Functional Recognition of the Modified Human tRNA^Lys3_UUU Anticodon Domain by HIV’s Nucleocapsid Protein and a Peptide Mimic

Abstract

The HIV-1 nucleocapsid protein, NCp7, facilitates the use of human tRNA^Lys3_UUU as the primer for reverse transcription. NCp7 also remodels the htRNA’s amino acid accepting stem and anticodon domains in preparation for their being annealed to the viral genome. To understand the possible influence of the htRNA’s unique composition of post-transcriptional modifications on NCp7 recognition of htRNA^Lys3_UUU, the protein’s binding and functional remodeling of the human anticodon stem and loop domain (hASL^Lys3) were studied. NCp7 bound the hASL^Lys3_UUU modified with 5-methoxycarbonyl methyl-2-thiouridine at position-34 (mcm⁵s²U₃₄) and 2-methylthio-N⁶-threonylcarbamoyladenosine at position-37 (ms²t⁶A₃₇) with a considerably higher affinity than the unmodified hASL^Lys3_UUU (K_d = 0.28 ± 0.03 and 2.30 ± 0.62 μM, respectively). NCp7 denatured the structure of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ more effectively than that of the unmodified hASL^Lys3_UUU. Two 15 amino acid peptides selected from phage display libraries demonstrated a high affinity (average K_d = 0.55 ± 0.10 μM) and specificity for the ASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ comparable to that of NCp7. The peptides recognized a t⁶A₃₇-modified ASL with an affinity (K_d = 0.60 ± 0.09 μM) comparable to that for hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇, indicating a preference for the t⁶A₃₇ modification. Significantly, one of the peptides was capable of relaxing the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ structure in a manner similar to that of NCp7, and therefore could be used to further study protein recognition of RNA modifications. The post-transcriptional modifications of htRNA^Lys3_UUU have been found to be important determinants of NCp7’s recognition prior to the tRNA^Lys3_UUU being annealed to the viral genome as the primer of reverse transcription.

Introduction

The lentivirus, Human Immunodeficiency Virus type 1 (HIV-1), is a highly adaptive virus that uses a complex system of proteolytic cleavages of its polyprotein precursor Pr55^Gag (Gag), to replicate and integrate its genome into host cell chromosomes. The Gag polyprotein contains two zinc finger motifs that are conserved among most retroviral proteins and is the only component of the HIV-1 virion that is required for virus particle assembly.^1–3 The life cycle of HIV-1 includes an assembly and budding step that results in the proteolytic cleavage of the Gag polyprotein by HIV-1 protease (PR) into six mature Gag proteins. Arguably one of the most vital proteins produced from this cleavage is the nuclear capsid protein 7 (NC or NCp7) (recently reviewed in reference 4).^1,2,4,5 NCp7 is a strongly basic, 55 amino acid protein containing two shortened zinc fingers (knuckles) (Figure 1a). NCp7 functions as a nucleic acid chaperone during various stages of viral progression.^4,6 Both theoretical and empirical studies indicate that approximately 1400 copies of NCp7 coat each of the two copies of HIV’s RNA viral genome in the capsid.^7–9

Figure 1.

HIV nucleocapsid protein, NCp7, human tRNA^Lys3_UUU, its anticodon stem and loop domain and its modified nucleosides. (a) NCp7 sequence of 55 amino acids and zinc finger structures. The single tryptophan is denoted, W37. (b) Sequence and secondary structure of the Primer Binding Site (PBS) and A-rich Stem and Loop 1 of HIV-1, sero(sub)type G, viral RNA. The secondary structure of the HIV-1 sero(sub)type-G, having the highest complementariness of the Loop I sequences, is drawn from that reported for the HIV-1 sero(sub)type B, the most prevalent serotype in the Americas, Europe and Oceania.¹⁷ The A-rich bulge of the viral serotype-G Loop 1 sequence (VGL; green) binds the U-rich anticodon domain of the htRNA^Lys3_UUU and is adjacent to the 18-residue primer binding site (blue). The single stranded Loop 1 sequence with significant complementarity to the anticodon domain was used alone and in a truncated construct (bolded) with the anticodon domain fragment for experiments with NCp7. (c) Sequence and secondary structure of human tRNA^Lys3 (htRNA^Lys3_UUU) with all of its known modified nucleosides: N²,N²-dimethylguanosine at position 10, m²₂G₁₀; dihydrouridine at positions 16, 20 and 48, D; pseudouridine-27, 39 and 55, ψ; 5-methoxycarbonylmethyl-2-thio uridine-34, mcm⁵s²U₃₄ (*U₃₄ in red); 2-methylthio-N⁶-threonylcarbamoyladenosine-37, ms²t⁶A₃₇ (*A₃₇ in red); N⁷-methyl-guanosine, m⁷G; 5-methylcytidine at positions 48 and 49, m⁵C; 2′-O-methylribothymidine-54 (2′-O-methyl-5-methyluridine), Tm; N¹-methyladenosine-58, m¹A. (d) Chemical structures of the hypermodified nucleosides within the anticodon stem and loop domain of htRNA^Lys3_UUU: mcm⁵s²U₃₄, and ms²t⁶A₃₇. (e) Modified anticodon stem and loop domain of htRNA^Lys3 (hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉). The modifications mcm⁵s²U₃₄ and ms²t⁶A₃₇ are in red. The ASLs were synthesized with a G₂₇–C₃₄ terminal base pair instead of the ψ₂₇–A₄₃ in order to stabilize the stem. The hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ with a 3′-terminal biotin was used for the peptide selection, whereas the hASL^Lys3-mcm⁵s²U₃₄;ms²t⁶A₃₇ was used for the characterization of the peptides. (f) The unmodified hASL^Lys3_UUU sequence. The unmodified hASL^Lys3_UUU was synthesized with and without fluorescein conjugated to the 5′-terminus (F) and dabsyl at the 3′-terminus (D). (g) The anticodon domain of htRNA^Lys1_CUU with the modification N⁶-threonylcarbamoyladenosine (t⁶A₃₇, in red), hASL^Lys1_CUU-t⁶A₃₇. The modified and unmodified dodecamer anticodon domain fragments used in experiments with NCp7 were comprised of residues 29–40 and modifications were introduced at U₃₄, A₃₇ and U₃₉).

NCp7’s chaperone activity is characterized by its ability to destabilize RNA structure, facilitating a conformational rearrangement, or remodeling, that is critical to the replication of HIV-1. NCp7 is involved in the specific use of the human host cell lysine tRNA isoacceptor 3, htRNA^Lys3_UUU, as the primer of HIV reverse transcription.^10,11 The htRNA^Lys3_UUU with the anticodon UUU is one of some 45 tRNAs, and the minor species of three lysine tRNAs in human cells. A conformational alteration of its canonical L-shaped tRNA structure is required for htRNA^Lys3_UUU to function as the primer for reverse transcription.¹² In facilitating the annealing of the primer to the HIV genomic RNA, NCp7 recognizes, binds and destabilizes the htRNA^Lys3_UUU.⁴ The Watson-Crick hydrogen bonding of the 3′-terminal nucleotides and that of the U-rich anticodon stem and loop (ASL) are destabilized and annealed to the complementary 18 nucleotides of the viral RNA’s primer binding site (PBS)^13,14 and approximately 12 nucleotides of the A-rich bulge of Loop 1,^15,16 respectively (Figure 1b). The htRNA^Lys3_UUU is packaged along with tRNA^Lys1,2,^18,19 the human lysyl-tRNA synthetase (KRS)²⁰ and the viral genome into the newly formed virions and used as the primer during a subsequent infection.⁴ NCp7’s ability to bind and manipulate both the viral genomic RNA and host cell’s htRNA^Lys3 for successful viral propagation make it a primary target for the development of antiviral therapeutics.²¹

