The Deubiquitin Enzyme USP28 Maintains the Viability of Triple Negative Breast Cancer Cells through Stabilizing the RecQ Family Helicases

Background: Triple-negative breast cancer lacks significant expression of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2. It is the more aggressive and malignant kind of breast cancers and is currently without any effective targeted therapies. Methods: We have screened for genes in the ubiquitin-proteasome system that are essential for the proliferation and survival of TNBC cells via CRISPR/Cas9-mediated gene editing. Growth of TNBC cells were assayed using cell and tumor xenograft models to validate the vital role of USP28 . We employed cell biology and biochemical methods to uncover the mechanisms underlying the requirement of USP28 for the proliferation of TNBC cells. Results: USP28 , a deubiquitin enzyme, is an essential gene for TNBC cells in vitro and in vivo. Compromising the function of USP28 causes TNBC cells to arrest in S/G2 phases with DNA damage checkpoint activation. We show that RecQ family helicases are regulated by USP28, which is more important in TNBC cells than in other breast cancer cells. We further showed that a small molecule inhibitor of USP28 displayed anti-tumor activities against xenografts derived from TNBC cells. Conclusion: Our data establish a critical role played by USP28 in supporting the proliferation and viability of triple negative breast cancer cells through stabilizing RecQ family helicases and support USP28 as a therapeutic target for TNBC. Secondary antibodies conjugated to horseradish peroxidase for western blot and Secondary antibodies anti-mouse, -rabbit for


Introduction
Triple-negative breast cancer (TNBC) lacks significant expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) [1]. Although it is only 10-20% of all breast cancers, TNBC is more aggressive and poorer in prognosis than other breast malignancies. Gene expression profiling of TNBC showed enrichment of genes involved in DNA damage checkpoint response [2,3], which is consistent with the finding of higher levels of endogenous DNA damage in TNBC cells in vitro and in vivo [4,5]. Moreover, about 20% TNBC patients carry mutation in BRCA1/2 [6,7]. While hormonal and targeted therapies are available for ER positive and HER2 positive breast cancers, triple negative breast cancer currently lacks targeted therapies (except BRCA1/2 mutated ones) and relies on chemotherapy [8].
Almost all these substrates, especially c-MYC, have been demonstrated to contribute significantly to the initiation and progression of various types of cancers. Consistent with that, Usp28 was shown to help the tumorigenesis of colorectoral cancer in mice [12].
More recently, USP28 was shown to be required by squamous cell carcinoma (SCC) due to its function in maintaining stability of Np63 [15].
Here we report that USP28 is required for the viability of triple negative breast cancer cells and this requirement is not only associated with its function in maintaining 4 the stability of c-MYC protein but more importantly with its role in protecting RECQL5 and other members of the RecQ family helicases from proteasome-mediated degradation.
These helicases are required for TNBC cells to deal with the elevated levels of endogenous DNA damage resulted from replication stress [5,16]. As a result, compromising the function of USP28 causes TNBC cells to arrest in S/G2 phases with DNA damage checkpoint activation. We further showed that a small molecule inhibitor of USP28 displayed anti-tumor activity against xenografts derived from TNBC cells. Our results point USP28 as a therapeutic target for triple negative breast cancer.

Materials and Methods
Cell culture

Reagents
The antibodies used in this study were as follows: USP28 (17707-1-AP, 1

Plasmids and lentiviruses
Plasmids used in the work were generated through standard cloning methods.
shRNAs were constructed in pLKO.1 with following sequences: The USP28 cDNA (wild-type or C171A mutant) and RECQL5 cDNA were cloned into lentiviral vector pHAGE. The c-MYC cDNA (wild-type or T58A nutant) were cloned into pCMV.
Lentiviruses-carrying overexpression, knockdown or knockout elements were produced in the lab and used to infect the above cell lines with MOI (multiplicity of infection) ＞1. The infected cells were selected with puromycin treatment (4 μg/ml for 2 days).

