Inhibition of Nuclear Pore Complex Formation Selectively Induces Cancer Cell Death

Nuclear pore complexes (NPC), the multiprotein channels that connect the nucleus with the cytoplasm, are the sole gateway to the genome (1). Together with nuclear transport receptors, NPCs control nucleocytoplasmic molecule exchange. Recent studies have shown that many cancer cells, in particular multidrug resistance cells and cells from aggressive tumors, have increased numbers of NPCs and higher nucleocytoplasmic transport rates, and become addicted to the nuclear transport system, suggesting that the nuclear transport machinery is a vulnerability of several cancers (2–7). Consistent with this idea, inhibitors of nucleocytoplasmic transport have been found to be highly effective at inducing cancer cell death and are the subject of multiple clinical trials (8, 9).

Because NPCs are essential cellular structures, directly targeting these channels has historically been viewed as an unfeasible therapeutic strategy. Notably, we and others previously uncovered that nondividing cells strongly downregulate the expression of key NPC components and maintain their assembled NPCs for years, maybe even decades, suggesting the formation of NPCs would be mostly restricted to cell proliferation (10, 11). These findings put forth the exciting possibility that preventing the assembly of new NPCs would decrease the number of these structures only in cells that are actively dividing. Moreover, cells that divide faster, require more nuclear pores, or have a stronger dependency on the nuclear transport process, such as transformed cells, would be expected to be more sensitive to the reduction in NPC numbers. But whether reducing NPC numbers can indeed be exploited to differentially induce cancer cell death, and how it affects the physiology of normal cells, have not been investigated. Here, we provide evidence for the selective requirement of NPC formation in proliferating cells and uncover that inhibition of nuclear pore formation induces cancer cell death while it triggers a reversible cell-cycle arrest in normal cells that allows them to survive a block in NPC production. We also show that reducing NPC numbers is sufficient to prevent the growth of melanoma and colorectal xenograft tumors and to induce tumor regression. Consistent with their multifunctional nature, we demonstrate that suppression of NPC assembly affects multiple cellular processes associated with transformation, resulting in cancer-specific cell death. The restricted nature of nuclear pore assembly, the selective sensitivity of cancer cells to the inhibition of this process, and the simultaneous alterations of multiple cellular processes that contribute to cancer cell death indicate that blocking nuclear pore formation represents a promising strategy for antineoplastic therapies.

As a first approach to investigate if NPC assembly represents a vulnerability of cancer cells, we performed an siRNA screen against most nucleoporins to identify the most critical components for NPC formation. For this, malignant human melanoma A375 (BRAF^V600E) cells were transfected with siRNA pools against 28 of the 32 nucleoporins (Fig. 1A), and NPCs at the nuclear envelope were quantified by immunofluorescence using the mAb414 NPC antibody recognizing four different nucleoporins (NUP62, NUP153, NUP214, and NUP358). Consistent with previously described functions of scaffold nucleoporins (12–19), depletion of components from the NUP107–160 and NUP93–205 scaffold subcomplexes most strongly inhibited NPC assembly (Fig. 1B). NPC biogenesis is a highly ordered process that involves the stepwise recruitment of different components, with the NUP107–160 and NUP93–205 complexes being required at different stages of assembly (20). The NUP107–NUP160 complex comes early during assembly and is required for the subsequent recruitment of the NUP93–NUP205 complex to the nuclear envelope (20). The NUP93–NUP205 complex is critical for the later recruitment of the central channel nucleoporins required for the initiation of nucleocytoplasmic transport (20). To inhibit NPC formation at different stages of the process, we chose one nucleoporin from each complex, NUP160 and NUP93. Downregulation of these nucleoporins using the pool of four siRNAs from our screen (Fig. 1C) or individual siRNAs (Fig. 1D) strongly decreased the NPC signal at the nuclear envelope. Consistent with the previously reported order of assembly, depletion of NUP160 resulted in the loss of all tested NPC components from the nuclear envelope (Fig. 1D). On the other hand, downregulation of NUP93 did not entirely block the recruitment of the early scaffold nucleoporin NUP133 to chromatin and the nuclear envelope but inhibited the association of the later nucleoporins recognized by mAb414, preventing the formation of mature/functional NPCs (Fig. 1D).

The strong downregulation in the expression of critical assembly nucleoporins during cell-cycle exit and the low turnover of NPCs in postmitotic cells suggest that the formation of these structures is mostly restricted to actively dividing cells (10, 11). To test this hypothesis, we analyzed the consequence of blocking NPC assembly in nonproliferating and proliferating cells. We performed these experiments with A375 melanoma cells, which have been found to be sensitive to alterations in nucleocytoplasmic transport (2, 21), and used the depletion of the essential nucleoporins NUP160 and NUP93 to inhibit NPC assembly. For this, we generated stable A375 cell lines expressing doxycycline-inducible control shRNA or shRNAs against NUP160 and NUP93 that allowed us to temporally control the depletion of these nucleoporins. Treatment of these cells with doxycycline led to a strong depletion of NUP160 and NUP93 and resulted in inhibition of NPC assembly comparable with siRNA downregulation (Fig. 1E–G; Supplementary Fig. S1A). A375 cells were allowed to grow to confluency before being treated with doxycycline for 72 hours to eliminate the synthesis of new nucleoporins. Cells were then either replated at 1:10 dilution (T = 0) to allow cell proliferation or maintained in a confluent nondividing state, both in the continuous presence of doxycycline. We found that blocking NPC assembly by depletion of these essential nucleoporins strongly reduced the number of dividing cells without significantly affecting nondividing/confluent cells (Fig. 2A). Analysis of RNA expression and NPC staining indicated that, even though nucleoporin expression was strongly downregulated in both conditions (Supplementary Fig. S1B), blocking nuclear pore formation decreased NPC numbers only in proliferating cells (Fig. 2B). The same results were observed when growth was slowed down by decreasing the serum concentration in the culture media (Supplementary Fig. S1C). Cells grown in low serum, which decreased their proliferation, showed a lower inhibition of NPC assembly compared with cells grown in high serum media (Supplementary Fig. S1D). To further confirm that NPC formation is required only during cell proliferation, the expression of NUP160 was eliminated in proliferating myoblasts or differentiated postmitotic myotubes, and NPC levels at the nuclear envelope were analyzed 3 or 10 days after shRNA induction, respectively. Eliminating the expression of the essential nucleoporin NUP160 in proliferating myoblasts significantly reduced the number of NPCs in these cells, whereas the number of NPCs in postmitotic myotubes remained unaffected (Supplementary Fig. S1E). These findings further indicate that NPC formation is only required during cell proliferation. To investigate if inhibition of NPC assembly is cytostatic or results in cell death (cytotoxic), proliferating A375 cells depleted of NUP160 or NUP93 were stained with trypan blue or the apoptotic marker Annexin V, and the percentage of viable cells and apoptotic/necrotic cells was quantified. Figure 2C and D show that blocking the formation of NPCs results in decreased cell viability and a significant increase in the levels of Annexin V.

