Aberrant activated APOBEC3B is associated with p53 mutant-driven refractory/relapsed Diffuse Large B-cell Lymphoma

and and in relationship between and induced expression


Abstract Background
Although treatment of diffuse large B-cell lymphoma (DLBCL) has progressed considerably in recent years, treatment failure still occurs in about 40% of patients who are refractory/relapse. Recent studies suggest that TP53 mutation may be an important cause of refractory/relapse in DLBCL, but the cause of TP53 mutation remains unclear.

Methods
In the present study, the correlation between TP53 mutation status and APOBEC3A and APOBEC3B expression in DLBCL specimens was searched by detecting the correlation between TP53 mutation and APOBEC3B expression in combination with database informatics analysis. Further, the relationship between APOBEC3B expression and TP53 mutation was analyzed by constructing APOBEC3B induced expression DLBCL cell lines. The effects of APOBEC3B-induced TP53 mutants on DLBCL cell proliferation and drug resistance were also tested.

Results
We identify APOBEC3B as a critical factor that regulates p53-mutant driven drug resistance of DLBCL. APOBEC3B induces TP53 mutations of DLBCL cells, and its mutation patterns are similar to those in DLBCL patients. Moreover, APOBEC3B-induced p53 mutants promoted growth of DLBCL cells as well as contributed to drug resistance. In human DLBCL, APOBEC3B is aberrantly activated and associated with p53 mutant-mediated refractory/relapsed DLBCL.

Conclusion
These ndings yield insights into the mechanism of refractory/relapsed DLBCL induced by p53 mutants and reveal APOBEC3B as a new therapeutic target.

Background
Diffuse large B-cell lymphoma (DLBCL), the most frequent subtype of lymphoid malignancy in adulthood, is a heterogeneous disease. Although current standard-of-care immunochemotherapeutic regimen of rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) has greatly improved the outcome of DLBCL patients, and durable remission is achieved in 50% of cases, DLBCL remains a signi cant clinical challenge, with approximately one-third of patients not being cured by this regimen, highlighting a need for more potent and safer therapies against novel targets [1][2][3] .
TP53 is an important tumor suppressor which participates in regulation of cell cycle, DNA repair, apoptosis and senescence 4 . TP53 deletion or mutation is frequent in B-cell malignancies and is associated with a low response rate 5,6 . Importantly, several hotspot mutations including 283,248,273,175,176 and 213, which are mainly located in DNA binding domain especially the loop-sheethelix and L3 motifs, are independent negative predictors for DLBCL [7][8][9] . However, the mechanism of TP53 mutation in DLBCL is still not fully understood. Interestingly, most of these mutations, especially the highrisk hotspot mutations, were G/C to A/T mutations 10 , which are similar to those induced by apolipoprotein B mRNA editing enzyme catalytic polypeptide-like (APOBEC)3 s family 11,12 . These facts prompted us to investigate whether APOBEC3s are responsible for the G/C to A/T mutations in DLBCL.
In the present study, through analysis of TP53 mutation in specimens from refractory/relapse DLBCL patients, we investigated the relationship between APOBEC3B and TP53 hotspot mutations in DLBCL. We expect to reveal the mechanism of TP53 G/C to A/T mutation and then hope to provide some indications for further monitoring and treatment of DLBCL.

