For the purpose of precise singlebase editing, a number of plasmids called base editors (BE) were developedduring 2016-17. These plasmids facilitate base editing (a transition) involving conversion ofcytosine into uracil (Fig. 3), leading to replacement of cytosine/guanine (C:G)base pair by thymine/adenine (T:A) base pair. Since these base editors weremeant for alteration of cytosine only, these could be better named as cytosinebase editors (CBE) as against adenine base editors (ABE) that were developedfor A®I(G)conversion later in 2017 (I = inosine). The first-generation C®U base editors (BE1) were developed using the rat cytidinedeaminase AID/APOBEC1 connected to a disabled Cas9 (dCas9) via a 16 base XTEN linker4 (Komoret al. 2016).
AID/APOBECs (activationinduced deaminase/ apolipoprotein BmRNA editing enzyme, catalytic polypeptide-like) used in this study represent afamily of naturally occurring cytidine deaminases, which use single-strandedDNA/RNA as a substrate11 (Knisbacheret al. 2016). The members of AID/APOBEC family were combined with theCRISPR/dCas9 system to perform targeted base editing.
This combinationimproved CRISPR/Cas9-mediated gene editing at single base precision, thus greatlyenhancing its utility. The original requirements for singlebase editing included the following components: (i) a disabled Cas9 (dCas9) fused toa cytidine deaminase; (ii) a gRNA that helps dCas9 to target a specificlocus associated with a protospacer adjacent motif (PAM)sequence available ~18-20 base pairsdownstream, and (iii) a target cytosine withina window of positions 4-8. These first generation base editors (BE1) were further improved leading to the development of aseries of base editors that were described as second generation, thirdgeneration and fourth generation base editors12 (BE2, BE3, BE4)(Table 1). In each case, high-throughput DNA sequencing (HTS) was used to quantify baseediting efficiency. Digenome seq (sequencing of digested DNA) was also used forassessment of off-target effects in human cells13 (Kim, D et al.2015). Improvement of BEsusing Uracil N-Glycolase Inhibitors (UGI) The major problem withthe first generation base editors (BE1) included the formation of undesired products due to the followingtwo reasons: (i) frequent removal of uracil by cellular N-glycosylase (UNG) and(ii) possible occurrence of more than one Cs within the base editing activitywindow of 4-8 bases, permitting base editing of non-target cytosines possible.
The enzyme UNG works during Base ExcisionRepair (BER) and therefore, will identify transitional edited base pair G:U asDNA damage and will excise U in G:U base pair, which is used for the conversionof G:C into T:A base pair. Keeping this in view and in order to increase in vivo editing efficiency, secondgeneration base editors (BE2) were developed, which carried a uracil glycosylaseinhibitor (UGI) fused with dCas9, so that the enzyme UNG will not be able toexcise U from the G:U base pair. The editing efficiency of these second-generationbase editors (BE2) was three-fold that of BE1 reaching a maximum of ~20%; indelformation was very low (<0.1%) both in BE1 and BE2, since the DNA was not directly cleavedas in case of CRTISPR-mediated genome editing.
The second problem of the occurrence of more thanone Cs in the editing window was partly resolved by reducing the size ofediting window to 1 or 2 base pairs (see later). The next stage ofimprovement of base editors was achieved by converting dCas9 to a nickase throughreplacement of either amino acid aspartate (D) by alanine (A) atposition 10 (D10A; also described as Cas9n), or replacement of amino acidhistidine (H) by alanine at position 840 (H840A). Cas9n and H840A both produce nicks inopposite strands, and have been suitably utilized in single base gene editing14(Ran et al., 2013). For instance D10A mutant of Cas9 retains a domain thatgenerates a single strand DNA nick in the non-target strand instead ofcreating double strand breaks at the desired site; this would simulate mismatchrepair, so that a unmodified opposite DNA strand would mimic a DNA strandundergoing synthesis, where the strand containing the edited base is used as a template(C ®U; Fig. 4), takingU as T. Therefore, BE3 had the following three components: (i) an AID/APOBEC1 deaminase,that was fused through a 16–amino acid linker to (ii) a Streptococcus pyogenes nickase Cas9n Cas9n(D10A), whichwas first disabled for its nuclease function and was later converted into anickase (Cas9n) and (iii) a UGI that was linked to Cas9n through a 4 aminoacids linker.
The importance of UGI in base editing was demonstrated by showing thatthe UGI-deleted BE3 (BE3-?UGI ) was less competent in base editing compared to originalBE3, and produced not only lower frequency of desired C®T editing,but also produced a higher frequency of unwanted indels. A number of improvedBE3 variants were also developed (Table 2), which resulted in much moreefficient conversion of the G:U intermediate to desired A:U and A:T products4,11(Komor et al., 2016, 2017). Another problem associated with BE1 and BE2 was the occurrence of morethan one Cs within the base editing activity window, so that the cytosinedeaminase will convert even a non-targeted C into U.
This problem was overcome by thedevelopment of BE3 with SpCas9 (NGG), where even the non-NGG PAM sequence couldbe used for base editing. It was also shown that addition of anothercopy of UGI to BE3 further reduced the frequency of indels, so that BEs werelater improved by having more than one copy of UGI associated with Cas9n andcytosine deaminase. These were described as fourth generation base editors, theBE4, which were found to be more efficient (Wang et al.
, 2017). BE4 or SaBE4 were further improved by addingGam to the cassette, so that the use of BE4-Gam resulted in a further 1.5 to2.0 fold decrease in the indel frequency (Table 1).