Functional Characterization of Non-Ankyrin Repeat Domains of Orientia tsutsugamushi Ank Effectors Reveals Their Importance for Molecular Pathogenesis

ABSTRACT

Orientia tsutsugamushi is a genetically intractable obligate intracellular bacterium, causes scrub typhus, and has one of the largest known armamentariums of ankyrin repeat-containing effectors (Anks). Most have a C-terminal F-box presumed to interact with the SCF ubiquitin ligase complex primarily based on their ability to bind overexpressed Skp1. Whether all F-box-containing Anks bind endogenous SCF components and the F-box residues essential for such interactions has gone unexplored. Many O. tsutsugamushi Ank F-boxes occur as part of a PRANC (pox protein repeats of ankyrin-C-terminal) domain. Roles of the non-F-box portion of the PRANC and intervening sequence region (ISR) that links the ankyrin repeat and F-box/PRANC domains are unknown. The functional relevance of these effectors’ non-ankyrin repeat domains was investigated. The F-box was necessary for Flag-tagged versions of most F-box-containing Anks to precipitate endogenous Skp1, Cul1, and/or Rbx1, while the ISR and PRANC were dispensable. Ank toxicity in yeast was predominantly F-box dependent. Interrogations of Ank1, Ank5, and Ank6 established that L1, P2, E4, I9, and D17 of the F-box consensus are key for binding native SCF components and for Ank1 and Ank6 to inhibit NF-κB. The ISR is also essential for Ank1 and Ank6 to impair NF-κB. Ectopically expressed Ank1 and Ank6 lacking the ISR or having a mutagenized F-box incapable of binding SCF components performed as dominant-negative inhibitors to block O. tsutsugamushi NF-κB modulation. This study advances knowledge of O. tsutsugamushi Ank functional domains and offers an approach for validating their roles in infection.

INTRODUCTION

Orientia tsutsugamushi is an obligate intracellular bacterium that causes scrub typhus, a febrile infection with a case fatality rate that can be at least 50% if untreated (1). The disease has threatened human health for centuries, dating as far back as 313 CE. It is indigenous to a 13,000,000-km² area in the Asia Pacific known as the Tsutsugamushi Triangle, where it accounts for as much as 19% of all illnesses and 23% of febrile illnesses (1, 2). O. tsutsugamushi is vectored by trombiculid mites, which maintain the organism in nature transovarially and by transmission to rodent reservoirs. Humans are accidental hosts and play no role in the bacterium’s enzootic cycle (3). O. tsutsugamushi infects monocytes, macrophages, dendritic cells, and endothelial cells (4). The recognized geographic range of scrub typhus has been expanding for decades. Non-travel-related cases along with seroprevalence studies and molecular detection of Orientia species DNA in rodents and mites have confirmed the presence of the disease and etiologic agent in the Arabian Peninsula, several African countries, Peru, and Chile (2, 5–10). In addition to fever, scrub typhus patients present with headache, chills, arthralgias, myalgias, and often an eschar prior to the emergence of a maculopapular rash following disease onset. In fatal cases, scrub typhus results in multiorgan failure with lesions in the lungs, kidneys, liver, and brain (2).

The ankyrin repeat is one of the most common protein-protein interaction motifs (11). Acquisition of ankyrin repeat-containing effectors (Anks) is linked to the evolution of virulence and host specificity of intracellular microbes (12, 13). Over the course of its reductive evolution as an obligate intracellular inhabitant of arthropods and mammals, O. tsutsugamushi retained a large repertoire of Anks that are type I secretion system effectors and most of which carry additional eukaryotic-like domains (14–16). The O. tsutsugamushi Ikeda strain, a patient isolate that causes severe disease (17), has nine ank pseudogenes and 38 ank open reading frames (ORFs) (15, 16). The Ikeda ank cohort is the fifth largest of nearly 2,000 bacterial genomes examined (11). The 38 ORFs consist of 11 single-copy and 8 multiple identical or nearly identical copies, equaling 19 distinguishable Anks. Each carries one to nine tandemly arranged N-terminal ankyrin repeats. O. tsutsugamushi strain Ikeda transcribes all ank genes during infection of mammalian cells (16). Translational expression of several Anks has been verified in scrub typhus patients and infected tissue culture cells (18–20). Thus far, functional roles have been ascribed to a small number of Anks based on their ability to recapitulate infection-associated phenomena when ectopically expressed. These include blocking nuclear accumulation of the proinflammatory transcriptional activator NF-κB (Ikeda Ank1 and Ank6), globally altering host cell transcription to counter the antimicrobial response (Ikeda Ank13), inhibiting the secretory pathway (Ikeda Ank9), and reducing cellular levels of elongation factor 1α (O. tsutsugamushi Boryong strain Ank1U5) (18–21). Hence, O. tsutsugamushi Anks are an effector family that establishes a myriad of incompletely characterized structure-function relationships with host cells to mediate scrub typhus molecular pathogenesis.

