Structural insights from active site variants and β-8 loop interactions in viperin-like enzymes

SUMMARY

Viperin and viperin-like enzymes (VLEs) are members of the radical SAM superfamily that perform radical-mediated dehydrations on nucleoside triphosphates to yield 3′-deoxy-3′,4′-didehydronucleoside triphosphates (ddhNTPs). Interestingly, viperin and VLEs demonstrate species-dependent substrate selectivity. Some fungal species have a second VLE and, while most viperin and VLEs contain an NΦHX₄CX₃CX₂CF motif, these secondary VLEs are catalytically hindered by a histidine to phenylalanine substitution, an NΦFX₄CX₃CX₂CF motif (NΦF). Herein, we utilize a combination of bioinformatics, enzymology, and X-ray crystallography to demonstrate that NΦF VLEs likely utilize CTP as a substrate. Based on these observations, we demonstrate that the β-8 loop in TvVip1 can be engineered with the β-8 loop from a CTP-selective viperin (Mus musculus) to “swap” substrate selectivity from UTP to CTP. These results provide insight into the determinants of substrate selectivity exhibited by VLEs and introduce a potential route for engineering viperin and VLEs to form alternative ddhNTPs.

In brief

Lachowicz et al. report viperin-like enzymes (VLEs) that have been inactivated by a His to Phe mutation and prefer different substrates. They also demonstrate that the β-8 loop is a major contributor to substrate selectivity by engineering a fungal VLE to select for CTP instead of UTP by swapping divergent β-8 loops.

Graphical Abstract

INTRODUCTION

VIPERIN (virus inhibitory protein, endoplasmic reticulum-associated, interferon-inducible; RSAD2/Cig5) is an interferon-inducible gene that was first discovered as an RNA transcript induced by human cytomegalovirus (HCMV) infection.¹ Since its discovery, human viperin has been demonstrated to exhibit antiviral properties for a wide range of viruses such as Zika virus,^2–6 West Nile virus,^2,7,8 human cytomegalovirus,⁹ dengue,² chikungunya,¹⁰ influenza A,¹¹ hepatitis C,^2,12–14 human immunodeficiency virus,¹⁵ rabies,¹⁶ and others.^17,18 These antiviral effects can be partially explained by viperin’s various published functions through interactions with both host and viral proteins involved in metabolism,^19,20 immune signaling,^21,22 lipid raft formation,¹¹ iron-sulfur cluster formation,²³ isoprenoid biosynthesis,^11,19,24 protein degradation,⁴ and macrophage polarization.²⁵ Interestingly, viperin has been demonstrated to have effects that seem independent of its antiviral activity through metabolic changes that enhance cancer progression²⁶ and its role in regulating β-oxidation.^27,28 These studies, however, have failed to provide any unifying function explaining viperin’s wide range of antiviral effects.

Viperin is a member of the radical S-adenosyl-L-methionine (SAM) superfamily and mammalian viperin was recently found to catalyze the formation of 3′-deoxy-3′,4′-didehydrocytidine triphosphate (ddhCTP) through a radical-mediated mechanism.² ddhCTP acts as a chain terminator for the RNA-dependent RNA polymerases of flaviviruses (Figure 1),² partially explaining viperin’s broad spectrum antiviral properties. Since the discovery of ddhCTP, other viperin-like enzymes (VLEs) were discovered to catalyze a wide variety of other 3′-deoxy-3′,4′-didehydronucleoside triphosphates (ddhNTPs) and possibly their unphosphorylated (ddhNs), di-(ddhNDPs), and monophosphate (ddhNMPs) derivatives.^29–31 These ddhNTPs act as chain terminators for phage polymerases,²⁹ thus demonstrating that viperin and VLEs are an ancient and conserved antiviral mechanism. Why viperin and VLEs evolved to catalyze the ddh-synthase reaction with a divergent array of nucleoside substrates is currently unknown.

Figure 1.

General viral restriction mechanism employed by mammalian viperin (PDB: 6Q2P) through the production of ddhCTP

It was previously reported that viperin and VLEs share a highly conserved NΦHX₄CX₃CX₂CF motif.³⁰ However, these enzymes are highly divergent in the β-8 loop, which has been demonstrated in crystal structures of Mus musculus viperin (MmVip),³² the Trichoderma virens VLE (TvVip1),³⁰ and the Nematostells vectensis VLE (NvVip) to have direct contacts with the substrate pyrimidine ring.^30,32 Interestingly, some other species, like T. virens, have a second VLE.³⁰ Instead of the conserved NΦHX₄CX₃CX₂CF motif, these VLEs have a phenylalanine in place of the conserved histidine, an NΦFX₄CX₃CX₂CF (NΦF) motif. The conserved histidine within the NΦH motif has been demonstrated to be important but not essential for catalysis, as Rattus norvegicus viperin and MmVip with histidine to alanine mutations at this position only reduced turnover rate of ddhCTP by ~6-fold.³⁰ The mechanistic role of the conserved histidine, whether it be for catalysis or substrate binding is currently unknown, but it is postulated to promote acid-assisted dehydration of the 3′-OH during viperin catalysis.^2,30 It was previously reported that T. virens (TvVip2), which contains the NΦF motif, was unable to form a detectable amount of ddhNTPs,³⁰ suggesting that these NΦF VLEs may be catalytically hindered due to His → Phe variation.

Herein, we provide bioinformatic evidence that these NΦF VLEs predominantly exist in fungi and have β-8 loops similar to viperins and VLEs that are CTP-selective. We further demonstrate through enzymatic and crystallographic studies of a representative NΦF VLE that these VLEs are catalytically hindered due to the NΦF motif and prefer to catalyze ddhCTP formation. This hindrance suggests a gene duplication and “inactivation” event by species with both a NΦF and NΦH VLE, rendering the NΦF VLE vestigial in nature. By leveraging this knowledge, we demonstrate that regions on or near the β-8 loop can be engineered to modify substrate preference by engineering TvVip1 to be CTP-selective by swapping regions from a CTP-selective viperin. These data provide not only evolutionary insights into ddhNTP selection, but also solidify our understanding of the structural basis of pyrimidine binding by VLEs. These data also provide foundational knowledge for viperin engineering, which can be utilized to determine the effects of different ddhNTP-producing viperins against human viruses.

RESULTS

Bioinformatic investigations of NΦF viperin-like enzymes

To begin investigating the role of VLEs that contain the NΦF motif, we performed two local blastp searches using default parameters (NCBI) from the sequences of human viperin (Uniprot: Q8WXG1) and a representative NΦF VLE from Colletotrichum fioriniae (41% sequence identity with human viperin; Uniprot: A0A010SDJ8). The union of hits from both blast searches yielded 4907 unique sequences. Using a prosite motif we constructed for NΦF VLEs (N-[FWY]-F-X(4)-C-X(3)-C-X(2)-C-F) and NΦH viperins (N-[FWY]-H-X(4)-C-X(3)-C-X(2)-C-F),³³ we further filtered out sequences lacking either motif, yielding 2713 unique sequences.

From these unique sequences, approximately 87% contained an NΦH motif and 13% contained an NΦF motif. Interestingly, almost all NΦF VLEs identified originate in fungi and diatoms. While it was previously noted that organisms containing an NΦF VLE always have an NΦH VLE,³⁰ after in-depth bioinformatic analysis, this is not the case. Only a small minority of fungal species have both NΦF and NΦH VLEs. It is interesting to note that even though no plant VLEs have been reported to date, we identified NΦF motif-containing sequences in the plant species, Quercus suber (GenBank: POE46900.1; POE67353.1), and an NΦH motif-containing sequence in Persea americana (GenBank: KAJ8607132.1). However, these are likely to originate from contaminating fungal DNA during DNA sequencing.

