Impact of H216 on the DNA Binding and Catalytic Activities of the HIV Restriction Factor APOBEC3G

Abstract

APOBEC3G has an important role in human defense against retroviral pathogens, including HIV-1. Its single-stranded DNA cytosine deaminase activity, located in its C-terminal domain (A3Gctd), can mutate viral cDNA and restrict infectivity. We used time-resolved nuclear magnetic resonance (NMR) spectroscopy to determine kinetic parameters of A3Gctd's deamination reactions within a 5′-CCC hot spot sequence. A3Gctd exhibited a 45-fold preference for 5′-CCC substrate over 5′-CCU substrate, which explains why A3G displays almost no processivity within a 5′-CCC motif. In addition, A3Gctd's shortest substrate sequence was found to be a pentanucleotide containing 5′-CCC flanked on both sides by a single nucleotide. A3Gctd as well as full-length A3G showed peak deamination velocities at pH 5.5. We found that H216 is responsible for this pH dependence, suggesting that protonation of H216 could play a key role in substrate binding. Protonation of H216 appeared important for HIV-1 restriction activity as well, since substitutions of H216 resulted in lower restriction in vivo.

INTRODUCTION

Human APOBEC3G (A3G) is a member of a family of Zn²⁺-dependent polynucleotide cytosine deaminases. This family was named after APOBEC1 (apolipoprotein B mRNA-editing enzyme catalytic polypeptide 1) and also includes the antibody gene diversification enzyme AID (activation-induced cytidine deaminase) (reviewed in references 1–5). A3G can restrict HIV-1 replication by packaging into assembling viral particles for delivery to target cells, where it deaminates cytosine to uracil in newly transcribed viral DNA. These cDNA uracils base pair with adenine during plus-strand synthesis and result in G-to-A hypermutation and, in turn, inactivation of the viral genome. A3G has two Zn²⁺ binding domains that span residues 1 to 196 and 197 to 384, but only the C-terminal domain is catalytically active (6–8). The N-terminal domain interacts with HIV-1 Vif, RNA, and single-stranded DNA (ssDNA) (e.g., see references 7 and 9–11). A3G predominantly deaminates the 3′ cytosine (underlined) in a 5′-CCC sequence, although the middle cytosine can also be deaminated in subsequent reactions following deamination of the 3′ cytosine (12–16). In longer ssDNA substrates with multiple 5′-CCC sites, A3G deamination exhibits a 3′→5′ spatial preference in vitro (9, 17, 18). In the present study, we use the catalytic domain of A3G (A3Gctd) to determine kinetic parameters. Our results provide kinetic constants for two independent deaminations within a 5′-CCC sequence, which explain A3G's catalytic site preference for the 3′ cytosine. We identify a strong pH dependence of the reaction speed, which implies that a histidine residue is involved in substrate binding. In addition, we identify the shortest-length ssDNA substrate for A3Gctd to be a pentanucleotide.

MATERIALS AND METHODS

Purification of A3Gctd.

The APOBEC3G C-terminal domain (A3Gctd), comprising amino acids 191 to 384, was expressed and purified as previously described (19). Briefly, the glutathione S-transferase (GST)-fused A3Gctd was expressed in Escherichia coli BL21(DE3) cells overnight at 17°C. After harvesting, the cells were resuspended in 50 mM sodium phosphate buffer (pH 7.4) and lysed by sonication. After ultracentrifugation at 25,000 × g for 10 min, the supernatant was added to glutathione (GSH)-Sepharose, which was subsequently washed. For kinetic analysis, the GST fusion protein was eluted from the Sepharose matrix with 100 mM GSH in phosphate buffer. By using filtration at 4,000 × g, the buffer was changed to a solution containing 50 mM phosphate (pH 7.4), 50 mM NaCl, 50 μM ZnCl₂, and 10% glycerol, and the protein was concentrated and frozen for storage. For pH analysis and measurements at pH 5.5, the buffer was composed of 75 mM sodium phosphate and 75 mM citrate, and the pH was adjusted accordingly. The ionic strength of different buffers was adjusted by using sodium chloride.

Kinetic analysis of A3Gctd's deaminase activity.

