Structural Basis of Ubiquitin Recognition by a Bacterial Ovarian Tumor Deubiquitinase LotA

ABSTRACT

Pathogenic bacteria have acquired a vast array of eukaryotic-protein-like proteins via intimate interaction with host cells. Bacterial effector proteins that function as ubiquitin ligases and deubiquitinases (DUBs) are remarkable examples of such molecular mimicry. LotA, a Legionella pneumophila effector, belongs to the ovarian tumor (OTU) superfamily, which regulates diverse ubiquitin signals by their DUB activities. LotA harbors two OTU domains that have distinct reactivities; the first one is responsible for the cleavage of the K6-linked ubiquitin chain, and the second one shows an uncommon preference for long chains of ubiquitin. Here, we report the crystal structure of a middle domain of LotA (LotA_M), which contains the second OTU domain. LotA_M consists of two distinct subdomains, a catalytic domain having high structural similarity with human OTU DUBs and an extended helical lobe (EHL) domain, which is characteristically conserved only in Legionella OTU DUBs. The docking simulation of LotA_M with ubiquitin suggested that hydrophobic and electrostatic interactions between the EHL of LotA_M and the C-terminal region of ubiquitin are crucial for the binding of ubiquitin to LotA_M. The structure-based mutagenesis demonstrated that the acidic residue in the characteristic short helical segment termed the “helical arm” is essential for the enzymatic activity of LotA_M. The EHL domain of the three Legionella OTU DUBs, LotA, LotB, and LotC, share the “helical arm” structure, suggesting that the EHL domain defines the Lot-OTUs as a unique class of DUBs.

IMPORTANCE To successfully colonize, some pathogenic bacteria hijack the host ubiquitin system. Legionella OTU-like-DUBs (Lot-DUBs) are novel bacterial deubiquitinases found in effector proteins of L. pneumophila. LotA is a member of Lot-DUBs and has two OTU domains (OTU1 and OTU2). We determined the structure of a middle fragment of LotA (LotA_M), which includes OTU2. LotA_M consists of the conserved catalytic domain and the Legionella OTUs-specific EHL domain. The docking simulation with ubiquitin and the mutational analysis suggested that the acidic surface in the EHL is essential for enzymatic activity. The structure of the EHL differs from those of other Lot-DUBs, suggesting that the variation of the EHL is related to the variable cleaving specificity of each DUB.

INTRODUCTION

Polyubiquitination is an important posttranslational modification that commonly regulates various cellular processes in eukaryotes (1–3). The addition of ubiquitin requires successive reactions of three types of enzymes: ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzymes (E2s), and ubiquitin ligases (E3s). Ubiquitin molecules are linked through isopeptide bonds between the C-terminal carboxy group and the N-terminal amino group or the side chain amino group of one of the seven lysine residues (K6, K11, K27, K29, K33, K48, or K63) (1, 4). The linkage-specific assembly of ubiquitin chains is mediated by a subset of E2s and E3s (5, 6).

Ubiquitination is reversibly regulated by deubiquitinases (DUBs) to disassemble ubiquitin chains by cleaving the isopeptide bond between ubiquitin molecules in a nonspecific or in a linkage-specific manner (7, 8). Human has approximately 100 DUBs, and they are classified into seven families: ubiquitin-specific proteases (USPs), the ubiquitin-C-terminal hydrolases (UCHs), ovarian-tumor domain DUBs (OTUs), Machado-Joseph domain DUBs (MJDs/Josephins), motif interacting with ubiquitin-containing novel DUB family (MINDYs), zinc finger-containing ubiquitin peptidase 1 (ZUP1), and Jab1/Mov34/Mpr1 Pad1 N-terminal domain proteases (JAMM/MPN). These DUBs are cysteine proteases except for JAMM/MPN (9).

The eukaryotic ubiquitin system can target intracellular pathogens in many aspects, serving as host innate and adaptive immunity and antimicrobial autophagy (10). Although bacteria do not encode ubiquitin, some pathogenic bacteria have evolved numerous strategies to hijack the host ubiquitin system to counteract the host defense. The relevant examples are bacterially encoded E3 ligases and DUBs (11–18). The various functions of bacterial E3 ligases and DUBs have been extensively analyzed for Legionella pneumophila (15, 16, 19, 20).