NCp7 binds to RNAs in an ionic strength dependent manner that is characterized by a range of affinities.²² The protein has a demonstrated preference for single stranded polyGs, especially in GU (and GT) stretches,²³ and appears to melt G•U wobble base pairs, first.^6,24,25 In its recognition and binding of htRNA^Lys3_UUU, NCp7 exhibits its highest affinity for the amino acid acceptor stem.^4,26 As many as eight to ten copies of the protein are thought to bind to the stem and loop regions of the tRNA in a non-specific manner.^27–29 In addition to the two zinc fingers, NCp7 has a number of positively charged amino acids that are suggestive of how NCp7 could attach to any RNA, including the different regions and conformations of htRNA^Lys3_UUU.^4,5,30–32 Sequence analysis of RNA-binding proteins has led to the identification of specific motifs that are responsible for protein recognition of RNA structure.³³ Many of these motifs contain positively charged residues, such as arginine, that interact in a non-specific manner with the negatively charged backbone of the RNA. However, any degree of specificity of a protein’s recognition of a particular RNA, such as NCp7 for htRNA^Lys3_UUU, would be dependent on the RNA’s nucleotide sequence and resulting conformation. Post-transcriptional modifications alter the conformation and dynamics of the RNA structure, specifically the loop regions of tRNAs where most modifications are found, thereby creating conformations that are favorable for a specific protein-tRNA interaction.³⁴ The modified nucleosides of htRNA^Lys3_UUU may affect NCp7 recognition of and binding to the tRNA.

Mature htRNA^Lys3_UUU contains the naturally occurring post-transcriptional modifications 5-methoxycarbonylmethyl-2-thiouridine (mcm⁵s²U₃₄) at wobble position 34 and 2-methylthio-N⁶-threonylcarbamoyladenosine (ms²t⁶A₃₇) at position 37 within the tRNA’s anticodon stem and loop domain (hASL^Lys3_UUU; Figure 1c,d). Pseudouridine, ψ₃₉ is located at position 39, 3′ to the anticodon loop. Post-transcriptional modifications within the anticodon domain are identity elements that are recognized by the cognate aminoacyl-tRNA synthetases for some tRNAs and are important for effective aminoacylation of the tRNA for many, particularly for those tRNAs having position 34 modified uridines (glutamic acid, glutamine and lysine).^35,36 The post-transcriptional modifications of U₃₄ and A₃₇ increase the affinity of the hASL^Lys3_UUU to bind the AAA and AAG codons on the ribosomal-bound mRNA,^37,38 and their bacterial counterparts increase the efficiency of translation and maintain the translational reading frame.³⁹ Thus, the naturally occurring post-transcriptional modifications present on htRNA^Lys3_UUU may assist in the recognition of htRNA^Lys3_UUU by NCp7, and other HIV-1 proteins, though the importance of these modifications to the tRNA’s recognition by HIV proteins has not been explored. The ability of NCp7 to disrupt the structure of the htRNA^Lys3_UUU anticodon domain for subsequent annealing to the A-rich bulge of the viral Loop 1 sequence could also be facilitated by the presence of post-transcriptional modifications in the ASL. Anticodon domain modifications are characterized, generally, by their ability to maintain an open ASL loop and yet enhance base stacking for codon binding.^40,41 As a consequence, the open loop would be more susceptible to conformational change by NCp7.

We have investigated the relationship between post-transcriptional modifications and the recognition of htRNA^Lys3_UUU by NCp7. Here, we report that the anticodon domain modifications considerably enhanced NCp7 binding affinity for the hASL^Lys3_UUU. Modifications facilitated NCp7 disruption of the ASL’s ordered conformation as a prelude to the structural rearrangement that precedes its annealing to the HIV genome. In addition, we describe the selection and characterization of a 15 amino acid peptide that both mimics the NCp7 recognition of the modified hASL^Lys3_UUU, and is also capable of disrupting the RNA structure.

RESULTS

NCp7 recognition of the modified anticodon domain of htRNA^Lys3_UUU

NCp7 binds to different RNAs to varying degrees.^4,6 Studies of NCp7 binding to RNA suggest that it binds to single stranded regions of RNA with little specificity.^4,6 When interacting with the unmodified transcript of the htRNA^Lys3_UUU and stable isotope labeled tRNA expressed in Escherichia coli, NCp7 appears to bind initially to the nucleotides at the position of the tertiary interactions at the center of the L-shaped structure from which the melting of acceptor stem/T-stem helices and anticodon domain would occur.^4,24,42,43 The induced conformational change and dynamics now favor the annealing of the disrupted acceptor and T stems to the complementary PBS and the tRNA’s U-rich anticodon domain to the virus’ A-rich bulge of the Loop 1 sequence.^15,16 In order to examine in some detail the extent to which the naturally occurring modifications within the hASL^Lys3_UUU may influence NCp7 recognition of the tRNA, the protein’s binding to various constructs of the hASL^Lys3_UUU was investigated. The ASL^Lys3_UUU and its fragments (residues 32–43) were chemically synthesized with and without the complete and incomplete human modifications mcm⁵s²U₃₄, ms²t⁶A₃₇, 2-thiouridine (s²U₃₄) and ψ₃₉, and the E. coli modification 5-methylaminomethyluridine (mnm⁵U₃₄).^38,44 NCp7 (75 nM) was titrated with the unmodified ASL^Lys3_UUU (0 – 3 μM) and the change in fluorescence, or quenching, of NCp7’s single tryptophan residue (W37) was recorded (Figure 2a).⁷ NCp7’s interaction with the unmodified hASL^Lys3_UUU yielded a K_d in the low μM range (K_d = 2.3 ± 0.62 μM; Table 1) and as such established a baseline for recognition of the RNA in the absence of modifications.

Figure 2.

NCp7 association with the anticodon stem and loop domain of htRNA^Lys3_UUU. The binding of hASL^Lys3_UUU to NCp7 was monitored by observing the change in fluorescence (quenching) of the protein’s one tryptophan (W37). (a) NCp7 (75 nM) titrated with increasing concentrations (0 – 3 μM of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (●) and the unmodified hASL^Lys3_UUU (■). (b) NCp7 binding of the HIV-1, serotype-G, viral Stem and Loop 1. The quenching of the NCp7 W37 was monitored in a single experiment when titrated with a truncated construct of the A-rich HIV-1, serotype-G, viral Stem and Loop 1 (VSL-1 with VGL).

Table 1.

Affinity of NCp7, and peptides P6 and P17 for hASL^Lys

NCp7 or Peptide	ASLLys, Fragment, or HIV Stem/Loop I	Dissociation Constant Kd (μM)
NCp7	Unmodified ASLLys3UUU	2.30 ± 0.62
	ASLLys3UUU-mcm5s2U34;ms2t6A37	0.28 ± 0.03
	ASLLys1CUU-t6A37	ND
	HIV Stem/Loop I (VSL-1)	0.3
	Unmodified Fragment29–40	0.62 ± 0.12
	Fragment29–40-s2U34	0.01 ± 0.01
	Fragment29–40-ψ39	0.48 ± 0.03
	Fragment29–40-mnm5U34	0.20 ± 0.07
	Fragment29–40-t6A37	0.15 ± 0.06
	Fragment29–40-mnm5U34;ψ39	0.25 ± 0.05
	Fragment29–40-mnm5U34;t6A37	0.26 ± 0.06
	A-rich Loop 1 dodecamer	0.06 ± 0.01
	Duplex Anticodon/Loop 1	1.11 ± 0.16
P6	Unmodified ASLLys3UUU	ID
	ASLLys3UUU-mcm5s2U34;ms2t6A37	0.50 ± 0.10
	ASLLys1CUU-t6A37	0.60 ± 0.09
P17	Unmodified ASLLys3	ID
	ASLLys3UUU-mcm5s2U34;ms2t6A37	0.60 ± 0.10
	ASLVal3ACU-m6A37	0.49 ± 0.31
	ASLLys1CUU-t6A37	1.80 ± 0.20
P21	Unmodified ASLLys3UUU	0.98 ± 0.03
	ASLLys3UUU-mcm5s2U34;ms2t6A37	1.30 ± 0.04