While the Ank ankyrin repeat regions facilitate protein-protein interactions, domains that lie C-terminal to this region are also important. The abilities of Ank1, Ank6, Ank9, and Ank13 to impair their target host cell pathways are linked to a C-terminal domain that is homologous to the eukaryotic F-box motif that interacts with S-phase kinase-associated protein 1 (Skp1) and cullin 1 (Cul1) of the SCF (Skp1-Cul1-F-box) E3 ubiquitin ligase (18, 19, 21, 22). This quaternary structure is comprised of Skp1, Cul1, RING (really interesting new gene) box protein 1 (Rbx1), and an F-box-containing protein (FBP) (22, 23). The SCF complex functions as follows. A eukaryotic FBP binds its target protein via a protein-protein interaction domain, while its F-box motif recruits Cul1 and Skp1. Cul1 binds Skp1 and Rbx1 to form the catalytic core that polyubiquitinates the bound substrate, destining it for 26S proteasomal degradation (22). Many intracellular bacterial pathogens encode FBPs that are critical for microbial fitness and virulence (24, 25). Our previous study found that 16 of the 19 O. tsutsugamushi Ikeda Anks carry putative C-terminal F-boxes, 10 of which were identified using the SMART (Simple Modular Architecture Research Tool) algorithm. The remaining six were detected manually (26). Notably, the 10 identified by SMART exist as part of a larger PRANC (pox protein repeats of ankyrin-C-terminal) domain (26), so named due to its discovery at the C termini of poxviral Anks (27, 28). The PRANC domain’s F-box portion mediates SCF complex nucleation and directs polyubiquitination/proteasomal degradation of host cell protein targets during poxviral infection (12, 29). However, the role of its non-F-box portion has not been investigated. The intervening sequence region (ISR) that lies between the ankyrin repeat and PRANC/F-box domains of O. tsutsugamushi Anks has been functionally assessed only for Ank1 and Ank6. Deletion of the ISR from either protein ablates the ability to impede NF-κB nuclear accumulation (21).

Despite these advances in understanding O. tsutsugamushi Ank function, several pressing questions pertaining to the contributions of the ISR, PRANC, and F-box domains persist. For instance, most of what is known about O. tsutsugamushi strain Ikeda Ank interactions with the SCF complex is based on pulldowns performed using overexpressed glutathione S-transferase (GST)-tagged Skp1 (26). Whether all the Anks are capable of binding endogenous Skp1, Cul1, and Rbx1 is unknown. Likewise, the F-box residues that are required for such interactions and that modulate host cell processes are undefined. The relevance of the non-F-box portion of the PRANC domain and remaining C-terminal amino acids to Ank function and to SCF complex interactions is unexplored. It is unclear if the abilities of F-box-containing Anks to modulate host cell pathways is universally F-box dependent. Finally, whether the manually identified F-boxes are indeed competent for nucleating the endogenous SCF complex is unestablished. Here, we sought to fill these knowledge gaps. Due to the genetic intractability of O. tsutsugamushi, we assessed ectopically expressed Anks and mutant versions thereof, performed phylogenetic analyses, and employed yeast toxicity screening to evaluate the relevance of the ISR, PRANC, F-box, and critical residues thereof to Ank function. Building on our findings, we used ectopically expressed functionally defective versions of Ank1 and Ank6 as dominant-negative competitors of their bacterially derived counterparts to validate their roles in infection. This report expands knowledge of O. tsutsugamushi Ank functional domains and establishes a strategy for confirming the roles of Anks and potentially other effectors in O. tsutsugamushi infection.

RESULTS

Deletion of the PRANC domain ablates the abilities of Ank1 and Ank6 to impede NF-κB nuclear accumulation.