Kinetic measurements were conducted in the same buffer at a glycerol concentration of 0.6% and with 10% deuterium oxide. All nuclear magnetic resonance (NMR) data were acquired on a 700-MHz Bruker NMR spectrometer equipped with a 1.7-mm cryoprobe. A series of ¹H spectra of the oligonucleotide substrates including 5′-ATTCCCAATT-3′, 5′-ATTCCUCAATT-3′, 5′-ATTCUUAATT-3′, 5′-ATTCCCA-3′, 5′-TTCCCA-3′, 5′-TCCCA-3′, 5′-CCCAA-3′, and 5′-TTCCC-3′ at concentrations ranging from 100 μM to 10 mM was recorded. The spectra were baseline corrected, and the H5 proton signals of the 3′ and middle uracils from the 5′-CCU and 5′-CUU products, respectively, were integrated; these appeared at 5.60 ppm and 5.69 ppm, respectively. Two groups of signals were used as an internal standard in order to determine the product concentration during the reaction, including oligonucleotide signals at 8.18 to 7.97 ppm and a doublet of doublets at 2.57 to 2.39 ppm originating from citrate buffer. The product signal area was converted to product concentration and plotted versus time of reaction. To determine the initial speed, this plot was fitted with linear regression, and the slope of the initial linear part was recorded as initial speed. Enzyme concentrations were determined by SDS-acrylamide gel electrophoresis. A dilution series by a factor of 5 of the protein of interest was prepared, and usually, three samples were loaded onto an 18% gel. To calibrate the measurement, commercial carbonic anhydrase (CA) was weighted to 10 mg/ml. A dilution series by a factor of 1.5 of the 1-mg/ml CA solution was loaded next to the protein of interest. After electrophoresis and staining, the gel was photographed, and the image was analyzed by using ImageJ (20).

Purification of full-length A3G.

Full-length A3G-mycHis, A3G H216A-mycHis, and A3G H216R-mycHis proteins were purified from transiently transfected 293T cells as described previously (21, 22). Briefly, 293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) with 10% fetal bovine serum (FBS) (Gibco), 50 units/ml penicillin, and 50 μg/ml streptomycin (Invitrogen) at 37°C in 5% CO₂. Forty-eight hours after transfection with TransIT-LTI (Mirus Bio), cells were harvested and lysed in a solution containing 25 mM HEPES (pH 7.4), 150 mM NaCl, 1 mM MgCl₂, 1 mM ZnCl₂, 1 mM EDTA, 0.5% Triton X-100, and 10% glycerol. The soluble cell lysates were harvested by centrifugation (14,000 rpm for 10 min), and A3G-mycHis proteins were purified by affinity chromatography with Ni-nitrilotriacetic acid (NTA) agarose (Qiagen). The purified proteins were subjected to SDS-PAGE/Coomassie blue R250 staining for purity and concentration determination.

FRET-based deamination assay.

Fluorescence resonance energy transfer (FRET)-based deamination assays were performed as described previously (22–24). Purified A3G-mycHis and mutant proteins were incubated with a dually labeled single-stranded DNA (5′–6-carboxyfluorescein [FAM]–AAATATCCCAAAGAGAGAATGTGA–6-carboxytetramethylrhodamine [TAMRA]–3′) (Biosearch Technologies, Inc.) and E. coli uracil-DNA glycosylase (NEB) in a 384-well black plate (Nunc) at 37°C for 2 h. The products with abasic sites were cleaved by adding 3 μl of 4 N NaOH. The pH of the reaction solution was regulated with 3 μl of 4 N HCl and 37 μl of 2 M Tris-Cl (pH.9), and the fluorescence was read in a Synergy Mx monochromator-based multimode microplate reader (BioTek).

Single-cycle HIV-1 infectivity assay.

Single-cycle HIV-1 infectivity assays were performed as described previously (25). 293T cells were plated into a 6-well plate and transfected with 0.22 μg pCS-CG, 0.14 μg pRK5/Pack1 (Gag-Pol), 0.07 μg pRK5/Rev, and 0.07 μg pMDG (vesicular stomatitis virus G protein) along with 4.4, 5.5, 7.3, 11, or 22 ng of pcDNA3.1-A3G-mycHis plasmid or mutant A3G plasmids by using TransIT-LTI (Mirus Bio). After 48 h of incubation, virus-containing supernatants were harvested to infect target 293T cells. The transduced target 293T cells were harvested after an additional 24-h incubation and subjected to flow cytometry (BD FACSCanto II) to measure the ratio of green fluorescent protein (GFP)-expressing cells.