ND = experiment not done

ID = Indeterminable, could not be calculated from the data

NCp7 was also titrated with the doubly modified hASL^Lys3_UUU having the naturally occurring modifications, mcm⁵s²U₃₄ at wobble position 34 and ms²t⁶A₃₇ at position 37, 3′-adjacent to the anticodon, hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (Figure 2a). Conditions were identical to those of the titration with the unmodified ASL^Lys3_UUU, described above. NCp7 exhibited a 10-fold increase in affinity for the modified ASL (K_d = 0.28 ± 0.03 μM; Table 1). The titration of NCp7 with the unmodified hASL^Lys3_UUU resulted in a maximum fluorescence quenching of only 68% as compared to 75% reduction achieved with the doubly modified hASL^Lys3_UUU (Figure 2a). When the htRNA^Lys3_UUU is annealed to the viral genome for the priming of reverse transcription, the U-rich ASL^Lys3_UUU anneals to the A-rich bulge of the viral stem and loop (Figure 1b) which also must be destabilized by NCp7. In order to compare NCp7’s binding of the modified and unmodified ASL^Lys3_UUU to that of the A-rich bulge of the viral RNA, the NCp7 was titrated with a 27 nucleotide, stem and loop construct of the HIV-1, serotype-G RNA (VSL-1) and the fluorescence of tryptophan 37 was monitored (Figure 2b). NCp7 bound the VSL-1 with an affinity comparable to its binding of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇, and to other stem/loop constructs of viral RNA (reviewed in reference 45), but considerably greater than that of its binding to the unmodified ASL^Lys3_UUU (Figure 2b; Table 1).

We also assessed the binding of the NCp7 to fragments of the hASL^Lys3_UUU that were composed of the single stranded, dodecamer oligoribonucleotide residues 29–40, and its single stranded complementary sequence of the viral A-rich Loop 1 (Figure 1b). The fragments were chosen to be non-self complementary, and thus could not form a stem and loop structure. Chemical constraints to the syntheses of these fragments in sufficient quantities limited us to the introduction of the incomplete modifications of s²U₃₄ and N⁶-threonylcarbamoyladenosine (t⁶A₃₇), along with ψ₃₉. In addition, the bacterial analog of mcm⁵U₃₄, mnm⁵U₃₄, was incorporated into the fragments. Because of its known preference for single stranded RNA, it is not surprising that NCp7 exhibited a higher affinity for the shorter, unstructured fragments of the anticodon and Loop 1 than for the unmodified hASL^Lys3_UUU hairpin (Table 1). In fact, NCp7 bound with high affinity to the single stranded unmodified anticodon fragment (K_d = 0.62 ± 0.12 μM) and even more so to the A-rich Loop 1 (K_d = 0.06 ± 0.01 μM) than to a double stranded RNA composed of these two fragments (K_d = 1.11 ± 0.16 μM) or the unmodified hASL^Lys3_UUU (K_d = 2.30 ± 0.62 μM).

Perhaps more interesting is the comparison of binding affinities between modified and unmodified fragments. Introduction of the modifications further enhanced the affinity of NCp7 for these fragments. In particular, the individual introductions of s²U₃₄, mnm⁵U₃₄ and t⁶A₃₇ significantly increased the protein’s affinity for the anticodon fragment (Table 1). NCp7 bound the modified dodecamer fragments of the hASL^Lys3_UUU with an affinity higher than that reported for DNA dodecamers.⁴⁶ These results for single stranded modified anticodon fragments along with those of the ASL study suggest that NCp7’s binding of htRNA^Lys3_UUU is considerably enhanced when modifications are present. The modifications may not be necessary for NCp7’s initial non-specific binding of the tRNA’s acceptor stem, but they appear to be of considerable importance for the protein’s recognition of the anticodon stem and loop domain.

Peptide recognition of the modified anticodon domain of htRNA^Lys3_UUU

Phage Display Library selection of peptides is an effective way of screening a large number of peptide-RNA interactions under biologically relevant conditions to determine the contribution of modifications to protein recognition of tRNAs, and the amino acids required.^47,48 Peptides that bound the hASL^Lys3_UUU-ms²t⁶A₃₇;mcm⁵s²U₃₄;ψ₃₉ were selected using two different phage display libraries and two different conditions for elution from microplates.⁴⁹ This triply modified hASL^Lys3_UUU was chemically synthesized with biotin at the 3′-end, and bound to streptavidin-coated, high capacity microplates. The first round of screening was conducted with the unmodified hASL^Lys3_UUU bound to streptavidin plates, followed by multiple rounds of screening with the triply modified hASL^Lys3_UUU. Phages were eluted from the plates using both alkaline and acid conditions.⁴⁹ Phage that demonstrated an affinity to bind the unmodified hASL^Lys3_UUU after four rounds of selection were catalogued and eliminated from further screening. In the subsequent step, phage that did not bind hASL^Lys3_UUU were selected using hASL^Lys3_UUU-mcm⁵s²U₃₄; ms²t⁶A₃₇;ψ_39. After four iterations of phage selection against the fully modified hASL^Lys3_UUU, 155 distinct colonies were chosen for DNA sequencing. Several sequences were found to occur repeatedly among the 155 colonies. Twenty-five peptides were chosen for further study, however five sequences either failed chemical synthesis or were insoluble due to a high degree of hydrophobicity. Thus, 20 peptides were synthesized with a conjugated fluorescein isothiocyanate (FITC) at the N-terminus for analysis of their interaction with the ASLs using fluorescence spectroscopy. Peptides 6 (P6), 15 (P15), 17 (P17) and 21 (P21) were chosen for further study as they appeared multiple times from separate colonies during the phage display selection (Table 2) suggesting that these four would bind to the hASL^Lys3_UUU-ms²t⁶A₃₇;mcm⁵s²U₃₄ with high affinity. Unfortunately, P15 lacked the solubility required of the assay and was eliminated from the study.

Table 2.

Peptides Generated from Phage Display Library Screening

Clone No.	15 and 16 Amino acid Peptides	Library/Elution	Times Appearing
1	FSVSFPSLPAPPDRS	Fuse5/basic	18
3	GRVTYYSCGVSLLFQ	Fuse5/basic	4
4	AGPVPLHSLSYYYNQ	Fuse5/basic	1
5	RAVMTVVWPVSFAGF	Fuse5/acidic	5
6	RVTHHAFLGAHRTVG	Fuse5/acidic	10
8	PAVASTSSLIIDGPF	Fuse5/acidic	2
9	PKAFQYGGRAVGGLW	Fuse5/acidic	1
10	AAHVSEHYVSGSLRP	Fuse5/acidic	1
11	ASVGPAPWAMTPPVS	Fuse5/acidic	1
12	APALWYPWRSLLPLY	Fuse5/acidic	1
13	ASLHPVPKTWFFLLS	Fuse5/acidic	1
14	WSHSRNTADVPVSML	Fuse5/acidic	1
15	HRGYCRDRWNCGEYF	F88-cys6/basic	8
17	PHRQCSAPAKSCKILP	F88-cys6/basic	8
19	TLPACHELPKHCKRRG	F88-cys6/basic	4
20	TLPACHELPKHCNEAR	F88-cys6/basic	1
21	NGPECNAYMVRCRGYH	F88-cys6/acidic	4
23	GNSNCPMLNEQCPWQD	F88-cys6/acidic	1
24	HTETCINIRNTCTTVA	F88-cys6/acidic	1
25	LKLPCKITINNCQLAG	F88-cys6/acidic	1

Though the biotin conjugated hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ was used for the selection of phage, its synthesis and purification in quantities required for characterizing peptide affinity and specificity was cost prohibitive. We and others have found that ψ₃₉ slightly improves the thermal stability of hASLs, but adds little to their chemistry and conformation, and nothing to their affinity for the cognate codon on the ribosome.^50,51 Therefore, the affinity of the fluorescein-conjugated peptides for the doubly modified hASL^Lys3-mcm⁵s²U₃₄;ms²t⁶A₃₇ was determined by monitoring the fluorescence quenching upon binding of the ASLs (Figure 3). Each of the peptides exhibited fluorescence quenching with increasing concentrations of doubly modified hASL^Lys3_UUU. Both P6 and P17 bound the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ with a high degree of affinity (K_d = 0.5 μM ± 0.1 and 0.6 ± 0.1 μM, respectively; Table 1), and comparable to that of NCp7.