To determine the diversity of VLEs with the NΦF motif and their conservation compared to other viperins and VLEs, we determined the taxonomy of each of the 2713 unique sequences using NCBIs E-Utilities tool and clustered the sequences using CD-HIT.³⁴ Specifically, we attempted to align only the regions encompassing the radical SAM domain by taking only 16 residues before and 184 residues after the NΦF or NΦH motif, and used a sequence identity cutoff of 70%.³⁴ We then aligned a representative sequence from each resulting cluster using clustalw,³⁵ and created a phylogenetic tree using iqtrees model finder,^36–38 which informed us to use the WAG substation matrix,³⁷ and maximum-likelihood function (Figure S1).³⁸

From the identified NΦF VLEs, almost all (292/347 sequences) originate from fungi and belong to either the Basidiomycota and Ascomycota phyla, which contain 106 and 186 of the NΦF VLEs, respectively. The next largest taxonomic group was the phylum Ochrophytes, which contains only 7 members. No prokaryotes contained sequences with the NΦF motif and only one archaeon was found to have an NΦF VLE; therefore, we removed prokaryotes to construct an unrooted eukaryotic phylogenetic tree (Figure 2). Notably, there are two large clusters of NΦF VLEs within the unrooted tree, suggesting that the NΦF motif evolved through two, separate evolutionary events.

Figure 2. Phylogenetic tree of eukaryotic NΦF and NΦH viperin and VLEs.

An unrooted Phylogenetic tree was constructed by combining the union of sequences between blast searches of human viperin and a representative NΦF viperin-like enzyme from Colletotrichum fioriniae, clustering the sequences using a sequence identity cut-off of 70%, filtering out sequences that did not contain the NΦF and NΦH motifs, and building the tree using maximum-likelihood. Clustered nodes containing NΦF VLEs as well as their respective branch are colored red. The major representative Kingdom/subkingdoms of major branchpoints are highlighted as well as a representative species and its respective Uniprot ID.

Since the NΦF VLE from T. virens, TvVip2, was previously reported as unable to form a detectable amount of ddhNTPs,³⁰ we performed a sequence alignment within the identified NΦF VLEs and specifically looked for differences within the β-8 loop region (Figure 3A). We previously demonstrated that TvVip1 contains an arginine on the β-8 loop and prefers to utilize UTP as a substrate.³⁰ Another viperin-like enzyme from Thielavia terrestris (TtVip) has a similar β-8 loop sequence and has been previously shown to also prefer UTP as a substrate.³¹ To our surprise, no NΦF VLE identified contained arginine at this position. Instead, this position was occupied by mostly lysine (60%) or cysteine (26%). This occupancy is similar to viperins and VLEs that prefer CTP as their substrate,^2,30 suggesting that these NΦF VLEs may selectively catalyze ddhCTP formation instead of ddhUTP.

Figure 3. β-8 loop comparison of fungal NΦF VLEs as well as competition assays of a representative NϕF VLE suggest selectivity for CTP.

(A) Sequence alignment between the β-8 loop regions of previously studied CTP and UTP-selective viperin and VLEs and representative NΦF VLEs. β-8 loop interacting residues from published crystal structures are highlighted in red. Substrate preferences for NΦH viperin and VLEs are shown in parentheses.

(B) Substrate competition assays with 10 μM of TvVip2 or TvVip2 F40H in the presence of 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, and 2 mM of deuCTP, UTP, GTP, and ATP. Data are presented as the percent of deu-5′-dA and 5′-dA within the reaction mixture after 60 min. Data are mean ± SD from three independent experiments.

(C) Turnover rate (V/E_T) of ddhCTP formation exhibited by TvVip2 and TvVip2 F40H. Turnover rates were calculated by the amount of ddhCTP catalyzed in 5 min with reactions with 10 μM (TvVip2 F40H) or 100 μM (TvVip2) enzyme mixed with 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, and 1 mM of CTP. Data are mean ± SD from three independent experiments.

The NΦF viperin-like enzyme in T. virens prefers CTP as a substrate

To determine if NΦF VLEs could selectively bind CTP and catalyze ddhCTP formation, we purified the NΦF VLE from T. virens (TvVip2), for enzymatic studies. The UV/vis spectrum of purified TvVip2 exhibited a broad shoulder around 400 nm, like other purified viperins and VLEs (Figure S2),^2,30 indicating that it bound a [4Fe-4S] cluster. To determine the substrate preference of TvVip2, we performed a substrate competition assay with deuterated CTP (deuCTP; 5-D1, ribose-3′,4′,5′,5^′′-D₄), UTP, GTP, and ATP, as previously described,³⁰ leveraging the knowledge that viperin selectively abstracts the 4′-hydrogen atom of the nucleotide substrate to initiate catalysis.² TvVip2 produced more deuterated 5′-deoxyadenosine (deu-5′-dA) than 5′-deoxyadenosine (5′-dA) over 1 h, indicating that TvVip2 prefers CTP over the other nucleotides (Figure 3B). TvVip2 catalyzes ddhCTP formation with a rate (V_max/E_total) of 0.023 ± 0.0044 min⁻¹¹ (Figure 3C), which is ~10-fold lower than the previously published rate of TvVip1 based on 5′-dA production (V_max/E_total = 0.20 min⁻¹¹).³⁰ Interestingly, this reduced turnover rate is accompanied by a large amount of abortive cleavage. This significant amount of excess 5′-dA produced by RS enzymes is likely due to perturbed substrate binding within the active site and subsequent quenching of the 5′-deoxyadenosyl radical (5′-dA·) by an adventitious hydrogen atom (Figure S3A).^39,40 Based on these data, we postulated that by creating the TvVip2 F40H variant, the ddhCTP turnover rate would increase to a rate more comparable with TvVip1. Indeed, the recombinant TvVip2 F40H enzyme preferred CTP over other nucleotides in competition assays (Figure 3B), exhibited a 6-fold increase in ddhCTP formation over TvVip2 (Figure 3C; V_max/E_total = 0.12 ± 0.013 min⁻¹¹), and eliminated the abortive cleavage of 5′-dA (Figure S3B).

Since histidine is within close contact with the 2′-OH group of the nucleotide substrate in published viperin and VLE crystal structures,^30,32 we speculated that TvVip2 may prefer 2′-deoxyCTP over CTP. When a competition assay was performed between deuCTP and 2′-deoxyCTP utilizing a physiological reducing system (E. coli Fpr/FldA), we observed excess 5′-dA production verses deu-5′-dA (Figure S3C), but a molecular ion corresponding to 2′,3′-deoxy-3′,4′-didehydroCTP was not found using previously published methods (Figure S3D).³⁰ We did observe, however, the formation of a molecular ion corresponding to deuterated ddhCTP (Figure S3D). The excess 5′-dA likely originates from abortive cleavage upon 2′-deoxyCTP binding similar to what was observed for CTP (Figure S3A). The lack of observable 2′,3′-deoxy-3′,4′-didehydroCTP formation but observable deuterated ddhCTP formation suggests that CTP is the endogenous substrate for TvVip2 rather than 2′-deoxyCTP.

Structural characterization of the NΦF viperin-like enzyme in T. virens

To elucidate the structural basis for CTP selectivity by TvVip2, we determined the crystal structure of both TvVip2 (PDB: 9DFN) and TvVip2 F40H (PDB: 9DFU) in the presence of CTP and SAM to resolutions of 1.5 Å and 2.3 Å, respectively (Table 1). Both enzymes form partial (β/α)₆-TIM barrels (Figure 4A) and contain strong electron density representing the radical SAM [4Fe-4S] cluster, SAM, and CTP (Figure S4).

Table 1.

