. performed all the cloning experiments and clonogenic assays using mouse tumor cells/organoids under the supervision of J.J. Y.Z. with the help of M.v.d.D. performed cloning experiments of cDNA-SMARCAD1. E.M.M. with help from C.S.Y.L. performed clonogenic assays with MRC5 cells and chromatin fractionations for RAD51. H.L. helped C.S.Y.L. and M.v.d.D. in cloning experiments in MRC5 cells. M.E.v.R., W.Z., and I.S. analyzed fluorescence microscopy data. The assistance to use high-content imaging microscope facility was provided by M.E.v.R and P.J.F. J.D. analyzed mass spectrometry data. J.G.S.C.S.G. analyzed TCGA ovarian BRCA data. D.W. analyzed RNA-seq data. J.A.M. supervised the iFRAP and chromatin fractionation experiments. A.R.C. supervised the iPOND experiments performed by C.M. and EM experiments performed by E.M.M. N.T. conceptualized the project, supervised it, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors. NCBI BioProject accession number is as follows: PRJNA609878.

Read the rest here:
SMARCAD1-mediated active replication fork stability maintains genome integrity - Science Advances