Figure 3.

The association of phage display selected peptides with hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇. (a) Peptide P6 binding of hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇. The fluorescence of the FITC-conjugated peptide 6 (P6) (percent quenching) was monitored with addition of the ASL. P6 titrated with hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (■); P6 titrated with the unmodified hASL^Lys3_UUU (●); P6 titrated with E. coli ASL^Val3_UAC-m⁶A₃₇ (◆). (b) Peptide P17 binding of hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇. P17 titrated with the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (■); P17 titrated with the unmodified hASL^Lys3_UUU (●); P17 titrated with ASL^Val3_UAC-m⁶A₃₇ (◆). (c) The anticodon stem and loop of E. coli tRNA^Val3_UAC with the modification N⁶-methyladenosine at position 37 (m⁶A₃₇) (ASL^Val3_UAC-m⁶A₃₇).

The specificities of P6, P17 and P21 for the unmodified ASL^Lys3_UUU were assessed and compared. In contrast to their binding of the modified ASL^Lys3_UUU, P6 and P17 demonstrated little to no interaction with this ASL, and a dissociation constant could not be determined (Figure 3a,b; Table 1). However, P21 bound the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ and the unmodified hASL^Lys3_UUU with considerable, and almost equal, affinity (K_d = 0.98 ± 0.03 μM and K_d = 1.30 ± 0.04 μM, respectively) demonstrating that this peptide’s binding of the ASLs is relatively non-specific. To determine if the peptides were non-specifically recognizing the five base-paired stem and/or the seven-residue loop of the ASL hairpin structure, we tested their abilities to bind the E. coli tRNA^Val3_UAC. The E. coli tRNA^Val3_UAC has a different stem, and an anticodon loop nucleoside sequence similar to that of hASL^Lys3_UUU (Figure 3c). The tRNA^Val3_UAC construct had its A₃₇ modified at N⁶, but with only a methyl group, i.e. m⁶A₃₇ (ASL^Val3_UAC-m⁶A₃₇). P6 did not bind the E. coli ASL^Val3_UAC-m⁶A₃₇ indicating that peptide P6 is specifically recognizing the modifications of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇. However, the fluorescence of peptide P17 was quenched when titrated with the ASL^Val3_UAC-m⁶A₃₇ indicating recognition of an N⁶-modified A₃₇ (Figure 3a,b).

The specificity and affinity of P6 and P17 for hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ may be due to only one of the two modifications rather than both. Therefore, we determined their affinities for the singly modified anticodon stem and loop of the human tRNA^Lys species 1 (hASL^Lys1_CUU-t⁶A₃₇) (Figure 1g). Though htRNA^Lys1_CUU has an unmodified C₃₄, A₃₇ is modified to t⁶A₃₇ and naturally lacks the ms²-moiety. Otherwise the stem and loop sequences of the ASLs are the same. Both peptides exhibited significant affinities for hASL^Lys1_CUU-t⁶A₃₇ (K_d= 0.60 ± 0.09 μM and 1.8 ± 0.20 μM, respectively; Table 1). Not only was the affinity of the peptides for the hASL^Lys1_CUU-t⁶A₃₇ considerable, but the fluorescence was quenched to over 50% when the peptide was saturated with hASL^Lys1_CUU-t⁶A₃₇, as compared to just 12% with the unmodified hASL^Lys3_UUU. Apparently, the single modification t⁶A₃₇ is an important recognition element.

NCp7 and a peptide mimic alter the hASL^Lys3 conformation in a modification dependent manner

NCp7 has the ability to disrupt RNA secondary structure to enable the annealing of the primer htRNA^Lys3_UUU to the HIV genomic RNA which is energetically favored. To test NCp7’s ability to disrupt the base stacking and hydrogen bonds of the hASL^Lys3_UUU, we designed an unmodified hASL^Lys3_UUU to exhibit a fluorescence (Förster) resonance energy transfer (FRET) between the donor, 5′-conjugated fluorescein, and the acceptor, a 3′-conjugated accepting quencher, dabsyl (Figure 1f). Because fluorescence emission from the 5′-terminal fluorescein is transferred to the neighboring 3′-terminal dabsyl, increases in fluorescence would be expected should the NCp7 disrupt the structure of the hASL^Lys3_UUU and thereby increase the distance between the two termini. In fact, the protein’s ability to denature the unmodified hASL^Lys3_UUU was similar to the thermal denaturation of the same ASL (Figure 4). At a molar ratio of ~2:1 (NCp7 to hASL^Lys3_UUU) and a temperature of ~40 °C, the ASL appears to be half melted. Having established that NCp7 will disrupt the structure of the unmodified hASL^Lys3_UUU, we were interested in knowing if the natural modifications had an effect on this function and whether the peptides were capable of mimicking NCp7’s denaturation of the RNA. However, for these experiments we were compelled to use a different approach because the synthesis of the fully modified hASL^Lys3_UUU with fluorescein and dabsyl was chemically difficult and inordinately expensive.

Figure 4.

NCp7 disruption of the htRNA^Lys3_UUU anticodon stem and loop domain compared to thermal denaturation. An unmodified anticodon stem and loop domain construct (hASL^Lys3_UUU) with a 5′-conjugated fluorescein and a 3′-conjugated dabsyl was titrated with NCp7 and the increase in fluorescence monitored (●). The hASL^Lys3_UUU was half “melted” at an NCp7 to RNA ratio of ~2. In addition, an unmodified hASL^Lys3_UUU (without fluorescein or dabsyl) was thermally denatured/renatured from 20 to 95 °C and the UV absorbance at 260 nm was recorded, averaged and normalized (percent absorbance, ●). The temperature at which the sample was half-melted was ~40 °C.

Changes in nucleic acid base stacking, and indirectly hydrogen bonding, caused by an alteration in the nucleotide sequence, changes in temperature, or by binding of ligands can be monitored by circular dichroism (CD) spectrapolarimetry.^52,53 The titration of an RNA with either a protein or peptide adds ellipticity to the CD spectrum of the nucleic acid, especially in the wavelength range where backbone interactions are observed, 220–250 nm. Fortunately, both proteins and peptides have a spectral null in the wavelength range of interest for observing base stacking interactions, 250–280 nm,⁵² enabling the recording of changes in the RNA conformation even in the presence of proteins. NCp7 alone exhibited the expected spectral null in the wavelength range of 260–280 nm (Figure 5a). The unmodified hASL^Lys3_UUU (1.5 μM) was titrated with NCp7 and its CD ellipticity monitored for a change in RNA conformation (Figure 5a). The CD spectra of the unmodified hASL^Lys3_UUU when titrated with increasing concentrations of NCp7 exhibited a decrease in ellipticity, indicating that the protein binds and relaxes the ASL structure (Figure 5a), which is consistent with decreased stacking interactions, as would be expected from the results of the prior FRET experiment.