Pertinent information for the 11 single-copy and each selected representative paralog of multicopy Anks is provided in Table 1. As a starting point for examining the functional relevance of the linear domains of O. tsutsugamushi F-box-containing Anks, we initially focused on Ank1 and Ank6. Thus far, both effectors had been shown to inhibit tumor necrosis factor alpha (TNF-α)-stimulated NF-κB p65 accumulation in the nucleus in F-box- and ISR-dependent manners when ectopically expressed (21). Contributions of each effector’s PRANC domain and remaining C-terminal residues had not yet been investigated. HeLa cells expressing Flag-tagged Ank1, Ank6, or deletion mutants thereof lacking the ISR (ΔISR), F-box (ΔF-box), PRANC (ΔPRANC), or residues that are C-terminal of the PRANC domain (ΔCterm) (Fig. 1) were exposed to TNF-α or vehicle alone and examined by confocal microscopy. Please note that in this study, so as to distinguish the functional contributions of the F-box and PRANC domains, the term “PRANC domain” corresponds to only the portion of each Ank that shares homology with poxviral ankyrin repeat-containing proteins but excludes the F-box motif (Fig. 1). Flag-tagged bacterial alkaline phosphatase (BAP), which exerts no inhibitory effect on NF-κB, and the IκBα super repressor (IκB αSR), which has Ser32 and Ser36 replaced with Ala to prevent its phosphorylation and degradation upon TNF-α exposure and thereby constitutively retains NF-κB in the cytoplasm (30), served as negative and positive controls, respectively. In the absence of TNF-α, p65 immunosignal was observed in the nuclei of ≤10% of transfected cells, confirming that merely ectopically expressing each Flag-tagged fusion is insufficient to stimulate the NF-κB response (Fig. 2 and 3). As previously observed, TNF-α promoted the accumulation of p65 in the nuclei of greater than 80% of cells expressing Flag-BAP and less than 10% of cells expressing Flag-tagged IκBα SR, Ank1, and Ank6 (21). As also reported before (21), Flag-Ank1 and Flag-Ank6 proteins lacking the ISR or F-box could not inhibit TNF-α-induced NF-κB nuclear accumulation. Flag-Ank1ΔPRANC and Flag-Ank6ΔPRANC were similarly incapable of impairing the NF-κB translocation into the nucleus, while Flag-tagged Ank1ΔCterm and Ank6ΔCterm proteins exhibited NF-κB-impairing phenotypes comparable to their wild-type counterparts. Thus, deletion of the PRANC domain, just like deletion of the F-box or ISR, but not deletion of the C terminus eliminates the abilities of Ank1 and Ank6 to block NF-κB p65 nuclear accumulation.

TABLE 1.

O. tsutsugamushi Ikeda strain Anks examined in these studies

Name	GenBank accession no.	PRANC	Predicted C-terminal F-box
Ank1_02a	OTT_RS03615	Yes	Yes
Ank2	OTT_RS00235	Yes	Yes
Ank3_08	OTT_RS05410	No	No
Ank4_01	OTT_RS00980	No	Yes
Ank5_01	OTT_RS01000	Yes	Yes
Ank6_02	OTT_RS05585	Yes	Yes
Ank7_02	OTT_RS09885	No	No
Ank8	OTT_RS01225	Yes	Yes
Ank9	OTT_RS01425	Yes	Yes
Ank10_01	OTT_RS01880	Yes	Yes
Ank11	OTT_RS02210	No	No
Ank12_01	OTT_RS02915	Yes	Yes
Ank13	OTT_RS04140	Yes	Yes
Ank14	OTT_RS04940	No	Yes
Ank15	OTT_RS05980	No	Yes
Ank16	OTT_RS06155	No	Yes
Ank17	OTT_RS07150	No	Yes
Ank19	OTT_RS07355	No	No
Ank20	OTT_RS07600	Yes	Yes

^aFor Anks that are followed by an underscore and number, the number indicates the representative paralog of a multicopy paralogous family. Hereafter, the paralog indicator will be excluded. Ank18 is excluded because it does not contain ankyrin repeats, a PRANC, or F-box domain.

FIG 1.

Schematics of Ank1, Ank6, and mutants thereof. Schematics of full-length wild-type Ank1 (A) and Ank6 (B) are presented. The four ankyrin repeats (ARs; blue arrows), ISR (orange box), PRANC domain (gray box), F-box (Fb; green box), and C-terminal region (C-term; white box), are indicated, as are the amino acids that encompass each region. Below the wild-type Ank schematics are schematics in which deleted (Δ) domains and corresponding domains are demarcated. The relative positions of Ank1 L285, P286, E288, I293, and D301 (A) and Ank6 L287, P288, E290, I295, and D303 (B), which correspond to L1, P2, E4, I9, and D17 of the O. tsutsugamushi Ank F-box consensus sequence, are denoted by yellow lines. These amino acids were replaced with A to generate indicated mutants.

FIG 2.

Deletion of the Ank1 ISR, F-box, or PRANC domain, but not the C terminus, abrogates its ability to inhibit NF-κB p65 nuclear accumulation. HeLa cells were transfected to express Flag-tagged BAP, Ank1, Ank1ΔISR, Ank1ΔF-box, Ank1ΔPRANC, Ank1ΔC-terminus (C-term), or IκBα SR. At 16 h, the cells were exposed to TNF-α or vehicle control for 30 min, after which they were fixed, screened with antibodies specific for the Flag epitope and p65, and examined by confocal microscopy. (A) Representative fluorescence images of cells viewed for Flag-tagged protein, p65, and merged images plus DAPI. (B) The mean ± SD percentage of transfected cells exhibiting p65 in the nucleus was determined. Triplicate samples of 100 cells each were counted per condition. Horizontal bars denote pairs for which statistical significance was assessed. Values that are statistically significantly different (****, P < 0.0001) are indicated. ns, not significant. Data are representative of three separate experiments with similar results.