Crystallography data collection and refinement statistics

	TvVip1 CTP-swap V2 (PDB: 9DFW)	TvVip1 w/CTP (PDB: 9DGW)	TvVip2 (PDB: 9DFN)	TvVip2 F40H (PDB: 9DFU)
Data Processing
Wavelength (Å)	0.9793	0.9201	0.9201	0.9201
Space group	P 31	P 1 21 1	P 21 21 21	P 21 21 21
Cell dimensions
α, β, γ (°)	90, 90, 120	90, 96.714, 90	90, 90, 90	90, 90, 90
a, b, c (Å)	85.131 85.131 110.471	51.569, 80.309, 105.613	36.808, 57.838, 157.692 90	37.137, 55.959, 152.81
Resolution (Å)a	61.323–1.677 (1.705–1.677)	28.59–1.72 (1.75–1.72)	32.58–1.5 (1.55–1.5)	33.40–2.3 (2.38–2.3)
Redundancya	9.8 (9.9)	3.9 (4.0)	6.7 (6.9)	5.1 (5.5)
Completeness (%)a	100.0 (100.0)	99.3 (98.8)	100.0 (100.0)	99.7 (100.0)
<I/σI>a	9.00 (1.0)	9.6 (2.7)	6.2 (1.3)	5.3 (2.0)
R merge	0.152 (2.332)	0.07 (0.445)	0.170 (1.246)	0.217 (0.799)
R measure	0.170 (2.612)	0.095 (0.587)	0.185 (1.347)	0.268 (0.976)
CC1/2a	0.997 (0.358)	0.992 (0.870)	0.992 (0.746)	0.979 (0.57)
Wilson B-factor (Å2)	21.60	23.41	14.05	23.10
Refinement
Resolution (Å)a	19.86–1.68 (1.74–1.68)	28.59–1.72 (1.781–1.72)	19.71–1.5 (1.54–1.5)	19.66–2.3 (2.48–2.3)
No. reflectionsa	101964 (3393)	89993 (8920)	54942 (3884)	14762 (2910)
Rwork/Rfree	0.1552 (0.310)/0.1879 (0.314)	0.2621 (0.3563)/0.2899 (0.3540)	0.16936 (0.304)/0.19674 (0.275)	0.2260 (0.2795)/0.2603 (0.3032)
No. atoms	7953	7261	2766	2462
Protein	7093	6770	2368	2333
Fe3+, S2	12, 12	12, 12	4, 4	4, 4
Water	710	299	360	91
B-factors (Å2)
Protein	28.70	56.38	18.55	23.10
Ligand/ion	26.56	37.53	13.89	18.48
Water	37.49	31.57	29.97	26.20
R.m.s. deviations
Bond lengths (Å)	0.015	0.012	0.011	0.013
Bond angles (°)	1.77	1.67	1.56	1.70
Ramachandran analysis
Favored (%)	97.78	97.59	98.93	97.12
Allowed (%)	2.22	2.41	1.07	2.52
Outliers (%)	0.00	0.00	0.00	0.36

Data were collected at NSLSII beamline 17-ID-2.

^aHighest resolution shell is shown in parenthesis.

Figure 4. Structural basis for CTP selectivity by TvVip2 and TvVip2 F40H.

(A) Monomer of TvVip2 with SAM, CTP, and the radical SAM [4Fe-4S] cluster.

(B) The β-8 loop of TvVip2 with CTP where K271 is shown making contacts with the C2-exocyclic oxygen and a mainchain carbonyl from M255 contacting the C4-exocyclic amine on CTP. Fo-Fc map of CTP density at 5σ is shown in green.

(C) The conserved histidine (H26) and asparagine (N24) within the TvVip1 (PDB: 7N7I) active site.

(D) phenylalanine (F40) and asparagine (N38) within the TvVip2 active site.

(E) histidine (H40) and asparagine (N38) within the TvVip2 F40H active site.

While the published TvVip1 structure contains an arginine that provides contacts with the C4-exocyclic oxygen of the uridine base, the β-8 loop of TvVip2 utilizes K271 to provide contacts with the C2-exocyclic oxygen (Figure 4B). Surprisingly, CTP binding is further supported by an interaction with the backbone carbonyl of M255 in the region adjacent to the β-8 loop, which provides contacts to the C4-exocyclic amine of CTP. Notably, these structures present a unique binding mode for CTP, as the structures of the CTP-selective viperin from Mus musculus and VLE from N. vectensis both utilize a cysteine and lysine located on the β-8 loop to interact with the pyrimidine moiety.^30,32

Within the substrate binding pocket of TvVip2, phenylalanine is positioned similarly to histidine in TvVip1 (Figure 4C) and is within proximity to both the 2′- and 3′-OH groups (Figure 4D). Also, the binding mode of the nucleotide substrate demonstrates similar bond distances with the ribose moiety and shows no major structural differences in substrate binding within the active site of TvVip1, TvVip2, and TvVip2 F40H (Figures 4C–4E and S4C–S4E).

The β-8 loop of the active T. virens viperin-like enzyme can be engineered to prefer CTP over UTP

Given the divergence of the β-8 loop and the presence of a NΦF VLE with an alternate substrate preference in T. virens, we hypothesized that the substrate selectivity of VLEs could be engineered by modifying regions along the β-8 loop—essentially reverse engineering how nature modulated substrate selectivity between TvVip1 and TvVip2. To start, we chose TvVip1 due to its high levels of expression, solubility, and stability, as well as its ability to be crystallized. We previously demonstrated that TvVip1 selectively catalyzed ddhUTP formation and contained a conserved arginine on the β-8 loop, providing contacts with the C4 exocyclic oxygen of the uridine base.³⁰ Crystal structures of MmVip and NvVip, which are CTP-selective, contain a lysine and cysteine that make direct contacts with the C2-exocyclic oxygen and C4-exocyclic amine on CTP, respectively.^30,32 We hypothesized that TvVip1 could be modified from being UTP-selective to CTP-selective by replacing the β-8 loop with that of a CTP-selective viperin or VLE, thus creating an engineered “loop swap” VLE. Therefore, we aligned the sequences of TvVip1 and the CTP-selective viperin, MmVip as a starting point. Between TvVip and MmVip, the β-8 loop-adjacent regions are similar, demonstrating a difference only within the 6–8 residues of the β-8 loop itself (Figure 5A). To create TvVip1 with the MmVip β-8 loop (TvVip CTP-swap) we swapped residues 313–320 of MmVip with 256–261 of TvVip1.

Figure 5. Substrate selectivity in TvVip1 can be modified by swapping the β-8 loop with MmVip.

(A) Regions swapped between the MmVip (magenta; PDB: 6Q2P) and TvVip1 (cyan; PDB: 7N7I) β-8 loop to create the corresponding loop swap enzymes.

(B) Competition assays with 10 μM of TvVip1, TvVip1 CTP-swap, or TvVip1 CTP-swap V2 in the presence of 50 mM HEPES pH 7.5, 150 mM KCl, 1mM SAM and 2mM of deuCTP, UTP, GTP, and ATP. Data are mean ± SD from two (TvVip1) or three independent experiments.

(C) Competition assays with 10 μM TvVip1 CTP-swap V2 with 1 mM deuCTP, 1 mM SAM, and either 1 mM, 3 mM, 5 mM, or 10 mM UTP. Data are presented as the percent of deu-5′-dA and 5′-dA within the reaction mixture after 30 min. Data are mean ± SD from two independent experiments.

Purified TvVip CTP-swap exhibited the same broad shoulder around 400 nm in the UV spectrum as other purified viperins and VLEs,^2,30 demonstrating proper binding of a [4Fe-4S] cluster (Figure S2). Competition assays were performed as previously described,³⁰ with the assay containing the deuterated nucleotide substrate (abbreviated deuNTP) corresponding to the engineered selectivity (deuCTP for TvVip CTP-swap). In competition assays, TvVip1 Wt exhibits more 5′-dA formed than deu-5′-dA (deu-5′ dA/5′-dA signal ratio of 0.06 for deuCTP) demonstrating that TvVip1 prefers a non-deuterated substrate (UTP, Figure S5) over deuCTP (Figure 5B). TvVip CTP-swap was assayed, more deu-5′dA was formed than 5′-dA (deu-5′dA/5′-dA ratio of 1.8 for deuCTP), indicating that the substrate selectivity has been modulated from UTP to CTP. While there is clear selectivity for CTP for TvVip CTP-swap, the selectivity is low compared to wildtype TvVip1 (deu-5’dA/5′-dA ratio of 15.8 for deuUTP (5-D₁, ribose-3′,4′,5′,5”-D₄); Figure S5).

