Proc. Nati. Acad. Sci. USA Vol. 85, pp. 1586-1589, March 1988 Genetics Targeted induced by a single acetylaminofluorene DNA adduct in mammalian cells and bacteria (shuttle vector/) MASAAKI MORIYA*, MASARU TAKESHITA*, FRANCIS JOHNSON*, KEITH PEDENt, STEPHEN WILL*, AND ARTHUR P. GROLLMAN *Department of Pharmacological Sciences, State University of New York at Stony Brook, Stony Brook, NY 11794; and tHoward Hughes Medical Institute Laboratory, Department of Molecular Biology and Genetics, Johns Hopkins University School of Medicine, Baltimore, MD 21205 Communicated by Richard B. Setlow, November 2, 1987

ABSTRACT Mutagenic specificity of 2-acetylamino- lems were obviated by initiating mutagenesis with a defined fluorene (AAF) has been established in mammalian cells and DNA adduct introduced at a specific site in the genome and several strains of bacteria by using a shuttle plasmid vector by detecting the full spectrum of mutations by oligodeoxy- containing a single N-(deoxyguanosin-8-yl)acetylaminofluo- nucleotide hybridization. rene (C8-dG-AAF) adduct. The nucleotide sequence of the Viral and plasmid vectors containing specifically located gene conferring tetracycline resistance was modified by con- DNA adducts have been described (5-8); some ofthese have servative codon replacement so as to accommodate the se- been tested for mutagenic properties in bacteria (8-10). We quence d(CCTTCGCTAC) flanked by two restriction sites, have constructed a shuttle plasmid containing a single Bsm I and Xho I. The corresponding synthetic oligodeoxynu- "bulky" adduct, N-(deoxyguanosin-8-yl)acetylaminofluo- cleotide underwent reaction with 2-(N-acetoxy-N-acetylamino)- rene (C8-dG-AAF). This modified plasmid was allowed to fluorene (AAAF), forming a single dG-AAF adduct. This replicate in mammalian cells and bacteria. In both experi- modified oligodeoxynucleotide was hybridized to its comple- mental systems, the adduct generated a mutagenic response mentary strand and ligated between the Bsm I and Xho I sites consisting primarily of transversions and single-base dele- of the vector. Plasmids containing the C8-dG-AAF adduct tions targeted to the site of chemical modification. were used to transfect simian virus 40-transformed simian kidney (COS-l) cells and to transform several AB strains of MATERIALS AND METHODS Escherichia coli. Colonies containing mutant plasmids were detected by hybridization to 32P-labeled oligodeoxynucleo- Construction of Plasmid Vector and 2-Acetylaminofluorene tides. Presence of the single DNA adduct increased the muta- (AAF)-Modified Oligodeoxynucleotide. The shuttle plasmid tion frequency by 8-fold in both COS cells and E. coli. Over pAG75 was constructed from the bacterial plasmid pKP772 80% of mutations detected in both systems were targeted and by inserting the simian virus 40 origin of replication (11), represented G-C -- CG or G-C- T-A transversions or single nucleotides 5155-119, at the EcoRI site. Plasmid pKP772 nucleotide deletions. We conclude that modification of a was derived from pBR322 by deletion of nucleotides deoxyguanosine residue with AAF preferentially induces mu- 1332-2520 (K.P., unpublished data). Oligonucleotide- tations targeted at this site when a plasmid containing a single directed mutagenesis (12) created unique Mlu I (nucleotide C8-dG-AAF adduct is introduced into mammalian cells or 945) and Xho I (nucleotide 978) sites in the tetracycline- bacteria. resistance (TcR) gene (Fig. 1). The 33-base-pair sequence between these restriction sites was replaced by a chemically synthesized oligonucleotide containing sites for SnaBI Chemical mutagenesis often involves formation of a covalent (nucleotide 956), Nhe I (nucleotide 964), and Bsm I (nucle- adduct between the and DNA (1, 2). Such lesions otide 968), creating pGM86 (Fig. 1). After digestion with activate cellular mechanisms involved in DNA repair and, Bsm I and Xho I, pGM86 was used as vector for the muta- unless repaired prior to replication, may lead to nucleotide genesis studies