Transient AID expression for in situ mutagenesis with improved cellular fitness

Activation induced cytidine deaminase (AID) in germinal center B cells introduces somatic DNA mutations in transcribed immunoglobulin genes to increase antibody diversity. Ectopic expression of AID coupled with selection has been successfully employed to develop proteins with desirable properties. However, this process is laborious and time consuming because many rounds of selection are typically required to isolate the target proteins. AID expression can also adversely affect cell viability due to off target mutagenesis. Here we compared stable and transient expression of AID mutants with different catalytic activities to determine conditions for maximum accumulation of mutations with minimal toxicity. We find that transient (3–5 days) expression of an AID upmutant in the presence of selection pressure could induce a high rate of mutagenesis in reporter genes without affecting cells growth and expansion. Our findings may help improve protein evolution by ectopic expression of AID and other enzymes that can induce DNA mutations.

Protein engineering is a powerful technique to improve protein activity, stability and other properties for industrial, diagnostic and therapeutic applications 1 . Protein engineering by rational design is very effective, but requires knowledge of the structure and function of the protein of interest 1,2 . By contrast, directed protein evolution performed by alternating rounds of random mutagenesis and selection can be employed in many cases where rational design is not feasible.
An interesting approach for autonomous mutagenesis in mammalian cells has been inspired by the germinal center reaction in which B cells overexpress the enzyme activation induced cytidine deaminase (AID), which is essential for diversification of antibody genes in vivo 3,4 . AID mutagenesis has been applied successfully for evolution of antibody and non-antibody proteins in both B and non-B cells [4][5][6][7][8][9][10] . AID mutagenesis coupled with high throughput screening in mammalian cells appears to be an ideal system for protein evolution because it avoids repetitive transfection and recloning and provides selection for proteins with better expression, activity and stability 9 . However, AID mutagenesis for protein evolution suffers from two major drawbacks; low mutation rate and genotoxicity manifested by widespread genomic instability and reduced fitness when harmful mutations accumulate in highly transcribed vital genes [11][12][13][14][15] . Wang et al. investigated the possibility of improving AID mutagenesis by increasing the specific activity of AID. Indeed, AID upmutants with increased activity were isolated, but they also led to increased genomic instability 16 and impaired cells expansion 14 , which limits their use for protein evolution applications.
To better understand how to effectively use AID upmutants for mutagenesis in mammalian cells, we investigated in situ mutagenesis of a red fluorescence reporter protein by AID mutants with different enzymatic activities. We found that an AID upmutant (m7.3) generates a spike of mutagenesis shortly after expression (less than 10 days). Longer expression did not produce additional mutations in the reporter gene but reduced cell fitness and expansion. Thus, our results suggest that ideally, AID upmutants should be expressed transiently to maximize mutagenesis in target transgenes and minimize off-target toxicity.

Results
Somatic hypermutation reporter system. Because AID can mutate highly transcribed genes [17][18][19][20] , fluorescence reporter proteins are often used to monitor mutation rates 4,21 . We therefore generated a 293FT cell line that stably expresses monomeric RFP (RFP1) (Fig. 1A). Expression of AID in these cells can induce mutations