Due to the difference in selectivity between wildtype TvVip1 and TvVip CTP-swap, we set out to further improve the substrate selectivity of TvVip CTP-swap. We hypothesized that proper positioning of the β-8 loop by recapitulating additional β-8 loop interactions with nearby regions would increase the selectivity for TvVip CTP-swap. We identified two amino acids in the CTP-bound MmVip crystal structure (PDB: 6Q2P) that may play a role in proper β-8 loop positioning: E255, which may have a salt bridging interaction with the β-8 loop arginine, and I254, which we propose may provide additional van der Waals contacts with CTP (Figure S6). We swapped these residues with the corresponding residues in TvVip CTP-swap to create the TvVip CTP-swap VT198IE variant (TvVip1 CTP-swap V2). TvVip1 CTP-swap V2 exhibited an elevated preference toward CTP, with a deu-5’dA/5′-dA ratio of 9.7, a ~ 5-fold increase in selectivity compared to TvVip CTP-swap (Figure 5B). Even with 10-fold excess UTP present (Figure 5C), TvVip1 CTP-swap V2 still maintains selectivity for CTP. Between the deuterated and non-deuterated forms of CTP and UTP, the kinetic isotope effect (KIE) exhibited between TvVip1 and TvVip1 CTP-swap V2 for each is marginal and would not be a significant factor that affects the conclusions drawn from competition assays (Figure S7). Additionally, TvVip1 CTP-swap V2 catalyzed coupled ddhCTP and 5′-dA formation, indicating proper binding of CTP (Figure S8).

Next, we measured steady-state kinetics and determined K_m and V_max for TvVip1 and TvVip CTP-swap V2 with UTP or CTP to identify if kinetics are responsible for the observed swap in substrate selectivity for the engineered enzyme (Table 2; Figure S9). While the K_m of TvVip1 CTP-swap V2 with CTP is within range of the reported K_m of Wt MmVip (K_m (CTP) = 6.0 μM compared to <10 μМ for MmVip with CTP),³² we expected that TvVip1 CTP-swap V2 would have lower K_m and higher Vmax for CTP compared to the wildtype enzyme. This is not the case as the K_m and V_max are comparable for both TvVip1 CTP-swap V2 (K_{m (CTP)} = 6.0 μM; V_max(CTP) = 0.082 min⁻¹) and TvVip1 Wt (K_{m (CTP)} = 2.4 μM; V_{max (CTP)} = 0.065 min⁻¹) with CTP. While it is expected that TvVip1 would demonstrate a higher V_max for its endogenous substrate, UTP, it is surprising that the K_m is higher than CTP (K_{m (UTP)} = 79 μM) and that the kinetic constants for TvVip1 CTP-swap V2 (K_{m (UTP)} = 14 μM, V_{max (UTP)} = 0.29 min⁻¹) demonstrate the same trend. These data indicate that binding and not kinetics drive selectivity in viperin and VLEs.

Table 2.

Steady state kinetic constants for TvVip1 CTP-swap V2 and TvVip1 Wt with CTP or UTP

	Km (CTP)	V/ET (CTP)	Km (UTP)	V/ET (UTP)
TvVip1 CTP-swap V2	6.0 ± 3.2 μM	0.082 ± 0.006 min−1	14 ± 8.3 μM	0.29 ± 0.027 min−1
TvVip1 Wt	2.4 ± 1.4 μM	0.065 ± 0.006 min−1	79 ± 21 μM	0.45 ± 0.030 min−1

Turnover rate is displayed as velocity/total enzyme concentration (V/E_T). Steady state kinetic measurements were determined from reactions containing 1 μm or 5 μm enzyme in the presence of 50 mM HEPES pH 7.5, 150 mM KCl, 1mM SAM, and varying amounts of nucleotide. Data are mean ± SEM.

To test if binding affinity may be responsible for the discrepancy between kinetic constants, we performed competition experiments with increasing amounts of CTP or UTP against 1 mM of the inverse perdeuterated nucleotide (deuUTP or deuCTP, respectively). By calculating the difference in deuterated 5′-dA formed in the presence of increasing amounts of competing NTP, we can calculate an IC₅₀ value for each nucleotide (Figure S10). With IC₅₀ values calculated, we could indirectly determine the affinity of substrates for our enzymes using the relationship between IC₅₀ and K_m described by Cheng and Prus-off,⁴¹ assuming UTP and CTP are competitive inhibitors of each other. This relationship would yield a K_i for CTP and UTP with TvVip1 and TvVip1 CTP-swap V2 (Table 3).⁴¹ Between TvVip1 Wt and TvVip1 CTP-swap V2, the K_i values clearly show that these enzymes are selective for UTP and CTP, respectively. The K_i for CTP in TvVip1 CTP-swap V2 is in the nanomolar range (K_i = 0.22 μM) whereas in TvVip1 Wt, K_i for CTP is in the micromolar range (>360 μM). These K_i values are reversed for the for UTP, where the K_i for TvVip1 CTP-swap V2 is in the micromolar range (>59 μM), and TvVip1 in the nanomolar range (0.045μM). While the ratio of K_i values of the preferred nucleotide for TvVip CTP-swap V2 is > 260 compared to >8,000 for TvVip1, TvVip CTP-swap V2 still demonstrates a substantial preference for CTP over UTP. Overall, these data demonstrate that binding affinity for the substrate accounts for the observed substrate preference swap.

Table 3.

IC₅₀ and K_i for TvVip1 CTP-swap V2 and TvVip1 Wt with CTP and UTP

	IC50 (CTP)	IC50 (UTP)	Ki (CTP)	Ki (UTP)
TvVip1 CTP-swap V2	16 ± 6.5 μM	>9,900 μM	0.22 μM	>59 μM
TvVip1 Wt	>4,900μM	19 ± 5.7 μM	>360 μM	0.045 μM

IC₅₀ values were calculated from the concentration of competing nucleotide that was able to reduce the formation of deuterated-5′-dA formation from 1 mM deuUTP or deuCTP by 50%. Reactions consisted of 10 μM enzyme with 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 1 mM of deuterated nucleotide (deuUTP or deuCTP) and varying amounts of competing nucleotide (CTP or UTP). K_i values were determined from average IC₅₀ and Km values using the Cheng-Prusoff equation with the assumption that UTP and CTP are competitive inhibitors. IC₅₀ values are mean ± SEM.

Given that we were able to swap the substrate preference in TvVip, we aimed to do the same in Mus musculus viperin (Figures 5A and S11) and the VLE from N. vectensis (NvVip; Figure S11). However, both engineered viperins did not exhibit any change in substrate preference. While engineered MmVip (MmVip UTP-swap) demonstrated modest 5′-dA formation, there was no preference for either deuUTP or deuCTP in competition assays, suggesting that 5′-dA formation is abortive. Product formation assays with ATP and GTP also did not show any formation of the ddhNTP product (data not shown). Engineered NvVip (NvVip UTP-swap) hindered 5′-dA and deu-5′-dA formation and attempts to swap an adjacent loop region (NvVip UTP-swap VIA183IVT) for proper β-8 loop positioning also failed (data not shown).

Structural characterization of the engineered T. virens viperin-like enzyme and the wildtype T. virens viperin-like enzyme with CTP

To elucidate the structural basis for substrate selectivity of TvVip1 CTP-swap V2, we determined the crystal structure with CTP and SAM bound to a resolution of 1.68 Å (Figure 6A; Table 1; PDB: 9DFW). As previously reported, TvVip1 can form 5′-dA in the presence of CTP but does not prefer it,³⁰ suggesting that it is also capable of ddhCTP formation. Therefore, we also determined the structure of TvVip1 with CTP and SAM bound to 1.72 Å (Figure S12; Table 1; PDB: 9DGW). Both TvVip1 CTP-swap V2 and TvVip1 with CTP and SAM crystallized with three molecules in the asymmetric unit, similar to the previously reported crystal structures of TvVip1 with UTP bound.³⁰ Compared to the UTP-bound TvVip1 structure, where R257 makes direct contacts with the C4 exocyclic oxygen on UTP (Figure 6B), in the CTP-bound TvVip1 structure, R257 is shown extended outward and making direct contacts with the C2-exocyclic oxygen on the cytidine moiety instead, demonstrating the flexibility of substrate binding in TvVip1 between pyrimidines (Figure S9). In the TvVip1 CTP-swap V2 structure, the secondary structure of the β-8 loop region is virtually identical to the β-8 loop of MmVip, exhibiting an RMSD of 0.75 Å (8 Cα) (Figures 6C and 6D). Like the MmVip structure with CTP bound, lysine and cysteine on the β-8 loop of TvVip1 CTP-swap V2 make direct contacts with the C2-exocyclic amine and C4-exocyclic oxygen, respectively. Thus, we were able to recapitulate the same β-8 loop interactions in TvVip1 CTP-swap V2 that exist in MmVip, demonstrating that the β-8 loop is a main contributor to substrate binding in viperin and VLEs and can be engineered to modulate substrate selectivity.