FIG 3.

Deletion of the Ank6 ISR, F-box, or PRANC domain, but not the C terminus, ablates its ability to inhibit NF-κB p65 nuclear accumulation. HeLa cells were transfected to express Flag-tagged BAP, Ank6, Ank6ΔISR, Ank6ΔF-box, Ank6ΔPRANC, Ank6ΔC-terminus (C-term), or IκBα SR. At 16 h, the cells were exposed to TNF-α or vehicle control for 30 min, after which they were fixed, screened with antibodies specific for the Flag epitope and p65, and examined by confocal microscopy. (A) Representative fluorescence images of cells viewed for Flag-tagged protein, p65, and merged images plus DAPI. (B) The mean ± SD percentage of transfected cells exhibiting p65 in the nucleus was determined. Triplicate samples of 100 cells each were counted per condition. Values that are statistically significantly different (****, P < 0.0001) are indicated. ns, not significant. Data are representative of three separate experiments with similar results.

Deletion of the ISR, PRANC, and F-box domains of Ank1 and Ank6 differentially impairs their abilities to interact with endogenous SCF complex components.

Our prior report (21) and data presented here demonstrate that Ank1 and Ank6 retention of p65 in the cytoplasm is F-box dependent, which indirectly suggests that each effector’s efficacy in modulating NF-κB is linked to its ability to nucleate the SCF E3 ubiquitin ligase. We also previously demonstrated that ectopically expressed versions of all Anks carrying a SMART or manually predicted F-box precipitate coexpressed GST-tagged Skp1 (26). The abilities of these Anks to interact with endogenous Skp1, Cul1, and Rbx1 were unknown. Moreover, whether deleting the ISR or PRANC domain of an Ank alters its ability to interact with SCF components had yet to be investigated. HeLa cells were transfected to express Flag-tagged BAP; Ank1 or Ank6; ΔISR, ΔPRANC, and ΔF-box mutants thereof; or deletion mutants thereof lacking the PRANC through the C terminus (ΔP-Cterm) (Fig. 1). Lysates were collected and incubated with Flag antibody-coated beads to immunoprecipitate the Flag-tagged proteins and their interacting partners. Western-blotted input lysates and eluted complexes were screened with Skp1, Cul1, and Rbx1 antibodies. Flag-Ank1, -Ank1ΔISR, -Ank6, and -Ank6ΔISR precipitated all three SCF proteins (Fig. 4). Flag-tagged BAP and Ank1 or Ank6 proteins lacking the F-box, PRANC, or PRANC through the C terminus did not. Hence, even though the ISR is required for these effectors’ abilities to antagonize the NF-κB pathway and exhibits the highest degree of conservation between Ank1 and Ank6 relative to the other O. tsutsugamushi Anks (see Fig. S1 in the supplemental material) (21), it is dispensable for interacting with the SCF complex. Notably, Ank1 and Ank6 lacking either the F-box or PRANC domain failed to precipitate Skp1, Cul1, and Rbx1 (Fig. 4), which begs the question of whether the PRANC domain directly interacts with the SCF complex or if its absence indirectly impairs the ability of the F-box to do so.

FIG 4.

Deletion of the ISR, PRANC, and F-box domains differentially impairs the abilities of Ank1 and Ank6 to interact with endogenous SCF complex components. HeLa cells were transfected to express Flag-tagged BAP, Ank1, Ank1ΔISR, Ank1ΔPRANC, Ank1 lacking the PRANC domain through the C terminus (Ank1ΔP-Cterm), or Ank1ΔF-box (A) or Ank6, Ank6ΔISR, Ank6ΔPRANC, Ank6ΔP-Cterm, or Ank6ΔF-box (B). Input lysates were subjected to Western blotting with Flag antibody to verify ectopic expression of the proteins of interest; antibodies against Skp1, Cul1, and Rbx1 to confirm their presence; and GAPDH antibody to validate that comparable amounts of protein were present in each sample. Whole-cell lysates were incubated with Flag antibody-conjugated agarose beads to immunoprecipitate (IP) Flag-tagged proteins and their interacting proteins. The resulting Western blot was probed with indicated antibodies to confirm recovery of the Flag-tagged proteins and assess for Skp1, Cul1, or Rbx1 coimmunoprecipitation. Data are representative of three or four experiments with similar results.

The F-box, but not the PRANC domain of O. tsutsugamushi Anks, is necessary and sufficient for interacting with the SCF complex.