described in this paper. Plasmid pGM87 was substitutions, deletions, and chromosome rearrangements constructed by replacement of the short Bsm I/Xho I frag- (3). Chemical are also reported to activate cellular ment with another synthetic oligodeoxynucleotide, protooncogenes (4). d(CCTTCGCTAC) (G-10). Base substitutions introduced by Primary chemical structures have been established for these procedures conserve the original amino acid sequence; different DNA adducts (1, 2). However, such information thus pGM86 and pGM87 retain a functional gene conferring alone does not distinguish between potential mutagenic TcR. This property is lost in plasmids containing a single- species; in fact, so called "minor" adducts may be more base deletion or G -+ C at position 974, the site of important in this respect than those detected in much larger the dG-AAF adduct. quantities (1). Oligodeoxynucleotide G-10 and its complementary strand, Experiments in which cells are exposed to a given chem- d(TCGAGTAGCGAAGGCT) (C-16), were synthesized by ical mutagen involve a number of factors, including trans- solid-state methods, using conventional phosphoramidite port, metabolism, and the site of adduct formation. These chemistry (13). After removal from the solid support, oligo- processes complicate interpretation of subsequent muta- deoxynucleotides were purified as their 5'-dimethyltrityl genic events. Furthermore, many methods used for detec- derivatives, using reversed-phase HPLC on a Bio-Sil ODS- tion of mutations, although sensitive and convenient, intro- 5S column (Bio-Rad) (300 x 4.5 mm), eluting with a linear duce selective bias into the mutagenic response and/or fail gradient of 15-30% acetonitrile in 0.05 M triethylammonium to detect silent mutations. In our experiments, these prob- acetate buffer (pH 7.0). The main UV-absorbing fraction was

The publication costs of this article were defrayed in part by page charge Abbreviations: AAAF, N-acetoxy-2-(acetylamino)fluorene; AAF, payment. This article must therefore be hereby marked "advertisement" 2-acetylaminofluorene; C8-dG-AAF, N-(deoxyguanosin-8-yl)ace- in accordance with 18 U.S.C. §1734 solely to indicate this fact. tylaminofluorene; TcR, tetracycline resistance.

1586 Downloaded by guest on September 29, 2021 Genetics: Moriya et al. Proc. Natl. Acad. Sci. USA 85 (1988) 1587 from plates containing ampicillin (80 ,ug/ml), were analyzed for TcR. Mutant plasmids were detected by oligodeoxynu- cleotide hybridization. Mutagenesis in Mammalian Cells. Simian virus 40- transformed simian kidney cells (COS-1) (17) were grown at 370C under 5% C02/95% air in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Cells were seeded at 3 x 105 cells per 60-mm plate. After 24 hr, cells were transfected with ligation mixture (50 ng of DNA) using Pst (2423)- DEAE-dextran (18). Forty-eight hours after transfection, plasmid DNA was extracted from the cells, digested with proteinase K for 4 hr, and then treated sequentially with phenol, phenol/chloroform, and chloroform (19). After eth- anol precipitation, the DNA was treated with Dpn I to degrade input DNA (19) and used to transform competent E. coli DH5 cells. Transformants were screened for mutations by oligodeoxynucleotide hybridization. Asp A/a Leu G1Y Rr Vol Lou Leu A/a Phe A/a 7hr/iR G/Y Oligodeoxynucleotide Hybridization to Bacterial Colonies. Ampicillin-resistant colonies were replica-plated onto What- pAG75 944GAC GCGCTG GGC TAC GTC TTG CTG GCG TTC GCGACG CGAGGC985 man 3MM paper and incubated overnight at 370C. After pGM86 GAC GCG TTG GGC TAC GTA TTG CTA GCA TTC GCG ACT CGA GGC fixation of plasmids, filters were baked at 80'C under vac- p~~~~~~~~~p GM87 GM7IGAC GCG TTG GGC TAC GTA TTG CTA GCC rrc GCT ACT CGA GGC uum for 2 hr, hybridized at 420C or 450C for 16 hr with 32P-labeled oligodeoxynucleotide probes (12), washed three FIG. 1. Structure of pGM86 and partial sequence of the modified times at 40'C with 6 x SSC buffer (20), dried, and subjected TcR genes of pAG75 and pGM87. The