Loss of function screening based on CRISPR-Cas9
For Ubiquitin proteasome system (UPS) -wide genetic screen, the human sgRNA library targeting UPS (containing 2197 sgRNAs targeting 1098 UPS related genes and 1 negative control) was constructed into pLKO-U6-sgRNA. 7 6 million of HCC1937 cells were infected with UPS sgRNA lentiviral pool at a MOI of 0.1-0.3 to achieve an average coverage of 300-fold of the library after puromycine selection. 3 million sgRNA expressing cells were infected with adenovirus expressing GFP or Cas9(>30 million copies). After 10 population doublings of the cells, a final pool of 3 million cells was harvested. Genomic DNA was extracted using TIANamp Genomic DNA Kit (Tiangen). sgRNA inserts were PCR amplified using KOD FX (TOYOBO) from 2 million genome equivalents of DNA to ensure an average coverage of 1000x of the library. Primer sequences used for amplification of sgRNA inserts are: The PCR products were purified and subjected to deep sequencing. sgRNA abundance in both GFP and Cas9 treated groups were calculated and analyzed.

Assays for cell proliferation
For MTS assay, after lentiviral infection and selection, the cells were trypsinized and re-seeded in 96-well plates at a density of 3,000 cells/well and cultured for the

Western blotting analysis
The cells were lysed with RIPA buffer (Applygen Technologies Inc., Beijing, China) supplemented with a protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany). Western blots were processed according to standard procedures and analyzed with Chemiluminescent imaging system LAS 500 (GE Healthcare).

Immunoprecipitation
HEK293T cells cultured in 10 cm dish were transfected with 5 μg Flag-RECQL5 volumes of NETN buffer. The cell lysates were incubated with Flag M2 beads overnight at 4℃ followed by washing and western blotting analysis.

Immunostaining
The cells after the indicated treatment were plated on coverslips, fixed with 4% paraformaldehyde for 15 min, permeabilized in PBS containing 0.5% Triton X-100 for 5 min, and blocked with 5% BSA in PBS for 1 h at room temperature, followed by incubation with primary antibodies at 4°C overnight. After 3 washes in PBS, the coverslips were incubated with secondary antibodies for 20 min at 37℃. All images were taken on a Nikon Ni-E microscope (Nikon Corporation, Tokyo, Japan), with identical exposure times for each sample.

Statistics
Statistics analyses were performed with GraphPad Prism9.0 and ImageJ.

Identification of USP28 as an essential gene in TNBC cell lines
We performed a CRISPR/Cas9-based genetic screen to identify genes essential for the survival/proliferation of triple negative breast cancer cells. We used the BRCA1 deficient line HCC1937 for the screening. The gRNA library was designed against genes

USP28 is required for the proliferation of triple negative breast cancer cells
To determine if knocking down USP28 expression with shRNA could also achieve the effect of the gene editing, we designed two shRNAs against the 3'-UTR region of USP28 and used them to deplete USP28 expression in a number of TNBC cell lines. Knocking down the expression of USP28 indeed could block the proliferation of these TNBC cell lines ( Fig. 1A and Fig. S2 A-D), but could block neither the proliferation of MCF7 (Fig S2E) nor that of MCF10A cells (Fig. S2F). Importantly, the effect of these two shRNAs on proliferation could be rescued by re-expression of USP28 ( Fig. 1B and C). Further, we synthesized the reported USP28 inhibitor AZ1 [17] and used it to treat TNBC cells. In HCC1806 cells, a 48-hour AZ1 treatment resulted in a dose-dependent suppression of proliferation ( Fig. 1D-E), and similar results were obtained with MDA-MB-231 and HCC1937 (Fig. 1F).
To demonstrate that USP28 is required for the proliferation of TNBC cells in vivo, we depleted its expression in HCC1806 cells via the two shRNAs above and inoculated 13 the cells in female nude mice. As expected, these cells could not grow as xenografts ( Fig.   2A). Further, we used AZ1 to treat mice carrying HCC1806 xenografts. Although AZ1 is not a potent USP28 inhibitor, it did show anti-TNBC tumor activities (Fig. 2B).
These results demonstrate that UPS28 is uniquely required by the triple negative breast cancer cells for proliferation in vitro and in vivo, but not required by ER positive breast cancer cells nor by the normal mammary epithelial cells, although USP28 is expressed at similar levels across these cell lines (Fig. S2G).