Mutagenesis of RFP1 transgene by stably expressed AID.
To investigate mutagenesis by AID, we examined either wild type AID (AID-WT) or three AID mutants; AID m7.3 (m7.3) which displays high catalytic activity 22 , AIDS38A (S38A) which displays about 30% of AID-WT activity 23 , as well as M6A which lacks AID activity 24 . AID mutants were cloned into a lentivirus expression plasmid (Fig. 1B) and recombinant lentiviral particles were used to stably infect 293FT/RFP1 cells. Expression of AID in 293FT/RFP1 cells was confirmed 3 days later by immunoblot analysis which detects the HA tag present on the recombinant AID proteins (Fig. 1C).
As expected, the high activity AID m7.3 mutant induced significantly more RFP negative cells on day 10 as compared to AID-WT, consistent with introduction of more mutations in the reporter gene ( Fig. 2A). The low activity S38A mutant induced fewer RFP negative cells while the inactive M6A mutant produced almost no loss of RFP florescence ( Fig. 2A), indicating that the percentage of RFP fluorescence loss can be used as a readout of relative mutation rates.
To study accumulation of mutations over time, we measured the percentage of RFP negative cells for one month (Fig. 2B). The m7.3 AID upmutant achieved maximum RFP florescence loss on day 10 but the frequency of RFP loss progressively decreased after this time, consistent with slower growth of cells that accumulated excessive harmful mutations. By contrast, cells that stably expressed AID-WT lost RFP fluorescence at a slower rate, which reached a plateau 24 days after infection. As expected, S38A cells accumulated RFP loss at a slower rate than AID-WT and maximum accumulation of RFP loss was not reached in 32 days. Transient expression of AID mutants. Since the maximum mutation frequency was achieved within 10 days after stable expression of m7.3, we asked whether transient transfection with this AID upmutant is sufficient  for inducing efficient somatic hypermutation. We first transiently expressed m7.3 and monitored its expression over time by Western blot analysis in HEK293FT and HEK293 cells (Fig. 3A). Although expression is very efficient after 2 days, it decreased dramatically over time probably due to plasmid dilution during cell replication. HEK293FT cells shows clear enhancement of expression after 2 days which may be attributed to replication of the plasmid (pAS3w.Ppuro), which contains the SV40 origin of replication (SV40 ori), in HEK293FT cells which express the SV40 large T antigen 25 .
We then compared expression level of m7.3 after transient and stable expression (Fig. 3B). Stable expression using lentiviral transduction produced continuous expression of AID over time, while transient transfection generated a large amount of the protein for a short time.
We then asked if we can benefit from the high efficiency of drug selection to extend AID expression by simply using short term drug selection to select cells that retain the plasmid longer. Indeed, we found that three days of drug selection was sufficient to enhance expression of the eGFP reporter by about three fold, which is expressed bicistronically with AID (Fig. 3C). Extension of AID expression was further confirmed by Western blot analysis of M6A, WT and m7.3 (Fig. 3D). AID protein was apparent at day 7 when drug selection is used, but decreased to undetectable levels by day 14 (Fig. 3D), indicating that expression is still transient, and stable transfectants were not selected by short-term exposure to puromycin.
Finally, we tested whether transient transfection with short term selection has an impact on inducing mutations in the reporter 293FT/RFP1 cells (Fig. 3E). When puromycin was not used, the mutation frequency was extremely low (Fig. 3F). By contrast, addition of puromycin for 2 or 4 days significantly enriched RFP fluorescence loss by 12 and 15 fold for AID-WT and m7.3, respectively (Fig. 3F). Enhancement of RFP loss by puromycin treatment could be due to preferential selection of cells that express AID for longer times.

Time course and multiple AID transient transfections.
Loss of RPF fluorescence was monitored for 4 weeks after transient transfection of AID mutants and selection in puromycin for 3 days. In all cases, maximum mutagenesis was achieved within 10 days after transient transfection, reaching around 3% and 8% for AID-WT and m7.3 respectively (Fig. 4A). The percentage of RFP negative cells did not change over a month follow up, consistent with the transient expression of AID and similar growth rates of the RFP positive and RFP negative cells. Limiting the time cells are exposed to AID may be important to balance between diversity generation and accumulation of deleterious mutations 26,27 .
We further examined if a second round of transient transfection can continue to produce mutations in the reporter gene, because repeated rounds of mutagenesis and selection are typically required for successful protein evolution. Indeed, the frequency of RFP negative cells significantly increased after a second round of transient transfection of AID (Fig. 4B,C), reaching around 13% of total cells with m7.3 AID.

Transient AID expression in a DsRed reversion assay.
To further test the usefulness of transient AID expression, we investigated reversion of a single fatal base pair in the red fluorescent protein (DsRed2) gene. We introduced a pre-mature stop codon within an AID hotspot motif, which renders the DsRed2 gene non-functional (Fig. 5A). Emergence and retention of DsRed2 fluorescence was studied over time after mutagenesis by AID-WT and m7.3 by either stable or transient transfection protocols. Stable expression of m7.3 (Fig. 5B) induced rapid emergence of red fluorescence, which reached a maximum frequency of 1.7% after two weeks and then gradually decreased over time. Stable expression of AID-WT required about 42 days to accumulate a similar number of DsRed2 positive cells. Transient transfection of m7.3 also produced about 1.8% DsRed2 positive cells after two weeks (Fig. 5C), but in contrast to stable expression of m7.3, there was no obvious outgrowth of DsRed2 negative cells at later times, indicating less damage to cells.
Furthermore, the reporter DsRed2 reporter system was validated by sequence analysis, which revealed that mutation load of the unsorted cells after 14 days of stable or transient expression of AID mutants correlates with the percentage of revertant cells in the DsRed revertant system (Fig. 5D,E). Mutations in all groups were predominantly transition mutations and mainly G to A mutations (Fig. 5F,G).