To resolve if the PRANC domain contributes to interactions with SCF complex components, we assessed the abilities of all PRANC-containing Anks aside from Ank1 and Ank6 versus ΔF-box versions thereof to coimmunoprecipitate endogenous Skp1, Cul1, and Rbx1. As observed for Ank1 and Ank6, Flag-tagged Ank2, Ank5, Ank8, Ank9, Ank10, Ank12, Ank13, and Ank20 pulled down Skp1, Cul1, and Rbx1 but failed to do so when the F-box was deleted and PRANC domain retained (Fig. 5A and Table 2). Thus, deletion of the F-box from all PRANC-containing Anks is sufficient to eliminate SCF complex interactions. Next, Flag-tagged Ank4, Ank14, Ank15, Ank16, and Ank17, all of which naturally lack the PRANC domain but have putative C-terminal F-boxes that had been manually identified instead of being delineated via the SMART algorithm (26), were compared alongside their ΔF-box counterparts for endogenous SCF component precipitation. Flag-Ank17 weakly pulled down Skp1 but robustly precipitated Cul1 and Rbx1 (Fig. 5B and Table 2). Flag-Ank4 weakly precipitated Skp1 but neither Cul1 nor Rbx1. Flag-tagged Ank14, Ank15, and Ank16 were unable to pull down any SCF component. Hence, the F-box but not PRANC domain is necessary and sufficient for O. tsutsugamushi Anks to nucleate the SCF complex. These results also establish that the predicted Ank14, Ank15, and Ank16 F-box sequences are nonfunctional for precipitating Skp1, Cul1, and Rbx1 and that Ank4 forms a weak interaction with only Skp1 (Table 2). In sum, 12 of the 19 O. tsutsugamushi Ikeda strain Anks (Ank1, Ank2, Ank4, Ank5, Ank6, Ank8, Ank9, Ank10, Ank12, Ank13, Ank17, and Ank20) have a functionally confirmed F-box.

FIG 5.

F-box, but not the PRANC domain of O. tsutsugamushi Anks, is necessary and sufficient for interacting with the SCF complex. HeLa cells were transfected to express Flag-tagged BAP (A and B), the indicated Anks that have a PRANC domain (A) or naturally lack a PRANC domain (B), or ΔF-box versions thereof (A and B). Input lysates were subjected to Western blotting with Flag antibody to verify ectopic expression of the proteins of interest; antibodies against Skp1, Cul1, and Rbx1 to confirm their presence; and GAPDH antibody to validate that equivalent amounts of protein were present in each sample. Whole-cell lysates were incubated with Flag antibody-conjugated agarose beads to immunoprecipitate (IP) Flag-tagged proteins and their interacting proteins. The resulting Western blot was probed with the indicated antibodies to confirm recovery of the Flag-tagged proteins and assess for Skp1, Cul1, or Rbx1 coimmunoprecipitation. Data are representative of three to six experiments with similar results.

TABLE 2.

Ank F-box annotation, SCF component binding, toxicity in yeast, and FBP designation

Name	C-terminal F-box prediction method	Endogenous SCF component binding	Functional F-boxa	Yeast toxicity		Revised FBP designation
				Full length	ΔF-box
Ank1 b	SMARTc	Skp1, Cul1, Rbx1	Yes	+++d	+++	FBP
Ank2	SMART	Skp1, Cul1, Rbx1	Yes	+++	+	FBP
Ank3	None	NAe	NA	−	NA	Not an FBP
Ank4	Manual	Skp1	Yes	+	−	FBP
Ank5	SMART	Skp1, Cul1, Rbx1	Yes	+++	−	FBP
Ank6	SMART	Skp1, Cul1, Rbx1	Yes	−	−	FBP
Ank7	None	NA	NA	−	NA	Not an FBP
Ank8	SMART	Skp1, Cul1, Rbx1	Yes	+++	−	FBP
Ank9	SMART	Skp1, Cul1, Rbx1	Yes	++++	+++	FBP
Ank10	SMART	Skp1, Cul1, Rbx1	Yes	+++	−	FBP
Ank11	None	NA	NA	−	NA	Not an FBP
Ank12	SMART	Skp1, Cul1, Rbx1	Yes	+++	++	FBP
Ank13	SMART	Skp1, Cul1, Rbx1	Yes	++++	−	FBP
Ank14	Manual	No	No	+	−	Inconclusivef
Ank15	Manual	No	No	−	−	Not an FBP
Ank16	Manual	No	No	+++	−	Inconclusive
Ank17	Manual	Skp1, Cul1, Rbx1	Yes	+	−	FBP
Ank19	None	NA	NA	−	NA	Not an FBP
Ank20	SMART	Skp1, Cul1, Rbx1	Yes	+	−	FBP

^aF-box functionality based on the ability to coimmunoprecipitate Skp1, Cul1, and/or Rbx1.

^bAnks functionally confirmed to interact with Skp1, Cul1, and/or Rbx1 are in boldface.

^cDetermined by Beyer et al. (26).

^d++++, very strongly toxic. +++, strongly toxic. ++, intermediate toxicity. +, mildly toxic. −, nontoxic.

^eNA, not applicable.