The requirement of USP28 in TNBC cells could not be bypassed with MYC expression
USP28 was shown previously to regulate the stability of c-MYC protein [9]. We therefore set out to determine if it does so in triple negative breast cancer cells as well. As shown in Fig. 3A, depleting USP28 did cause decreases in c-MYC levels in TNBC cells.
However, c-MYC levels did not decline in MCF7 cells (Fig. 3A), which is consistent with the finding that USP28 loss of function, either through Cas9-mediated editing (Fig.   S1D) or shRNA-mediated depletion (Fig. S2E), had no effect on their proliferation. Thus, it is possible that the depletion of USP28 in triple negative breast cancer cells reduces c-MYC to a level insufficient to support proliferation, making USP28 essential in these cells. To determine if that is the case, we first analyzed cell cycle distribution in USP28 depleted cells. Given the function of c-MYC in promoting G1/S transition, it was expected that USP28-depleted TNBC cells would show a G1 arrest. However, we saw an increase in G2/M fraction instead (Fig. 3B), while MCF7 cells did not show any cell cycle disturbance (Fig. 3B). Furthermore, in both MDA-MB-231 and HCC1937 cells, S 14 phase fraction also increased upon USP28 depletion (Fig. 3B). These results suggest USP28 regulate more than c-MYC in TNBC cells to help their proliferation.
We next sought to determine if re-express c-MYC could rescue USP28 depletion.
However, we were unable to express wildtype c-MYC in USP28-depleted cells, as the wildtype c-MYC protein relies on USP28 for stability. To circumvent that, we expressed c-MYC T58A , a mutant form resistant to ubiquitination catalyzed by FBXW7 [18,19]. c-MYC T58A was overexpressed in MDA-MB-231 cells first, and USP28 expression was subsequently depleted via shRNA (Fig. 3C). Cell proliferation assay on these cells demonstrated that restoring c-MYC expression could not prevent these cells from losing the ability to proliferate due to USP28 depletion (Fig. 3C). However, the exogenously expressed c-MYC T58A did not cause an increase in the rate of proliferation neither (Fig.   3C), which made us wander whether the c-MYC T58A expressed was functional. To rule that out, we expressed the same c-MYC T58A in human normal diploid fibroblasts IMR90.
Both wildtype c-MYC and c-MYC T58A could enhance the proliferation of IMR90 cells, and c-MYC T58A showed a stronger effect than the wildtype (Fig. S4). Thus, the c-MYC T58A expression construct was functional. It is likely that MDA-MB-231 cells express already high enough levels of c-MYC, and thus further increasing its expression levels would not be able to further stimulate proliferation. Taken together, these results indicate that the reduction in c-MYC stability resulted from USP28 depletion is not the reason for triple negative breast cancer cells to cease proliferation. 15 The cell cycle profiles of USP28-depleted TNBC cells (Fig. 3B) suggest DNA damage checkpoint activation that stalled the cells in S and G2 phases. Together with our previous report of increased levels of DNA damage in TNBC cells due to elevated levels of replication stress [5], we decided to examine DNA damage levels in USP28-depleted TNBC cells. Staining of H2AX showed that indeed the endogenous DNA damage levels increased significantly, from 30% in the control to more than 60% of H2AX positive cells when USP28 was depleted (Fig. 4A). Consistent with that, the level of phosphorylated CHK1 also increased significantly (Fig. 4B), indicating activation of DNA damage checkpoint. These results provide an explanation for the observed stall in S and G2 phases (Fig. 3B).