Previous studies have identified that cancer cells from different tumor types including melanoma (2, 21), colorectal cancer (7, 22), lung cancer (3), leukemia, and lymphomas (9, 23) have high sensitivity to nuclear transport inhibitors. The susceptibility of cancer cells to nuclear transport inhibitors suggests that they could also be more sensitive to inhibition of NPC formation than normal cells. In agreement with this prediction, blocking NPC formation in MOLM-13 (acute myeloid leukemia) and HT-29 (colorectal cancer) cells rapidly reduced the number of proliferating cancer cells, whereas normal proliferating IMR90 and HPF primary fibroblasts showed a much smaller reduction in cell growth after several days of inhibition (Fig. 2E). Because IMR90 and HPF fibroblasts grow slower than most cancer cells analyzed (doubling time: ∼35–50 vs. 18–20 hours, respectively; Supplementary Fig. S1F), we examined the effect of inhibiting NPC assembly in normal RPE1 retina pigment epithelial cells (doubling time: ∼20 hours) to account for the differences in growth rates. Even though the inhibition of NPC assembly also reduced the number of proliferating RPE1 cells (Fig. 2F), we noticed that unlike A375 cancer cells that undergo rapid cell death after approximately 3 to 4 days of NPC formation inhibition, these normal cells stopped proliferating without dying (Fig. 2G and H; Supplementary Movie). The same differential cell death response to inhibition of NPC assembly was observed between the HT-29 and MOLM-13 cancer cells and normal HPF cells (Fig. 2H), indicating that the selective sensitivity to the reduction in NPC numbers is a common feature of several cancer cells. Altogether, these findings suggest that, different from cancer cells that die in response to inhibition of NPC assembly, normal cells might undergo cell-cycle arrest.

To further investigate the response of normal and cancer cells to inhibition of NPC assembly, we performed cell-cycle analyses on NUP160-depleted A375 and RPE1 cells at different times after inducing knockdown. We identified that by day 4 of NPC assembly inhibition, the majority of normal RPE1 cells were arrested in the G₁ phase of the cell cycle (Fig. 3A) and showed no significant cell death (≤2%). On the other hand, when NPC formation was blocked in melanoma A375 cells, ≥50% of the cells were labeled with the Zombie cell death marker, and the remaining Zombie-negative cells (considered live cells) displayed a massive expansion of the sub-G₁ population which represents dying cells (Fig. 3A; Supplementary Fig. S2A). A similar cell death response was observed when inhibition of NPC formation was performed by depletion of NUP93 (Supplementary Fig. S2B). After 7 days of NPC inhibition, the small number of A375 cells remaining were almost all (≥87%) stained with the Zombie cell death marker, whereas RPE1 cells showed no significant cell death and remained arrested in the G₁ phase (Fig. 3B and C; Supplementary Fig. S2A). Like RPE1 cells, normal HPF primary fibroblasts also arrested in G₁ after NPC assembly was inhibited (Supplementary Fig. S2C). Cell cycle–arrested RPE1 cells showed no staining for β-galactosidase (Fig. 3D), indicating they did not become senescent and suggesting a reversible cell-cycle arrest. To investigate if reestablishing NPC assembly could restore the growth of RPE1 cells, doxycycline was eliminated from the culture media after 6 days of treatment, and cell number was measured over time. We observed that RPE1 cells grew back when the expression of NUP160 was reestablished (Fig. 3E). However, no increase in A375 cell number was observed under these conditions (Supplementary Fig. S2D). To determine if other normal cells respond similarly to inhibition of NPC formation, we depleted NUP160 in the immortalized human melanocyte cell line H3A, normally used as a control for the A375 melanoma cells (24), and in H9 human embryonic stem cells (ESC; ref. 25), which are most closely related to cancer cells. For these experiments, H3A stably expressing control or NUP160 shRNAs were treated with doxycycline to reduce NUP160 levels, whereas control or NUP160 siRNAs were used to inhibit NPC assembly in H9 ESCs. Both treatments resulted in strong downregulation of NUP160 expression levels, decreased NPC numbers, and increased p21 (CDKN1A) expression, a marker of G₁–S cell-cycle arrest (refs. 26, 27; Supplementary Fig. S3A and S3B). Release from NPC inhibition in H3A cells was done by removing doxycycline from the media on day 6, whereas the release in ESCs was caused by the standard progressive decrease of siRNA-induced knockdown due to the dilution of siRNA levels over time (average duration of >70%–80% siRNA knockdown is ∼5–7 days according to manufacturer; Supplementary Fig. S3C). In the same way as RPE1 cells, inhibition of NPC assembly in these cell types resulted in a strong growth arrest that was relieved when NUP160 expression was restored, and these cells regained their ability to proliferate with no major impact on cell survival (Supplementary Fig. S3D and S3E).