Toxicity of Stable AID expression.
Both transient and stable expression of AID can effectively introduce mutations into reporter genes, but stable expression of AID appeared to adversely affect cell viability, in agreement with studies demonstrating toxicity induced by uncontrolled expression of AID [11][12][13][14][15] . To confirm the adverse effect of AID expression on cell expansion, we stably or transiently expressed M6A, AID-WT or m7.3 in 293FT cells, seeded similar number of cells and counted viable cells every 2 days for 6 days. Indeed, we found that stable Toxicity of AID was further confirmed in 3T3/TetOn-AID, a mouse cell line stably expressing a tetracycline-inducible AID expression cassette 10 . Controllable induction of AID expression was confirmed by Western blot analysis (Fig. 6C), and eGFP reporter (Fig. 6D). Continuous induction of AID expression in these cells for 6 days shows 50% growth inhibition compared to cells cultured in doxycycline-free medium. Growth retardation can be partially attributed to doxycycline itself, but this is a minor effect because parental 3T3 cells are only inhibited by 16% after exposure to the same dose of doxycycline for the same time period.

Discussion
We report a simple and reproducible protocol to generate somatic hypermutations in transgenes with reduced toxicity using the m7.3 hyperactive AID mutant. The protocol relies on transient bicistronic expression of m7.3 and a puromycin resistance gene combined with short-term puromycin selection. The devised protocol was based on the finding that the mutation rate declines quickly with concurrent decline of cell proliferation rate after stable expression of m7.3. We predicted that limiting the expression time of m7.3 could reduce its adverse effects. Indeed, transient expression of m7.3 for 3-5 days under puromycin selection was an efficient to introduce mutations in a reporter gene with almost no adverse effects on cell fitness. We confirmed that mutagenesis can be controlled by puromycin selection, mutagenesis is very efficient when AID is introduced under selection and removal of puromycin greatly reduced AID level and mutation rate. Taken together, our study reports a simple way to diversify highly transcribed genes with reduced toxicity and without the need for library construction, protein engineering or recombinant antibody production.
Protein evolution requires repetitive cycles of introducing mutations and selection 28 . The efficiency of mutagenesis is a key factor for fast and efficient protein evolution 29,30 . In vitro mutagenesis by error prone PCR or DNA shuffling generates high mutation rates, but requires repeated cloning and transfection. In vivo mutagenesis methods require minimal intervention because selection and mutagenesis are coupled in the same system. Most in vivo mutagenesis methods suffer from low mutation rates and increased off-target mutagenesis 31 . Therefore, there is a need for an in vivo mutagenesis systems which can diversify a chosen gene of interest with a high frequency.
Natural somatic hypermutation is an efficient system for antibody diversification. In vivo, about one point mutation is generated in an immunoglobulin gene per cell division 32 . This requires the enzyme activation-induced cytidine deaminase (AID) which deaminates deoxycytidine residues to uracil in actively transcribed genes [33][34][35][36] . Uracil is mutagenic when paired with guanosine in DNA, since dU mimics dT during replication, and the U:G mismatch triggers error-prone DNA repair in B cells. Thus, AID introduces somatic hypermutation (SHM) by converting dC to dU [37][38][39][40] . The power of SHM for protein evolution was first shown in a cell line that constitutively expresses AID (Ramos cells); Cumbers et al. were able to evolve endogenous IgM from unknown specificity to selectively bind streptavidin with nanomolar affinity after 19 rounds of FACS sorting 41 . Since that time, several studies reported using hypermutating cell lines like Ramos and DT40 for protein evolution 4,42,43 . Interestingly, ectopic expression of AID in non-B cells also turns on somatic hypermutation and diversifies highly transcribed transgenes 18 . Successful antibody evolution was reported after ectopic AID expression in human lung carcinoma 44 , HEK293 10,45-47 and CHO 48 cells.
A major drawback of inducing mutations by endogenous AID is that mutations are still far below that observed in vivo 20,49,50 . Ectopic overexpression of AID can partially solve this problem, but it increases off-target mutagenesis of highly expressed genes and transgenes such as AID itself 20,51 . Hypermutation rate in vivo has been estimated to be approximately 10 −3 mutations per base pair per generation; this frequency seems to be optimal as a lower frequency might generate poor diversity, whereas a higher mutation rate might cause increased cellular toxicity [52][53][54][55][56][57] .
