Past exam of the biochemistry course of the University of Cambridge / 2024 / Paper 5 / Section 1

The viral DNA is copied by semiconservative replication and its labelling kinetics broadly follow those of host DNA during viral latency. At the earliest time the newly labelled material is mainly hybrid: an old light strand has acquired a heavy complementary strand. Fully heavy duplexes accumulate later, after heavy strands themselves serve as templates. Conservative copying would instead produce a separate fully heavy population from the first copying round, while dispersive copying would produce progressively shifting intermediate densities rather than these three discrete classes.

For an ideal synchronous population starting entirely light, one complete round gives only hybrid duplexes; after $n\ge 1$ rounds the hybrid fraction is $2^{1-n}$ and the fully heavy fraction is $1-2^{1-n}$. The actual culture is asynchronous, so these integer-generation proportions cannot simply be imposed at $t/23$ hours. The similar viral and cellular bars support replication coordinated with cell proliferation rather than rapid autonomous amplification. They do not establish that every episome initiates exactly once in every individual biological cell, or prove an initiation time within S phase. Persistent light material also warns against treating the population as a perfectly uniform set of cycling biological cells. The strand-accounting principle is the one tested in ^[ref].

Base composition changes both the initial buoyant density and the size of the BUdR-induced density increment. The fraction-number axis runs toward lower density. The unlabelled viral peak lies at a lower fraction number than the corresponding host peak, consistent with viral DNA having a higher GC content. BUdR replaces thymidine in newly synthesized strands, so relatively AT-rich DNA incorporates more brominated nucleoside per unit length and undergoes a larger density increase.

Consequently the more AT-rich host DNA can overtake the viral DNA in the fully substituted class: the host fully heavy peak is slightly farther toward the dense end, although host unlabelled DNA was less dense. The hybrid peaks are intermediate. A useful qualitative model is ${\rho }_{LL}={\rho }_0(g)$, ${\rho }_{LH}={\rho }_0(g)+\Delta \rho (g)/2$ and ${\rho }_{HH}={\rho }_0(g)+\Delta \rho (g)$, where $g$ is GC fraction, ${\rho }_0$ increases with $g$, and $\Delta \rho$ decreases as the thymidine fraction decreases. Unequal substitution efficiencies could contribute, but different replication mechanisms are unnecessary. These are equilibrium density positions, not molecular-length measurements.

The DNA-binding domain supports site occupancy, while the region for homotypic protein interaction permits much larger assemblies. With the truncated protein, the faster complex 4 appears at low protein:DNA ratios and gives way to the slower complex 3 as protein increases. A natural assignment is occupancy of one and then both recognition repeats by functional binding units. A binding unit could itself be a protein dimer; a native gel does not determine the number of polypeptide chains from mobility alone. Loss of free probe with increasing protein supports binding, and failure to shift the zero-repeat probe supports sequence preference under these conditions.

Full-length protein instead produces the very slow complex 1 already at relatively low ratios, followed by more of complex 2 at high ratios. homotypic protein interactions can make cooperative higher-order assemblies and can bridge different probe molecules. At protein excess, separate saturation of available sites can reduce productive bridging, explaining why the largest species need not increase monotonically. Complex 2 is compatible with a smaller saturated full-length assembly; assigning its exact stoichiometry requires an independent measurement. protein size also contributes to the difference between full-length and truncated complexes, but size alone does not explain the concentration-dependent redistribution into the exceptionally slow species. The experiment supports assembly dependent on protein oligomerization rather than a unique structure for every numbered band.

At low concentration, the array with more and higher-affinity binding sites is preferentially occupied, giving one visible protein cluster on an otherwise extended DNA molecule. At higher concentration the weaker origin array also becomes occupied. homotypic protein interactions between proteins at the two arrays can bridge them, trapping the intervening DNA through DNA looping. This explains why the electron-dense protein cluster lies at the loop junction rather than uniformly coating the entire molecule.