To determine if primary cells also arrest in response to inhibition of NPC formation, we generated a conditional NUP160 knockout mouse line in which exon 4 is flanked by loxP sites (Nup160^f/f) and crossed it with a mouse line constitutively expressing a tamoxifen-regulated Cre recombinase (R26-CreER^T2; ref. 28), which allows temporal control of NUP160 ablation. Hematopoietic progenitors isolated from these mice were treated with tamoxifen to eliminate NUP160 and inhibit NPC assembly, and cell proliferation, survival, and cell-cycle progression were analyzed over time (Fig. 4A; Supplementary Fig. S4A and S4B). As shown in Fig. 4B and C, knocking out NUP160 in primary hematopoietic progenitors was associated with decreased cell numbers but did not strongly increase cell death (<10% dead cells in the population), supporting the idea that primary hematopoietic cells also arrest in response to inhibition of NPC assembly. To confirm this, cell-cycle and EdU incorporation analyses were performed in control and NUP160 knockout cells at 72 hours of tamoxifen treatment. Consistent with a cell-cycle arrest, NUP160-depleted cells, but not control cells, showed a significant decrease in the number of EdU-positive cells and a strong increase in the G₁ population (Fig. 4D and E). Altogether, our data indicate that when NPC formation is inhibited, normal cells reversibly arrest in G₁, whereas transformed cells undergo massive cell death. These findings confirm the selective sensitivity of cancer cells to alterations in NPC formation.

Proliferating cells have two mechanisms of NPC formation. One occurs at the end of mitosis when nuclear envelopes re-form around the daughters' genomes (postmitotic NPC assembly), and the other one takes place during interphase when cells double their number of NPCs to prepare for the next cell division (interphase NPC assembly; ref. 29). Even though both assembly processes result in the formation of the same structure, they occur through fundamentally different mechanisms (20, 29). During mitotic nuclear envelope breakdown, the NPCs of the mother cell are disassembled into stable subcomplexes that are reused for the formation of new nuclear pores during postmitotic NPC assembly. If all nucleoporins are recycled and distributed evenly between daughter cells, each cell is expected to receive enough components to assemble half of the nuclear pores that the mother cell had when entering M phase. Inhibiting interphase NPC assembly, for example by blocking the expression of new NPC components, should not prevent cells from assembling nuclear pores using preexisting components at the end of mitosis but would preclude cells from forming new NPCs during interphase when their numbers need to double. This would lead to a progressive reduction in the number of NPCs with each cell division. Because cancer cells die in response to NPC assembly inhibition, one possibility is that when NUP160 expression is inhibited, these cells continue to divide, therefore diluting their NPC numbers to levels that are incompatible with cell survival. To test this hypothesis, A375 and RPE1 cells were allowed to reach confluency and then treated with doxycycline for 72 hours to eliminate NUP160 as described before. Cells were then diluted in presence of doxycycline to maintain the inhibition of NPC formation during cell proliferation, and the number of times each cell divided before either dying or arresting was analyzed by live imaging. Notably, most A375 and RPE1 cells divided 3 times before undergoing cell death or cell-cycle arrest, respectively (Fig. 5A). If anything, it was more common to see cells that divided more than 3 times in the RPE1 population than in the A375 (Fig. 5A). To confirm this using a different synchronization and quantification method, cells were arrested in G₁–S using thymidine in the presence of doxycycline to reduce NUP160 expression, labeled with CellTrace Violet dye (CTV), which allows us to quantify the number of times cells divide (30), and released in the presence of doxycycline to maintain the inhibition of NPC formation during cell division. Consistent with our imaging assays, we found that most NUP160-depleted cells divided a maximum of 3 times, whereas control cells kept dividing until the CTV dye was undetectable (Fig. 5B). Also, EdU incorporation assays on day 4 after inhibition of NPC assembly showed a similar reduction in the percentage of actively proliferating RPE1 and A375 cells, further indicating that A375 melanoma cells do not divide more than normal cells when NPC formation is inhibited (Fig. 5C; Supplementary Fig. S5A). Even more, when cells were analyzed 16 hours after the EdU pulse, none of the sub-G₁ cells showed EdU incorporation, confirming that dying cells are not undergoing cell division at this stage (Supplementary Fig. S5B).

Another possibility is that cancer cells are more dependent on interphase NPC assembly and that inhibition of this process results in a stronger dilution of nuclear pores with each cell division. To investigate this, we quantified the NPC signal at the nuclear envelope of A375 and RPE1 cells at different times after induction of control or NUP160 shRNAs. Consistent with a progressive dilution of NPC numbers, downregulation of NUP160 expression resulted in a gradual decrease of nuclear pore signal in both A375 and RPE1 cells (Fig. 5D and E). Notably, the treatment resulted in the same reduction of NPCs relative to controls in both cell types, with RPE1 cells having slightly fewer NPCs at the end of treatment consistent with their lower starting number of NPCs compared with A375 cells (Supplementary Fig. S5C). Altogether, these findings indicate that the death of cancer cells in response to the inhibition of NPC formation is not due to a higher dilution of NPCs at the nuclear envelope but due to an increased dependency on nuclear pores, and provide additional evidence for the selective susceptibility of cancer cells to alterations in the nuclear transport machinery.