Cellular localization of APOBEC3s
Immuno uoscence staining of APOBEC3s-HA were described previously 15  supplemented with 10% fetal bovine serum and penicillin/streptomycin. Hek293 and 293T cell lines were purchased from ATCC, and cultured in DMEM medium supplemented with 10% fetal bovine serum and penicillin/streptomycin. All cell lines were maintained at humidi ed, 5% CO 2 atmosphere at 37℃. 2.6.2. APOBEC3A/APOBEC3B-inducible DLBCL cell lines APOBEC3A/APOBEC3B-ag cDNA was cloned into pLV-Ptight-puro vector (Clontech, Takara Bio, China) to construct inducible APOBEC3A/APOBEC3B-ag/pLV-Ptight-puro plasmid. Pfeiffer, OCI-LY10, 293T and Hek293 were transfected using Polyjet In Vitro DNA Transfection Reagent (SignaGen, USA) according to manufacturer's protocol. The 293T cells were transfected with APOBEC3A/APOBEC3B-ag/pLV-ptightpuro vector as well as PCL and 10A1. Seventy-two hours after transfection, virus supernatants were collected and puri ed by centrifuge and 0.45 um lter. Then the virus supernatant was stored at -80℃ for the subsequent experiments. Pfeiffer, OCI-LY10 and Hek293 cells were seeded in 6-well culture plate, 24 hours later the medium was replaced with fresh medium and virus supernatant were added into the medium as well as 8ug/ml polybrene (Qiagen, USA). After infection for 12 hours, the medium was replaced with fresh medium. Forty-eight hours later, 3ug/ml puromycin (Qiagen, USA) was added to the medium for selection of infected clones. Inducible expression of APOBEC3A/APOBEC3B-ag was con rmed by Western blot using anti-ag antibody (Cat No. M185-3L, MBL, Japan).
After selection by puromycin for 14 days, the remainder cells were seeded at concentration of one cell/well in 96-well plate and checked by Western Blot after cultured with medium containing 4ug/ml doxycycline (Sigma, USA). Positive clones were maintained and used in the following steps.

Detection of APOBEC3A/APOBEC3B-induced G/C to A/T mutations in TP53 exon8
After 14 days of induction by doxycycline, the APOBEC3A/APOBEC3B-ag-inducible cells as well as empty pLV-pTight-puro vector cells were collected and total DNA were extracted using Qiagen DNA mini kit (Qiagen, USA). The G/C to T/A mutations were detected and analyzed by 3D-PCR based Sanger sequencing method as described above.
2.8. Analysis of proliferation and drug sensitivity of p53mutants-drived DLBCL cells: 2.8.1 Colony formation assay was performed as previous described 16 . Brie y, APOBEC3B-inducible DLBCL cells were induced by 4ug/ml doxycycline for 14days, then the doxycycline was removed and cultured for another 24 hours until APOBEC3B-Flag disappeared. APOBEC3B-expression cells and control cells were seeded in 6-well culture plates containing semi-soft agarose at a concentration of 200 cells per well. Then the cells were cultured for another 14 days and colonies were stained by Giemsa and calculated under microscope. Doxorubicin sensitivity was measured by MTT assay as previous The Ly10-TP53-Mutants was constructed by lentiviral infection which performed accordingto the standard procedures. Brie y, the 293T cells were co-transfected with viral packaging vectors PCL and 10A1 (Clontech, Takara Bio, China), along with a lentiviral construct expressing vector or the empty vector as control, using Polyjet transfection reagent (SignaGen, Rockville, MD, USA). The transfection medium was replaced after 6 hours with fresh complete DMEM, and 48 hours later the viral supernatants were collected. Then the viral supernatants were added in Ly10 cells and accompany with 5 µg/mL Polybrene. The medium was replaced after 24 hours with fresh RPMI 1640 supplement with 10%FBS, and 72 hours later 2 µg/mL puromycin was added to the infected cells for selection. Doxorubicin sensitivity was measured by MTT assay as previous described 17 .

Statistical Analysis:
The statistical analyses were performed with IBM SPSS Statistics 20. Frequency tables and descriptive statistics (mean, median, minimum and maximum) were used for summarizing characteristics of the patients. Differences between compared groups of patients were assessed by Maximum Likelihood Chisquare test and Fisher exact test in categorical variables and by Mann-Whitney test in continuous variables. PFS was de ned as time to disease progression, relapse or death. PFS was estimated using the Kaplan-Meier survival curves method, and compared using the log-rank test. The Kaplan-Meier method was used for univariate survival analysis. Multivariate Cox proportional hazard models were used to evaluate whether TP53 mutation was independent prognostic factors for progress free survival (PFS). Colony formation data were reported as mean values ± SEM and were analyzed by independent t-test. Doxorubicin IC50 analysis were analyzed by GraphPad Prism 5 (San Diego, California, USA). P values < 0.05 were considered statistically signi cant.