The high-affinity array can stabilize weak-site occupancy through tethering: once one partner is anchored, the second site experiences a high effective local concentration and several contacts can contribute avidity. This reconciles stable looped assemblies with the vulnerability of the isolated weak site to nonspecific competitor DNA. The images support intramolecular DNA bridging, but selected micrographs do not measure its frequency or exclude some intermolecular associations. A high local density of protein need not imply biomolecular phase separation. The printed scale-bar unit appears implausible for a kilobase-scale DNA substrate; the mechanistic inference uses topology and site positions rather than that unit.

Latent viral DNA has regularly spaced nucleosome protection, whereas much of the nuclear viral DNA during productive infection lacks that regular protection. In the latent sample, increasing digestion resolves a ladder of protected fragments, as expected when micrococcal nuclease cuts exposed linker DNA between nucleosomes. In the productive sample, viral signal is lost rapidly into heterogeneous small material rather than accumulating as a comparable ladder. That is consistent with a large pool of newly replicated or otherwise differently packaged viral DNA being more accessible.

The total-DNA ladders show that host chromatin remains nucleosomal and that the assay can recover protected fragments in each cellular condition. The virus-specific Southern blot distinguishes viral DNA from this abundant host background, and the uninfected lane controls probe background. The conclusion concerns the dominant detectable viral population: productive cultures can contain both chromatinized templates in viral latency and accessible replication products. Absence of a ladder does not establish absence of every histone molecule, and differences in DNA recovery or nuclease accessibility should be considered before assigning a precise fraction of chromatinized genomes.

The cytoplasmic viral signal stays predominantly at high molecular size despite longer nuclease treatment. The viral DNA is protected from digestion, plausibly inside viral capsids or assembled particles. This differs from a nucleosomal ladder: a capsid can shield a whole genome instead of producing repeated small protected units. Release into the cytoplasm during productive infection makes particle-associated DNA a reasonable interpretation.

The virus-negative lane argues against nonspecific hybridization. Protection alone nevertheless does not prove that all detected particles are mature or infectious; an inaccessible compartment or another protective assembly could also shield DNA. Comparing intact material with material whose protective envelope or capsid has been disrupted would distinguish enclosure from intrinsic nuclease resistance of the DNA. Fractionation controls would address nuclear contamination. The result is about physical accessibility of viral DNA, not a quantitative infectivity assay.

The naked labelled DNA runs fastest. Reconstitution creates a slower nucleosome band, with a little residual unassembled DNA. Adding NVNA1 progressively generates several additional, more slowly migrating labelled species and reduces the unbound nucleosome population. These are consistent with ternary protein–DNA complexes containing DNA, histones and NVNA1, with increasing occupancy or altered assembly. The naked-DNA titration also shows several NVNA1–DNA species, reflecting different occupancy states on the repeat-containing fragment.

A mobility shift reports changes in charge, size and shape together. It cannot by itself establish that histones remain in every shifted species, or identify exact protein stoichiometries. In particular, similarity to a band in the naked-DNA lanes could indicate histone loss rather than merely another occupancy state. The specific-competitor experiment in the next part supplies the important compositional test: removal of NVNA1 can regenerate a nucleosome band, demonstrating retention of histones in the starting ternary population.

With nucleosomes alone, neither competitor substantially removes the main nucleosome band, although some free labelled DNA is present. With nucleosome–NVNA1 complexes, increasing specific competitor DNA removes the extra shifts and restores the nucleosome band. The starting shifted complexes therefore contained histones as well as NVNA1: competitor recognition sites draw NVNA1 away while leaving labelled DNA associated with histones.

nonspecific competitor DNA gives a different endpoint. The shifted population changes to a band matching the far-right NVNA1–naked-DNA control, rather than returning mainly to the nucleosome position. The labelled DNA remains bound to NVNA1 while histones can transfer to the excess competitor DNA. NVNA1 binding destabilizes the nucleosome against competitor-induced histone displacement. Remaining shifted DNA is not evidence that the histone octamer remained: its mobility must be compared with both controls. The nucleosome-only competition lanes show that competitor DNA alone is insufficient to cause the same conversion. Histone-specific detection would confirm the inferred component loss. A closely analogous ternary-complex and competition analysis was demonstrated for EBNA1; it provides a methodological precedent without establishing the identity of NV.^[ref]