nucleotide sequence and to autoradiography. Oligodeoxynucleotides used to detect corresponding amino acid sequence from residues 287-300 are mutations in these experiments were identical to C-16 except shown for pAG75 together with the nucleotide substitutions of the nucleotide at position 974 was changed to deoxyguano- pGM86 and pGM87. The bracketed sequence represents the in- serted G-10 oligonucleotide. The site of AAF modification (nucleo- sine (G-16), deoxyadenosine (A-16), or deoxythymidine (T- tide 974) and nucleotides changed are indicated. SV40, simian virus 16), or it was deleted (D-16). Colonies that failed to hybridize 40; ori, origin; Ap, ampicillin; Tet, tetracycline; bp, base pairs. to these probes were tested with L-16 and R-15, oligodeoxy- nucleotides that contain sequences complementary to the collected and then rechromatographed as the deprotected junctions between insert and vector. This procedure detects species, eluting with 5-15% acetonitrile. all mutant plasmids containing correctly inserted sequences. G-10 (0.1 mg) and N-acetoxy-2-(acetylamino)fluorene Plasmid DNA was prepared (21) from the mutants, including (AAAF) (0.5 mg) were dissolved in 1.0 ml of sodium citrate colonies that hybridize to L-16 and R-15 but fail to hybridize buffer containing 10%6 ethanol and allowed to stand for 90 to C-16, G-16, A-16, T-16, or D-16, and the DNA sequence min at 370C under nitrogen in the dark. HPLC analysis of the was determined by the dideoxynucleotide chain-termination reaction products, eluted over 10 min with a linear 0-20% method (22). gradient of acetonitrile in triethylammonium acetate buffer followed by an isocratic 20o elution, revealed a new UV- RESULTS absorbing peak, eluting at 23.5 min, which was clearly resolved from unmodified G-10, eluting at 21.6 min. The UV Transformation Efficiency and Mutation Frequencies of spectrum of this modified exhibited an AAF-Modified Plasmids. The presence of a single AAF oligodeoxynucleotide trans- maximum at 270 nm with a shoulder at 300 nm. adduct in pGM87 reduced the efficiency of bacterial absorption formation. an ratio of After phosphorylation with (y -32P]ATP and T4 polynucleo- Ligation mixtures, using insert/vector 10, were used directly in these experiments. All bacterial tide kinase, the modified product (G-10-AAF) could be transformants contained the G-10 oligodeoxynucleotide resolved from the slightly faster-migrating G-10 by electro- while, in COS cells, 6% of transformants in the control phoresis on a 20%o polyacrylamide gel. experiments lacked a single inserted sequence. In experi- G-10-AAF was analyzed by degradation with spleen phos- ments using plasmids modified with AAF, the frequency of phodiesterase (10 mM MgCl2) and alkaline phosphatase (14) this event in COS cells increased to 25%. These plasmids followed by HPLC. The mobile phase was provided by a were not analyzed further, and their numbers are not in- 0-60%o linear gradient of acetonitrile in triethylammonium cluded in the data presented. acetate buffer. The column was eluted over 30 min with a The overall mutation frequencies observed when AAF- flow rate of 0.7 ml/min and the major product, appearing at modified plasmids were used to transform AB1157, AB1886 the same position as authentic C8-dG-AAF (27.5 min), was (uvrA), and AB2463 (recA) were 6.8%, 6.6%, and 2.9%, clearly resolved from N-(deoxyguanosin-8-yl)aminofluorene respectively; the mutation frequency for the unmodified (28.7 min) and 3-(deoxyguanosin-N2-yl)acetylaminofluorene plasmid was 0.8%. In COS cells, presence of the adduct (25.5 min). Treatment of G-10-AAF with trifluoroacetic acid increased the mutation frequency from 1% to 8%. (15), followed by HPLC, produced a single product that Mutations in the Region of the AAF Adduct. Approxi- cochromatographed with guanyl-AAF. mately 400 ampicillin-resistant colonies from each experi- Mutagenesis in Escherichia coli. The unmodified and AAF- ment, selected at random, were analyzed for sequence alter- modified