^fAnk14 and Ank16 F-box functionality is inconclusive because although neither coimmunoprecipitate SCF complex components, the toxicity of both in yeast is F-box dependent.

The ISR and PRANC domains potentially influence structure-function relationships of O. tsutsugamushi Anks.

To correlate if the differential abilities of Ank1 and Ank6 proteins having the ISR or PRANC deleted to bind the SCF complex and modulate NF-κB could be due to changes in overall protein structure, the sequences of Ank1, Ank6, and their ΔISR and ΔPRANC counterparts were analyzed by AlphaFold (31, 32). Ank1ΔF-box and Ank6ΔF-box were included as controls for having truncations anticipated to not disrupt tertiary structure. The resulting Ank1 and Ank6 models were consistent with typical ankyrin repeat domain architecture: their four tandemly arrayed ankyrin repeats form exposed concave surfaces that present amino acids 2, 3, 5, 12, 13, and 14 of each repeat (Fig. 6A and C). Residues at these positions are well-established as being responsible for mediating Ank-target protein interactions (33, 34). The Ank1 and Ank6 F-boxes are predicted to form helices that adopt a fold archetypal of the F-boxes of human Skp2 and Legionella pneumophila AnkB crystal structures (23, 35). Furthermore, Ank1 and Ank6 F-box residues that correspond positionally to those of the Skp2, AnkB, and O. tsutsugamushi Ank9 F-boxes that bind the SCF complex (23, 26, 35, 36) are predicted to be accessible. These AlphaFold predictions agree with our working model that Ank1 and Ank6 each recruit the SCF complex via the F-box and an unknown host protein involved in the NF-κB pathway via the ankyrin repeat domain to negatively modulate NF-κB.

FIG 6.

Predicted structures for Ank1, Ank6, and ΔF-box versions thereof. AlphaFold was used to predict models for Ank1 (A), Ank1ΔF-box (B), Ank6 (C), and Ank6ΔF-box (D). Ankyrin repeats are colored blue. The ISR, PRANC, and F-box domains are colored orange, dark gray, and green, respectively. Residues at positions within each ankyrin repeat and the F-box that mediate protein-protein interactions are colored purple.

As expected, removing the F-box through the C terminus is not predicted to structurally disrupt Ank1 or Ank6 (Fig. 6B and D). However, deleting the ISR putatively changes their tertiary conformations. Indeed, the F-box and ankyrin repeat residues of Ank1ΔISR that mediate protein-protein interactions remain accessible, but their altered orientations relative to each other could prevent the F-box-recruited SCF E3 ubiquitin ligase complex from ubiquitinating an ankyrin repeat-bound target (Fig. 7A and B). Along with the possibility that the ISR of Ank1 and Ank6 each contributes to target specificity, this scenario offers a functional explanation for why Ank1ΔISR retains the ability to bind the SCF complex but cannot inhibit NF-κB nuclear accumulation. Similarly, the inability of Ank6ΔISR to regulate NF-κB could be because its F-box and ankyrin repeat motif project inward, facing each other (Fig. S2A and B), which might result in the docked SCF complex occluding the ankyrin repeat domain from binding its target. The F-box domains of both Ank1ΔPRANC and Ank6ΔPRANC are predicted to be more closely juxtaposed to the ankyrin repeat domain interface and, in the case of Ank1ΔPRANC, are at least partially occluded by a flanking helix (Fig. 7C and D and Fig. S2C and D). These models are consistent with removal of the PRANC domain from Ank1 and Ank6, nullifying their interactions with the SCF complex and, consequently, their abilities to modulate NF-κB.

FIG 7.

Deletion of the ISR or PRANC domain is predicted to alter the tertiary structure of Ank1. Models generated for Ank1 (A and C), Ank1ΔISR (B), and AnkΔ1PRANC (D) using AlphaFold are presented. Ankyrin repeats are colored blue while the ISR, PRANC, and F-box domains are colored orange, dark gray, and green, respectively. Residues at positions within each ankyrin repeat and the F-box that mediate protein-protein interactions are colored purple.

Next, AlphaFold was used to examine representative Anks that naturally lack the PRANC domain and are either capable (Ank17) or incapable (Ank14 and Ank16) of precipitating SCF components. These three were also selected because they have numbers of ankyrin repeats (Ank14, five; Ank16, three; Ank17, four) similar to those of Ank1 and Ank6 (both four). Although the predicted conformations of Ank14, Ank16, and Ank17 differ from each other, the ankyrin repeat and F-box domains of each are arranged such that they are available to mediate protein-protein interactions (Fig. S3). Based on these models, orientation of the F-box does not underlie the differential abilities of Ank17 versus Ank14 and Ank16 to nucleate the SCF complex. Overall, these data represent the first step toward understanding how F-box accessibility, which is likely structurally altered in Ank1ΔPRANC and Ank6ΔPRANC, influences the structure-function relationships of F-box-containing O. tsutsugamushi Anks.

