Skip to PREreview

PREreview of Mitotic adaptations shape acquired resistance and vulnerabilities to KIF18A inhibition in cancer

Published
DOI
10.5281/zenodo.24406057
License
CC BY 4.0

Preprint review of “Mitotic adaptations shape acquired resistance and vulnerabilities to

KIF18A inhibition in cancer”

Summary:

This manuscript investigates the mechanisms by which cancer cells acquire resistance

to KIF18A-inhibiting drugs - a class of drugs that have potential to specifically target

chromosomally unstable cancers. The authors provide strong evidence for rapid acquisition of

resistance to KIF18Ai by treating a panel of seven cell lines with KIF18Ai for six weeks. These

drug-adapted cells had increased relative viability in drug, less severe defects in proliferation,

decreased proportion of cells in mitosis, and shorter mitotic delays assessed via live-cell

imaging as compared to naïve cells. The live-cell imaging experiments show that cells entering

a second mitosis had shorter first arrests than cells failing to undergo a second division,

supporting the authors' hypothesis that KIF18Ai resistance likely arises through selection of pre-

existing subclones in the population. The authors provide several lines of evidence indicating

KIF18Ai resistance is not due to KIF18A mutations or overexpression.

The authors instead provide compelling evidence supporting multiple alternative

mechanisms contributing to KIF18Ai resistance. The genome-wide CRISPR screen in TP53-null

RPE1 cells finds that genes associated with both the spindle assembly checkpoint (SAC) and

microtubules are genetic determinants of KIF18Ai response. The authors validate hits using

both genetic approaches (single-gene knockout competition assays) and chemical inhibitors

(low-dose MPS1i and taxol), providing robust evidence that alterations in the SAC or

microtubule dynamics can contribute to KIF18Ai resistance. The study also employs super-

resolution STED microscopy to assess chromosome alignment and spindle organization, finding

that KIF18Ai-exposed cells somehow compensate for the inactivation of KIF18A to achieve

narrower metaphase plates, more defined kinetochores, and altered microtubule flux. This

supports the claim that resistant cells can restore mitotic functions independent of KIF18A,

although the mechanism for this remains unknown.

The third claim that reducing PP2A or APC/C activity potentiates KIF18Ai sensitivity is

also supported by a combination of genetic and chemical perturbations. The presence of PP2A

and APC/C complex genes among dropouts in the earlier CRISPR screen connects this work to

other studies showing that APC/C complex weakening marks sensitivity to KIF18Ai. The APC/C-

mutant cell line experiments also provide interesting nuance to this. CDC23-mutant HT29 cells

did not have clear differences in KIF18Ai sensitivity relative to CDC23-WT, while the CDC16-

mutant, MPS1i-adapted HCT116 cells exhibited longer mitotic arrests upon KIF18Ai treatment

as compared to wild-type. This result indicates that resistance to MPS1i (or tolerance to SAC

weakening) can make cells more sensitive to KIF18A inhibition. Together, these data suggest

that the APC/C complex may modulate response to KIF18Ai in a context-dependent manner

rather than serving as a universal feature of sensitivity to this drug class as previously

described.

In summary, this manuscript performs a thorough investigation of mechanisms of

KIF18Ai resistance in multiple cell types using a wide array of approaches. The authors both

discover novel characteristics of KIF18Ai-resistant cells (mitotic timing, etc.) and also reproduce

key findings from prior work about genetic factors influencing KIF18Ai sensitivity. Therefore, this

work nicely adds to the growing body of literature that examines the new drug class of KIF18A

inhibitors. Overall, the data provided support the authors’ main claims, though there are some

sections that would benefit from further study or clarification.

Major points:

1) The authors verify that KIF18A itself is not altered or overexpressed in the exposed cells

relative to the naïve cells by Sanger sequencing and western blotting. They also confirm2) 3) 4) 5) that there are no remarkable differences in ploidy between exposed and naïve cells.

However, these do not exclude many other possible mechanisms of KIF18Ai resistance,

such as alterations in SAC and/or microtubule-associated genes that are identified and

investigated later in the manuscript. A more thorough comparative analysis of the

genomic and/or transcriptomic differences between the drug-naïve and drug-exposed

cells would provide critical insight into whether the genes of interest remain unaltered in

the exposed cells, and whether this could explain the mitotic adaptations reported by the

authors to dictate KIF18Ai response. For example, CENP-E is highly functionally

redundant with KIF18A; reasonably, upregulation of CENP-E could compensate for the

loss of KIF18A activity in the exposed cells. Gene expression analysis of other mitotic

kinesins and SAC genes via qPCR in the exposed cells relative to the naïve cells could

provide valuable insight into the mechanisms by which drug-resistant cells compensate

for the loss of KIF18A activity upon drug treatment. Additionally, these data would help

provide context for cell line-to-cell line differences in KIF18Ai response that remain

unknown, as they could potentially harbor or acquire genomic alterations in genes that

dictate KIF18Ai response (such as in hit genes identified in the CRISPR screen).

The ploidy analysis comparing DNA content in the naïve and exposed cells was also

useful, though interpretation is limited because these models are already triploid and

may not tolerate further whole-genome doubling (WGD). Full exclusion of WGD as a

resistance mechanism would require these experiments to also be performed in diploid-

range cell lines, such as the RPE1 cells used later in the study.

The authors make an important point that KIF18Ai-resistant cells do not become

dependent on KIF18Ai for short-term survival, evidenced by comparable rates of

chromosome missegregation in drug-naïve and drug-exposed cells after a drug holiday.

Further, the resistance is stable after drug holiday, which is consistent with the selection

of stable resistant subclones of cells during the initial six-week drug treatment. However,

further genomic and/or transcriptomic profiling is necessary to truly confirm these claims.

The manuscript performs a genome-wide CRISPR screen in TP53-null RPE1 cells +/-

KIF18Ai, and the results nicely align with those from other screens aiming to identify

genetic determinants of KIF18Ai sensitivity (doi10.1038/s44318-024-00031-6). It would

benefit the field as a whole if the authors include a correlative analysis of their results

with those from prior screens. This would help further characterize consistencies across

studies and identify genotype-phenotype interactions that are truly ubiquitous across

contexts.

The authors perform the drug evolution experiments in Figure 1a at 250 nM because it

induces KIF18A relocalization from kinetochores to spindle poles in each cell line used.

However, the intrinsic level of sensitivity to the drug for each cell line may be variable

and affect how much each cell line must adapt to prolonged treatment. If some cell lines

are only minimally affected at 250 nM with little cell death while others are severely

affected with most cells dying, it could confound the results. Including dose-response

experiments or annexin-V staining (for apoptotic cells) at a few doses for each cell line

would more completely characterize the response to long-term KIF18Ai treatment and

how that differs among the models used in this study.Minor points:

6) Overall, the manuscript would be improved by including more details in the figure

captions. The captions for a few figure panels (Figures 1d, 2a, 2d-h, 3a, 3d-f, 3h, 3k, 4b-

c, 4e, 5b-h, S1a, S3a) are missing information about the doses of KIF18Ai that are used

for those experiments. The addition of these details in each figure itself or within the

figure captions would help clarify what is being tested in each experiment and make the

work easier to follow.

7) It would also be helpful to include details about the experiment that generated the data

for Figures 1f and 1g, as right now there is ambiguity. Including the drug dose and

experimental details within the figure captions would markedly improve readability of the

manuscript.

I reviewed this preprint as part of the Transparent and Accountable Peer Review training

program. This report was written with the support of my research mentor.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.