RESULTS

Time-resolved NMR analysis of DNA cytosine deamination catalyzed by A3Gctd.

As shown in Fig. 1a, the progress of A3Gctd-catalyzed deamination of 5′-ATTCCCAATT to 5′-ATTCCUAATT was monitored by using a series of ¹H NMR spectra. The rising product signal at 5.6 ppm is sufficiently resolved to allow quantitative analysis. By using appropriate enzyme and substrate concentrations, the speed of the reaction can be adjusted so that we can monitor the reaction course on a time scale of several hours. As shown in Fig. 1b, the speed of the catalytic reaction is constant for the first 25 data points. Thus, we can determine the initial speed of the reaction under the given conditions. This determination allows us to characterize the enzyme kinetics of A3Gctd by measurements at several different substrate concentrations. Our first attempt was performed at a physiological pH of 7.4. Although we used very high substrate concentrations of up to 10 mM, we were not able to saturate the enzyme. Instead, the initial speed of the reaction increased linearly with substrate concentration (Fig. 2a). We tested both wild-type A3Gctd and a soluble variant containing five substitutions, L234K, C243A, F310K, C321A, and C356A, called A3Gctd-2K3A (26) and found that both proteins exhibited the same linear dependence of reaction speed on substrate concentration. This finding suggested that either the reaction cannot be described by a Michaelis-Menten-type mechanism or saturation of the enzyme occurs at concentrations higher than 10 mM under the conditions used for these experiments.

Fig 1.

Time-resolved NMR analysis of A3Gctd-catalyzed ssDNA cytidine deamination. (a) An enlarged region of a ¹H spectrum of 5′-ATTCCCAATT-3′ is compared at several time points during a deamination reaction to illustrate the increase in the H5 proton signal of uridine due to formation of the product 5′-ATTCCUAATT-3′. The spectrum of a synthesized decanucleotide containing the product sequence is provided at the top, and that of the substrate is shown at the bottom. An 818 μM concentration of oligonucleotide was mixed with a 7.7 μM concentration of A3Gctd in 50 mM phosphate buffer (pH 7.4) with 10% D₂O at 25°C. After addition of enzyme and preparation of the spectrometer, a ¹H spectrum was recorded every 220 s. (b) The H5 proton signal of uridine at 5.60 ppm was integrated, calibrated, and plotted versus time. The straight line is a linear fit to the data. (c) Same as panel a, but the substrate is 5′-ATTCCUAATT-3′, and the product is 5′-ATTCUUAATT-3′. A 237 μM concentration of oligonucleotide was incubated with 10 μM A3Gctd in 75 mM citrate buffer (pH 5.5) with 10% D₂O at 25°C. A ¹H spectrum was recorded every 460 s. (d) The H5 proton signal of uridine at 5.69 ppm was integrated, calibrated, and plotted versus time. The straight line is a linear fit to the data.

Fig 2.

Speed of A3Gctd-catalyzed deaminations. (a) Speed of A3Gctd-catalyzed 5′-CCC-to-5′-CCU deamination at pH 7.4. The initial speed of the deamination was determined for different substrate concentrations for wild-type A3Gctd (circles) or the 2K3A variant (stars). (b) pH dependence of the speed of A3Gctd-catalyzed deamination. 5′-ATTCCCAATT-3′ oligonucleotides (300 μM) were incubated with wild-type A3Gctd (circles) or the 2K3A variant (stars), and the initial speed of the reaction at the indicated pH was determined. The ionic strength of the different buffers was adjusted to similar levels by addition of sodium chloride. (c) Substrate concentration dependence of A3Gctd-catalyzed 5′-CCC-to-5′-CCU deamination at pH 5.5. Initial speed of the deamination at 25°C was determined for several substrate concentrations using wild-type A3Gctd (circles) and the 2K3A variant (stars). For panels a to c, the data were collected from four independent enzyme preparations and calibrated to the same enzyme concentration of 820 nM. (d) Substrate dependence of wild-type A3Gctd-catalyzed 5′-CCU-to-5′-CUU deamination at pH 5.5. Initial speed of catalysis was determined by using wild-type A3Gctd at a 2.4 μM concentration. To compare these data with those for the 5′-CCC-to-5′-CCU reaction in panel c, the speed of the reaction was calibrated to an 820 nM enzyme concentration. For panels c and d, dashed lines show reaction curves generated by double-reciprocal (Lineweaver-Burk) analysis of the data.