TP53 exon8 of refractory/relapse DLBCL samples contain more G/C to A/T mutations
Using 3D-PCR based direct sequencing method, G/C to A/T mutations were detected in R/R DLBCL samples ( Fig. 1 and Figure S1), and most of these nucleotide mutations resulted in changes of amino acid in p53 protein ( Fig. 2A and Table S1). The distribution of these mutation points was not random, several mutation points, such as R273C, R282W, R282Q, R283C and R290H (Fig. 2B), were more frequent than others. Most of these points were previously reported hotspot mutations including R273C, R282Q, R282W, and R283C. TP53 exon8 of R/R DLBCL samples contained more G/C to A/T mutations than that of non-R/R DLBCL samples. TP53 exon8 G/C to A/T mutations were detected in 83.33% of R/R DLBCL patients (15/18), whereas only 23.26% (10/43) of non-R/R patients contained such mutations (X 2 = 18.93, P < 0.001) (Fig. 2C). Hotspot mutations were detected in 14 of 18 R/R DLBCL patients, but only 5 of 43 non-R/R DLBCL patients (X 2 = 25.89, P < 0.001) (Fig. 2D).
After following-up for 12 months, the complete response (CR) rate of different groups, based on TP53 exon8 status, were calculated and compared. The overall CR rate for the current cohort was 68.90%, which was similar with that of previous reported 3,18 . But the CR rates in different groups, based on TP53 status, were dramatically different. For the TP53 exon8 wild type group, the CR rate was 81.08% (30/37) (Fig. 3A). The TP53 exon8 mutation group had the CR rate of 28.57% (8/28) (compared with wild type group, X 2 = 18.1, p < 0.01, data not shown). Among the TP53 exon8 mutation group, those who containing the hotspot mutation had the lowest CR rate, which is only 21.05% (4/19)(compared with wild type group, Fig. 3A), however, those who containing non-hotspot mutation had the CR rate of 80% (4/5)(compared with hotspot group, Fisher Exact Test, p = 0.028, compared with wild type group, X 2 = 0, p = 1, Fig. 3A). Thus, the TP53 wild type group and non-hotspot mutation group could be merged as one wildtype & non-hotspot mutation group, which had a higher CR rate compared with that of TP53 hotspot mutation group (X 2 = 23.203, p < 0.01, data not shown). The hotspot mutation group had a median PFS of 6 months (95% CI: 3.245 to 8.755, Fig. 3B). While after follow-up for 12 months, PFS of non-hotspot mutation group and wild type group both were not reached (Fig. 3B). Among these two groups, the patients' characteristics were similar except for gender and B symptoms (Table 1).  (Fig. 3D, p = 0.08). These results indicated that TP53 mutation status had a prognostic value that is not currently captured by widely used IPI model.
To assess whether TP53 mutation was an independent prognostic factor, risk factors were rstly screened by univariate survival analysis using Kaplan-Meier method ( In summary, these data suggest that TP53 mutation, especially hotspot mutation, contribute to the poor outcome of DLBCL when treated with R-CHOP.