About 80% of full-length-expressing biological cells have two conspicuous nuclear foci, whereas roughly 90% of biological cells expressing the deletion protein have none. Both proteins show nuclear fluorescence, so the result is not explained simply by complete exclusion of the deletion protein from the nucleus. The DNA-binding domain is needed for efficient focal accumulation at a small number of host sites. A plausible explanation is recognition of repeated host DNA sequences resembling the viral binding sites; clustered repeats and homotypic contacts would concentrate the protein enough to create visible foci.

Two spots are compatible with one locus on each homologous chromosome in a diploid biological cell. They do not prove diploidy, identify a chromosome, or exclude unresolved sister loci and cell-cycle effects. An alternative is DNA-dependent recruitment through a host complex. Matching protein abundance and confirming stability of the deletion construct would make the domain comparison stronger. Punctate fluorescence alone does not demonstrate a liquid condensate or show that a focus identifies an origin of replication.

Use fluorescent dCas9 to ask whether NVNA1 foci coincide with a defined host DNA locus. Targeting a candidate repeated region with CRISPR locus labelling would produce a locus marker in a second colour. Reproducible colocalization with full-length NVNA1, but not with a labelled unrelated locus, supports sequence-associated accumulation. A non-targeting guide and biological cells lacking NVNA1 control fluorescence and targeting background; matched expression and nuclear localization control the protein comparison. The inactive Cas9 is important because cleavage itself could create the chromosome damage being investigated.

A complementary causal test would alter the candidate host recognition region and ask whether focal accumulation is reduced while general nuclear NVNA1 remains. Independent edited clones, verification of the host alteration, and an appropriate restored host-site control would distinguish loss of recognition from clone-specific damage or copy-number changes. Colocalization localizes the protein but does not alone prove direct base-specific contact; the biochemical binding evidence supplies that separate argument. Fluorescent CRISPR marking of a repeated host locus has a direct precedent in the EBNA1 study.^[ref]

The normal patterns are A3, B2 and C4. In each, two principal DNA fluorescence in situ hybridization signals lie at corresponding positions on the two replicated sister chromatids. A single chromosome in mitosis has already replicated its DNA, so a unique locus normally has one copy on each sister, rather than two signals because two homologous chromosomes are being shown.

The other examples show a single unresolved or missing partner signal, extra fragmented signals, or an abnormal distribution. The images are small and relative intensities differ, especially in row C, so the appropriate criterion is the paired principal locus signals at matching positions, not perfect equality of brightness. Counts from many chromosomes, adequate optical resolution and hybridization controls are necessary before classifying borderline examples.

The locus frequently appears as an abnormal, fragile-site-like structure rather than a clean replicated pair. A1, B3 and C2 have one dominant signal instead of a clearly resolved sister pair; A2/A4, B1/B4 and C1/C3 show extra, split or irregularly distributed signal components. These patterns are compatible with incomplete replication, abnormal condensation of sister chromatids, or splitting and fragmentation of the repeat-containing region. A single signal can also result from unresolved sisters or inefficient hybridization, so it is not automatically a deletion.

The several small signals should not simply be counted as newly created intact copies: stretching, gaps and fragmentation can divide the fluorescent image of an existing locus. Conversely, amplification would require independent copy-number evidence. The fact that aberrant structures occur in non-infected biological cells indicates intrinsic vulnerability of this host region, rather than a requirement for NVNA1 in their initial appearance. These images alone do not distinguish an undercondensed but continuous DNA region from a genuine DNA double-strand break. Their sensitivity to replication inhibition provides the next mechanistic clue.

aphidicolin increases the abnormal-locus fraction from roughly 40% to about 55–60% in all three lines. Inhibition of replicative DNA polymerases increases replication stress at an intrinsically difficult-to-complete region. Tandem repeats can complicate fork progression or chromosome organization. Slower or incomplete synthesis leaves underreplicated DNA, unresolved replication intermediates or abnormal condensation as biological cells approach mitosis; these can appear as split signals or gaps and can predispose to actual breakage.