In order to find better conditions for kinetic analysis, we tested the influence of pH, salt concentration, and the presence of Mg²⁺, Ca²⁺, and Zn²⁺ in the buffer solution on the reaction speed of A3Gctd deamination. Only in the case of varied pH did we observe a strong influence on the reaction speed. At low substrate concentrations, the reaction speed increased significantly with more acidic pH. As shown in Fig. 2b, wild-type A3Gctd reaches a maximum enzymatic activity at pH 5.5, where the speed of reaction is ∼30 times higher than that at pH 7.5 (Fig. 2b, circles). Catalytic activity of the 2K3A variant peaked at pH 6.0 (Fig. 2b, stars).

We therefore monitored deamination reactions of wild-type A3Gctd for several different substrate concentrations at pH 5.5. As shown in Fig. 2c, at pH 5.5, we were able to saturate the enzymes. The speed of enzymatic catalysis first increases with substrate concentration but then reaches a plateau (Fig. 2c). Accordingly, we fit the double-reciprocal data with a straight line to determine a wild-type K_m of 0.57 ± 0.09 mM and a k_cat of 5.80 ± 2.20 deamination events per minute for the 5′-ATTCCCAATT-to-5′-ATTCCUAATT reaction (Table 1). We also observed Michaelis-Menten-type behavior for A3Gctd-2K3A and obtained similar constants (K_m = 1.4 ± 0.7 mM; k_cat = 4.6 ± 2.9 min⁻¹).

Table 1.

Linear regression analysis of wild-type A3Gctd deaminase reactions^a

Parameter	Mean value for reaction ± SD
	CCC to CCU	CCU to CUU
Km (mM)	0.57 ± 0.09	3.60 ± 0.72
kcat (1/min)	5.80 ± 2.20	0.80 ± 0.29

^aWild-type A3Gctd deaminase reactions included 5′-ATTCCCAATT-to-5′-ATTCCUAATT and 5′-ATTCCUAATT-to-5′-ATTCUUAATT reactions. The constants originate from a linear fit of the function y = a × x + b to the data, with y as the reciprocal of the rate of production and x as the reciprocal of the substrate concentration, where k_cat equals 1/b and K_m equals a/b.

The product of the initial deamination reaction, 5′-ATTCCUAATT, was further deaminated to 5′-ATTCUUAATT. This second product exhibited a distinct doublet NMR signal at 5.69 ppm (Fig. 1c and d). We monitored wild-type A3Gctd's catalytic reactions with different 5′-ATTCCUAATT substrate concentrations to determine the initial speed of each reaction, which was plotted against the corresponding substrate concentration in Fig. 2d. The speed of this secondary deamination initially increases with increasing substrate concentrations but then plateaus. We determined Michaelis-Menten constants by using the linear fit of double-reciprocal data to find that this reaction is significantly slower, with a k_cat of less than 1 deamination event per minute (Table 1). In addition, a higher K_m was measured, which indicates that 5′-ATTCCUAATT binds A3Gctd with lower affinity than 5′-ATTCCCAATT. Moreover, the value of k_cat/K_m can be regarded as a substrate specificity constant (27). This value was calculated to be 10 M⁻¹ s⁻¹ for 5′-ATTCCCAATT and 0.22 M⁻¹ s⁻¹ for 5′-ATTCCUAATT. Therefore, A3Gctd exhibited a 45-fold greater preference for the 5′-CCC substrate over the 5′-CCU substrate. It should be noted that we have previously shown that A3Gctd produces undetectable levels of 5′-CUC or 5′-UCC products from the 5′-CCC substrate using an electrochemical deamination assay (28).

Determination of a minimum substrate for A3Gctd.