Figure 5.

NCp7 and peptide P6 remodeling of the modified and unmodified hASL^Lys3_UUU as monitored with circular dichroism. (a) The circular dichroism spectra (225 – 325 nm) of the unmodified hASL^Lys3_UUU alone and with increasing concentrations of NCp7. Increasing concentrations of NCp7 decreased the CD spectral ellipticity (255–275 nm) of the hASL^Lys3_UUU. The hASL^Lys3_UUU (1.5 μM) was titrated with NCp7 (0.0 – 3.65 μM). A spectrum of the NCp7 alone (1.50 μM; purple) demonstrates that the peptide has approximately a zero ellipticity between 255 and 275 nm. (b) The circular dichroism spectra (225 – 325 nm) of the triply modified hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ alone and with increasing concentrations of NCp7. Increasing concentrations of NCp7 (0.0 – 3.65 μM) decreased the CD spectral ellipticity (255–275 nm) of the ASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ (1.5 μM). A spectrum of the NCp7 alone (1.50 μM; purple) demonstrates a zero ellipticity between 255 and 275 nm. (c) Changes in ellipticity of unmodified ASL^Lys3_UUU and modified hASL^Lys3_UUU when titrated by NCp7. CD spectral changes were monitored at the wavelength maximum during titration of hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ (■) and the unmodified hASL^Lys3_UUU (●). (d) Changes in ellipticity of unmodified ASL^Lys3_UUU and hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ when titrated by peptide P6. CD spectral changes were monitored at the wavelength maximum during titration of hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ (■) and the unmodified hASL^Lys3_UUU (●).

The triply modified hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ was also titrated with NCp7 and the protein again appeared to disrupt the base stacking interactions and to relax the ASL^Lys3_UUU structure (Figure 5b). We compared the changes in ellipticity caused by NCp7 for unmodified ASL^Lys3_UUU and modified hASL^Lys3_UUU at the wavelength maximum and found that hASL^Lys3 _UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ required less protein to cause denaturation of the ASL structure (Figure 5c). The amount of NCp7 required to achieve the point at which half of the ASL was melted was 0.3 μM for the modified hASL^Lys3_UUU, and 0.46 μM for the unmodified hASL^Lys3_UUU (Figure 5c). Unlike the binding experiments monitored by fluorescence, the CD experiments are not indicative of the degree to which binding occurs. At concentrations of NCp7 above the ratio of 1, NCp7 to ASL, there was little change in CD ellipticity with addition of more protein. Since we did not perform a kinetic analysis, we cannot say that at any snapshot in time, there would be more NCp7 on the modified ASL versus the unmodified ASL under conditions in which each is saturated with protein. However, these data confirmed that the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ is bound and its secondary structure more effectively disrupted by NCp7 than is that of the unmodified hASL^Lys3_UUU.

The hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ and unmodified hASL^Lys3_UUU constructs were also titrated with the P6 peptide to determine if it was able to mimic the ability of NCp7 to bind and unfold the ASL structure. The spectra indicated that the peptide mimics NCp7 in its ability not only to bind hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ better than the unmodified hASL^Lys3_UUU, but also in its ability to relax the ASL^Lys3 structures (Figure 5d). Substantive differences were evident in the P6-facilitated denaturation of the triply-modified hASL^Lys3_UUU in comparison to that of the unmodified hASL^Lys3_UUU. When changes in ellipticity were compared for increasing concentrations of P6 at the wavelength maximum, it was apparent that the peptide completely disrupted the structure of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ at a concentration of 2.15 μM (Figure 5d). In contrast, the P6 denaturation of the unmodified hASL^Lys3_UUU was incomplete even at a concentration of 5.15 μM (Figure 5d). The concentration at which P6 had disrupted half the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇;ψ₃₉ was ~0.90 μM. Thus, P6 was able to mimic NCp7’s ability to melt the RNA, and the presence of modifications were important to how little of the peptide was needed to do so. It is also noteworthy that the modified hASL^Lys3_UUU was half-melted with an NCp7 concentration of 0.3 μM or by P6 at 0.9 μM, very much like their respective K_ds of 0.28 and 0.5 μM.

Discussion

The results presented here suggest that NCp7 recognition and remodeling of htRNA^Lys3_UUU in vivo may be more specific than originally revealed, since it is enhanced by the presence of post-transcriptional modifications. The recruitment of htRNA^Lys3_UUU for the priming of HIV-1 reverse transcription involves the Gag and the Gag-Pol polyprotein, and the human KRS, or probably a composite of these proteins.⁴ The NCp7 domain of Gag is an RNA chaperone of htRNA^Lys3_UUU and facilitates its annealing to the viral RNA.⁴ The anticodon domain of tRNA^Lys species is an identity determinant for the synthetase.^54,55 The synthetase is involved in the unacylated tRNA’s recruitment to the viral RNA. It is proposed that the synthetase may be recognizing an htRNA^Lys3_UUU-like element (TLE) of the viral RNA that has sequence homology with the U-rich anticodon region and thereby directs the tRNA to the complementary viral sequences for annealing.⁴ The complexity of protein components and the detailed mechanisms by which they recognize, recruit and anneal htRNA^Lys3_UUU to the viral genome have yet to be described.

NCp7 facilitates the use of the htRNA^Lys3_UUU species as the primer for reverse transcription. The protein remodels the tRNA conformation, thus facilitating its annealing to the viral genome for the initiation of reverse transcription during the next round of infection.^4,56 The HIV-1 capsids are packaged with viral RNA and concentrate htRNA^Lys3_UUU some 60-fold to a ratio of tRNA to viral genome of approximately 6:1 to 8:1.^4,57 The two zinc fingers of NCp7 provide the protein with a highly structure-specific binding of RNA, whereas a large number of basic residues produce a high degree of promiscuity in its binding to a variety of nucleic acids with varying affinities in vitro.⁴ NCp7 exhibits some sequence discrimination,^6,24,25 however, prior studies did not include the potential influence of the naturally occurring, post-transcriptional modifications that are found in the host cell tRNA^Lys3_UUU. As would be expected for a protein involved in the recognition and remodeling of the htRNA^Lys3_UUU conformation, NMR spectroscopy revealed that NCp7 bound to the base paired regions inside of the canonical L-shape of the acceptor stem of tRNA wherein lies tertiary structure hydrogen bonding and the G6•U67 and T54•A58 pairs.²⁴ With ¹⁵N-labeling, the imino protons of the base paired stems of a htRNA^Lys3_UUU product of E. coli cloning and thus with bacterial modifications were observable, but those of the anticodon loop were in too fast an exchange to be observed. In experiments with unmodified D- and T-domain fragments of the tRNA, NCp7 exhibited an affinity for the D-domain fragment comparable to the association that we observed with the anticodon domain.⁵⁸

Modifications had a considerable impact upon NCp7 recognition of ASL^Lys3_UUU. In fact, the affinity of NCp7 for the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (K_d = 0.28 ± 0.03 μM) was equivalent to the affinity reported for its binding to the native bovine tRNA^Lys3_UUU (K_d = 0.31 μM) using, as we did, the inherent fluorescence of the protein’s W37 and under conditions of comparable ionic strength.²⁸ NCp7 appears to bind nucleic acids in a salt-effected manner as would be expected of an electrostatic interaction with the phosphodiester backbone contributing significantly to the binding constant.⁵⁹ The binding of the stem loop structures of HIV’s ψ-domain contribute significantly to our knowledge of its affinities for these structures. NCp7 has a high affinity for model hairpins (20mers) of the ψ-domain stem and loops SL2 and SL3 (K_d ≈ 0.1–0.2 μM) and a lower affinity for SL4 (K_d ≈ 14 μM).^60–62 The affinities of NCp7 that we have observed for the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ was comparable (K_d= 0.28 ± 0.03 μM) and acquired under similar salt conditions. The equilibrium binding constant of the bacterial modified Fragment_29–40-mnm⁵U₃₄;t⁶A₃₇ (0.26 ± 0.06 μM) was comparable to that of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (0.28 ± 0.03 μM), and little different than that of the fragment with t⁶A₃₇ only (0.15 ± 0.06 μM) especially when one takes into consideration the errors of the binding study.