APOBEC3B but not APOBEC3A increases in DLBCL compared with normal tissue
APOBEC3s family has seven members including APOBEC3A, APOBEC3B, APOBEC3C, APOBEC3DE, APOBEC3F, APOBEC3G and APOBEC3H. Among these members, APOBEC3A, APOBEC3B, APOBEC3F and APOBEC3G exhibit cytidine deaminase that induces G/C to A/T mutation in single DNA strand.
APOBEC3F and APOBEC3G cause hypermutation of HIV genome, but both of them are localized in cytoplasm ( Figure S2). APOBEC3A is localized both in cytoplasm and nucleus, whereas APOBEC3B is predominantly localized in nucleus ( Figure S2) and could access to genomic DNA. Previous studies showed that APOBEC3A and APOBEC3B were candidates for host DNA mutation. Based on Oncomine database analysis, Brune lymphoma database revealed that APOBEC3A mRNA level was similar in DLBCL with that of normal tissue (Fig. 4A), whereas APOBEC3B mRNA level was higher in DLBCL than that of normal tissue (Fig. 4B). Using TCGA based GEPIA database, analysis of APOBEC3B and APOBEC3A expression level in DLBCL showed the similar result (data not shown) as that of Brune database. But this result was not con rmed in Compagno lymphoma database ( Fig. 4D and 4E). Further analysis showed that expression levels of APOBEC3B were higher than that of APOBEC3A in DLBCL tissues both in Brune lymphoma and Compagno lymphoma database ( Fig. 4C and 4F).
Then expression levels of APOBEC3A and APOBEC3B in DLBCL were further compared in GEO database.
In GEO database, the results were sorted by number of samples from high to low. And the rst 13 data were analyzed. APOBEC3A expression level was higher than that of APOBEC3B in 3 of 13 data including the biggest one ( Figure S3A, S3D and S3I). While APOBEC3B expression level was higher than that of APOBEC3A in 10 of 13 data ( Figure S3). In the biggest sample number data GSE117556, RNA was extracted from FFPE tumor tissue. So APOBEC3A and APOBEC3B expression levels were analyzed in different sample resources in GSE19246, which contains both frozen tumor tissue and FFPE tumor tissue. The results showed that APOBEC3A expression level was similar between these two kinds of sample resources ( Figure S4A). While APOBEC3B expression level in frozen tumor tissue was higher than that of FFPE tumor tissue ( Figure S4B). These results indicated that FFPE tumor tissue maybe result in underestimating of APOBEC3B expression. And APOBEC3B expression level was higher than that of APOBEC3A in most of data based on frozen tumor tissues.
Therefore, analysis of online database suggested that APOBEC3B, instead of APOBEC3A, is higher expressed in DLBCL cells.

APOBEC3B level was higher in R/R DLBCL samples than that of non-R/R ones
Given the fact that APOBEC3B induces G/C to A/T mutation in human cancers such as breast cancer, and the fact that APOBEC3B is up-regulated in DLBCL, we next measured the protein level of APOBEC3B in R/R and non-R/R DLBCL samples using IHC method. Interestingly, we found that APOBEC3B protein levels were higher in R/R DLBCL samples (Fig. 5A-5F) than that of non-R/R ones (Fig. 5G-5K). As expected, APOBEC3B protein was predominantly localized in nucleus. If set the cut-off value of APOBEC3B positive rate at 20%, analysis of APOBEC3B expression level in different TP53 mutation status groups showed that APOBEC3B protein was higher in TP53 mutation group (58.33%) than that of TP53 wild type group (27.03%) (X 2 = 4.657 p = 0.038) (Table S3). More interestingly, APOBEC3B positive rate in TP53 hotspot mutation group (68.42%) was higher than that of TP53 wildtype & non-hotspot group (26.19%) (X 2 = 9.776 p = 0.004) (Table S4). These data suggested that over expression of APOBEC3B is associated with TP53 mutation, especially hotspot mutation, and refractory/relapsed in DLBCL.