L. pneumophila is the main etiological agent of Legionnaires' disease, which is a severe form of pneumonia (21). Legionella inhabit anywhere in moist environments including freshwater and soil. When humans inhale water droplets (aerosol) contaminated with Legionella into the lung, this bacterium can replicate within alveolar macrophages (22). Legionella secrete various effector proteins for survival in host cells. More than 300 effector proteins are translocated from L. pneumophila into host cells via the Dot/Icm (defect in organelle trafficking/intracellular multiplication) system, which is a bacterial secretion apparatus (23, 24). Most of the effector proteins are enzymes that affect the functions of host cellular proteins and other effector proteins translocated into host cells in various ways (20, 25). By utilizing the functions of the effector proteins, L. pneumophila modulates host cellular processes to survive in host cells.

Recently, a novel class of bacterial DUBs, which possesses domains similar to eukaryotic OTU DUBs, has been identified in L. pneumophila. This DUB class was named Legionella OTU-like-DUBs (Lot-DUBs) and includes LotA (Lpg2248), LotB (Ceg23 or Lpg1621), LotC (Lem27 or Lpg2529), and Ceg7 (Lpg0227) (26–30). Among the Lot-DUBs, LotA has a unique architecture; two OTU domains (OTU1 and OTU2), each of which contains a catalytic cysteine (C13 and C303), are tandemly aligned. In addition, a phosphatidylinositol triphosphate [PI (3)P]-binding domain locates at its C-terminal region (26). Therefore, LotA can be divided into three regions, LotA_N, LotA_M, and LotA_C, which include OTU1, OTU2, and PI (3)P-binding domain, respectively (Fig. 1A; see Fig. S1 in the supplemental material). Translocated into the infected host cells via the Dot/Icm system, LotA contributes to the removal of ubiquitin from the bacterial vacuole by its DUB activity (26). LotA exhibits an activity to cleave K6-linked diubiquitin depending on C13 in OTU1 (Fig. 1A). The DUB activity of LotA is shown to be modest against other diubiquitin species, but longer K48- and K63-linked chains can be efficiently cleaved depending on C303 in OTU2 (Fig. 1A) (26). LotB, also called Ceg23 (31), has an OTU domain at its N terminus and two putative transmembrane helices in the C-terminal region (27, 28) (see Fig. S1 in the supplemental material). LotB preferentially cleaves K63-linked chains of both diubiquitin and polyubiquitin (27, 28, 30), whereas LotC shows broader specificity for substrates such as K6-, K11-, K33-, K48- and K63-linked di-ubiquitin chains (29). A recent analysis revealed that LotB contributes to modulating soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNARE)-pairing on the Legionella-containing vacuole by its activity to deconjugate K63-linked ubiquitin from a v-SNARE, Sec22b (28). The crystal structures of the catalytic domains of LotB and LotC have been recently reported (27, 29, 32). These studies revealed that the two bacterial OTUs have a unique substructure designated ”extended helical lobe (EHL),” which is anticipated to provide a ubiquitin-binding site, S1. The result of the activity-based probe assay suggested that LotB, but not LotC, has an additional ubiquitin-binding site (S1’), which may enable LotB to exhibit a high specificity toward K63-linked chains (29).

FIG 1.

Constitution of LotA derivatives and in vitro DUB assay. (A) The schematics of LotA_full, LotA_NM, and LotA_M used in this study. The fragment obtained by digestion with trypsin (residues from 294 to 544) was used for crystallization. The OTU domains are shown as yellow bars, and the PI (3)P binding domain [PI(3)PBP] is shown as a dark gray bar. (B) K48‐linked pentaubiquitin (K48‐Ub5) and K63‐linked pentaubiquitin (K63‐Ub5) were reacted with 0.5 μM purified LotA derivatives at 37°C for 20 min. The reaction products were analyzed by SDS-PAGE. Ubiquitin (Ub) and LotA derivatives were detected by immunoblotting using antiubiquitin (P4D1) (upper) and anti-LotA (lower) antibodies, respectively. (C) The intensity of Ub2 and Ub5 bands were quantified from three independent experiments shown in panel B. The Ub2/Ub5 ratio was represented as the DUB activity.

In this paper, we report the crystal structure of the middle part of LotA (LotA_M), which includes OTU2. The single OTU domain was sufficient to exhibit the DUB activity against K48- and K63-linked polyubiquitin chains. Our crystal structure revealed that LotA_M possesses the EHL, like LotB and LotC. We further identified a possible ubiquitin-binding site, S1, in the EHL. These findings give insights into the working mechanism of bacterial OTUs, which includes the ubiquitin recognition mediated by the EHL.