Figure 6. Crystal structure of TvVip1 CTP-swap V2 demonstrates the structural basis for CTP selectivity.

(A) Monomer of TvVip1 CTP-swap V2 with the engineered regions highlighted in magenta.

(B) The β-8 loop region of TvVip1 (PDB: 7N7I), arginine (R257) makes a direct interaction with the C4-exocyclic oxygen of UTP.

(C) β-8 loop of TvVip1 CTP-swap V2. Lysine and cysteine are shown making direct contacts to the C2-exocyclic oxygen and C4-exocyclic amine, respectively. A salt bridging interaction can be seen between glutamate (E199) and the arginine (R245) of the β-8 loop.

(D) Structural alignment of the β-8 loop region of TvVip1 CTP-swap V2 and MmVip (pink; PDB: 6Q2P). The β-8 loops between these two structures are almost virtually identical with a calculated RMSD of 0.75 Å (8 Cα).

DISCUSSION

We recently reported that mammalian viperin catalyzes the formation of ddhCTP from CTP and we and others demonstrated through X-ray crystal structures that the β-8 loop is divergent and provides direct contacts with the pyrimidine substrates in viperin and VLEs.^2,30,32 We also previously noted that some species have a second VLE that has been seemingly inactivated by a histidine to phenylalanine point mutation.³⁰ Herein, we sought to expand on the current knowledge of viperin and VLE structure and function by providing bioinformatic, biochemical, and structural evidence for NΦF VLEs, all characterized by the NΦF motif. We demonstrate that NΦF VLEs have β-8 loop sequences that are comparable to other CTP-selective viperins and VLEs, and provide evidence that a representative NΦF VLE, TvVip2, prefers to bind CTP instead of UTP in T. virens. While the observed turnover rate of TvVip2 for ddhCTP is ~13-fold slower than TvVip1 for ddhUTP at similar substrate concentrations, the apparent increase in turnover rate by TvVip2 F40H suggests this mutation may have occurred to inhibit ddhCTP production, rendering NΦF VLEs vestigial in nature.

Why fungi would prefer to produce ddhUTP over ddhCTP, and why certain fungal species chose to inactivate CTP-selective VLEs, is unknown, but it may be that ddhUTP is less toxic to the host or that mycoviral polymerases are more sensitive to ddhUTP. It is also possible that there may be toxicity if two different ddhNTPs are produced upon viral infection. Additionally, these NΦF VLEs may have an alternate function and there may be another substrate besides NTPs. This is not out of the question, as others have reported that TvVip1 and another fungal viperin-like enzyme from Thielavia terrestris can conjugate 5′-dA to UDP-glucose and isopentyl pyrophosphate, respectively, though characterization of these reactions was not performed,^24,42 and these proposed reactions have not been found in a cellular setting. In addition, some VLEs appear in defense islands next to genes that may encode proteins that modify their ddhNTP products for unknown roles in pathogen defense.²⁹

Interestingly, performing a sequence alignment of fungal NΦH VLE β-8 loop regions from our bioinformatic dataset demonstrates divergence in this region, with the predominant pyrimidine-interacting positions occupied by either lysine (50%), arginine (32%), and cysteine (8%); suggesting that the substrate selectivity of fungi is more widespread than previously thought and many NΦH VLEs are similar to the β-8 loops of CTP-selective NΦF VLEs reported herein (Figure S13A). An overlay with the crystal structure of TvVip2 and a representative AlphaFold model of a lysine-containing NΦH fungal VLE shows similar positioning of the β-8 loop, demonstrating that these VLEs are likely CTP-selective (Figure S13B).^43,44 Based on our data present here, the seemingly widespread conservation of CTP-selective β-8 loop residues, and the apparent lack of conserved neighboring genes that could indicate an alternate product (Figure S14),⁴⁵ we believe that NΦF VLEs are vestigial and do not catalyze an alternate reaction.

Given the presence of a CTP-selective NΦF VLE in T. virens and the divergence of the β-8 loops between TvVip1 and TvVip2, we further applied this knowledge to create an engineered VLE with modulated substrate selectivity and engineered the substrate selectivity of TvVip1 by swapping the β-8 loop from mouse viperin to create a CTP-selective TvVip1. Competition assays combined with X-ray crystal structures demonstrated that our engineered TvVip1 CTP-swap V2 preferred CTP over UTP, which supported our hypothesis that the β-8 loop drives substrate selectivity. However, when we determined the V_max and K_m for UTP and CTP for both the Wt and engineered enzymes, the relative catalytic efficiency was essentially unchanged for both enzymes. This led us to believe that substrate affinity and not kinetics dictate substrate preference in viperin and VLE enzymes. Because RS enzymes are highly oxygen sensitive, it is difficult to determine binding constants using standard affinity measurements such as ITC or SPR. Therefore, we decided that we could indirectly determine affinity of substrates for our enzymes using the relationship between inhibition (IC₅₀) and K_m.⁴¹ To do this, we used aset concentration of either deuUTP or deuCTP and varied that concentration of CTP or UTP, respectively. Using this technique, we can monitor how effective the light nucleotide is at inhibiting production of heavy ddhNTPs by observing deuterated 5′-dA, thereby producing an IC₅₀ curve. By combining these data with our K_m data, we can calculate an apparent K_i for each NTP. Indeed, our data showed that TvVip1 preferred UTP over CTP by > 8,000-fold (K_{i (CTP)} > 360 μM vs. K_{i (UTP)} > 0.045 μM) and our engineered TvVip1 CTP-swap V2 preferred CTP over UTP by > 260-fold (K_{i (CTP)} 0.22 μM vs. K_{i (UTP)} > 59 μM). One assumption with this technique is that it requires the KIE to be close to 1 to avoid biasing the reaction in favor of one kinetic outcome over the other. Based on our KIE for our enzymes with each substrate being between 0.9 and 1.1, this assumption is true. Lastly, we attempted to engineer the viperin from Mus musculus and VLE from N. vectensis using this same technique. However, both engineered enzymes could not produce any detectable ddhNTPs and only abortively produced 5′-dA, suggesting that swapping the β-8 loop in these enzymes largely deprived them from utilizing CTP as a substrate. We speculate that promiscuous viperin or VLEs may be easier to engineer, and the reason that TvVip1 was amenable to engineering is due to its apparent recent evolutionary divergence from CTP-selective fungal VLEs.

The conclusions we have drawn from our studies have been corroborated by a recent bioinformatic study. In this report, they showed that CTP selectivity in eukaryotes evolved from ancient, more promiscuous viperin/VLEs from Asgard Archaea that lack a definitive preference for a CTP, UTP, or GTP.⁴⁶ In addition, they were able to show that the promiscuous eukaryotic viperin from Aurellia sp. could be made selective for UTP or CTP by a combination of point mutations in the β-8 loop. Based on our engineering efforts and that of this report, it is possible that viperins and VLEs have evolved to a point where their selectivity is highly specialized. As a result, engineering these enzymes to select for alternate substrates through a rational approach may be difficult. Therefore, it might only be feasible to engineer more evolved and selective viperin and VLEs, such as human viperin, to select other NTPs using non-rational protein engineering technology such as PACE.⁴⁷

Overall, our data provide insights into how organisms may have chosen their “optimum” antiviral ribonucleotide throughout evolution and provide additional structural evidence describing how viperins and VLEs exhibit divergent substrate selectivity, despite their high conservation between the three domains of life.³⁰ However, a large gap still exists in our understanding of why each species evolved to catalyze different ddhNTP products, the antiviral effects of different ddhNTPs against pathogens, and if there are other roles for NΦF VLEs in fungi.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Tyler L. Grove (tyler.grove@einsteinmed.edu).

Materials availability

This study did not generate any unique reagents. Please email the lead contact regarding any requests.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Protein expression was carried out using commercially available chemically competent BL-21 (DE3) E. coli.