oligodeoxynucleotides, G-10 and G-10-AAF, were ations within the immediate vicinity of the AAF-modified phosphorylated at the 5' termini, hybridized to C-16, and deoxyguanosine residue (Table 1). In control experiments ligated between the Bsm I and Xho I sites of the pGM86 using an unmodified DNA insert, the G-C base pair located vector. Aliquots of the ligation mixture, containing 10 ng of at position 974 was unaffected. When the adduct was pre- vector and a 10-fold molar excess of the oligodeoxynucleo- sent, 85-90% of mutants isolated from E. coli and 80% of tide insert, were used to transform (16) E. coli strains AB mutants recovered from COS cells contained alterations at 1157, AB1886 (uvrA), and AB2463 (recA) obtained from B. J. this site. In E. coli, 70-80% of the changes observed at Bachmann (Yale University). Transformants, recovered position 974 were nucleotide substitutions, the remainder Downloaded by guest on September 29, 2021 1588 Genetics: Moriya et al. Proc. Nat!. Acad. Sci. USA 85 (1988) Table 1. Sequence changes observed in E. coli and COS-1 cells after transformation with AAF-modified DNA Colonies Targeted mutations* Plasmid Host analyzed GC -* C-G G-C -- T-A G-C -* A-T Deletions Othert pGM87 AB1157 3% 0 0 0 0 3 pGM87-AAF AB1157 383 10t 4 2 7 4 AB1886 380 11 7 0 4 3 AB2463 387 4 3 0 3 1 pGM87 COS-1 406 0 0 0 0 4 pGM87-AAF COS-1 323 9 10t 1 2 6 *Number of colonies with altered base at site of modification. tSpecific alterations are shown in Fig. 2. *One mutation was accompanied by an alteration at another site and is listed accordingly. being deletions. The majority of alterations in mutants Double-nucleotide substitutions involving base substitu- isolated from COS cells in each of several experiments tion mutations at and adjacent to the site of modification represented nucleotide substitutions; 10%6 were deletions. Of were found in AB1157 and COS cells. The C&G pair 3' to the nucleotide substitution mutations observed in E. coli, position 974 mutated to APT in pGM87-transformed AB1157. >90%o were transversions, G-C - C G predominating over In AB1886, using pGM87-AAF, addition of a single C-G pair G-C -+ TEA. In COS cells, 95% of these mutations were occurred at the same position. transversions; the two types occurred with similar fre- quency. DISCUSSION Background Mutations and Mutations Occurring at Sites Other than Nucleotide 974. While =1% of the colonies We have developed an experimental system that should be recovered after transformation of E. coli or transfection of widely applicable to studies of mutagenic specificity in COS cells with pGM87 contained mutations within the mammalian cells and bacteria. Due to the diversity of specified region of the TcR gene, no changes were found at adducts produced by certain mutagens, it is desirable to position 974 (Fig. 2). A similar spectrum of changes was relate the mutations produced to the structure and position found in experiments with pGM87-AAF, although the ma- of a given lesion. In conducting such experiments, it is jority of mutations in this case occurred at the site of the essential to establish the degree of homogeneity and purity adduct (Table 1). For example, in AB1157 and COS cells, of the mutagenic species so that experimental results are not C-G -* T*A and C-G -* A*T changes occurred at position 973 confounded by the presence of minor components (23). For in the presence and absence of AAF modification. An example, the frequency of untargeted mutations created by increased number of transitions at this position was ob- synthetic oligonucleotides in our experiments is apparently served in COS higher than in the naturally occurring bases of the TcR gene. cells after transfection with plasmids modi- This increased frequency of background mutations fied by AAF. In -- may bacteria, several COG G-C transversions reflect the error rate reported (24) when automated solid- and a base-pair deletion were detected at the same position. state chemical methods are used to prepare DNA. Over 80%o of the