RESULTS

The second OTU-domain in LotA is sufficient for cleavage of K48- and K63-linked polyubiquitin chains.

Initially, we purified full-length LotA (LotA_full) and LotA without the C-terminal PI (3)P-binding domain (LotA_NM) (Fig. 1A; see S2A in the supplemental material) and tried crystallizing them. However, no crystal was obtained. Therefore, we performed limited proteolysis of LotA_NM (∼67.6 kDa) with trypsin to determine the structurally stable core region of LotA for crystallization. LotA_NM was readily digested into two fragments with apparent molecular masses of 27 and 30 kDa (see Fig. S2B in the supplemental material). We purified the 30 kDa fragment by size exclusion chromatography and determined its molecular mass to be approximately 28,800 Da by MALDI-TOF mass spectrometry. We compared the molecular mass with the theoretically derived values of trypsin-digested fragments and concluded that the 28,800 Da fragment comprises residues 294 to 544 of LotA (M_w = 28,779), which corresponds to the middle part of LotA (287 to 543, LotA_M) that includes OTU2 containing C303. We constructed an expression system of LotA_M and purified it for the biochemical analysis (Fig. 1A).

To examine whether LotA_M retains the DUB activity as observed in LotA_full, we analyzed the polyubiquitin-cleavage activity of purified LotA_M (Fig. 1B and C). LotA_M cleaved K48-linked pentaubiquitin to the extent similar to LotA_full (Fig. 1C, upper panel). The DUB activity of LotA_M against K63-linked pentaubiquitin is also significant but to a lesser extent than that against K48-linked pentaubiquitin (Fig. 1C, lower panel). These results indicate that the C303-mediated DUB actively of LotA is retained in the fragment with the single OTU domain and suggest that the activity prefers K48 linkage to K-63 linkage.

Structure of LotA_M.

The crystal structure of LotA_M was determined at 2.0 Å resolution (Fig. 2; see Table S1 in the supplemental material). The crystal belongs to the space group of C222 and contains a single molecule in an asymmetric unit. The final model includes D296 to T457 and L465 to R544. The residues T458 to T464 were invisible in the electron density map and therefore were not modeled. The structure of LotA_M consists of 10 α-helices and three β-strands and folds into two subdomains, D1 (α1 to α4, α9, α10, and β1 to β3) and D2 (α5 to α8) (Fig. 2A and B; see Fig. S3 in the supplemental material). D1 is made up of two substructures, D1a and D1b, attached on a backbone structure formed by an 8-turn α-helix (α4) and a 3-turn α-helix (α9). D1a consists of α10, the N-terminal half of α1, and a β-sheet formed by β1–3. D1a and the backbone helices are structurally conserved in the catalytic domain of human OTUB1 (hOTUB1) as well as those of other human and viral OTU DUBs (Fig. 2C to F), although α4 is much longer than the corresponding helices of these DUBs. Despite the structural similarity, the sequence identity of D1a between LotA_M and hOTUB1 is rather low (20%). C303 locates at the N-terminal end of α1 in D1, and the catalytic cysteines of other OTU DUBs are present at the N-terminal end of the corresponding helices (Fig. 2; see Fig. S3 in the supplemental material). Therefore, D1 is deemed to be a catalytic subdomain of OTU2 (Fig. 2B). D1b is composed of α2, α3, and the C-terminal half of α1. A similar helical substructure exists in hOTUB1 (Fig. 2D) and a Legionella OTU-DUB, LotC (Fig. 3C), but not in other human and viral OTU DUBs (Fig. 2E and F), suggesting that D1b is not directly involved in the catalytic function of LotA. D2 is a helical subdomain attached on the C-terminal side of α4. This subdomain does not exist in human and viral OTU DUBs (Fig. 2D to F). D2 resembles the EHL domain identified in another Legionella OTU-DUB LotB (Fig. 3A and B; see Fig. S4 in the supplemental material) (27, 29). Thus, we assigned D2 as the EHL of LotA_M. LotC also has an EHL (29). However, the number and arrangement of the helices are different from those in the LotA_M and LotB EHLs (Fig. 3C; see Fig. S4 in the supplemental material). The loop connecting α6 and α7 in D2 contains short helical segments. One of the helical segments (P408 to D410) locates at the corresponding position to the short helix termed “helical arm” in the EHLs of LotB and LotC (28). The helical arm serves the primary-ubiquitin binding site (S1) in LotB and LotC (Fig. 3B and C) (29), suggesting that the helical segment provides the S1 site in LotA_M.