METHOD DETAILS

Materials

Materials utilized in this study are described previously.³⁰ Synthetic ddhCTP and ddhUTP were kind gifts from the Laboratory of Lawrence Harris at the Ferrier Research Institute, Victoria University of Wellington,⁵⁶ and were utilized to quantify ddhNTP production.

Cloning of the viperin and VLEs used in this study

The TvVip1 (Uniprot ID: G9MQB8), TvVip2 (Uniprot ID: G9N0G3), TvVip2 F40H, TvVip1 CTP-swap, MmVip UTP-swap, and NvVip UTP-swap were commercially synthesized by IDT and contained ligation independent cloning (LIC) sites with complementarity to vector pNIC28-Bsa4 and were ligated using standard LIC methods.^57,58 The resulting construct contains an additional 22 amino acids to the N-terminus, and includes a 6x-His-tag and a TEV cleavage site (see Table S1, underlined and italicized). Due to reports that the removal of the N-terminal amphipathic domain is important for production of soluble recombinant viperin from human, rat, and mouse,^2,59,60 we identified potential signal peptides utilizing Pred-Signal.⁶¹ Pred-Signal suggested that TvVip2 contained a signal peptide (residues 1-31) for cellular localization, which was removed. Although Pred-Signal did not indicate a signal peptide on TvVip1, we removed the N-terminal region similar to TvVip2 (residues 1-16). The remainder of our targeted proteins did not possess a predicted amphipathic sequence. Mutagenesis PCR was performed to yield the TvVip1 CTP-swap V2 and NvVip UTP-swap VIA183IVT constructs.

Computational

Local blastp searches used default parameters (NCBI) and were searched against the entire non-redundance sequence database (downloaded in October of 2023). Sequence identities were calculated using the Expasy – SIM Alignment tool for protein sequences with 20 computed alignments, a gap open penalty of 12, a gap extension penalty of 4, and using the BLOSUM62 comparison matrix. The taxonomy of each of the 2713 unique sequences was determined using NCBIs E-Utilities tool and the sequences were clustered using CD-HIT.³⁴ The radical SAM domain was aligned by taking only 16 residues before and 184 residues after the NΦF or NΦH motif, and using a sequence identity cutoff of 70%.³⁴ A representative sequence from each resulting cluster was aligned using clustalw³⁵ and a phylogenetic tree was created using iqtrees model finder,^36–38 which informed us to use the WAG substation matrix,³⁷ and maximum-likelihood function.³⁸ Sequence alignments focused on differences with the β-8 loop were computed using the Clustal Omega extension⁵³ with default settings and alignments were visualized using Jalview.⁵² Phylogenetic trees were constructed using iTOL.⁵⁴ Data availability — all viperin and viperin-like enzyme sequences, cluster consensus sequences, and compiled datasets are available at - https://github.com/jakelachowicz/FungalVLEs. The naming convention for each tree branch is Kingdom/subkingdom (determined by NCBI E-Utilities tool), Cluster number (see https://github.com/jakelachowicz/FungalVLEs for a list of sequences within each cluster), total sequences in cluster, number of sequences with NΦH motif, and number of sequences with NΦF motifs. Clusters containing NΦF VLEs are indicated with three dashes (---) at the end of the cluster label. Sequences from Archaea and Bacteria were not included as NΦF VLEs are predominantly in eukaryotes. Genome neighborhood networks were created utilizing the EFI – Genome neighborhood tool using default settings.^55,62

Expression and purification of wildtype and engineered viperin and VLEs

Transformed BL-21 E. coli were directly grown in 20 mL LB containing 50 μg/mL kanamycin and either 34 μg/mL chloramphenicol (pPH151)⁴⁸ or 100 μg/mL ampicillin (pDB1282)⁴⁹ overnight. This was used to inoculate a 2 L PYREX media bottle with either the defined autoinduction media, PA-5052,⁶³ or LB broth, with 100 μM FeCL3 and ~2mL antifoam solution (Spectrum). Cultures were grown with a constant sparge of 0.22 μm filtered air at 37°C until OD 0.6-0.8 (approximately 5 hours). After ~5 hours, 1.2mM cysteine was added to the cultures (along with 0.2% L-arabinose for the pDB1282 plasmid to induce expression of Fe-S cluster biogenesis), the temperature was reduced to 22°C, and the cultures were incubated for ~20 hours before being harvested by centrifugation at 10,000 x g. E. coli cell pellets were stored in liquid N₂ until purification. The following purification steps were performed anaerobically in an MBraun anaerobic chamber (<1.2 ppm O₂). Approximately 30 mL of lysis buffer containing 50mM HEPES, pH 7.5, 300 mM KCl, 5% glycerol, 5 mM imidazole, 14.3 mM 2-mercaptoethanol (BME), and 0.5% Triton x305 detergent was added to the top of the frozen cell pellet and was subsequently sonicated. During sonication, cells were cooled in an ice bath while being subjected to cycles of 3 seconds on and 20 seconds off, for a total of 4 full minutes of sonic disruption (80% output; approximately 30 minutes total when combining on/off times). The resulting cell lysate was centrifuged for 1 hour at 10,000 × g at 4°C. Protein purification was performed with the ÄKTA express FPLC system. The resulting supernatant was passed over a 5 mL Fast-Flow Ni-Sepharose column (GE Biosciences) which was equilibrated in lysis buffer (without detergent). After loading, the column was washed with 20 column volumes of lysis buffer (without detergent) and an on-column reconstitution was performed where the column was washed with a buffer containing 600 μM FeCl₃, 400 μM Na₂S, 57 mM BME, 100 mM HEPES pH 7.5, 300 mM KCl and 10% glycerol. Reconstituted protein was eluted with two column volumes of elution buffer (lysis buffer without detergent and with 300mM imidazole). The resulting eluent with high UV 280nm absorbance was collected and loaded onto a HiPrep 16/60 S-200 size-exclusion column equilibrated with 20 mM HEPES, pH 7.5, 300 mM KCl, 10% glycerol, and 5 mM DTT. The resulting fractions displaying a high UV 280 nm absorbance and brown color were pooled and concentrated using an Amicon centrifugal filter unit (10 kDa molecular weight cut off). The final purified protein concentration was estimated by UV-Visible (A₂₈₀) using the extinction coefficient calculated based on the amino acid sequence (Expasy ProtParam).

Substrate competition assays

Reactions contained in a total volume of 150 μL: 10 μM enzyme, 50mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 2 mM of nucleotides (ATP, GTP, CTP and UTP, where one of these are deuterated), and 5 mM dithionite. Titration reactions between deuCTP and UTP contained in a total volume of 150 μL: 10 μM enzyme, 50mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 1 mM of deuCTP, either 1 mM, 3 mM and 5 mM or 10 mM of UTP, and 5 mM dithionite. Reaction mixtures lacking dithionite were incubated for 5 minutes at room temperature, and a 20μL aliquot was removed (t=0) and added to 20 μL of a solution containing 100 mM H₂SO₄ and 200 μM L-tryptophan (internal standard) to yield a final internal standard concentration of 100 μM. Reactions were initiated with dithionite (5 mM) and incubated for either 60 minutes or 30 minutes (Figure 5B) before being quenched in acid as described above. TvVip2 competition assays between 2′-deoxyCTP and deuCTP (Figure S3B) utilized the E. coli Fpr/FldA reducing system and contained 30 μM of enzyme, 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 2 mM nucleotide, 5 μM Fpr, 25 μM FldA, and 1 mM NADPH. FldA/Fpr reactions were initiated with SAM. The samples were centrifuged at 4,000 x g to remove the precipitated protein. 5′-dA and deu-5′-dA amounts were quantified by mass spectrometry using Method A. Data are presented as the percentage of deuterated 5′-deoxyadenosine and 5′-deoxyadenosine out of total 5’-deoxyadenosine (deuterated plus non-deuterated) within the reaction mixture at the indicated timepoint. Percentages were calculated utilizing the internal standard response ratios of deu-5′-dA and 5′-dA (ratio of signal to that of 100 μM tryptophan as the internal standard). All assays were performed in duplicate at minimum.