Increased levels of endogenous DNA damage in USP28-depleted TNBC cells
USP28 was reported to regulate the stability of CLASPIN [14] which is involved in DNA damage response in S phase. We therefore hypothesized that such a regulation is critical in TNBC cells to relieve replication stress and prevent subsequent DNA damages.
However, the levels of CLASPIN did not change upon USP28 depletion (Fig. 4C), suggesting other substrates of USP28 are more important in this regard in TNBC cells.
We showed previously that the function of RECQL5 was critical in triple negative breast cancer cells [5] and therefore decided to look into the possibility that USP28 regulates its stability. As shown in Fig. 4D, depletion of USP28 led to a significant decrease in the levels of RECQL5 in TNBC and non-TNBC (T47D) cells. Furthermore, the other family members of RECQL5, namely the Bloom syndrome gene product (BLM), the Werner syndrome gene product (WRN), and RECQL4, all depended on the function of USP28 to maintain their levels (Fig. 4E). The only exception is RECQL1 which seemed unaffected by USP28 status. Furthermore, AZ1 treatment recapitulated the effects of USP28 depletion including the activation of CHK1 (Fig. 4F). Taken together, these results suggest that it is very likely that maintaining protein stability of RecQ family helicases underlies the requirement of USP28 in triple negative breast cancer cells.

USP28 regulates the stability of RECQL5
Having shown the requirement of USP28 for maintaining RECQL5 levels, we next sought to demonstrate that it regulates RECQL5 stability. To that end, we first determined the half-life of RECQL5 in control and USP28-depleted cells. As shown in The results above suggest that USP28 is a deubiquitin Enzyme (DUB) for RECQL5. To demonstrate that, we first set out to determine if the two proteins interact with each other. As shown in Fig. 6A, immunoprecipitating RECQL5 could bring down USP28 and vice versa. Next, we looked at the levels of ubiquitin attached to RECQL5 to see if they were affected by USP28 status. Flag-tagged RECQL5 and shRNA against USP28 were expressed in 293T cells, and the cells were then treated with and without MG132 to block proteasome activities. Immunoprecipitates obtained with anti-Flag antibodies from these cells under denaturing conditions were separated in a SDS-gel and 17 blotted for detection of ubiquitin. It is clear that the depletion of USP28 resulted in much higher levels of ubiquitin associated with RECQL5, but only if the cells were treated with MG132 (Fig. 6B). Otherwise, the ubiquitinated RECQL5 was degraded.
Opposite to depleting USP28, overexpression resulted in a decrease of ubiquitination of RECQL5 (Fig. 6C), and as expected, overexpression of USP28 C171A did not cause such a decrease (Fig. 6C). Taken together, these results demonstrate that USP28 is the DUB that prevents excess ubiquitination and degradation of RECQL5 and likely other RecQ helicases as well. It has been reported previously that Recql5-deficient cells are sensitive to replication stress inducers such as camptothecin [20]. We expect that USP28 depletion would also result in enhanced sensitivity to replication stress in non-TNBC cells. 18 Therefore, we treated control and USP28-depleted breast cancer cells T47D with 5-FU or CPT and monitored cell viability. Indeed, USP28 knockdown cells became much more sensitive to the two chemo agents (Fig. 7B).