FIG 2.

Crystal structure of LotA_M. (A) Ribbon drawing of LotA_296–544 (LotA_M) is colored rainbow from the N terminus (blue) to the C terminus (red). (B) Ribbon drawing of LotA_M. D1, which contains the region conserved in OTU DUBs, and D2, which corresponds to the EHL, are colored green and orange, respectively. The two substructures of D1, D1a, and D1b, are colored in lime green and dark green, respectively. (C to F) Ribbon drawing of LotA_M (C), human OTUB1 (PDB ID: 4DDI) (D), human OTUD1 (PDB ID: 4BOP) (E), and Crimean-Congo hemorrhagic fever virus OTU DUB (PDB ID: 3PRM) (F). The conserved catalytic domains are rainbow colored. The side chains of the catalytic residues for the DUB activity are indicated by spheres (A, C to F). D1b is shown with a pink circle (C).

FIG 3.

Structural comparison of LotA_M with other Lot-DUBs. Ribbon drawing of LotA_M (A), LotB (PDB ID: 6KS5) (B), and LotC (PBD ID: 6YK8) (C). The conserved catalytic domains are rainbow colored. The side chains of the catalytic residues for the DUB activity are indicated by spheres. The regions containing the “helical arm” in the EHLs are shown with pink circles.

The interface between LotA_M and ubiquitin.

To identify possible ubiquitin-binding sites in LotA_M, we compared the structure of LotA_M with those of OTU-DUBs in complex with ubiquitin. We found 11 DUB/ubiquitin complex structures, three DUB/linear diubiquitin complex structures, and five DUB/ubiquitin/ubiquitin complex structures in the Protein Data Bank (PDB) (see Fig. S5 in the supplemental material). Among them, the structure of hOTUB1 (PDB ID: 4DDI) in the hOTUB1/UbcH5b ubiquitin/ubiquitin complex (one ubiquitin was covalently attached to an E2 [UbcH5b]) (33) is most similar to that of LotA_M (Fig. 4). The arrangement of the two ubiquitin molecules in the hOTUB1/UbcH5b∼ubiquitin/ubiquitin complex structure (Fig. 4E) is reminiscent of a K48-linked diubiquitin, which is the confirmed substrate of the hOTUB1 DUB (33). LotA_M cleaves K48- and K63-linked polyubiquitin chains, suggesting the presence of two ubiquitin-binding sites for distal and proximal ubiquitin. Therefore, we built a model of the LotA_M/ubiquitin/ubiquitin complex by superimposing the LotA_M structure onto the hOTUB1/ubiquitin/ubiquitin complex structure (Fig. 4A). Both ubiquitin molecules (proximal and distal ubiquitin) from the hOTUB1/ubiquitin/ubiquitin complex structure were able to be placed without serious collision with LotA_M. Then, we carried out energy minimization on the LotA/ubiquitin complex model. After converging the calculation, proximal ubiquitin drew apart from LotA_M, and therefore we were not able to identify the proximal ubiquitin-binding site in LotA_M. In contrast, distal ubiquitin significantly interacts with LotA_M mainly with the EHL (Fig. 4B). LotA_M recognizes the positively charged surface of ubiquitin composed of R42, R72, and R74 in the distal ubiquitin using the acidic surface composed of D407, D410, and D412 on and around the “helical arm” in the EHL (Fig. 4B and C, and 5A; see Fig. S3 in the supplemental material), which does not exist in hOTUB1. Instead, hOTUB1 binds the positively charged surface via the negatively charged surface formed by E195 and D216, and the corresponding surface is not present in LotA_M (Fig. 4F and G and 5B).

FIG 4.