The observation that A3Gctd binds two different substrates with significantly different affinities and rate constants (Table 1) motivated us to assay for the minimal substrate length required for A3Gctd catalysis. As shown in Fig. 3, the speed of deamination of a decadeoxynucleotide substrate (5′-ATTCCCAATT) (10 nM s⁻¹) is similar to that of a pentadeoxynucleotide substrate (5′-TCCCA) (9 nM s⁻¹). We also monitored the deamination of a 30-deoxynucleotide substrate to find that the speed of catalysis was comparable to that of the shorter substrates (data not shown). We tested pentadeoxynucleotide substrates with a 5′-CCC hot spot shifted by 1 nucleotide in either direction and found that the speed of catalysis is strongly decreased compared to that with a centrally located 5′-CCC hot spot (5′-TCCCA, 5′-CCCAA, or 5′-TTCCCA in Fig. 3). The measurements were also done at pH 7.4, with similar results (data not shown). These data indicate that tetramer or shorter ssDNAs are not good substrates for A3Gctd, and we concluded that the minimal substrate is a pentadeoxynucleotide containing a 5′-CCC hot spot that is flanked on each side by a single deoxynucleotide.

Fig 3.

Speed of deamination catalysis as a function of substrate sequence. Initial speed was determined for the conversion of 600 μM substrate of the indicated sequence and A3Gctd-2K3A at pH 5.5. The enzyme concentration for all experiments was 820 nM, except for the 5′-CCCAA-3′ determination, for which 8.2 μM enzyme was used. The product concentration of 5′-CCCAA-3′ is calibrated to the 820 nM enzyme concentration.

Effect of H216 on the pH dependency of A3Gctd deamination.

To test whether H216 is responsible for the pH dependency of A3Gctd deamination, we mutated H216 to alanine (A3Gctd-2K3A-H216A) and to arginine (A3Gctd-2K3A-H216R) and measured reaction speeds at pH 7.3 and pH 6.0. A3Gctd-2K3A-H216A was catalytically dead at pH 7.3 (data not shown), which is consistent with previous E. coli assay results (26). In comparison, A3Gctd-2K3A-H216R showed a reaction speed similar to that of A3Gctd-2K3A at pH 7.3 (Fig. 4b). The difference was revealed at pH 6.0, as A3Gctd-2K3A's reaction speed was 8-fold higher than that at pH 7.3, while A3Gctd-2K3A-H216R showed only a modest 1.4-fold increase in the reaction speed (Fig. 4b).

Fig 4.

Role of H216 in A3Gctd's catalytic reaction. (a) Locations of 10 histidine residues in the three-dimensional structure of A3Gctd (Protein Data Bank [PDB] accession number 3IR2). A Zn atom (pink) is coordinated by H257, C288, and C291 (blue). H216 is shown in green, whereas all other histidines are gray. Two loops spanning residues 207 to 217 and 313 to 320 are shown in pink. (b) pH dependence of the speed of A3Gctd-2K3A-catalyzed deamination. 5′-ATTCCCAATT-3′ oligonucleotides (150 μM concentration) were mixed with a 1.5 μM concentration of the A3Gctd-2K3A (black diamonds) or A3Gctd-2K3A-H216R mutant (red circles) in sodium-phosphate buffer at pH 7.3 or pH 6.0, and the initial speed of the reaction was determined. Three independent experiments were conducted for each pH, and error bars show standard errors.

pH dependency of full-length A3G deamination.

We next used a FRET-based deamination assay to ask whether wild-type full-length A3G deamination is similarly modulated by pH. Figure 5 shows that A3G has 4-fold-higher deamination activity at pH 5.5 than at pH 7.4 (values were compared at a 33.7 nM protein concentration). Since we observed increased deamination activity at lower pH, we generated H216A (A3G-H216A) and H216R (A3G-H216R) variants of full-length A3G in order to find whether H216 is responsible for the pH dependence of deamination activity. Figure 5 shows that A3G-H216A was catalytically dead at pH 7.4 as well as at pH 5.5, but A3G-H216R had deamination activity similar to that of wild-type A3G at pH 7.4. Interestingly, A3G-H216R showed only 1.5-fold-higher deamination activity at pH 5.5 than at pH 7.4 (values were compared at a 33.7 nM protein concentration) (Fig. 5), further indicating that protonation of H216 is a key part of the mechanism by which A3G shows increased deamination activity at pH 5.5.