APOBEC3B could induce G/C to A/T mutation in DLBCL cell lines
To further investigating whether over-expressed APOBEC3B could induce TP53 mutation in DLBCL, we performed an in vitro assay of TP53 mutation via inducible APOBEC3B-expression DLBCL cell clones. Using APOBEC3B-ag/pLV-Ptight-puro plasmid, we constructed an APOBEC3B-inducible DLBCL cell clones in TP53 wild type DLBCL cell lines Pfeiffer and OCI-LY10. APOBEC3B-ag expression was con rmed by western blot after induction by doxycycline. TP53 exon8 was ampli ed via 3D-PCR and c-MYC exon2, which is reported less mutated in DLBCL, was also ampli ed as control. As the denature temperature low-down from 94℃ to 87℃, the TP53 exon8 fragment was detected at 88℃ and 87℃ in APOBEC3B-inducible cells but not in the control cells (Fig. 6A). While MYC exon2 fragment was not detected at denature temperatures lower than 94℃ (Fig. 6B). Also, over expression of APOBEC3A in DLBCL cells could not result in detection of TP53 exon8 fragment under 89℃ via 3D-PCR ( Figure S5). Mono-clone sequencing of the TP53 exon8 PCR products at 87℃ indicated that over 20% clones contain more than one G/C to A/T mutation compared with the control group and the 94℃ PCR products. The major mutation patterns were TG and GC. The mutation pattern and sites were similar with that of DLBC samples (Fig. 6D). Further analysis showed that there were several known hotspot amino acid mutations including R273C, R282Q, R282W and R283W (Fig. 6D). These data suggested that APOBEC3B could induce TP53 G/C to A/T mutation, including those hotspot mutations, in DLBCL cells.

APOBEC3B-induced p53 mutants promoted cell proliferation and caused resistance to doxorubicin
As previous showed that inducible-expression of APOBEC3B could induce p53 mutants, then we analysis the in uence of APOBEC3B-induced p53 mutants on proliferation and drug sensitivity of DLBCL cells (Pfeiffer) using previous APOBEC3B-expressed cells. Semi-soft colony formation assay showed that after induced expression of APOBEC3B for 14 days, the APOBEC3B-expression cells generated more colonies that that of control group although APOBEC3B was restored ( Fig. 7A and 7B). After induced expression of ABPOEC3B for 14 days, APOBEC3B-expression cells showed resistant to doxorubicin, with a 4.65-fold increase of IC50 value as compared with control (6.484ug/ml versus 1.395ug/ml) (Fig. 7C). In vitro doxorubicin sensitivity assay of DLBCL cells (Ly10) expressing p53 mutant also found that Ly10 cells carrying p53 mutants R273C and R282Q had reduced sensitivity to doxorubicin and increased IC50 by 2.28-fold and 2.23-fold, respectively (Fig. 8). These data suggested that APOBEC3B-induced p53 mutants could both promote the proliferation and cause drug resistance in DLBCL cells.