In recent years, it has become evident that NPCs play critical roles in many cellular processes. Thus, reducing their numbers is likely to result in multiple alterations contributing to the decreased survival of cancer cells. Consistent with their canonical role in regulating nucleocytoplasmic molecule exchange, using two different transport reporters, we found that inhibiting the formation of NPCs was associated with a gradual decline in nuclear import efficiency in both cell types (Fig. 6A; Supplementary Fig. S6A and S6B). For RPE1 cells, nuclear import rates recovered when cells were released from NPC inhibition and cell proliferation was restored (Supplementary Fig. S6B). RNA-sequencing (RNA-seq) analyses of gene expression at different times after the inhibition of NPC assembly showed a progressive deregulation of gene activity that paralleled the alterations of nucleocytoplasmic transport (Fig. 6B). Even though the reduction in NPC numbers was associated with decreased nuclear import rates and the deregulation of genes in both cell types, A375 cells showed much larger alterations in these processes with a massive deregulation of gene activity by day 4 of NUP160 knockdown, supporting the higher dependency of these cells on the nuclear transport machinery. Because NPCs also play a key role in maintaining the nuclear permeability barrier, it is also a possibility that NUP160 depletion could lead to nuclear permeability alterations and the leaking of cytoplasmic and nuclear molecules that could allow normal cells to partially compensate the nuclear transport reduction. To investigate this possibility, we performed two complementary approaches. In the first one (Supplementary Fig. S7A and S7B), we investigated the diffusion of two inert fluorescent proteins of different sizes, Dronpa (∼26 kDa) and 2x-GFP (∼52 kDa), into the nucleus of control and NUP160-depleted RPE1 cells using fluorescence recovery after photobleaching (FRAP) as previously described (31, 32). In the second approach (Supplementary Fig. S7C), we analyzed nuclear permeability in vitro using a combination of two fluorescent dextrans of different sizes above the NPC passive diffusion limit of 40 to 60 kDa in plasma membrane–permeabilized cells (10, 33). Using these methods, we found that depletion of NUP160 was not associated with alterations in nuclear permeability (Supplementary Fig. S7A–S7C), consistent with the NUP107–160 complex being required at the very early stages of NPC assembly before the nucleocytoplasmic connection is established (34).

In addition to regulating nucleocytoplasmic transport, NPCs have been found to play important roles in DNA damage response (DDR; refs. 35–39). These findings suggest that the reduction in NPC numbers could lead to an altered DDR and increased accumulation of DNA damage. Alterations in the DNA-repair machinery and higher levels of genome damage are a hallmark of cancer cells (40). Due to the increased mutation rates and fast proliferation, cancer cells generally show higher levels of DNA damage and replication stress and are more sensitive than normal cells to inhibition of the mechanisms that modulate these processes (40, 41). Consistent with cancer cells having more DNA damage than normal cells, A375 cells showed higher endogenous levels of total and phosphorylated (p) ATR (Ser428) and CHK1 (Ser345), two key components of DDR, as well as increased basal levels of the DNA damage marker γH2AX compared with RPE1 cells (Fig. 6C; Supplementary Fig. S8A). Remarkably, we found that the reduction in NPC number by depletion of NUP160 or NUP93 led to decreased levels of pATR, pCHK1, and total CHK1 in both cell types but increased the levels of γH2AX only in A375 cells (Fig. 6C; Supplementary Fig. S8B). NPC formation inhibition resulted in more A375 cells with γH2AX foci that were larger in size and intensity when compared with control-treated cells (Fig. 6D and E; Supplementary Fig. S8C and S8D). Similarly, blocking NPC assembly increased the fraction of cells with large 53BP1 foci in the A375 population, but not in RPE1 cells (Fig. 6F; Supplementary Fig. S8E). Notably, many γH2AX and 53BP1 foci observed in NUP160 knockdown cells were localized to the nuclear periphery (Supplementary Fig. S8F), supporting the previously described local role for NPCs in the resolution of DNA damage at the nuclear envelope (42, 43). Even though RPE1 cells do not show increased DNA damage when NPC assembly is inhibited, these cells show decreased nuclear import rates and transcriptional alterations (Fig. 6A and B). It is a possibility that when these cells are released from NPC assembly inhibition, re-entering S-phase with these alterations could lead to DNA damage and genome instability. To test this possibility, we analyzed DNA damage in RPE1 and H3A cells 6 days after restoring NUP160 expression as described in Fig. 3E. We found that after regaining cell proliferation, these cells showed no increase in γH2AX staining compared with control cells (Supplementary Fig. S8G).