This is an increase of approximately 15–20 percentage points, not proof of a 15–20-fold change. The two independent measurements per condition support a reproducible direction, but the graph does not provide enough information for a precise significance calculation or a quantitative model of fork failure. aphidicolin need not bind NVNA1 or the repeat sequence directly. Because these are uninfected biological cells, the result establishes susceptibility of the host locus to replication stress independently of viral protein expression.

The large breakage frequencies occur mainly at supraphysiological NVNA1 levels; the probability at the mean latent level is not measured directly. At 20 ng/ml doxycycline, NVNA1 is already about twice the latent amount, and breakage is about 8%. At 50 ng/ml, the protein is approximately eightfold above latent levels and breakage about 25%; at 100 ng/ml, approximately sixteenfold and about 45–50%. Higher induction reaches roughly 60% while expression rises to about 21–24 times the latent level. The western blot loading control supports a protein-abundance comparison, although quantitative interpretation also requires unsaturated immunoblot detection.

Latent expression, defined as a ratio of one, lies below the lowest nonzero measured dose. Extrapolating a straight line from baseline and the twofold point would suggest a small frequency, but neither linearity nor a threshold is established. The justified conclusion is lower expected short-term breakage than in the strongly induced conditions, with a potentially important effect of modest upward fluctuations. Cell-to-cell variability, duration of expression and repeated divisions could make uncommon events consequential during a persistent infection. A percentage of broken chromosomes in this assay is not a patient's lifetime cancer probability. Doxycycline and tag controls are also needed to attribute the induced-cell phenotype specifically to NVNA1.

The evidence supports NVNA1 as a sequence-specific organizer of viral chromatin, with possible additional roles in nuclear retention and episome partitioning. Binding to the origin of replication and DNA bridging explain assembly at separated viral repeat arrays. Access to recognition sites in nucleosomes and weakening of histone–DNA association provide a plausible way to make an origin accessible to cellular replication machinery. The density-labelling result is consistent with a latent genome using a host-coordinated copying programme; it does not establish that NVNA1 is itself a DNA polymerase or DNA helicase.

Association with host chromosomal DNA could also retain episomes in the nucleus and help distribute them during biological cell division. Such tethering is a plausible maintenance function, but the experiments here do not directly visualize a physical bridge between an episome and a host chromosome. Binding at a vulnerable host repeat could instead be an incidental consequence of recognition specificity. The enrichment of host chromosome rearrangements supports a damage connection; it does not show that chromosome breakage is required for viral replication or that integration is part of the normal latent programme. chromatin opening and host-locus interference are mechanistic interpretations of the presented binding data, not proof of one obligatory replication pathway.

NVNA1 could contribute to cancer by increasing the chance of host chromosome damage and heritable rearrangements. Persistent occupancy of a chromosomal fragile site could obstruct replication or normal chromatin reassembly. Incompletely replicated or damaged DNA can break, mis-segregate and be repaired into deletions, gene fusions or copy-number alterations. Particular surviving rearrangements might remove a tumour suppressor gene or deregulate an oncogene, after which clonal selection could expand that lineage. A viral protein can therefore promote genomic instability without the viral genome having to integrate at every breakpoint.

The tumour-sequencing association is consistent with this mechanism but does not establish causality by itself. tumour type, pre-existing instability, infection detection sensitivity and cellular composition can confound an NV-positive/NV-negative comparison. The induction experiment strengthens the case that NVNA1 can damage the locus, while its high doses limit direct estimation of risk during ordinary viral latency. Very common infection is compatible with rare cancer because susceptible biological cell types, expression fluctuations, time and cooperating alterations all matter. A documented biological analogue links EBNA1 occupancy of a chromosome 11 repeat cluster to chromosome breakage and rearrangement.^[ref]