The simulated structure models of the LotA_M/ubiquitin complex and the crystal structure of the human OTUB1/ubiquitin complex. (A) Ribbon drawing of the LotA_M with two ubiquitin molecules constructed by the superimposition of LotA_M on the human OTUB1/UbcH5b∼ubiquitin/ubiquitin complex structure. Green, structural elements of LotA_M conserved with human OTUB1; orange, LotA_M EHL; gray, the other structural elements of LotA_M; yellow, proximal ubiquitin; cyan, distal ubiquitin. (B) The complex structure after energy minimization. Proximal ubiquitin was omitted. (C, D) Magnified view of electrostatic (C) and hydrophobic (D) interaction sites between LotA_M EHL (orange) and distal ubiquitin (cyan). (E) Ribbon drawing of the human OTUB1/UbcH5b∼ubiquitin/ubiquitin complex (PDB ID: 4DDI). The UbcH5b moiety is omitted. Green and gray, human OTUB1; yellow, proximal ubiquitin; cyan, distal ubiquitin. The extended N-terminal helix (HA) and H2 and H3 helices of human OTUB1 are labeled (see the text). (F) Distribution of charged residues involved in the interaction between human OTUB1 and distal Ub. The view is the same as (E). Proximal ubiquitin is omitted. (G) Magnified view of the electrostatic interaction sites between human OTUB1 (green) and distal ubiquitin (cyan). (H) Magnified view of the hydrophobic interaction site between human OTUB1 (green and gray) and proximal ubiquitin (yellow). The catalytic cysteine and key residues for the interaction with ubiquitin are shown as spheres (B, F) or sticks (C, D, G, H).

FIG 5.

Comparison of the ubiquitin-interaction surfaces between LotA_M and OTUB1. The ubiquitin interaction surfaces of LotA_M (A) and human OTUB1 (B). Top, ribbon models; middle, electrostatic surfaces; bottom, hydrophobic surfaces. The electrostatic surfaces are colored blue and red for positive and negative charges, respectively. The hydrophobic surfaces are colored yellow according to the hydrophobicity. The conserved electrostatic or hydrophobic surfaces are shown in squares, and the hydrophobic surface specific for LotA_M/ubiquitin complex is shown in a circle.

Our model also suggests that the hydrophobic surface around V398 in the EHL contributes to the binding of LotA_M to the hydrophobic I44 patch of the distal ubiquitin (Fig. 4B and D and 5A). The C-terminal chain of the distal ubiquitin stretches into the cleft between the catalytic D1 and the D2 (EHL) domains in LotA_M and forms intermolecular hydrogen bonds with both domains. The C terminus of ubiquitin is in close proximity to the catalytic center. The distance between the Cα atom of G76 in ubiquitin and that of C303 in LotA_M is 3.7 Å (Fig. 4C). Thus, the distal ubiquitin is in an orientation capable of being cleaved by LotA_M in our model. R72 in the C-terminal chain of ubiquitin electrostatically interacts with D410 on the “helical arm” in the EHL (Fig. 4C). This interaction is likely to be a determinant for fixing the orientation of the C-terminal chain of ubiquitin suitable for the catalytic reaction. To address this possibility, we replaced D410 with arginine (D410R) and analyzed the DUB activity of the derivative. As expected, the DUB activity against K48- and K63-linked pentaubiquitin was almost abolished by the mutation (Fig. 6A and B). We further examined the significance of D410 in interaction with ubiquitin by utilizing ubiquitin propargylamide (Ub-PA) which can covalently bind to the catalytic residue of DUBs (34) (Fig. 6C). LotA C13S, but not LotA C13S C303S, formed a covalent bond with Ub-PA (Fig. 4C upper panels) as previously reported (26), showing that the presence of C303 is sufficient for making the Ub-PA conjugate. LotA_M also formed the covalent bond with Ub-PA, while LotA_M D410R abolished the interaction (Fig. 4C lower panels). These results indicate that the electrostatic interaction between the EHL and ubiquitin via D410 is essential for the LotA_M function and supports our structural model of the LotA_M/ubiquitin complex.

FIG 6.

D410 is crucial for interaction with ubiquitin and for the DUB activity of LotA_M. (A) K48‐linked pentaubiquitin (K48‐Ub5) and K63‐linked pentaubiquitin (K63‐Ub5) were reacted with 0.5 μM the purified LotA derivatives at 37°C for 20 min. The reaction products were analyzed by SDS-PAGE. Ubiquitin (Ub) and LotA derivatives were detected by immunoblotting using antiubiquitin (P4D1) (upper) and anti-LotA (lower) antibodies, respectively. (B) The intensity of Ub2 and Ub5 bands were quantified from three independent experiments shown in panel A. The Ub2/Ub5 ratio was represented as the DUB activity. (C) The suicide probe Ub-propargylamide (Ub-PA) was reacted with 1 μM the purified LotA derivatives at 23°C for 60 min. The reaction products were analyzed by SDS-PAGE followed by silver staining (upper) and by immunoblotting using antiubiquitin (P4D1) antibody (lower). Arrows indicate the proteins conjugated with the probe.