Fig 5.

pH dependence of the deamination activities of A3G, A3G-H216A, and A3G-H216R. (a) Purities of A3G, A3G-H216A, and A3G-H216R, as shown by SDS-PAGE and Coomassie blue R250 staining. These purified proteins were used for deaminase activity assays shown in panel b. BSA, bovine serum albumin. (b) pH dependence of deaminase activities of A3G at pH 5.5 (red diamonds), A3G-H216A at pH 5.5 (red squares), A3G-H216R at pH 5.5 (red triangles), A3G at pH 7.4 (white diamonds), A3G-H216A at pH 7.4 (white squares), and A3G-H216R at pH 7.4 (white triangles). Deaminase activities are shown by using fluorescence intensity emitted by products in relative fluorescence units (RFU).

Impact of H216 mutations on HIV-1 restriction activity.

Since H216 is important for A3G's deaminase activity, A3G-H216R and A3G-H216A were tested for their capability to restrict HIV infection. A3G-H216A showed almost no restriction capability, but A3G-H216R maintained restriction of HIV infection (Fig. 6). The restriction activity of A3G-H216R was slightly weaker than that of wild-type A3G, consistent with biochemical data presented above and prior studies demonstrating the importance of DNA deaminase activity for full levels of HIV-1 restriction (25, 29, 30).

Fig 6.

HIV-1 restriction activities of A3G-H216A and A3G-H216R. HIV-1 ΔVif infectivity was determined by measuring the ratio of GFP-expressing target 293T cells by flow cytometry. Each data point represents the average of three independent reactions. The standard deviations are shown as error bars. The expression of A3G-mycHis and mutant proteins in producer 293T cells was detected by Western blotting using an A3G monoclonal antibody (catalog number 7105; ImmunoDiagnostics). Gag (p55) was detected by using a p24 monoclonal antibody, and β-tubulin is shown as a loading control.

DISCUSSION

A3G predominantly deaminates 3′ cytosine in a 5′-CCC sequence (12–16). The middle cytosine can be deaminated, but its deamination speed is much lower than that of the primary 3′ cytosine. Previous studies failed to distinguish A3G's preference for the primary 3′ cytosine. We used time-resolved NMR, which was originally proposed by Furukawa and coworkers (12), to determine kinetic parameters for A3Gctd-catalyzed deamination. The reaction speed was maximal at pH 5.5, and it followed Michaelis-Menten-type kinetics. This pH dependency may imply involvement of histidine residues for substrate binding and/or the catalytic reaction, since the pK_a value of the conjugated acid of imidazole is around pH 6.0. A3Gctd contains 10 histidine residues, including a Zn²⁺-coordinating histidine, H257. Protonation of a Zn²⁺-coordinating histidine is likely to destabilize the protein structure, and this is probably the reason for the decline of catalytic activity observed at pH 4.5. E. coli cytidine deaminase, which has a Zn²⁺-coordinating histidine, did not show a maximum reaction speed at pH 5.5 to 6.0 (31). Therefore, H257 is not likely the cause of the observed pH dependency of A3Gctd deamination. Among the remaining histidines, alanine mutants of H216 and H345 have shown reductions in mutation frequency of 30-fold and 10-fold, respectively, whereas substitution of the others maintained >25% of wild-type catalytic activity (26, 32). H216 was one of the five most perturbed residues upon titration of a 21-mer single-stranded DNA containing a 5′-CCC hot spot, identified by monitoring main-chain ¹H-¹⁵N chemical shifts (33). Figure 4a shows positions of all histidine residues in a three-dimensional structure of A3Gctd. Loops spanning residues 207 to 217 and 313 to 320 are also shown in color, since they have been suggested to play important roles in DNA binding (9, 18, 26, 33–35). H216 is located in the loop at residues 207 to 217, and it is spatially close to the catalytic site (Fig. 4a). Indeed, amino acid substitutions of H216 modulated catalytic activity of A3Gctd-2K3A, as A3Gctd-2K3A-H216A was catalytically dead, whereas A3Gctd-2K3A-H216R, as well as A3Gctd-2K3A, was active at pH 7.3, and it showed only a modest increase in catalytic speed at pH 6.0 (Fig. 4b). Therefore, it is plausible that H216 interacts with substrate DNA using a mechanism which is enhanced by protonation of its imidazole ring. Importantly, wild-type full-length A3G also showed a significant increase in deamination activity at pH 5.5 (Fig. 5), indicating that the results obtained with A3Gctd reflect the DNA binding and deamination mechanism of the full-length enzyme. It should also be noted that full-length A3G binds 10-mer ssDNA as strongly as 69-mer ssDNA, with apparent K_d (dissociation constant) values of 50 nM and 52 nM, respectively (9), suggesting that shorter substrates make all essential contacts required for binding and catalysis. Furthermore, our data indicate that the observed pH dependency is independent of substrate length, as shorter substrates (decamer in NMR experiments) and longer substrates (24-mers in FRET-based experiments) behaved similarly. There were small but significant decreases in viral restriction with A3G-H216R compared to wild-type A3G, although A3G-H216R's deamination activity was very similar to wild-type A3G activity at pH 7.4 (Fig. 5 and 6). These data imply that H216 may be protonated to increase A3G's catalytic activity inside the viral capsid during reverse transcription.