Product formation assays

Reactions contained in a total volume of 150 μL (ddhCTP formation): 10 μM or 100 μM (TvVip2) enzyme was mixed with 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 1 mM of CTP, and 5 mM dithionite. Reaction mixtures lacking dithionite were incubated for 5 minutes at room temperature, and an aliquot was removed (t=0) and added to a solution containing 100 mM H₂SO₄ and 200 μM L-tryptophan (internal standard) in a 1:1 ratio to yield a final internal standard concentration of 100 μM. Reactions were initiated with dithionite (5 mM) and incubated at the indicated timepoints before being quenched in acid as described above. All product formation assays were quantified by mass spectrometry using Method C except for 5′-dA concentration in Figure S7, which was quantified using Method A, and 5′-dA concentration in Figure S3B, which was quantified using Method D. All assays were performed in duplicate at minimum.

Steady-state kinetics of TvVip1 CTP-swap V2 and TvVip1

Reactions contained in a total volume of 90 μL: A 5x master mix was made containing either 5 μM or 25 μM enzyme with 250 mM HEPES pH 7.5, 750mM KCl, and 5mM SAM. The 5x master mix was diluted to create a 1x dilution mix and a 1x substrate mix containing 1mM of NTP. The 1x substrate mix was serial diluted with 1x dilution mix to create assay mixtures with varying NTP concentration. Reaction mixtures lacking dithionite were incubated for 5 minutes at room temperature, and an aliquot was removed (t=0) and added to a solution containing 100 mM H₂SO₄ and 200μM L-tryptophan (internal standard) in a 1:1 ratio to yield a final internal standard concentration of 100 μM. Reactions were initiated with dithionite (5 mM) and turnover rate was calculated from either 30 or 90 minutes using 1 μM final enzyme concentration, or 10 minutes using 5 μM final enzyme concentration. ddhCTP or ddhUTP concentration was determined by mass spectrometry utilizing Method B. Curve fitting was performed using GraphPad Prism. All assays were performed in duplicate at minimum.

Calculating kinetic isotope effects of deuterated substrates

Reactions contained in a total volume of 100 μL: 10 μM enzyme was mixed with 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 1 mM of nucleotide or deuterated nucleotide, and 5 mM dithionite. Reaction mixtures lacking dithionite were incubated for 5 minutes at room temperature, and an aliquot was removed (t=0) and added to a solution containing 100 mM H₂SO₄ and 200 μM L-tryptophan (internal standard) in a 1:1 ratio to yield a final internal standard concentration of 100 μM. Reactions were initiated with dithionite (5 mM) and quenched in acid after 15 minutes. 5′-dA and deu-5′-dA amounts were quantified by mass spectrometry using Method A. All assays were performed in duplicate at minimum.

Determining IC50 and K_i for CTP and UTP

Reactions contained in a total volume of 60 μL: 10 μM enzyme was mixed with 50 mM HEPES pH 7.5, 150 mM KCl, 1 mM SAM, 1 mM of deuterated nucleotide (deuUTP or deuCTP), varying amounts of competing nucleotide (CTP or UTP), and 5 mM dithionite. Reaction mixtures lacking dithionite were incubated for 5 minutes at room temperature, and an aliquot was removed (t=0) and added to a solution containing 100 mM H₂SO₄ and 200 μM L-tryptophan (internal standard) in a 1:1 ratio to yield a final internal standard concentration of 100 μM. Reactions were initiated with dithionite (5 mM) and incubated for 10 minutes before being quenched in acid. Fractional activities were determined from the amount of deu-5′-dA produced by a control without competing nucleotide (v0) compared to the amount produced with competing nucleotide (v). deu-5′-dA amounts were quantified by mass spectrometry using Method A. All assays were performed in duplicate at minimum. K_i was determined based on the Km and IC₅₀ values for CTP and UTP with TvVip1 and TvVip1 CTP-swap V2 based on the Cheng-Prusoff equation (Equation 1) with the assumption that UTP and CTP are competitive inhibitors.

$$K_i=\frac{I{C}_{50}}{1+\frac{\left[S\right]}{K_m}}$$ (Equation 1)

Metabolite quantification by mass spectrometry

Method A

High-performance liquid chromatography (HPLC) with detection by mass spectrometry (LCMS) was conducted on an Agilent Technologies (Santa Clara, CA) 1200 system coupled with an Agilent Technologies 6490 triple Quadrupole mass spectrometer. The Agilent MassHunter software package was used for data collection and analysis. Supernatants were injected (4 μL) onto an Agilent Sb-Aq RRHD column (3.0 x 50 mm, 1.8 μm) equilibrated in 99.2% buffer A (0.1 M Formic Acid) and 0.8% buffer B (acetonitrile) with a column compartment temperature of 37°C and a flow rate of 0.4 mL/min. A gradient of 0-20% B was applied from 0-2 minutes, and 20-60% B from 2-2.75 minutes, and returning to 0.8% B from 2.75-3 minutes. The column was allowed to re-equilibrate for 1 minute under initial conditions before subsequent sample injections. Retention time and fragmentation products measured by LC/MS are indicated in Table S2.

Method B

LCMS was conducted on an Agilent Technologies (Santa Clara, CA) 1290 Infinity II system coupled with an Agilent Technologies 6545 Q-TOF mass spectrometer. The Agilent MassHunter software package was used for data collection and analysis. Assay mixtures were injected (3 μL) on an Agilent InfinityLab Poroshell 120 HILIC-Z column (2.1 x 150mm, 2.7μm, PEEK Lined) equilibrated in 90% buffer B (10mM ammonium acetate pH 9, 5μM medronic acid, 10:90 v/v water:acetonitrile) and 10% buffer A (10mM ammonium acetate pH 9, 5μM medronic acid, 90:10 v/v water:acetonitrile) with a column compartment temperature of 50°C and a flow rate of 0.4 mL/min. A gradient of 10-24% A was applied from 0.5-1.5 minutes, 24-27% A from 1.5-5 minutes, and 27-40% A from 5-5.5 minutes, before returning to 10% A from 6.0-6.5 minutes. The column was allowed to re-equilibrate for 1.5 minutes under initial conditions before subsequent sample injections. Collision energy was calculated using the formula (Slope)*(m/z)/100+Offset where slope=3 and offset=10. Retention time and fragmentation products measured by LC/MS are indicated in Table S2.

Method C

Same parameters as method B except solvent content of Buffer A was adjusted. Buffer B: 10 mM ammonium acetate pH 9, 5 μM medronic acid, 10:90 v/v water:acetonitrile. Buffer A: 10 mM ammonium acetate pH 9, 5 μM medronic acid, 100% water. This method was performed on either a 6490 QQQ or 6545 Q-TOF. Retention time and fragmentation products measured by LC/MS are indicated in Table S2.

Method D

High-performance liquid chromatography (HPLC) with detection by mass spectrometry (LCMS) was conducted on an Agilent Technologies (Santa Clara, CA) 1290 Infinity II system coupled with an Agilent Technologies 6545 Q-TOF mass spectrometer. The Agilent MassHunter software package was used for data collection and analysis. Supernatants were injected (2 μL) on to an Agilent EclipsePlusC18 RRHD column (2.1 x 50 mm, 1.8μm) equilibrated in 99.2% buffer A (0.1 M Formic Acid) and 0.8% buffer B (acetonitrile) with a column compartment temperature of 50°C and a flow rate of 0.4 mL/min. A gradient of 0-20% B was applied from 0-2 minutes, and 20-60% B from 2-2.75 minutes, and returning to 0.8% B from 2.75-3 minutes. The column was allowed to re-equilibrate for 1 minute under initial conditions before subsequent sample injections. Retention time and fragmentation products measured by LC/MS are indicated in Table S2.