DISCUSSION

Functions of most cellular proteins are ingeniously controlled by the interplay between enzymes catalyzing ubiquitination and deubiquitination. Deubiquitination is mediated by various types of DUBs. Some pathogenic bacteria hijack the ubiquitin system by various effector proteins. LotA is one of those effector proteins of L. pneumophila and works as a DUB relying on its two OTU-like domains (26). OTU family DUBs of humans and bacteria generally target specific substrate proteins and exhibit cleaving activity for a limited type(s) of ubiquitin linkage. Our study demonstrated that LotA_M, a fragment composed of the second OTU domain in LotA, is solely responsible for the cleavage of K48- and K63-linked ubiquitin chains, suggesting that the first and second OTU domains work in a mechanistically independent manner.

To understand how LotA_M recognizes ubiquitin, we determined the crystal structure of LotA_M and compared it with the structures of the human OTU/ubiquitin complexes and with the structures of other Legionella OTUs. We found that LotA_M is composed of the catalytic domain (D1) and the Legionella OTU-specific EHL domain (D2). LotB has two distinct ubiquitin-binding sites, S1 and S1', which are primary and additional sites, respectively, to accept different ubiquitin chains, whereas LotC has only S1 (29). LotB and LotC utilize the EHL to bind ubiquitin in S1. We found that the LotA_M EHL plays a crucial role in ubiquitin binding and identified the possible S1 site in the EHL from the energy-minimized model of the LotA_M/ubiquitin complex (Fig. 4B to Fig. 4D).

In the hOTUB1/UbcH5b∼ubiquitin/ubiquitin complex structure, the hydrophobic I44 patch of proximal ubiquitin (conjugated with Ubc5H) mediates interaction with the extended N-terminal helix (HA) of hOTUB1 (33), which is similar to the interaction mediated by ubiquitin-interacting motifs (UIMs) in Vps27p (33, 35) (Fig. 4E and D). Proximal ubiquitin is also in contact with the loop connecting H2 and H3 helices of hOTUB1 on the β-sheet of the catalytic OTU domain of hOTUB1 (Fig. 4E) (33). However, LotA_M does not have the helices corresponding to HA, H2, and H3 (Fig. 4A). The EHL of LotA_M is close to the proximal ubiquitin, but there was no significant interaction between them in the model. Therefore, we were not able to identify the proximal ubiquitin-binding site in LotA_M.

Most OTUs contain a catalytic triad consisting of cysteine, histidine, and an acidic residue in the active site (see Fig. S3B in the supplemental material). Human OTUB1 has a triad of C91, H265, and D267 in the catalytic OTU domain (36). The structure of LotB has shown that C29, H270, and D21 are arranged appropriately to form a catalytic triad, but a biochemical assay has revealed that D21 is inessential for the reaction (27). LotA_M has a histidine residue, H535, near the catalytic C303 (Fig. 4C) but no acidic residues in close proximity to H535. Some cysteine proteases utilize serine or asparagine as a substitute for the acidic residue (37). S531 and N532 are located at a suitable position to form a catalytic triad with H535 and C303 in LotA_M, and either residue may align the histidine base during the reaction.

We found that LotA_M has the “helical arm” in the position corresponding to those identified in the EHLs of LotB and LotC (Fig. 3). D410 in the “helical arm” of LotA_M is essential for the DUB activity against K48- and K63-linked ubiquitin. D410 interacts with R72 in the C-terminal stretched chain of ubiquitin and may adjust the orientation of the C-terminal chain where the cleavage site is located. The previous study on LotC reported that E153 in the “helical arm” of LotC is important for the DUB activity of LotC (29). E153 is close to R74 of ubiquitin in the LotC-ubiquitin complex structure (32). Although the position of D410 in the “helical arm” is slightly different from that of E153, these two acidic residues may share a similar role in the adjustment of the orientation of the C-terminal chain of ubiquitin for the DUB activity.