The K_m value for 5′-CCC-to-5′-CCU deamination by A3Gctd is on the same order of magnitude as mononucleoside deamination by E. coli cytidine deaminase (0.12 mM), whereas A3Gctd's k_cat value is much lower than the value of 299 s⁻¹ determined for E. coli cytidine deaminase (36). A3Gctd's k_cat value is similar to that determined for human RNA-editing adenosine deaminase 2 (ADAR2) (0.88 min⁻¹) (38). A common feature between ADAR2 and A3G is that both require binding to polynucleotide substrates prior to deamination, as ADAR2's substrate is duplex RNA (39, 40) and A3G's substrate is single-stranded DNA, whereas E. coli cytidine deaminase deaminates the mononucleoside cytidine. Therefore, A3Gctd and ADAR2 may take additional time to position their substrates before catalysis can occur, which may cause their speed of deamination to be lower than that of E. coli cytidine deaminase. The k_cat value of A3Gctd is on the same order of magnitude as the reported k_cat value of wild-type full-length A3G (9.78 ± 2.4 min⁻¹) (13), while the K_m value of wild-type A3G was reported to be 0.82 ± 0.06 nM (13), which is 10⁵ times lower than that of A3Gctd. The nonspecific ssDNA binding of wild-type A3G, with a dissociation constant of 52 to 238 nM, is much stronger than that of A3Gctd, which has a dissociation constant of 130 to 400 μM (9, 10, 12, 13, 18, 26, 41). Thus, the large difference in K_m values may be partly due to the stronger binding capability of the N-terminal domain of A3G, which was not present in our NMR studies because full-length A3G aggregates at higher protein concentrations. It should be noted that the above-mentioned K_m value was determined by using an 80-mer oligonucleotide substrate containing one 5′-CCC hot spot (13; see also references 9 and 18), suggesting that longer substrates are more efficiently deaminated than shorter substrates, since deamination predominantly occurs when A3G approaches a 5′-CCC target with a 3′→5′ orientation. Therefore, it is plausible that 3′→5′ sliding enhances the likelihood of binding in a catalytically active orientation of substrate DNA, which could explain the lower K_m value determined by using a longer ssDNA substrate (13).

Although RNA- and ssDNA-dependent oligomerization of A3G has been observed, atomic force microscopy studies showed that A3G was able to bind and slide over ssDNA as a monomer (9, 18, 22, 42–46). In addition, a double mutant (F126A/W127A) that was predominantly monomeric in solution bound ssDNA and catalyzed deamination with 3′→5′ directionality similarly to wild-type A3G (17). Therefore, monomeric A3Gctd is a useful model for studying the kinetic parameters of the full-length enzyme.

The minimum substrate length for A3Gctd was found to be a pentamer containing a 5′-CCC hot spot flanked on both sides by one nucleotide, suggesting that key interactions for substrate specificity may involve these five nucleotides. This result is consistent with substrate specificity studies using nucleoside analogues, which revealed that A3G was sensitive to nucleotides at positions −3 to +1 of the primary target cytosine (14). The present study provides fundamental knowledge of A3G-catalyzed cytosine deamination and may aid in the design of ssDNA for high-resolution structure determination of an A3G-substrate complex.