mutations detected in our study were pGM87 transversions or single-base deletions occurring exclusively at the AAF-modified base. While C-G -- T-A transitions 5' to Cost {a i the modified base arise as background mutations, the in- creased number of mutations at the same site may relate to -C-G-C- T-A -C-T- the presence of the adduct. These results should be com- AB1157 A A pared to the mutagenic spectrum described by Koffel- Schwartz et al. (25) in which a fragment ofpBR322, modified randomly with AAF, was used to transform E. coli. More pGM87-AAF than 90%o of the changes reported in this study were frame- shift mutations. T Fuchs and co-workers also reported that mutations cre- T ated by AAF adducts arose only when adducts were intro- Cos 1 T duced in both strands and host cells were irradiated with T ultraviolet light (26). In their experiments, the modified i strand contains an average of three AAF adducts and leads to a significant decrease in transformation efficiency. The -C-G-C-Tr-A-C-T- presence of several AAF adducts located on a single tem- G-5 C C I plate strand is likely to impede DNA synthesis (27); under AB1157 14 T these conditions, the unmodified complementary strand ap- ,T pears to be selectively copied (26). In our experiments, we observed an increase in mutation frequency without marked diminution of transformation efficiency. There was no ap- AB1886 { +C parent requirement for induction of SOS functions. A lesser increase of mutation frequency was recorded for the single recA strain tested (28). The nucleotide sequence in which the C8-dG-AAF adduct AB2463 f G is located may play an important role in determining the nature of mutations produced. The single adduct in our FIG. 2. Mutations occurring at positions other than the site of modified plasmid was introduced at position 974; this site is modification. Two base-pair substitutions are indicated by arrows. located in a pyrimidine-rich sequence. The effect of neigh- Downloaded by guest on September 29, 2021 Genetics: Moriya et al. Proc. NatL. Acad. Sci. USA 85 (1988) 1589 boring bases on the mutagenicity of C8-dG-AAF or other M. (1984) Proc. Natl. Acad. Sci. USA 81, 13-17. arylamine adducts may be ascertained more readily in se- 6. Stohrer, G., Osband, J. A. & Alvarado-Urbina, G. (1983) quences containing more than one deoxyguanosine residue. Nucleic Acids Res. 11, 5093-5102. 7. Johnson, D. L., Reid, T. M., Lee, M. S., King, C. M. & Mutations resulting from treatment of bacteria with carci- Romano, L. J. (1986) Biochemistry 25, 449-456. nogenic agents generally occur opposite premutational le- 8. Bhanot, E. & Ray, A. (1986) Proc. Natl. Acad. Sci. USA 83, sions (29). In our experiments, the majority of mutations 7348-7352. created by the bulky adduct C8-dG-AAF are also targeted to 9. Loechler, E. L., Green, C. L. & Essigmann, J. M. (1984) Proc. the site of the lesion. Our assay detects phenotypic changes Natl. Acad. Sci. USA 81, 6271-6275. within the TcR gene that arise from mutations outside the 10. Mitchell, N. & Stohrer, G. (1986) J. Mol. Biol. 191, 177-180. 11. Myers, R. M. & Tjian, R. (1980) Proc. Natl. Acad. Sci. USA targeted region; there was no increase in the number of such 77, 6491-6495. mutations in either COS cells or bacteria. The few untarget- 12. Inouye, S. & Inouye, M. (1987) in Synthesis and Applications ed mutations observed in the vicinity of the AAF adduct are ofDNA and RNA, ed. Narang, S. (Academic, New York), pp. concentrated at the position immediately 5' to the modified 181-206. base. 13. Matteucci, M. D. & Caruthers, M. H. (1981) J. Am. Chem. The mutagenic response to C8-dG-AAF in mammalian Soc. 103, 3185-3191. cells was similar to that observed in bacteria. This observa- 14. Yamasaki, H. P., Pulklabeck, P., Grunberger, D. & Weinstein, I. B. (1977) Cancer Res. 37, 3756-3760. tion suggests that these mutations are governed by a mis- 15. Tang, M. S. & Lieberman, M. W. (1983) 4, pairing or misreading phenomenon, possibly created by 1001-1006. conformational distortion of DNA (30, 31). Our results 16. Hanahan, D. (1983) J. Mol. Biol. 166, 557-580. predict a model in which an AAF-modified deoxyguanosine 17. Gluzman, Y. (1981) Cell 23, 175-182. residue preferentially directs incorporation of dGTP and, to 18. McCutchan, J. H. & Pagano, J. S. (1968) J. Natl. Cancer Inst. a lesser extent, dATP, into DNA. Such misincorporation of 41, 351-357. deoxynucleotides could be facilitated if the deoxyguanosine 19. Peden, K. W. C., Pipas, S. M., Pearson-White, S. & Nathans, moiety of the adduct adopted the syn conformation (32, 33). D. (1980) Science 209, 1392-1396. 20. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular AAAF reacts at the C-8 position of deoxyguanosine to Cloning: A Laboratory Manual (Cold Spring Harbor Labora- form a stable covalent adduct with DNA. This AAF adduct tory, Cold Spring Harbor, NY), pp. 458-459. could conceivably be deacetylated by cellular enzymes to 21. Ish-Horiwicz, D. & Burke, J. F. (1981) Nucleic Acids Res. 9, form N-(deoxyguanosin-8-yl)aminofluorene or depurinated 2989-2998. to produce an abasic site (34). We cannot exclude the 22. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. possibility that some of the mutational changes observed in Acad. Sci. USA 74, 5463-5467. our experiments arise from these lesions. These questions 23. Borowy-Borowski, H. & Chambers, R. W. (1987) Biochemis- can now be explored systematically by directly incorporat- try 26, 2465-2471. ing N-(deoxyguanosin-8-yl)aminofluorene adducts (7) and 24. McClain, W. H., Foss, K. & Mittelstadt, K. L. (1986) Nucleic Acids Res. 14, 6770. abasic sites (35) into oligodeoxynucleotides and testing their 25. Koffel-Schwartz, N., Verdier, J.-M., Bichara, M., Freund, mutagenic specificity by the methods described in this A.-M., Daune, M. P. & Fuchs, R. P. (1984) J. Mol. Biol. 177, paper. 33-51. 26. Koffel-Schwartz, N., Maenhaut-Michel, G. & Fuchs, R. P. We dedicate this report to the memory of Dr. Tsuneo Kada of the (1987) J. Mol. Biol. 193, 651-659. National Institute of Genetics, Japan. These studies were initiated 27. Moore, P. D., Rabkin, S. D., Osborne, A. L., King, C. M. & while one of us (A.P.G.) was working in the laboratory of Dr. Daniel Strauss, B. S. (1982) Proc. Natl. Acad. Sci. USA 79, 7166- Nathans. His advice and encouragement during the course of this 7170. investigation are gratefully acknowledged. We thank Dr. Gail Arce 28. Schmid, J. E., Dause, M. P. & Fuchs, R. P. P. (1982) Proc. for authentic standards of AAF-modified deoxyguanosine, Carol Natl. Acad. Sci. USA 79, 4133-4137. Abdelhamid and Robert Rieger for technical support, and Susan 29. Miller, J. (1982) Cell 31, 5-7. Rigby for preparing the manuscript. This work was supported in part 30. Fuchs, R. P. P., Lefevre, J. F., Pouyet, J. & Daune, M. P. by Grants CA17395 and ES04068 from the National Institutes of (1976) Biochemistry 15, 3347-3351. Health. 31. Grunberger, D. & Weinstein, I. B. (1979) in Chemical Carcin- ogens and DNA, ed. Grover, P. (CRC, Boca Raton, FL), Vol. 1. Osborne, M. R. (1984) in Chemical , ACS Mono- 2, pp. 59-94. graph, ed. Searle, C. E. (Am. Chem. Soc., Washington, DC), 32. Hingerty, B. & Broyde, S. (1982) Biochemistry 21, 3243-3252. No. 182, 485-524. 33. Rein, R., Shibata, M., Kieber-Emmons, T., & Zielinski, T. J. 2. Hemminki, K. (1983) Arch. Toxicol. 52, 249-285. (1985) in Molecular Basis ofCancer, Part A: Macromolecular 3. Wurgler, F. E. (1984) in Mutation, Cancer and Malformation, Structure, Carcinogens and Oncogenes, ed. Rein, R. (Liss, eds. Chu, E. H. Y. & Generoso, W. M. (Plenum, New York), New York), pp. 357-368. pp. 113-155. 34. Drinkwater, N. R., Miller, E. C. & Miller, J. A. (1980) 4. Vousden, K. H., Bos, J. L., Marshall, C. J. & Phillips, D. H. Biochemistry 19, 5087-5092. (1986) Proc. Natl. Acad. Sci. USA 83, 1222-1226. 35. Takeshita, M., Chang, C.-N., Johnson, F., Will, S. & Groll- 5. Green, C. L., Loechler, E. L., Fowler, K. W. & Essigmann, J. man, A. P. (1987) J. Biol. Chem. 262, 10171-10179. Downloaded by guest on September 29, 2021