In normal cells, activation of the DNA damage checkpoint results in mTOR inhibition, which leads to reduced protein translation and G₁–S cell-cycle arrest (44). In many cancer cells, mTOR activity is deregulated and not inhibited by DNA damage (44, 45). In fact, sustained mTOR activity in these cells has been found to promote CHK1 gene expression and to be required for cell survival in the presence of increased DNA damage (45). Consistent with a role for mTOR in alleviating DNA damage accumulation in cancer cells, pharmacologic inhibition of this signaling pathway leads to decreased CHK1 protein levels (45, 46), accumulation of γH2AX (45, 47), increased levels and activity of 53BP1 (47), and reduced viability in several cancer cell types including A375 cells (48). mTOR regulation of CHK1 gene expression occurs through the activation of S6 kinase (45). Because we identified that the reduction in NPC numbers in A375 and RPE1 cells is associated with decreased CHK1 protein (Fig. 6C; Supplementary S8A) and mRNA levels (Supplementary Fig. S9A), as well as with the accumulation of γH2AX and 53BP1 foci (Fig. 6C–F), we investigated if blocking NPC assembly affected mTOR signaling by analyzing the phosphorylation levels of mTORC1 and its downstream effector S6. Although mTOR phosphorylation on Ser2448 was not strongly affected by NUP160 downregulation, we found that reducing NPC numbers was associated with a strong reduction in S6 phosphorylation (Ser235/6; Fig. 6G). These findings suggest that inhibiting NPC assembly prevents cells from properly signaling through mTOR, which leads to decreased CHK1 expression. Our findings indicate that inhibition of NPC assembly affects ATR and CHK1 signaling in both RPE1 and A375 cells, but only induces cell death in the latter. If decreased activity of CHK1 and/or ATR selectively induces the death of A375 cells when NPC numbers are reduced, we could expect that these cells should be more sensitive to inhibition of this DDR pathway. To test this hypothesis, A375 cells and RPE1 cells were incubated with the ATR inhibitor AZD6738 or CHK1 inhibitor CCT245737, and cell viability was quantified over 7 days. Figure 6H shows that both inhibitors selectively induced cell death in A375 cells but had minimal effect on RPE1 cell survival. Moreover, we determined that depletion of NUP160 in A375 cells did not further increase the sensitivity of these cells to ATR or CHK1 inhibition, suggesting that NPCs and ATR/CHK1 are part of the same pathway (Supplementary Fig. S9B). Together, these findings suggest that inhibition of NPC assembly induces the death of A375 cells, at least in part, through the modulation of ATR/CHK1 signaling.

MAD1 is a critical component of the spindle assembly checkpoint (49), and its displacement from NPCs was previously shown to lead to mitotic timing defects, mitotic checkpoint alterations, and genomic instability (50, 51). Notably, we found that the reduction of NPC numbers was associated with the loss of MAD1 from the nuclear periphery of interphase cells (Fig. 6I; Supplementary Fig. S9C). In addition to its mitotic roles, MAD1 was recently found to work with MAD2 and ATR to facilitate DNA damage repair at the nuclear periphery during interphase (52), which suggests that loss of MAD1 from NPCs might contribute to the increased DNA damage associated with the reduction in NPC number. Altogether, our findings indicate that inhibition of NPC assembly and the consequent reduction in the number of these channels results in multiple cellular alterations that contribute to the selective death of cancer cells.

The higher dependency of cancer cells on the nuclear transport machinery and their differential cell death response to inhibition of NPC assembly compared with normal cells suggest that targeting this process could represent a promising strategy for anticancer therapies. To investigate if inhibition of NPC formation is sufficient to affect tumor growth in vivo, A375 and HT-29 cells carrying inducible control or NUP160 shRNAs were used for xenograft studies. Cells were injected subcutaneously in NOD/SCID immunodeficient mice, and tumors were allowed to grow to 100 to 125 mm³ (∼9–14 days) before shRNA induction. Control or NUP160 shRNAs were induced by doxycycline, and tumor volume was measured over time. As shown in Fig. 7A and B, inhibition of NPC formation in established tumors by depletion of the essential nucleoporin NUP160 prevented tumor growth and led to significant tumor regression. Similarly, depletion of NUP93 was recently described to delay the growth of breast cancer tumors in vivo (53). To confirm that NUP160 depletion in vivo prevents NPC formation, tumors were collected at the end of treatment, sectioned, and stained with antibodies against the NPC component NUP98 (Fig. 7C; Supplementary Fig. S10A). NUP160-depleted tumors showed drastically reduced levels of the cell proliferation marker Ki-67, increased necrosis, and higher terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) staining indicative of apoptotic cell death (Fig. 7D and E; Supplementary Fig. S10B and S10C). To investigate if inhibition of NPC assembly could also be exploited for the treatment of liquid tumors, we performed xenograft studies using MOLM-13 acute myeloid leukemia cells. To follow the growth of leukemia cells in vivo, MOLM-13 cells carrying control or NUP160 shRNAs were transduced with lentivirus-expressing luciferase and selected to generate stable cell lines. Cells were transplanted into NOD/SCID immunocompromised mice, and animals were followed by in vivo bioluminescence imaging. Similar to A375 and HT-29 cells, inhibition of NPC assembly in MOLM-13 cells strongly inhibited tumor growth (Fig. 7F; Supplementary Fig. S10D). Altogether, these findings demonstrate that inhibition of NPC formation could represent an effective treatment for solid and liquid tumors.

NPCs are essential cellular channels and the only gateway into the genome. Besides controlling the transport of molecules across the nuclear envelope, these structures perform multiple transport-independent functions that are critical for cell homeostasis. The formation of NPCs is tightly regulated in cells, and alterations in the numbers and function of these structures have been proposed to contribute to malignant transformation and cancer development. Here, we demonstrate that NPC assembly is predominantly required in proliferating cells, and that inhibiting this process selectively affects the survival of cancer cells. We found that although reducing the number of NPCs leads to a reversible cell-cycle arrest in several normal cells, it induces the rapid death of cancer cells, providing evidence of how normal and transformed cells differentially respond to the reduction of nuclear pore numbers. The selective response of cancer cells to the inhibition of NPC assembly exposes the potential targetability of this process for cancer therapies.