X-ray crystallography

Crystallization of TvVip1 CTP-swap V2, TvVip1, TvVip2, and TvVip2 F40H:

Diffraction-quality crystals were obtained by sitting-drop vapor diffusion at 20°C in an anaerobic chamber maintained at <0.1 ppm oxygen (MBraun). Proteins were TEV-cleaved without subsequent purification steps before crystallization. Drops of 0.4 μL of TvVip1 CTP-swap V2 [10 mg/mL in 10 mM HEPES (pH 7.5), 100 mM KCl, 1 mM SAM, 1 mM CTP] were mixed with 0.4 μL of precipitant (0.1 M HEPES (pH 7.0), 10% (w/v) PEG 6K) and equilibrated against a solution of 0.5 M LiCl. Drops of 0.4 μL of TvVip1 [10 mg/mL in 10 mM HEPES (pH 7.5), 1 mM SAM, 1 mM CTP] were mixed with 0.4 μL of precipitant (0.1 M Bis-Tris pH 6.5, 0.2 M NaCl, 25% (w/v) PEG 3350) and equilibrated against a solution of 0.5 M LiCl. Drops of 0.4 μL of TvVip2 [10 mg/mL in 10 mM HEPES (pH 7.5), 100 mM KCl, 1 mM SAM, 1 mM CTP] were mixed with 0.4μL of precipitant (0.1 M Bis-Tris (pH 6.5), 20% (w/v) PGG MME 5K) and equilibrated against a solution of 0.5 M LiCl. Drops of 0.4 μL of TvVip2 F40H [1 mg/mL in 10 mM HEPES (pH 7.5), 100 mM KCl, 1 mM SAM, 1 mM CTP] were mixed with 0.4 μL of precipitant (0.1 M HEPES (pH 7.5), 25% (w/v) PEG 3350) and equilibrated against a solution of 0.5 M LiCl. Crystals were mounted in nylon loops and flash cooled in liquid nitrogen inside of the anaerobic chamber and were stored in liquid nitrogen prior to data collection. Diffraction data were collected at NSLSII (Brookhaven National Lab, Brookhaven, NY) on beamline 17-ID-2 (FMX). All diffraction data were integrated and scaled using the HKL3000 suite or XDS and aimless. Diffraction data from the TvVip1 CTP-swap V2 crystal dataset was consistent with the trigonal space group P3₁ and a resolution of 1.676 Å with unit cell parameters consisting of a=85.131 Å, b=85.131 Å, and c=110.47 Å (a=b=90°, c=120°) with three molecule per asymmetric unit. Diffraction data from the TvVip1 crystal dataset was consistent with the monoclinic space group P2₁, with unit cell parameters a=51.569 Å, b=80.309 Å, and c=105.61 Å (a=c=90°, b=96.7174°), with 3 molecule per asymmetric unit. Diffraction data from the TvVip2 crystal dataset was consistent with the trigonal space group P2₁2₁2₁ and a resolution of 1.5 Å with unit cell parameters consisting of a=36.808 Å, b=57.838 Å, and c=157.69 Å (a=b=c=90°) with one molecule per asymmetric unit. Diffraction data from the TvVip2 F40H crystal dataset was consistent with the monoclinic space group P2₁2₁2₁ and a resolution of 2.3 Å with unit cell parameters consisting of a=37.137 Å, b=55.959 Å, and c=152.81 Å (a=b=c=90°) with one molecule per asymmetric unit. Phases were determined by molecular replacement using the previously solved TvVip1 structure (PDB: 7N7I). Molecular replacement was performed using PHENIX,⁵⁰ with manual model building using Coot,⁵¹ and refinement using phenix.refine and Refmac5 (CCP4).^50,64 Data was cut and scaled for TvVip1 with CTP, TvVip2, and TvvVip F40H using Scala (CCP4).⁶⁵ The final model of TvVip1 CTP-swap V2 consists of three molecules in the asymmetric unit with chains A and B consisting of residues 18-311, and Chain C consisting of residues 18-300. The final model of TvVip1 consists of residues 17-295 (chain A), 17-298 (chain B), and 18-297 (chain C). Chain C of the final TvVip1 with CTP structure demonstrated significant static disorder. Therefore, chain C of TvVip1 with CTP was modeled as an NCS copy of chains A and B and refinement was performed utilizing previously fixed TLS restraints not written to the coordinates. The final model of TvVip2 consists of one molecule in the asymmetric unit consisting of residues 29-311. The final model of TvVip2 F40H consists of one molecule in the asymmetric unit consisting of residues 32-311. Each structure contained 4 iron ions, 4 sulfide ions, SAM, and CTP within each molecule in the asymmetric unit. All figures were produced using PyMOL (Schrodinger, LLC). Data collection and refinement statistics are shown in Table 1.

QUANTIFICATION AND STATISTICAL ANALYSIS

All graphing, curve fitting, and standard deviation/error calculations were performed using GraphPad Prism. K_i in Table 3 was determined based on the Km and IC₅₀ values for CTP and UTP with TvVip1 and TvVip1 CTP-swap V2 based on the Cheng-Prusoff equation (Equation 1) with the assumption that UTP and CTP are competitive inhibitors. All assays were performed with n=2 at minimum. Specific statistical details for experiments can be found within the figure legend. X-ray data collection and refinement statistics are shown in Table 1.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.str.2025.04.009.

Highlights.

• Some fungal species contain an active and inactive viperin-like enzyme (VLE)

• Inactive fungal VLEs prefer CTP as a substrate

• Engineered a VLE to switch substrate preference

• Structures of engineered VLEs provide a foundation for broader substrate engineering

Data and code availability

• All viperin and viperin-like enzyme sequences, cluster consensus sequences, and compiled datasets are publicly available as of the date of publication at - https://github.com/jakelachowicz/FungalVLEs. Crystal structures have been deposited in the PDB under the accession codes 9DFN, 9DFU, 9DFW, and 9DGW and are publicly available as of the date of publication. Accession codes are also listed in the key resources table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
deuCTP	Cambridge Isotopes	Cat#DLM-8924-CA-20
deuUTP	Cambridge Isotopes	Cat#DLM-8925-CA-20
Deposited data
Model for molecular replacement for X-ray structures	Lachowicz & Gizzi et al. (2021)30	PDB: 7N7I
X-ray crystal structure of the second viperin-like enzyme from Trichoderma virens with bound CTP and SAM	This study	PDB: 9DFN
X-ray crystal structure of the second Viperin-like enzyme from T. virens variant F40H with bound CTP and SAM	This study	PDB: 9DFU
X-ray crystal structure of an engineered Viperin-like enzyme from T. virens with bound CTP and SAM	This study	PDB: 9DFW
X-ray crystal structure of the Viperin-like enzyme from T. virens with bound CTP and SAM	This study	PDB: 9DGW
Bioinformatics data on NΦF viperin-like enzymes generated in this study.	This study	https://github.com/jakelachowicz/FungalVLEs
Oligonucleotides
Primer (IDT): TvVip1 CTP-swap V2 Forward (5′→3′): GTTTCCAGGTCCTTATTATCGAGGGGGAGAATGACTCTGAC	This study	N/A
Primer (IDT): TvVip1 CTP-swap V2 Reverse (5′→3′): GTCAGAGTCATTCTCCCCCTCGATAATAAGGACCTGGAAAC	This study	N/A
Primer (IDT): NvVip UTP-swap V2 Forward (5′→3′): GTGTTTCAGTGTCTTATTGTGACCGGCGAAAACTCCGGTGAGGAC	This study	N/A
Primer (IDT): NvVip UTP-swap V2 Reverse (5′→3′): GTCCTCACCGGAGTTTTCGCCGGTCACAATAAGACACTGAAACAC	This study	N/A
Recombinant DNA
Plasmid-pPH151	Hänzelmann et al. (2004)48	N/A
Plasmid-pDB1282	Zheng et al. (1998)49	N/A
Software and algorithms
Phenix	Adams, P.D. et al. (2010)50	https://phenix-online.org/
CCP4	Emsley, P. et al. (2010)51	https://www.ccp4.ac.uk/
Jalview	Waterhouse, A.M. et al. (2009)52	https://www.jalview.org/
CD-HIT	Li, W., and Godzik, A. (2006)34	https://www.bioinformatics.org/cd-hit/
clustalw	Thompson, J.D. et al. (1994)35	http://www.clustal.org/clustal2/
iqtrees model finder	Nguyen, L.-T et al. (2015)38	http://www.iqtree.org/
Clustal Omega	Sievers, F. et al. (2011)53	http://www.clustal.org/omega/
iTOL	Letunic, I., and Bork, P. (2019)54	https://itol.embl.de/
EFI – Genome neighborhood tool	Zallot, R. et al. (2023)55	https://efi.igb.illinois.edu/efi-gnt/

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.