To improve the efficiency of AID mutagenesis, Wang et al. generated human AID upmutants with improved activity, but unfortunately the increase in catalytic activity was accompanied with increased adverse effects on DNA instability and cells fitness 22 . However, the increased genomic instability associated with the AID upmutants was assayed by stable expression of AID upmutants in B cells 22 . By contrast, AID is transiently expressed for short periods in B cells during antibody affinity maturation 26,58 . Therefore, we decided to test transient expression of AID to induce mutations in transgenes with reduced cellular toxicity.
Transient transfection of AID for 3-5 days under puromycin selection was ideal to generate SHM with reduced toxicity. By contrast, transient transfection without any selection pressure produced extremely low mutation frequency. Since the puromycin resistance gene is a highly transcribed transgene, it has a chance to be targeted by AID. We used the shortest possible time for drug selection to avoid killing the cells that accumulate more mutations in the puromycin-resistance gene, which are probably the cells with better SHM activity. Puromycin was useful for pulse selection because it can kill the cells that did not express the gene in as short as 1 or 2 days using low drug concentration [59][60][61] .
An interesting application of the findings in this study is to mimic the natural cancer progression and metastasis mechanisms. There is a mounting evidence that mutagenesis by AID and other closely related deaminases play important roles in cancer evolution and metastasis [62][63][64] . It was hypothesized that transient but not constitutive mutagenesis facilitates cancer progression 12,65 . Indeed our data support this hypothesis because it seems that continuous mutagenesis might have eventual adverse effects on cancer fitness due to continuous accumulation of mutations, which might adversely affect cancer growth. This concept might be useful for researchers interested in studying cancer pathogenesis. For example, a cytidine deaminase can be expressed for a short time and cells with interesting phenotype can be selected. This could be a useful approach to allow researchers to isolate customized cancer clones based on the selection criteria they want.
In this study, we provide a simple solution for fast, controllable, low toxicity mutagenesis of highly transcribed genes by using transient hyperactive AID expression. AID gene transfection generates a library of mutants for the gene of interest inside the cells without the need for using error-prone PCR or DNA shuffling. Proteins with improved properties can be isolated, propagated and further mutagenized by additional rounds of mutagenesis by AID upmutant. Besides selection for improved protein function, mutants can also be selected for other desired properties such as solubility, expression, and post-translational modification.
Generation of Reporter cells. 293FT/mRFP1 and 293FT/DsRed2s reporter cells were generated to assess mutation rates. The mRFP1 gene was amplified from mRFP1 pRSET B plasmid which was a kind gift from Dr. R. Y. Tsien (University of California, San Diego, USA). mRFP1 was amplified by PCR with a 5′ primer 5′-ctagcggcgcgccgccatggtgagcaagggcgagg-3′ that encodes an AscI restriction site and a 3′ primer 5′-gagaggttaacttaggcgccggtggagtggcggc-3′ that encodes a HpaI site. mRFP was ligated into the backbone of pLKO_AS3w-Hygro (National RNAi Core Facility, Institute of Molecular Biology/Genomic Research Center, Academia Sinica, Taiwan) to generate pLKO_AS3w-RFP1 lentiviral vector for mammalian cell expression.
Lentiviral plasmids encoding mRFP1 and DsRed2s were co-transfected with pCMVΔR8.91 packaging plasmid and pMD.G VSV-G envelope plasmid in HEK293FT cells to produce lentiviral particles 10 . Two days later, the SCIeNTIfIC REPORTS | (2018) 8:9413 | DOI:10.1038/s41598-018-27717-2 culture medium was filtered, mixed with 8 μg/ml polybrene, and incubated with HEK293FT cells at a multiplicity of infection (MOI) of 0.1 to ensure single copy of reporter gene per cell. Stable cell lines were selected with 200 μg/ ml hygromycin for two weeks. AID gene amplification. Ramos cells (1 × 10 6 ) were lysed with 1 ml Super RNApure reagent and RNA was extracted according to the manufacturer's instructions (Genesis Biotech Inc., Taipei, Taiwan). Five micrograms of RNA were reverse transcribed using the Superscript III first-strand synthesis system (Thermo Fisher Scientific). Phusion high fidelity DNA polymerase was used to amplify the AID gene using primers 5′ ctagtgctagcgccaccatggacagcctcttgatgaac-3′ and 5′-cgtgagttaacacctccaagtcccaaa-gtacgaa-atgc-3′. AID mutant construction. PCR products were cloned by Zero Blunt PCR cloning kit (Thermo Fisher Scientific) and sequenced using standard M13 forward and reverse primers. Multiple site mutagenesis was performed as described 66 in the wild-type human AID gene to generate the AID m7.3 mutant (K10E, T82I, and E156G) with increased catalytic activity 22 . Site directed mutagenesis using QuikChange ™ Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) was used to generate the AID S38A mutant with compromised activity 67 and AID-M6A with total loss of activity 24 . Wild type AID and mutants AID genes were cloned into a lentiviral vector pAS3w.Ppuro 10 by the HpaI and NheI restriction sites (New England BioLabs, Ipswich, MA, USA). An HA tag epitope (YPYDVPDYA) was fused to the C-terminus of AID and AID mutants to allow detection by immunoblot analysis.