Alterations in the nuclear transport machinery have long been observed in cancer cells (54). Recent evidence showing that many cancer cells become addicted to nuclear transport has fueled the idea that the nuclear transport machinery represents a promising target for cancer (3, 6, 7). Consistent with this idea, inhibitors of the main nuclear export factor CRM1/XPO1 have been found to be highly effective in inducing cancer cell death and are the focus of more than 40 different clinical trials encompassing both solid tumors and hematologic malignancies (8). Different from nucleocytoplasmic transport, our findings show that the formation of NPCs is primarily required in proliferating cells. As a consequence, blocking NPC assembly might achieve the same result as inhibition of nuclear transport but without affecting nondividing cells such as neurons and muscle. Our data also indicate that although cancer cells undergo rapid cell death in response to inhibition of NPC assembly, normal cells undergo a reversible cell-cycle arrest and recuperate their ability to divide once NPC formation is restored. Notably, we observed that cancer cells die even though the dilution of NPC numbers is not stronger than in normal cells, indicating that they need more NPCs to survive and further confirming the increased dependency of cancer cells on the nuclear transport machinery. This selective sensitivity of cancer cells to the inhibition of NPC formation suggests that blocking this process could be exploited for cancer treatments. In support of this idea, we present evidence that blocking NPC formation inhibits tumor growth and promotes tumor regression. Collectively, our data expose the process of nuclear pore formation as a novel susceptibility of cancer cells and put forth the exciting possibility that reducing the number of these structures might represent a viable strategy for cancer therapies. The fact that many different cancer cells are highly sensitive to nuclear transport alterations suggests that a therapy based on inhibition of NPC formation could be exploited for the treatment of multiple cancer types. Finally, we found that besides reducing nuclear transport rates, the inhibition of NPC assembly results in multiple cellular alterations that contribute to cancer cell death, including alterations in gene expression regulation, DNA damage repair, cell signaling, and cell-cycle progression. Replication stress is another likely contributor to the decreased survival in NPC-depleted cancer cells. Consistent with the previously described role of NPCs in DNA replication (55, 56), we identified that dilution of NPC numbers results in decreased ATR signaling, a key responder to replication stress, and increased γH2AX levels in cancer cells, a marker of stalled replication forks and double-strand breaks (57). Because cancer cells normally have higher levels of replication stress and aberrant responses to DNA damage, the ATR signaling pathway becomes critical for their survival. Consequently, the decreased activity of the ATR-mediated response that results from lowering NPC numbers is likely to selectively affect the survival of cancer cells. The simultaneous alteration of multiple cellular processes that contribute to cell death is a strong therapeutic advantage that would make the emergence of resistance to the inhibition of NPC formation more challenging for cancer cells.

In this work, we inhibited NPC assembly by depletion of essential nucleoporins. Therapeutically, this could be achieved by indirectly modulating the transcription, translation, or degradation of individual nucleoporins. Directly targeting nucleoporins, on the other hand, could represent a more difficult technical challenge. Most of these proteins have a very limited set of domains, generally lacking functional or catalytic domains that could be targeted by small molecules (1, 58). Moreover, the crystal structure for most structural components of the mammalian NPC has not been solved yet (58), limiting the ability to develop structure-based rationally designed inhibitors. As an alternative approach, NPC formation could be inhibited by modulating the activity of cellular factors that regulate the assembly process. This could also avoid the undesired phenotypes that could result from the tissue-specific and NPC-independent functions described for several nucleoporins.

A375 and HT-29 cells were obtained from the ATCC. HPF cells (Catalog #3310) were obtained from ScienCell. RPE1 cells engineered to knock out puromycin were generated and kindly provided by Dr. Andrew Holland (John Hopkins University). Primary human myoblasts, IMR-90, MOLM-13, Hermes 3A (H3A), and H9 cells were kindly provided by Drs. Alessandra Sacco, Pier Lorenzo Puri, Anirunda Deshpande, Ze'ev Ronai, and Evan Snyder respectively (Sanford Burnham Prebys Medical Discovery Institute). Cell lines not obtained from the ATCC have been authenticated by the SBP Genomics Core Facility, which utilizes the GenePrint10 (Promega) system for short tandem repeat analysis on genomic DNA. All cells were cultured at a humidified 37°C with 5% CO₂ in media supplemented with 10% heat-inactivated FBS (Sigma-Aldrich) and 1 unit/mL penicillin–streptomycin (GE Healthcare Life Sciences) unless otherwise noted. Cells were routinely tested for Mycoplasma contamination (MycoAlert Mycoplasma Detection Kit LT07–118). A375 and HPF cells were cultured in DMEM (Mediatech). HT-29, MOLM-13, and H3A cells were cultured in RPMI-1640 medium (GE Healthcare Bio Science, Hyclone). IMR-90 cells were cultured in Eagle Minimum Essential Medium (Quality Biological) with 20% FBS and no penicillin–streptomycin. RPE1 cells were cultured in DMEM/F-12 with GlutaMAX (Gibco). Primary human myoblasts were cultured on plates coated with collagen (Corning, Rat Collagen I) in proliferation media (Ham's F-10 Nutrient Mix with 15% FBS) or differentiation media (DMEM with 2% Horse Serum, Gibco) and no penicillin–streptomycin. H9 cells were cultured in complete mTesR1 medium (STEMCELL Technologies) with no penicillin–streptomycin on Matrigel (Cell Applications)-coated tissue culture plates. For low-serum experiments, cells were grown in complete media with 0.6% FBS. For irradiation-induced senescence, cells were irradiated with 10 Gy using an RS2000 (Rad Source Technologies). For inhibitor treatments, cells were grown with CHK1 inhibitor CCT245737 (Selleck Chemicals) or ATR inhibitor VE-821 (R&D Systems) or AZD6738 (Cayman Chemical Company) for the specified times and counted directly or imaged using a Celigo Imaging Cytometer (Nexcelom Biosciences).

Cells were plated at the indicated amounts and counted daily. To count cells, conditioned media were collected, and adherent cells were washed once with PBS and incubated with trypsin-EDTA (0.25%, Life Technologies) for 5 minutes. Cells in trypsin were pooled with conditioned media and spun down. Cells were resuspended in PBS and counted 50% in Trypan Blue Solution, 0.4% (Life Technologies) with a TC20 cell counter (Bio-Rad Laboratories). Doubling time was calculated by dividing relative change in cell number by time when cells were in log phase growth.