The hydrophobic interaction plays an important role in ubiquitin binding to S1 in LotB and LotC. F143 and M144 in the LotB EHL and Y119 and Y149 in the LotC EHL were identified as key residues to form the hydrophobic interaction with ubiquitin. F143 and M144 of LotB were proposed to interact with F45 and A46 of ubiquitin, and Y119 and Y149 of LotC with the hydrophobic I44 patch of ubiquitin (29, 32). In the LotA_M/ubiquitin complex model, V398, which is present on the hydrophobic surface in the EHL, directory interacts with the hydrophobic I44 patch. The model suggested another hydrophobic interaction between D1a and ubiquitin. L492 in α10 interacts with L8 and V70 in ubiquitin (Fig. 5A). L8 and V70 interact with the hydrophobic surface composed of F1190, F1193, I1218, and I1221 in hOTUB1 (Fig. 5B). Interestingly, L492 locates at the same position as I218 in the corresponding helix.

Collectively, the role of the EHL in LotA_M is crucial for the proper positioning of ubiquitin for its catalytic activity. Although the EHL is also present in LotB and LotC, their structural alterations may define the role of the EHL for variable cleaving specificity of each DUB. A number of effector proteins having ubiquitin ligase or DUB activities have been identified in bacterial pathogens. Among them, LotA represents an unprecedented feature possessing two DUB activities with distinct catalytic preferences. Why the two DUB domains are encoded in a single polypeptide remains elusive. Future studies to identify the target proteins of LotA will clarify the significance of dual reactions which should be coordinately executed in the process of Legionella infection.

MATERIALS AND METHODS

Bacterial strains, media, and plasmids.

The bacterial strains and plasmids used in this study are listed in Table S2 in the supplemental material. E. coli cells were cultured in LB medium (Lennox; Nacalai tesque). If needed, ampicillin was added at final concentrations of 50 μg ml⁻¹. To construct the plasmids, PCR-amplified DNA fragments, by using the DNA primes listed in Table S3, were cloned into pET15b plasmid vector by using restriction enzymes (NdeI and BamHI; TaKaRa) and T4 DNA ligase (New England Biolabs). The point mutation in the plasmid was introduced by QuikChange site-directed mutagenesis (Agilent Technologies). Transformation of E. coli was performed using a standard heat shock method.

Purification of LotA proteins.

BL21 (DE3) cells carrying pNH1941, pNTI1, or pNTI2 were cultured in LB broth at 37°C. After reaching an optical density at 600 nm of 0.2 to 0.5, 0.5 mM IPTG was subsequently added to the culture, and the culture was prolonged for about 20 h at 18°C. Cells were collected by centrifugation (6,700 × g) and suspended in TN buffer (50 mM Tris-HCl [pH 8.0], 200 mM NaCl) containing cOmplete EDTA-free (Roche) and lysozyme (Wako). The cells were then disrupted by sonication and centrifuged at 20,000 × g for 10 min to remove cell debris. The supernatant was ultracentrifuged at 100,000 × g for 15 min. The supernatant was loaded into a HisTrap HP column (GE Healthcare) equilibrated with TN buffer, and the protein-bound column was washed with TN buffer containing 5 mM imidazole. The proteins were subsequently eluted using a linear gradient of imidazole up to 200 mM in TN buffer within 16 column volumes. The protein was further purified by size exclusion chromatography using a Superdex 200 10/300 GL or a Superdex 75 10/300 GL column (GE Healthcare). The expression and purity of the proteins were checked by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE).

To prepare the LotA_M sample for crystallization, purified LotA_NM was digested by trypsin at the LotA_NM/trypsin ratio of 5/1 (wt/wt) for 90 min at 27°C, and the products were separated by size exclusion chromatography. The fraction containing LotA_M was collected and concentrated using an Amicon Ultra 10 K device (Merck Millipore). The molecular masses of the digestion products were examined using Matrix assisted laser desorption time of flight (MALDI-TOF) mass spectrometer (Kratos AXIMA-Performance, Shimazu) and were compared with calculated masses of possible tryptic fragments.

The selenomethionine derivative protein was prepared from E. coli BL21-CodonPlus(DE3) RIL-X carrying pNTI1. The cells were cultured in SeMet minimal medium [0.1% (wt/vol) NH₄Cl, 0.3% (wt/vol) NH₂PO₄, 0.3% (wt/vol) Na₂HPO₄, 2% (wt/vol) glucose, 0.03% (wt/vol) MgSO₄, 0.001% (wt/vol) Fe_{2 (}SO₄)₃, 0.001% (wt/vol) thiamine, 0.005% (wt/vol) seleno-l-methionine]. Then, the LotA_M was purified in the same way as the native protein.