The human tRNA^Lys3_UUU has 18 nucleotides at the 3′-terminus, most of which are engaged in A-form RNA duplexes of the amino acid acceptor stem and the T-stem. These nucleotides are complementary to the viral PBS and must first be freed from their internal hydrogen bonding to participate in binding to the PBS. Though this sequence is an important distinctive property of the tRNA and critical to its priming function, the combination of modified nucleosides found in the anticodon loop of htRNA^Lys3_UUU are a chemically rich and exclusive attribute to that tRNA. The modifications s²U₃₄ and t⁶A₃₇ are found in other tRNAs.^63,64 Therefore, it is interesting to note that htRNA^Lys3, htRNA^Lys1,2, htRNA^Ile, and htRNA^Asn are among those tRNAs most frequently found within the HIV viral capsid.^18,19,65 The htRNA^Lys1,2, htRNA^Ile, and htRNA^Asn have the nucleoside modification t⁶A₃₇. We have found that NCp7 binds with high affinity to the hASL^Lys1_CUU with t⁶A₃₇, as well as to the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇, but not to the unmodified hASL^Lys3_UUU. Thus, the modifications are themselves recognition determinants for NCp7. Alternatively, the anticodon domain conformation resulting from the modifications enhances recognition. We hypothesize that this recognition mechanism provides another level of specificity to protein-nucleic acid interactions in vivo.

NCp7 disruption of the hydrogen bonding of the tRNA’s amino acid accepting stem and the stem of the T-domain is well documented and is an accepted functional mechanism of facilitating the binding of the tRNA to the viral PBS.⁴ Presumably, NCp7 is also involved in disrupting the structure of the U-rich anticodon stem and loop domain to facilitate the subsequent binding of this sequence to the A-rich bulge of the Loop 1 region of the viral RNA.^29,42 Using a FRET analysis between the 5′- and 3′-termini of the hASL^Lys3_UUU, we found that when titrated with NCp7, the hASL^Lys3_UUU structure is disrupted. This denaturation was analogous to the thermal denaturation of the ASL.

The interaction of NCp7 with the full length transcript of tRNA^Lys3 had been monitored previously by circular dichroism and a change in the protein’s spectrum was reported.⁶² However, little to no change in the tRNA’s spectrum was observed with the addition of three mole-equivalents of NCp7 to one of RNA, ~1/25 of NCp7 to nucleotides.⁶⁶ This result contrasts to that of other reports^29,42 and also to our own CD results with the ASLs. The protein’s denaturation of the hASL^Lys3_UUU was observed by monitoring CD spectra while increasing the concentration of NCp7. The presence of modifications in the hASL^Lys3_UUU facilitated NCp7’s denaturation of the hASL^Lys3_UUU. The heptadecamer ASL was half melted at a stoichiometry of approximating 2:1, NCp7 to ASL, consistent with the reported ratio of one NCp7 per 6–8 nucleotides. At first this stoichiometry appears to contrast with the one-site, non-linear curve fitting of the binding curves to the ASL. The protein does have a considerably greater affinity for the single stranded anticodon domain fragments than for the ASL hairpin. Thus, the second site binding of NCp7 to the ASLs could be weaker than that of the first and not observed under the conditions of our fluorescence studies. The 12mer fragments would have one or possibly two NCp7 bound per RNA, the heptadecamer RNA stem and loops would have two molecules of NCp7 bound and the 27mer VSL construct (Figure 2b) being considerably larger could bind more NC proteins.

A stoichiometric binding of NCp7, one per 6–8 nucleotides, would decrease the possibility of annealing the ASL to the viral RNA until the local amount of the protein is reduced, perhaps through the kinetics of on/off rates. In other systems, (e.g., trans-activating region RNA interaction, TAR-TAR, and the dimer initiation signal interaction, DIS-DIS), it appears that only a small amount of nucleocapsid protein is sufficient to produce annealing and structural rearrangement (e.g., from kissing dimer to extended duplex).^45,67,68 In fact, our CD experiments that monitored disruption of base stacking and not melting indicated that NCp7 at a ratio a of 0.2:1 (NCp7/ASL) and peptide at 0.5:1 were sufficient to dissolve half the base stacking interactions of the modified hASL^Lys3_UUU (Figure 5c,d). Though the ratio seems low, CD results average all of the interactions taking place in the experiment even when the amount of protein is below that of the ASL. Thus, we see base stacking being significantly disrupted with as little as 0.2:1 ratio of NCp7/ASL, ie. 20% of the ASL is bound by the NCp7. We are not alone in this observation of NCp7 interactions with htRNA^Lys3_UUU (personal communication, K. Musier-Forsyth).^45,67,68 We found little difference in the concentrations of NCp7 required for the half-maximal denaturation of the triply modified and unmodified hASL^Lys3_UUU, “destablization K_d” determined by CD (Figure 5c). This is in contrast to the significant difference in the binding K_d (Table 1). NCp7 affinity for the modified hASL^Lys3_UUU was comparable to that for NCp7 binding of viral RNA and significantly higher than that for the unmodified hASL^Lys3_UUU. Although there are some similarities between the mechanisms of action of DNA single stranded binding protein (SSB) and NCp7, NCp7’s faster kinetics and ability to aggregate, as well as destabilize, RNA and DNA appear to be most important for its functions.⁶⁹ Thus, destabilization of RNA occurs at sub-saturating NCp7 concentrations, whereas annealing/aggregation requires saturating levels of NC.⁷⁰

The effects of modifications on the chemistry, structure and conformational dynamics of the anticodon loop of htRNA^Lys3_UUU increased the binding affinity of NCp7 and of peptide P6, and facilitated disruption of the ASL structure. The major recurring RNA chemistries available as identity determinants for protein recognition are the nucleobase functional groups (NH₂ and CO), the ribose 2′-OH and the phosphate backbone. But, these per se do not create a unique identity. Thus, with some notable exceptions among the most studied of RNA-binding proteins,⁷¹ there is little understanding of how proteins specifically interact with their target RNA to form native complexes. Peptide-RNA interactions are more easily studied in detail, such as the thoroughly investigated Tat-TAR⁷² and Rev-RRE systems.⁷³ Previously, we had used Peptide Phage Display libraries to effectively screen and select peptides that bind specifically to modified nucleosides with high affinity and specificity.^47–49 In this study, peptides were selected for their abilities to bind specifically to the modified ASL region of the human tRNA^Lys3 (hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇). The peptides selected from this screen were analyzed in detail for their abilities to bind the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇. Two peptides were found to have affinities and specificities for ASL^Lys3-mcm⁵s²U₃₄;ms²t⁶A₃₇ that were comparable to those of NCp7. In addition, peptide P6 could also mimic NCp7’s ability to denature the RNA’s hairpin conformation, albeit with somewhat reduced efficiency. Potentially, P6 could facilitate the annealing of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ to the A-rich bulge of the HIV’s stem and Loop 1. In binding to hASL^Lys1_CUU-t⁶A₃₇, P6 and P17 appeared to favor the t⁶A₃₇ over the mcm⁵s²U₃₄ as the most important of the modifications in their recognition of the modified hASL^Lys3_UUU. Perhaps in using mcm⁵s²U₃₄ and ms²t⁶A₃₇ and the t⁶-moiety in particular as identity elements, the two peptides bind to the anticodon, the 3′-side of the loop and the 3′-side of the stem.