O. tsutsugamushi Ank toxicity in yeast is predominantly F-box dependent.

The few O. tsutsugamushi Anks that have been characterized thus far negatively regulate their target host cell processes in an F-box-dependent manner (18, 19, 21). We recently validated this phenomenon for Ank13 using a yeast toxicity screen (18), a proven model for studying bacterial effectors due to the conservation of cellular processes between mammalian cells and Saccharomyces cerevisiae (37–44). We examined all 19 O. tsutsugamushi Ikeda strain Anks as well as ΔF-box versions of each that had either a SMART- or manually identified C-terminal F-box motif. Each ORF was inserted into the galactose-inducible vector pYesNTA-Kan. S. cerevisiae W303 transformants were grown in uracil dropout media, serially diluted 10-fold, and spotted onto dropout agar containing 2% glucose or 2% galactose. Yeast carrying empty vector was a negative control. Ank13 and Chlamydia trachomatis CT694 served as positive controls for toxic effectors (18, 43). The Ank13ΔF-box was a positive control for F-box-dependent toxicity (18). As expected, CT694 and Ank13 nearly ablated yeast growth while toxicity was nullified in yeast expressing Ank13ΔF-box (Fig. 8 and Table 2). Of the 12 Anks that have functionally confirmed F-boxes, 11 inhibited yeast growth, 8 of which exhibited strong or very strong toxicity. The only ones that were mildly toxic were Ank4, Ank17, and Ank20. The mild toxicity of Ank4 is notable given that it is also the only Ank that weakly binds Skp1 and is incapable of interacting with Cul1 and Rbx1 (Fig. 5B). Deleting the F-box reduced toxicity approximately 10- to 100-fold for Ank2, Ank4, Ank5, Ank8, Ank9, Ank10, Ank12, Ank14, Ank16, Ank17, and Ank20 (Fig. 8 and Table 2). Ank1 was the only F-box-containing effector that displayed F-box-independent toxicity. While Ank6 carries a functional F-box, it does not interfere with yeast growth. Ank3, Ank7, Ank11, and Ank19, which lack an F-box, and Ank15, which carries a manually predicted F-box-like sequence that was functionally invalidated in this study, were also nontoxic. Ank14 carries a manually predicted F-box that fails to interact with Skp1, Cul1, and Rbx1 (Fig. 5B), but its mild toxicity in yeast is F-box dependent (Fig. 8 and Table 2). Likewise, although Ank16 carries an F-box sequence that fails to precipitate the SCF complex (Fig. 5B), it exhibits strong toxicity in yeast that is ablated when this sequence is removed (Fig. 8 and Table 2). Overall, these data indicate that the toxicity of most O. tsutsugamushi Anks is linked to the ability to interact with Skp1, Cul1, and Rbx1.

FIG 8.

O. tsutsugamushi Ank toxicity in yeast is primarily F-box-dependent. S. cerevisiae W303 was transformed with pYesNTA-Kan constructs for expressing O. tsutsugamushi Anks, ΔF-box versions of all F-box containing Anks, C. trachomatis CT694, or vector alone. Transformants were diluted to an optical density at 600 nm of 0.2 and spotted as 10-fold serial dilutions onto dropout media containing 2% glucose (noninducing conditions) or 2% galactose (inducing conditions). Data are representative of two experiments with similar results.

Identification of O. tsutsugamushi Ank F-box residues essential for interacting with endogenous SCF components.

Aligning the F-boxes of each Ank, including those defective for SCF component binding, yielded ₁LPXEXXXXILXXLXXXDLXXX₂₁ as the consensus sequence (Fig. S4A). We sought to determine the contributions of L1, P2, E4, I9, and D17 to binding native Skp1, Cul1, and Rbx1 using alanine substitution. The rationale for targeting these five residues was based on precedents established by us and other researchers. We had found that replacing F-box consensus residues L1, I9, and E17 (occupies the same position as D17 in the consensus) of GFP-Ank9 with alanine eliminates its ability to precipitate GST-Skp1 (26). A study by Price et al. demonstrated that alanine replacement of L9 and P10 of the L. pneumophila AnkB F-box, which corresponds to L1 and P2 of the O. tsutsugamushi F-box consensus, abolishes AnkB binding to Skp1 and recruitment of polyubiquitinated proteins (36). The crystal structure of the Cul1-Rbx1-Skp1-F-box^Skp2 quaternary complex determined by Zheng et al. revealed that P113 and E115 of Skp2, which correlate with P2 and E4 of the O. tsutsugamushi F-box consensus and are conserved among most eukaryotic F-box proteins (45), interact with Cul1 (23). Finally, the Anks that do not or poorly interact with the SCF complex (Ank4, Ank14, Ank15, and Ank16) have differences in one or two of the five F-box residues of interest (Fig. S4A).