Discussion
In this study, we for the rst time demonstrate that APOBEC3B is a critical factor that induces p53mutation and drug resistance of DLBCL. P53 is an important key factor that regulating of cell proliferation, DNA repair, and apoptosis. TP53 somatic mutations were identi ed in many types of cancer and were regarded as an important carcinogenesis and drug-resistant mechanism in many cancers 10 .
Recent studies identi ed that TP53 mutation rate is about 20%-30% in DLBCL, with a similar incidence in germinal center B-cell-like (GCB) and activated B-cell-like (ABC) subtypes [7][8][9]19 . While in present study, 3D-PCR-based sequencing method could identify TP53 exon8 mutation rate at about 35.9% in whole cohort, which is higher than previous data. Even in non-R/R DLBCL cases the TP53 exon8 mutation rate is 23.26%. This maybe could be explained by selective ampli cation of DNA fragments containing G/C to A/T mutation 20 . These results suggest that 3D-PCR could detect the G/C to A/T mutation in a higher sensitivity than that of regular PCR method. But this method also has a shortcoming of missing of other mutation types. While as previous reported that G/C to A/T mutation type is the main mutation type in TP53 gene, and most of hotspot mutations were G/C to A/T mutation 10 , so 3D-PCR also could be used as a valuable method for TP53 mutation detection.
As an important tumor genetic factor, TP53 mutation was included as a poor prognostic factor for acute myeloid leukemia (AML) and chronic lymphocytic leukemia (CLL) in National Comprehensive Cancer Network (NCCN) guidelines as well as for multiple myeloma in Mayo Clinic sMART system. In recent years, several studies indicated that TP53 mutation is also a poor prognostic factor for DLBCL [7][8][9] . However, this relationship has not been consistent because of some controversial data that failed to demonstrate any correlation between TP53 mutations and prognosis 21,22 . In present study, results showed that TP53 exon8 G/C to A/T mutation was higher in R/R DLBCL patients than that of non-R/R group. Most of those mutations were missense mutation and caused amino acid change. And the predictive effect of TP53 exon8 mutation was also studied. Results showed that TP53 mutation group had a lower CR rate compared with that of wild-type group when treated with standard R-CHOP. Then the prognostic value of TP53 mutation status was analyzed in different IPI groups, which is the most wildly used prognostic factors model. Results showed that TP53 hotspot mutation also could discern patients with signi cantly distinct outcomes in IPI > 2 group. On the other side, the group with high risk IPI but wildtype & non-hotspot TP53 had the similar outcome with those IPI 0-2 group. Also, multivariate Cox model showed that TP53 hotpot was the strongest prognostic factor for PFS among TP53 mutation, IPI, LDH, stage and gender. These data indicate that TP53 hotspot mutation status provide robust prognostic information that is not captured by IPI.
Similar with previous reports, our data also showed that TP53 mutation could be regarded as a prognostic factor for DLBCL. In present study, only exon8 was sequenced and analyzed, this may underestimate the proportion of TP53 mutations. TP53 mutations of other exons also need to be detected to perfect this relationship. But this does not prevent the role of TP53 mutations in the prognosis of DLBCL.
Among the TP53 missense mutations, about a third of these mutations are located in six "hotspot" residues that are p.R175, p.G245, p.R248, p.R273, and p.R282 23 . Most of the previous reports did not analyze the differential prognostic value of different TP53 mutation positions in DLBCL. Young 9 . These indicated that hotspot mutations maybe more important for p53 activity and were more valuable for prognostic prediction. In present study, we also found that some hotspot mutation rate, such as 273, 282 and 283, were even higher in R/R DLBCL than that of non-R/R DLBCL. Those nonhotspot mutation patients had the similar CR rate and PFS with wild type group. But this need to be further studied because the non-hotspot mutation group was small in present study. These hotspot mutations, instead of non-hotspot mutations, were strongly associated with lower CR rate and shorter PFS in DLBCL. Previously data showed that these hotspot mutations could affect the DNA binding activity of p53 protein and result in failure of regulating of target genes [24][25][26] . Some of these hotspot p53 mutants were responsible for gaining of function in carcinogenesis and drug resistance [27][28][29][30] . For example, R273C and R273H mutants were reported to confer a more aggressive phenotype on cancer cells, as well as enhance resistance to DNA damaging drugs 31 . More recently, Boettcher S also found that TP53 missense mutations, including R273H and R282W, showed a dominant-negative effect, rather than gain of function, and caused drug resistance in AML cells 32 . Our data found that the hotspot mutations instead of all the mutation sites of TP53 were associated with worse survival of DLBCL. Which hotspot mutations are more important for prognostic evaluation of DLBCL and how these hotspot mutations affect the role of p53 in tumorigenesis and drug resistance are still unclear.