In vitro DUB assay.

K48‐linked pentaubiquitin (no. D1400) and K63‐linked pentaubiquitin (no. D2400) were purchased from UBPBio. LotA or its derivatives (10 pmol) were preincubated in 10 μl DUB activation buffer (25 mM Tris‐HCl [pH 7.5], 150 mM NaCl, 10 mM DTT) at 23°C for 10 min. The activated enzymes were reacted with 0.5 μg of the pentaubiquitin in 20 μl DUB buffer (50 mM Tris‐HCl [pH 7.5], 50 mM NaCl, 5 mM DTT) at 37°C for 20 min. The reaction was stopped by adding 20 μl 2×SDS sample buffer [125 mM Tris-HCl (pH 8.0), 4% (wt/vol) SDS, 20% (wt/vol) glycerol, 2% (vol/vol) 2-mercaptoethanol, ∼0.4% (wt/vol) Bromophenol Blue (Wako)] and heating the mixture at 95°C for 5 min. The samples were analyzed by SDS-PAGE followed by immunoblotting using antiubiquitin (P4D1) antibody (sc-8017, Santa Cruz) and anti-LotA antibody raised against LotA protein purified using an E. coli expression system. Intensity values of protein bands were quantified with ChemiDoc System Image labs software (Bio-Rad). Values represent the mean from three independent experiments with Student’s t tests.

Modification of DUBs using a suicide probe.

Ub-PA was purchased from UbiQ (UbiQ-057). The modification was performed as described previously (26). Briefly, the purified LotA proteins (20 pmol) were preincubated in 10 μl DUB activation buffer and reacted with 2 μg Ub-PA in 20 μl DUB buffer at 23°C for 60 min. The reaction was stopped by adding 20 μl 2× SDS sample buffer and heating the mixture at 95°C for 5 min. The samples were analyzed by SDS-PAGE followed by silver staining (Silver Stain MS kit, FUJIFILM Wako) and by immunoblotting using antiubiquitin (P4D1) antibody (sc-8017, Santa Cruz).

Crystallization and structure determination.

Crystallization was carried out using the sitting-drop vapor-diffusion method. Crystallization drops were prepared by mixing 0.5 μl of ca. 10 mg ml⁻¹ LotA_M solutions with 0.5 μl of the reservoir solutions. Initial screening was carried out using the following screening kits: Wizard Classic I and II, Wizard Cryo I and II (Rigaku), and Crystal Screen I and II (Hampton Research), and then the conditions were optimized. Crystals appeared within a few weeks. The crystals used for X-ray data collection were grown at 20°C from drops prepared by mixing 0.5 to 1 μl protein solution (ca. 10 mg ml⁻¹) in TN buffer with the equivalent volume of reservoir solution containing 39% (wt/vol) polyethylene glycol 200, 100 mM sodium acetate (pH 4.5) and 100 mM NaCl.

The X-ray diffraction data were collected at synchrotron beamline BL41XU in SPring-8 (Harima, Japan) with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal No. 2019B2550 and 2019B2551). Crystals were frozen in liquid nitrogen and mounted in nitrogen gas flow at 100 K for X-ray data collection. The diffraction data were processed with MOSFLM (38) and were scaled with AIMLESS (39). The statistics of the diffraction data are summarized in Table S1. The experimental phase was calculated using the SAD data of the selenomethionine derivative with a program Phenix (40). The atomic model was built with Coot (41) and refined to 1.97 Å resolution with Phenix (40). The final refinement R factor and the free R factor were 19.8% and 24.6%, respectively. The refinement statistics are summarized in Table S1.

Structural comparison and modeling.

OTU-like DUB proteins that show structural similarity to LotA_M were searched using the Dali server (http://ekhidna2.biocenter.helsinki.fi/dali/). After the superimposition of the structure of LotA_M onto that of the human OTUB1/UbcH5b∼ubiquitin/ubiquitin complex (PDB ID: 4DDI), an energy minimization calculation was performed using the relax application in Rosetta (42–45) to build the model of the LotA_M/ubiquitin complex.

Data availability.

The atomic coordinates have been deposited in Protein Data Bank (https://www.wwpdb.org/) (PDB ID code 7F9X).