Cells were seeded in polymer coverslip 8-well chamber slides (IBIDI) and grown for the indicated times. All subsequent steps were performed at room temperature (RT). Cells were fixed in 2% to 4% paraformaldehyde (methanol-free) in PBS for 5 minutes and blocked using IF buffer (1x PBS, 10 mg/mL BSA, 0.02% SDS, 0.1% Triton X-100) for 30 minutes. Cells were then incubated with primary antibody in IF buffer for 1 hour at RT or overnight at 4°C and washed with IF buffer. Cells were then incubated with secondary antibody in IF buffer for 1 hour at RT and washed with IF buffer. Cells were then incubated with 1 μg/mL Hoechst 33342 (Life Technologies) in PBS for 5 minutes and washed with PBS before imaging.

Cells were seeded with doxycycline in polymer coverslip 8-well μSlide dishes (IBIDI) and grown in a humidified 37°C with 5% CO₂ stage top incubator (H301, Okolab) controlled by the Oko-Touch (Okolab). Cells were imaged on a Leica DMi8 microscope at 10X magnification one frame per 20 minutes for 5 days. Twenty to 26 cells were manually tracked, and number/time of cell divisions was recorded.

Analytical cytometry was performed in the Sanford Burnham Prebys Flow Cytometry Core using a BD LSRFortessa (BD Biosciences) or a BD LSRFortessa X-20 (BD Biosciences). For proliferation, live cells were labeled with 5 μmol/L CTV (Life Technologies) for 10 minutes at 37°C, washed with cold DMEM supplemented with 0.5% FBS and 10 mmol/L HEPES, and cultured in complete medium at 37°C. CTV dilutions were analyzed at the indicated times by flow cytometry. For viability staining, live cells were stained with Zombie NIR Fixable Viability dye (BioLegend) in PBS at 1:500 for 10 minutes. For cell-cycle staining, cells were fixed with 70% cold ethanol in PBS for 1 hour at -20°C and stained with 2.5 μg/mL Hoechst (Invitrogen) in PBS with 2% FBS. For EdU staining, cells were incubated with 10 μmol/L EdU for the indicated times, and Click-iT Plus EdU Alexa Fluor 647 Flow Cytometry Assay (C10634, Invitrogen) was used according to the manufacturer's directions. For Annexin V staining, live cells were stained with 2.5 μL Alexa Fluor 488 Annexin V (Life Technologies A13201) in 100 μL 1X Binding Buffer (140 mmol/L NaCl, 10 mmol/L HEPES, and 2.5 mmol/L CaCl₂) for 15 minutes before flow cytometry.

Nup160 exon 4 floxed (fl) mice were generated by Cyagen US Inc. in the C57BL/6N background using Turboknockout technology. Nup160^fl/f animals were crossed with R26-CreER^T2 mice (Stock No. 008463, The Jackson Laboratory) to generate Nup160^fl/fl-CreER^T2 mice. Lineage negative hematopoietic progenitor cells were isolated from mouse bone marrow (BM) of these animals. BM cells were flushed from femurs and tibia into PBS and 1.2% FBS, and passed through a 40 μm nylon mesh to obtain a single-cell suspension. After red blood cell lysis, Lineage-negative hematopoietic progenitor cells were isolated by negative selection using EasySep Mouse Hematopoietic Progenitor Cell Isolation Kit (STEMCELL) according to the manufacturer's instructions. Briefly, BM cells were incubated first with rat serum and biotinylated antibodies directed against nonhematopoietic stem cells and nonprogenitor cells [CD5, CD11b, CD19, CD45R/B220, Ly6G/C(Gr-1), and TER119], and then with streptavidin-coated magnetic particles. Labeled cells were captured using a magnet, and the Lin-negative cells were collected. The efficacy of enrichment was verified by flow cytometry. Purified Lin-negative cells were cultured in 6-well plates at 0.2 × 10⁶ cells/mL in DMEM, 15%FBS, 20 ng/mL SCF, 10 ng/mL IL6, and 6 ng/mL IL3 for 24 hours at 37°C and subsequently treated with DMSO vehicle or 4-hydroxytamoxifen (0.5 μmol/L, Sigma) for 72 hours. Cells were counted daily with Trypan Blue.

Cells were transfected with the plasmids pcDNA6.2/C-Lumio expressing NES-Tomato-NLS, 2x-GFP and 2x-GFP-NLS (59), or Dronpa (Addgene # 54682) using Lipofectamine RNAiMAX (Life Technologies), according to the manufacturer's instructions. Twenty-four to 72 hours after transfection, cells were imaged on a Leica SP8 confocal microscope. For FRAP, nuclear signal was photobleached with 50% power 488 nm (GFP) or 552 nm (Tomato) laser for 3 seconds, and transport of fluorescent signal was recorded for 2 minutes using the LAS X software. For Dronpa diffusion, entire cell Dronpa signal was inactivated with 50% power 488 nm laser for 3 seconds, a 10 μm² area of cytoplasm was activated using 25% power 405 nm laser for 100 ns, and diffusion of Dronpa signal out of the cytoplasm and into the nucleus was recorded for 12 seconds using LAS X software. A total of 7 to 9 nuclei per group were analyzed. Data were normalized by calculating the difference between the nuclear and cytoplasmic background-subtracted signal and normalizing to the average maximal recovery of the control.