The short sequence (15 amino acids) of peptide P6 and its lack of structure as determined by CD effectively mimics the modification-facilitated binding and functionality of the larger and structured NCp7 (55 amino acids and two zinc fingers). The binding of P6 to the anticodon region of the modified hASL^Lys3_UUU was confirmed with preliminary data obtained from mass spectrometry (MS). MS experiments were conducted with the specific aim of observing complex formation between the peptides and modified nucleosides. Electrospray ionization-MS (ESI-MS), an analytical technique that is used to observe non-covalent complex formation between nucleic acids and proteins in solution at nM concentrations,⁷⁴ has been used to investigate the specific interactions between NCp7 and the different domains of the HIV-1 packaging signal.^45,75,76 For this reason, we employed ESI-MS as the most cost effective screening technique to observe the interaction between peptide P6 and hASL^Lys3_UUU-mcm⁵s²U₃₄; ms²t⁶A₃₇. As a control, the modified hASL^Lys3 was analyzed by ESI-MS in the absence of the peptide. Spectra indicate a −4 charge state for the RNA (Figure 6). Peptide P6 was analyzed alone and under the same conditions. Both the modified and unmodified hASL^Lys3_UUU were titrated with P6. A comparison of the resulting spectra led to the observation of a decrease in the abundance of free peptide in solution, along with an increase in the abundance of its complex with hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇. With equimolar amounts of RNA and peptide, ~50% of the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ was in a complex with P6, comparable to the results obtained when the fluorescein-conjugated peptide’s binding of the modified hASL^Lys3_UUU was monitored by fluorescence. At a ratio of 1:5, RNA to peptide, ~100% of the modified hASL^Lys3_UUU had been bound by P6 and the free ASL was no longer observed (Figure 6b). In these preliminary experiments, small amounts of the unmodified hASL^Lys3_UUU were bound by P6, and only 50% of E. coli ASL^Val3_UAC was in a complex with P6 when the peptide was in five-fold excess (data not shown). Thus, peptide P6 bound the modified hASL^Lys3_UUU preferentially.

Figure 6.

Mass spectrometry of hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ bound by P6. (a) ESI-MS of hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ (relative intensity vs. mass/charge ratio) (b) ESI-MS of hASL^Lys3-mcm⁵s²U₃₄;ms²t⁶A₃₇ bound to P6. The ratio of P6 to ASL is 5/1.

Anticodon loop modifications create an architecture optimized for the tRNA’s binding to the mRNA codon by reducing thermal stability and conformational dynamics.^40,41,77 The modifications at wobble position 34, mcm⁵s²U₃₄ and 3′-adjacent to the anticodon at position 37, ms²t⁶A₃₇, reduce thermal stability by negating intra-loop hydrogen bonds (Figure 7). They alter the nucleobase’s hydrogen acceptors and donors known to engage in non-canonical, as well as canonical base pairing. Their prominent physical volume widens the loop. However, modifications at purine-37 also increase the base stacking of the nucleosides on the 3′-side of the loop and thus increase the order within the loop.^41,77 Therefore, the modifications would contribute a physicochemical advantage, as well as their being chemically distinct identity determinants, for the NCp7 denaturation of the htRNA^Lys3_UUU. In contrast to an unmodified hASL^Lys3_UUU, two NCp7 molecules could ostensibly bind sequentially, perhaps cooperatively, to the hASL^Lys3_UUU-mcm⁵s²U₃₄;ms²t⁶A₃₇ and thus prepare the ASL for its annealing to the viral RNA (Figure 7). The introduction of the natural modifications provides the perfect model for recognition of htRNA^Lys3_UUU by viral proteins such as NCp7 when studied in vitro, and may replicate the higher degree of specificity appreciated in vivo. The abilities of peptides such as P6 to discriminate between modified and unmodified htRNA^Lys3, and for P6 also to be a functional mimic of NCp7 in its ability to disrupt ASL structure, make these peptides prime candidates for the development of HIV-1 antiviral therapeutics. The selection of peptides that can effectively function as mimics of RNA binding proteins, could potentially lead to the development of tools in drug discovery and therapeutics that would silence viral gene expression in vivo.

Figure 7.

Natural modifications provide identity determinants and a physicochemical advantage to NCp7 remodeling of htRNA^Lys3_UUU. (a) NCp7 (blue oval) interaction with the modified anticodon domain of htRNA^Lys3_UUU. The modifications of the anticodon stem and loop domain of htRNA^Lys3_UUU, mcm⁵s²U₃₄ (*U₃₄) and ms²t⁶A₃₇ (*A₃₇), provide unique identity determinants for NCp7. (b) NCp7 interaction with the unmodified anticodon domain of htRNA^Lys3_UUU. In contrast to the modified anticodon domain of htRNA^Lys3_UUU, the unmodified domain lacks the identity determinants and has intra-loop hydrogen bonds (----) between the invariant U₃₃ and A₃₇.⁴⁰

Materials and Methods

RNA sample preparation

The unmodified, heptadecamer oligoribonucleotide corresponding to the anticodon stem and loop of the human tRNA^Lys3_UUU (hASL^Lys3_UUU) and that of the Escherichia coli tRNA^Val_UAC isoacceptor 3 (ASL^Val3_UAC) were chemically synthesized by Dharmacon (Thermo Fisher, Lafayette, CO) using “ACE” chemistry.⁷⁸ The hASL^Lys3_UUU-ms²t⁶A₃₇; mcm⁵s²U₃₄;ψ₃₉ with biotin conjugated at the 3′-end, the hASL^Lys3 _UUU -mcm⁵s²U₃₄;ms²t⁶A₃₇, and the ASL^Val3_UAC with the natural modification N⁶-methyladenosine, ASL^Val3_UAC-m⁶A_37, were synthesized by Integrated DNA Technologies BVBA using phosphoramidite chemistries.⁷⁹ The t⁶A₃₇-modified ASL of the human tRNA^Lys_CUU isoacceptor 1 (htRNA^Lys1_CUU-t⁶A₃₇) was synthesized by the North Carolina State University Nucleic Acid Facility with the phosphoramidite of t⁶A₃₇ having the ribose 2′ and 5′ hydroxyls and the threonine carboxyl and hydroxyl suitably protected, and with standard ribonucleoside phosphoramidites and little change in standard synthesis protocols. The unmodified ASL^Lys3_UUU with a 5′-conjugated fluorescein and a 3′-dabsyl, a 27 nucleotide construct of the HIV A-rich bulge stem and loop, and 12 nucleotide anticodon domain fragments were synthesized by the NCSU Nucleic Acid Facility using standard 2′-O-TBDMS protected ribonucleoside phosphoramidites (ChemGenes, Wilmington, MA).⁸⁰ The newly synthesized ASLs were purified by anion exchange HPLC (Nucleogen 60–7 DEAE; 250 mm × 10 mm column), and desalted (Waters Corporation Sep-pak columns, Milford, MA).⁸¹ The purified ASLs were analyzed for their nucleoside composition by HPLC,⁸² and mass spectrometry, and the modifications observed by NMR,⁴⁰ and X-ray crystallography bound to codon on the E. coli 30S ribosomal subunit (Vendeix, A.F.P., Murphy, F.V. 4^th and Agris, P.F., unpublished data).