TP53 mutation could result in resistance to chemotherapy agents such as doxorubicin 32 and cisplatin [33][34][35][36] , which are usually used for DLBCL. Although there were some attempts for overcoming of TP53 mutation but this still needs further study. Lack of understanding of TP53 gene mutation mechanism is one of the important obstacles to overcome p53 mutant-mediated drug resistance. Among the reported TP53 mutation types, G/C to A/T is the most common type, especially some hotspot mutation sites. Most of the reported carcinogens, such as PAH (B[a]P), AA, a atoxin B1, Vinyl chloride and 3-NBA, doesn't induce G/C to A/T mutation 10 . Except for the UV radiation was reported could inducing CC to TT mutation in skin cancer 37 . But the mechanism of the majority G/C to A/T mutation in TP53 is still unclear. Previously our group and other groups had identi ed that some APOBEC3s family members could induce G/C to A/T mutation in viral genome including human immunode ciency virus (HIV) 38-40 , human T-lymphocytic leukemia virus (HTLV-1) [41][42][43][44] , and hepatitis B virus (HBV) 15,45−47 . In recent years, APOBEC3B is reportedly one of the most extensive candidate factors for studying G/C to A/T mutations in a variety of human cancers 14,48−50 . Moreover, some reports proved that APOBEC3s could be upregulated by interferon 51,52 , which is an important cytokine involved in in ammation. And chronic in ammation was regarded as a potential factor in lymphoma carcinogenesis. So, we raise the hypothesis that APOBEC3s, especially APOBEC3B, may be responsible for TP53 G/C to A/T mutation.
Among the seven family members (from APOBEC3A to APOBEC3H), APOBEC3B localized mainly in nucleus, APOBEC3G, APOBEC3F and APOBEC3DE localized mainly in cytoplasm, and APOBEC3A and APOBEC3C localized both sides 15,47,53 . Based on bio-informatics analysis of APOBEC3s expression in DLBCL, we found that APOBEC3B not APOBEC3A, was up-regulated in DLBCL in several lymphoma database. This suggests that APOBEC3B may be more possibly responsible for TP53 mutations in DLBCL.
Using inducible expression system, we over express APOBEC3B in DLBCL cells, and found that G/C to A/T mutation could be induced in TP53 exon8. More importantly, these in-vitro mutation patterns were as the same as that of in-vivo R/R DLBCL samples. Those hotspot mutants were also could be detected in APOBEC3B-induced p53 mutants as the similar pattern of that of DLBCL samples. While over expression of APOBEC3A could not induces G/C to A/T mutation in TP53. Also, c-MYC, which is less mutated in DLBCL 54-56 , was not mutated after APOBEC3B expression in DLBCL cells. These results indicated that APOBEC3B, instead of APOBEC3A, may responsible for the TP53 G/C to A/T mutation in DLBCL. APOBEC3B could selectively induce G/C to A/T mutation in special genes through an unknown mechanism in DLBCL. Further study found that these APOBEC3B-induced mutants could improve the proliferation of DLBCL cells as well as cause resistant to doxorubicin, one of the main drugs of CHOP.
This drug resistance maybe could be explained by previous work by Li J 31 and Boettcher S 32 . These results indicated that APOBEC3B-induced p53 mutants maybe responsible for the refractory and resistant of DLBCL. As APOBEC3B target different sequence in different cells, how APOBEC3B selective the targeting sequence and which factors were involved in this process needed to be further investigated.
Recently, p53 protein was reported could regulate the expression of APOBEC3B 57 , and APOBEC3B expression also increases APOBEC signature mutation in p53-defective cells 58 . So, the detailed crosstalk between p53 and APOBEC3B, during carcinogenesis and drug resistance, is still needed to be further investigated.

Conclusion
In this study, we for the rst time demonstrate that APOBEC3B is a critical factor that induces TP53 mutation and leads to drug resistance in DLBCL. As a DNA mutator, APOBEC3B may serve as a potential target to reduce the rate of TP53 mutation and improve the prognosis of DLBCL. TP53 mutants especially those hotspot mutants had the potential value for prognostic evaluation of DLBCL in the era of R-CHOP. We also provide a more sensitive method for TP53 mutation detection in tumor tissue DNA, and this may be helpful for further study of TP53 mutation not only in DLBCL but also in other caner types.

Declarations
Ethics approval and consent to participate This study was approved by the ethics committee of the hospital in accordance with principles of the Declaration of Helsinki. All the patients provide written informed consent at enrollment.

Consent for publication
All authors reached an agreement to publish the study in this journal.

Availability of data and materials
Not applicable.

Competing interests
The authors declare that they have no competing interests.      In uence of TP53 exon8 mutation on clinical outcome of DLBCL.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.