RPE1 cells transfected with control or NUP160-specific siRNAs were seeded on 8-well μSlide IBIDI dishes to achieve 60% to 80% confluency at 96 hours after transfection. At 96 hours of knockdown, cells were washed with PBS twice and placed on ice. Cells were equilibrated with permeabilization buffer (20 mmol/L Tris-HCL pH 7.5, 110 mmol/L potassium acetate, 5 mmol/L magnesium chloride, and 0.25 mol/L sucrose) for 5 minutes before being permeabilized in permeabilization buffer plus 20 μg/mL digitonin (Millipore CHR-103) for 4 to 7 minutes. Cells were washed 3 times with assay buffer (20 mmol/L Tris-HCL, pH 7.5, 110 mmol/L potassium acetate, 5 mmol/L sodium chloride, 2 mmol/L magnesium chloride, and 0.25 mol/L sucrose) for 1, 5, and 10 minutes. A mixture of 65 to 85 kDa (Sigma T-1162) and 500 kDa (Sigma FD-500S) fluorescent dextrans, 0.02 mg/mL each, plus Hoechst 5 mg/mL were added, and cells were imaged on a Leica SP8 confocal microscope. The intranuclear signals for the 65 to 85 kDa and 500 kDa dextran were quantified using Image J and normalized. For each image, the intranuclear dextran intensity was normalized to the extracellular signal and graphed as a relative intensity of 65 to 85 kDa/500 kDa.

All mice were housed under 12-hour light/12-hour dark cycles and fed ad libitum, and all experiments started between 6 and 12 weeks of age. Immune-compromised animals were bred and housed in sterile caging and maintained under pathogen-free conditions. Doxycycline hyclate (Alfa Aesar) was administered in drinking water at 0.5 mg/mL for the indicated time. The tumor injections and measurements were performed by the Sanford Burnham Prebys Animal Facility Core. In 100 μL 50% Matrigel (Cell Applications) and 50% DMEM, for each condition, 2 × 10⁶ cells were injected subcutaneously into both the right and left flank. Length and width were measured, and volume was calculated using the formula (length × width²)/2 as previously described (3). When average tumor size reached 100 to 125 mm³, doxycycline was administered. For MOLM-13, 2 × 10⁶ cells were injected via tail vein. To measure bioluminescence, 250 μL luciferin (PerkinElmer, 15 mg/mL) was i.p. injected, and radiance was measured by Xenogen IVIS Spectrum imager. When luciferase signal reached 25,000 average radiance (p/s/cm²/sr), doxycycline was administered. All animals were housed at the Sanford Burnham Prebys Medical Discovery Institute, and all experiments were approved by the Institutional Animal Care and Use Committee and were performed in accordance with institutional guidelines and regulations.

On the final day of the experiment, mice were euthanized by isoflurane overdose, and tumors were measured and frozen in OCT (Tissue-Tek) using isopentane equilibrated to liquid nitrogen. Tumors were cut into 10 μm frozen sections using a cryostat (Leica Biosystems). Frozen sections were thawed and processed for immunofluorescence or histology. Hematoxylin and eosin staining and TUNEL assay were performed by the Sanford Burnham Prebys Histology Core.

Microsoft Excel was used to record and calculate data. GraphPad Prism 8 software v8.2 (GraphPad Software, Inc.) was used to prepare graphs and to perform statistical analyses. Outcomes were compared using a two-tailed unpaired Student t test when comparing two groups. Multiple t tests with corrections for multiple comparisons or two-way ANOVA were used when comparing greater than two groups. Microscopy data were collected with a DMi8 Leica SP8 confocal microscope and analyzed using the Leica Application Suite X software v3.1.5.16308 (Leica Microsystems) or ImageJ v2.0.0-rc-54/1.51h (NIH). Relative mAb414 intensity was calculated on single Z section images in ImageJ. For each cell, a polygon was drawn around the nuclear periphery, mean mAb414 fluorescence was calculated, and values were expressed relative to control. γH2AX foci were calculated on max projection images in ImageJ by finding maxima on after minimum threshold. MAD1 line intensities were extracted from single Z section images in ImageJ. Flow cytometry data were collected using the BD FACSDIVA Software (BD Biosciences) and analyzed using FlowJo software v10.0.8r1 (Tree Star, Inc.), and cell-cycle distribution was calculated with the Dean-Jet-Fox model. qPCR data were collected in a CFX384 Real-Time PCR Detection System (Bio-Rad). Celigo images were acquired with Celigo Software Version 2.1 (Nexcelom Biosciences).

M.A. D'Angelo reports a patent for PCT/US2018/041461 pending. No disclosures were reported by the other authors.

S. Sakuma: Formal analysis, validation, investigation, writing-original draft, writing-review and editing. M. Raices: Formal analysis, investigation, writing-review and editing. J. Borlido: Formal analysis, validation, investigation. V. Guglielmi: Validation, investigation. E.Y.S. Zhu: Validation, investigation. M.A. D'Angelo: Conceptualization, supervision, funding acquisition, investigation, writing-original draft, project administration, writing-review and editing.

We thank Drs. S. Heynen-Genel and P. Aza-Blanc for assistance with the siRNA screen. We also thank Drs. A. Holland, A. Sacco, P.L. Puri, A. Deshpande, Z. Ronai, and E. Snyder for providing cell lines used in this work. This work was supported by a Pew Biomedical Science Scholar Award and Research Scholar Grant RSG-17-148-01-CCG (to M.A. D'Angelo) from the American Cancer Society and by the NIH (awards RO1AR065083 and R01AI148668). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. This work was additionally supported by the NCI Cancer Center grant P30 CA030199, which supports the Animal, Flow Cytometry, Genomics, Bioinformatics and High Content Screening cores, and the Conrad Prebys Center for Chemical Genomics at SBP.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.