Isolation and purification of NCp7

The 55 amino acid NCp7 expression clone, pRD2, and NCp7 protein were the gifts of Dr. Michael Summers (University of Maryland, Baltimore County). Additionally, NCp7 protein was expressed using BL21 (DE3) E. coli cells and purified as described previously.^61,83

Phage display selection of peptides and peptide synthesis

Peptides that bound to the fully modified anticodon stem and loop domain of htRNA^Lys3 (hASL^Lys3_UUU-ms²t⁶A₃₇;mcm⁵s²U₃₄;ψ₃₉) were selected using two different phage display libraries, fuse5 and f88-cys6, gifts from Dr. George Smith (University of Missouri-Columbia). The fuse5 library is that of a completely random 15 amino acid sequence containing more than 10⁷ different phage.⁸⁴ The f88-cys6 is a 16 amino acid sequence randomized for every residue except for the locations of two cysteines and containing more than 2.7 × 10⁸ phage.⁸⁵ The 3′-biotin conjugated hASL^Lys3 _UUU -ms²t⁶A₃₇;mcm⁵s²U₃₄;ψ₃₉ was bound to 96-well, streptavidin-coated, high capacity microplates (Pierce) in TTDBA buffer (1 mg/ml BSA and 0.02% NaN₃ in 200/1 TBS/Tween, vol/vol; TBS buffer, 50 mM Tris HCl, pH 7.5, 150 mM NaCl). The plates were incubated with phage for 4 hr at 4°C, and were then washed 5 times with a TBS-tween solution in order to remove any unbound phage. To elute the bound phage from the plates with either acidic or alkaline conditions, the plate wells were rinsed with acid elution buffer (Acidic conditions) or alkaline elution buffer (Basic conditions) on a shaker with gentle agitation for 10 minutes.⁴⁹ The elution mixture containing eluted phage was transferred to a microtube containing the respective pH-neutralizing buffer.⁴⁹ The selected and isolated phage were amplified and the selection was reiterated for a total of five rounds. A screen was conducted with the unmodified hASL^Lys3_UUU bound to strepavidin plates. Phage that demonstrated an affinity to bind hASL^Lys3_UUU were catalogued and eliminated. The final round of selection was concluded with a serial dilution of the isolated phage and quantified to determine yield. Phage clones were then constructed on starved E. coli K91BluKan cells. Of the phage that demonstrated an affinity for the fully modified hASL^Lys3_UUU and not the unmodified hASL^Lys3_UUU, 155 distinct colonies were chosen, DNA was isolated and sequenced.⁴⁹ The resulting sequences were used to select the peptides to be synthesized. Some sequences were found to occur repeatedly among the 155 colonies (Table 1). Twenty-five sequences were chosen, but five sequences either failed chemical synthesis or were insoluble due to a high degree of hydrophobicity. Thus, 20 peptides were synthesized with a conjugated fluorescein isothiocyanate (FITC) at the N-terminus (Table 1).

The 20 peptides were chemically synthesized (Sigma-Aldrich) with fluorescein isothiocyanate (FITC) conjugated to the N-terminus for assaying peptide interaction with the ASLs by fluorescence spectroscopy. The final concentration was determined by the Bradford assay using a standard curve generated from bovine serum albumin, and a peptide of known concentration. Peptides of low solubility were eliminated from further consideration.

Fluorescence spectroscopy

The importance of anticodon domain modified nucleosides to NCp7’s recognition of htRNA^Lys3_UUU was characterized by monitoring the changes in the intrinsic fluorescence properties of the protein’s one tryptophan residue (position 37, Figure 1) upon the addition of RNA. For binding of the ASL’s oligonucleotide fragments, tryptophan fluorescence was monitored with a QuantaMaster^TM Model C 61 spectrofluorometer (Photon Technology International Inc.). The titrations were performed at 25 °C with increasing amounts of RNA added to a fixed concentration of 75 nM NCp7 in 400 μL buffer (50 mM HEPES, 50 mM NaCl, 0.04% PEG 8000 buffer, pH 7.5; or with equivalent results 20 mM phosphate buffer, 20 mM Na₂HPO₄ and 20 mM KH₂PO₄ of equal volume, pH 6.8). NCp7 fluorescence was measured at 340 nm emission (290 nm excitation) in a 1 cm path length cuvette. In monitoring the fluorescence of fluorescein 3′-conjugated hASL^Lys3_UUU with a 5′-dabsyl, fluorescence was measured at 518 nm emission (490 nm excitation) in a 1 cm path length cuvette. Experiments were repeated three times with fluorescence detection collected over 10 seconds after an initial 0.5 min period of equilibration. In assessing the binding of NCp7 to the heptadecamer ASLs, a microplate spectrofluorometer was used (Molecular Devices Spectramax Gemini XS with dual scanning monochrometers, Sunnyvale, CA). Under the conditions of 20 mM phosphate buffer, the tryptophan fluorescence was observed at 360 nm (285 nm excitation). Experiments were performed in triplicate and repeated at least twice. The resulting spectra were normalized and the percentage change in fluorescence was plotted against respective RNA concentrations. An equilibrium binding constant (K_d) was derived from the analysis of the curve using a one site binding, non-linear regression model (Prism, GraphPad Software, San Diego, CA). To assess and correct for inner-filter effects and background, fluorescence intensities were adjusted for dilution, buffer fluorescence, and screening effects due to the presence of RNA. The RNAs inner-filter effect was minimal, and the buffer chosen had insignificant effects on NCp7 tryptophan fluorescence. Photobleaching was determined to be only 3% of total fluorescence during a period of 10 min. Peptide interaction with the ASLs was monitored through changes in the fluorescence of the N-terminal FITC that was observed at 524 nm (486 nm excitation). The peptides (0.5 μM) were titrated with varying concentrations of ASLs in 20 mM phosphate buffer. The K_ds for each fluorescence experiment were assessed using the one site binding equation Y = [Bmax][X]/[K_d+X]. As a control a non-RNA binding peptide (P31; AGPVPLHSLSYYYNQ) was used as a baseline (data not shown). The data from the P31 was not represented on the graph because of the high rate of error that occurred with this non-specific interaction.

UV-monitored, thermal denaturations

Human ASL^Lys3_UUU was dissolved to a concentration of 2 μM (20 mM phosphate buffer). Thermal denaturations and renaturations, performed in triplicate in cells of 1 cm path length, were monitored by UV absorbance (260 nm) using a Cary 3 spectrophotometer as previously described.⁵⁰ Data points were averaged over 20 seconds and collected three times/minute with a temperature ramp of 1 °C/min from 5 – 90 °C. Data from denaturations and renaturations were treated similarly. No hysteresis was observed.

Circular dichroism (CD)

CD experiments, performed in triplicate, were conducted with a Jasco J815 spectropolarimeter and an interfaced computer.^86,87 RNA samples (1.5 μM) were prepared in 10 mM phosphate buffer, pH 6.8, and spectra collected from 225 nm to 325 nm. To maintain RNA and buffer concentrations and to minimize volume changes, the RNA samples were added to lyophillized protein or peptide samples of known quantity resulting in a range of protein concentrations from 0.05 μM to 3.65 μM and peptide concentration from 0.05 μM to 5.15 μM. The CD spectra were collected at 4 °C, at a rate of 10 nm per minute, a resolution of 1 nm, and the spectra were averaged over 6 runs. Normalization of spectra was performed by calculating molar circular dichroism Δε (cm²/mmol) using the formula Δε = (θ/32980)(C)(L)(N), where θ = raw CD amplitude (mdeg), C = concentration (mol/L), L = path length (cm) and N = number of nucleotides in RNA. Data were plotted and a one-site (NCp7) or two-site (P6), non-linear regression analysis was performed using Prism 5 (GraphPad Software, San Diego, CA).

Abbreviations

NCp7

HIV nucleocapsid protein

hASL^Lys_UUU

anticodon stem and loop of human tRNA^Lys3_UUU

mcm⁵s²U₃₄

5-methoxycarbonylmethyl-2-thiouridine at anticodon wobble position 34

ms²t⁶A₃₇

2-methylthio-N⁶-threonylcarbamoyladenosine at position 37

ψ₃₉

pseudouridine at anticodon stem position 39