Discussion
Triple negative breast cancer is the more aggressive form amid all breast cancers and currently lacks effective therapies. We show here that USP28 is a potential therapeutic target of TNBC. Loss of its function prevents TNBC cells from proliferation in vitro and in vivo. Given the role of USP28 in maintaining c-MYC stability [9] and the importance of MYC in cell proliferation, we reasoned that the requirement of USP28 by TNBC cells for proliferation might be related to its function as a DUB for c-MYC. As expected, depletion of USP28 could lead to destabilization of c-MYC in TNBC cells (Fig.   3A). However, restoring c-MYC levels through overexpression could not rescue the lethality resulted from USP28 depletion (Fig. 3C), indicating that the lethality is a result of more than just a decrease in c-MYC levels (although which might still be a contributing factor).
We have shown previously that TBNC cells suffer unusually high levels of replication stress that require the function of RECQL5 to relieve [5]. Although it is unclear the source of replication stress in TNBC cells, it is likely that single strand lesions are the culprit. Interestingly, depletion of USP28 could worsen the replication stress in TNBC cells (Fig. 4A), despite the fact the DUB has been shown not to play any significant roles in DNA damage responses [13]. In addition, we did not see 19 destabilization of CLASPIN upon USP28 depletion in TNBC cells but did observe enhanced activation of CHK1. These results suggest that USP28 regulates other proteins involved in DNA metabolism. Given the phenotypic similarities between USP28 depletion and that of RECQL5 in TNBC cells, we suspected that USP28 might regulate RECQL5 stability. Indeed, RECQL5 is a substrate of USP28 (Fig. 6). Depletion of USP28 resulted in more ubiquitination and destabilization of the helicase. Furthermore, we showed that not only RECQL5, but its family members are also regulated by USP28, which explains, at least in part, the failure of our attempts to rescue USP28 depletion with overexpression of RECQL5 (data not shown). Taken together, our results uncovered the link between USP28 and the RecQ family helicases. The protection by USP28 on these helicases seems dispensable as mice lacking Usp28 are more or less normal [12,13], but it becomes indispensable in triple negative breast cancer cells (Fig. 7C). In normal cells, although loss of USP28 does cause decreases in the levels of RecQ helicases, such decreases would presumably still leave sufficient amounts of these helicases behind to deal with the low level (if any) of replication stress. However, in the cells with induced replication stress, USP28 is no longer dispensable (Fig. 7C).
USP28 was first identified due to its function in DNA damage response through protecting CLASPIN [14]. We found here that depletion of USP28 in TNBC cells did not lead to a decrease in CLASPIN levels (Fig. 4C), suggesting that whether USP28 functions as CLASPIN's DUB is cell context dependent. Previous studies also showed that USP28 was dispensable in DNA damage response, as Usp28-deficient MEFs were not more sensitive than wildtype cells to DNA damage agents including ionizing 20 radiation [13]. However, we show here that USP28 is required for cells to cope with replication stress, either arose internally such as in TNBC cells, or induced by chemotherapeutic agents (Fig. 7B).
It is interesting but not unexpected that depletion of USP28 did not lead to a reduction in c-MYC levels in MCF7 (Fig. 3A)

Conclusions
While USP28 has been shown to regulate the stability of a number of other oncogene products, such as LSD1 [10], NOTCH1 [12], etc., our data establish a new critical role played by USP28 in supporting the proliferation and viability of triple negative breast cancer cells through stabilizing RecQ family helicases and support USP28 as a therapeutic target for TNBC.         The precipitates were then subjected to immunoblotting analysis with antibodies against HA, Flag, and USP28.   Targeting USP28 impedes the growth of HCC1806 xenografts. (A) HCC1806 cells infected with shNC or shUSP28 lentiviruses were innoculated subcutaneously into female Balb/c nude mice. The tumors were collected three weeks after the inoculation. Tumor weight and size were measured. Data are mean ± SD (n = 5 per group). Student's t test: *, p<0.05; **, p<0.01; ***, p<0.001. (B) HCC1806 cells were innoculated subcutaneously into female Balb/c nude mice. When the average tumor size reached 100 mm3, the mice were divided randomly into two groups (n=7 per group). One group were treated with vehicle, another with AZ1 (200 mg/kg, oral gavage) every day. The tumor sizes were measured every three days and the tumors were harvested 12 days after the treatment. Data are mean ± SD. Student's t test: *, p<0.05.