Transient expression of AID. Transient transfection was performed using TransIT-LT1 Transfection
Reagent (Mirus Bio) according to the manufacturer's instructions. Briefly, 293FT, 293FT/RFP1 or 293FT/DsRed2s cells were seeded in 6-well plates in DMEM containing 10% FBS without antibiotics one day before transfection, such that they were 70-90% confluent at the time of transfection. On the day of transfection, 2.5 µg of AID lentiviral vectors and 7.5 µL of TransIT-LT1 were combined in 500 µL of Opti-MEM Reduced Serum Medium (Thermo Fisher Scientific). The mixture was incubated for an additional 25 min at room temperature and then added to each well, and the cells were incubated at 37 °C in a CO 2 incubator for 24 hours. Cells were then refreshed with DMEM containing 10% FBS for one day, and then selected with 2 µg/mL puromycin for 2 or 4 days.
Immunoblotting. To measure AID protein levels, 5 × 10 6 cells were lysed in 0.5 mL RIPA buffer (1% NP-40, 150 mM NaCl, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0) for 1 h at 4 °C. Fifty μg of total protein from the clarified lysates were separated on a 12.5% reducing SDS-PAGE, transferred to nitrocellulose paper and sequentially stained with biotinylated goat anti-HA antibody (Vector Laboratories, CA, USA) or rabbit anti-tubulin α antibody (NeoMarkers, CA, USA) followed by streptavidin-HRP and goat anti-rabbit Ig-HRP, respectively (Jackson ImmunoResearch Laboratories, PA, USA). Bands were visualized by ECL detection (Thermo Fisher Scientific) and analyzed with a LAS-3000 Mini Fujifilm imaging system (FujiFilm, Tokyo, Japan). Cell proliferation assay. Two days after transient or stable transfection of HEK293FT cells with AID mutants, cells were selected in medium containing 2 µg/mL puromycin for 4 days, and refreshed with puromycin-free media for 1 day. Similar numbers of cells were seeded in 6 well plates and viable cells were counted every 2 days by trypan blue dye exclusion using a Countess Automated Cell Counter (Thermo Fisher Scientific).
Stably infected 3T3/TetOn-AID was reported previously. Cells were seeded in 6 cm plates with or without 0.5 µg/mL doxycycline, and viable cells are counted every 2 days for 6 days.
Sequencing. Mutations in the DsRed2 gene were characterized by sequencing of genomic DNA that was PCR-amplified from 10 6 cells 2 weeks after transfection. The PCR product was ligated to zero-blut vector (Thermo Fisher Scientific) and sequenced to confirm the mutations.