For these interrogations, we focused on Ank1, Ank5, and Ank6 because their F-boxes are more closely related to each other than to those of the other 12 (Fig. S4B), and functional roles have been ascribed to Ank1 and Ank6 (21). Residues 1, 2, 4, 9, and 17 of the F-box consensus sequence are identical among Ank1, Ank5, and Ank6. Their actual amino acid positions are presented in Fig. 9A. Flag-tagged versions of these three effectors bearing alanine substitutions of F-box consensus residues L1, I9, and D17 (F-boxAAA); L1, P2, I9, and D17 (F-boxAAAA); and L1, P2, E4, I9, and D17 (F-boxAAAAA) (Fig. 9B) were assessed alongside their wild-type and ΔF-box counterparts for the ability to coimmunoprecipitate endogenous Skp1, Cul1, and Rbx1. Consistent with the known contributions of L1, P2, I9, and D17 to Skp1 binding (24, 26), all three alanine substitution mutants of Ank1, Ank5, and Ank6 were unable to precipitate Skp1, similar to the ΔF-box controls. In agreement with P2 and E4 in facilitating F-box interactions with Cul1 (23), the F-boxAAA versions of Ank1, Ank5, and Ank6 retained Cul1 and Rbx1 binding. However, these interactions were not as robust as those observed for the wild-type proteins (Fig. 10). Per the roles of P2 and E4 in binding Cul1, Flag-Ank1-F-boxAAAA and Flag-Ank5-F-boxAAAA poorly precipitated Cul1 and Rbx1 (Fig. 10A and B). Moreover, the abilities of Flag-Ank1-F-boxAAAAA and Flag-Ank5-F-boxAAAAA to pull down Cul1 were ablated and nearly so for Rbx1 (Fig. 10A and B). The same trends were observed for Flag-Ank6-F-boxAAAA and Flag-Ank6-F-boxAAAAA, except that each retained the ability to precipitate Cul1 and Rbx1, albeit inefficiently (Fig. 10C).

FIG 9.

Mutations made to the F-boxes of Flag-tagged Ank1, Ank5, and Ank6. (A) Alignment of Ank1, Ank5, and Ank6 F-box sequences. Asterisks demarcate the five residues that exhibit the highest conservation among all Anks and were targeted for alanine substitution. Identical residues are denoted by white text and black highlighting. (B) Schematics of the three alanine substitution mutants made in the F-box of Ank1, Ank5, and Ank6. F-boxAAA proteins have A substituted for L1, I9, and D17 in the F-box consensus. F-boxAAAA proteins have A substituted for L1, P2, I9, and D17 in the F-box consensus. F-boxAAAAA proteins have A substituted for L1, P2, E4, I9, and D17 in the F-box consensus. Residues replaced by A are indicated by gray highlighting. (C) Schematics of Ank6-F-boxAAAAA-F293M, Ank6-F-boxAAAAA-D300S, Ank6-F-boxAAAAA-D302T, Ank6-F-boxAAAAA-DD300, 302ST, and Ank6-F-boxAAAAA-FDD293, 300, 302MST. Substituted amino acids are indicated by gray highlighting.

FIG 10.

L1, P2, E4, I9, and D17 of the F-box consensus are essential for Ank1, Ank5, and Ank6 to each interact with endogenous SCF components. HeLa cells were transfected to express Flag-tagged BAP, Ank1 (A), Ank5 (B), Ank6 (C), or the indicated mutants thereof bearing alanine substitutions in F-box consensus residues L1, I9, and D17 (F-boxAAA); L1, P2, I9, and D17 (F-boxAAAA); or L1, P2, E4, I9, and D17 (F-boxAAAAA). Input lysates were subjected to Western blotting with Flag antibody to confirm ectopic expression of the proteins of interest; antibodies targeting Skp1, Cul1, and Rbx1 to confirm their presence; and GAPDH antibody to verify that equal amounts of protein were present in each sample. Whole-cell lysates were incubated with Flag antibody-conjugated agarose beads to immunoprecipitate (IP) Flag-tagged proteins and interacting proteins. The resulting Western blot was screened with indicated antibodies to confirm pulldown of the Flag-tagged proteins and assess for Skp1, Cul1, or Rbx1 coimmunoprecipitation. Data are representative of at least five experiments with similar results.

The Ank1 and Ank5 F-boxes have M, S, and T at consensus positions 7, 14, and 16, respectively, versus Ank6, which has F, D, and D (corresponding to Ank6 residues F293, D300, and D302, respectively) (Fig. 9A). To determine if any of these differences account for the residual ability of Ank6-F-boxAAAAA to bind Cul1 and Rbx1, the coimmunoprecipitation experiment was performed with Flag-Ank6-F-boxAAAAA proteins, having F293M, D300S, and/or D302T as additional mutations. Flag-Ank6 and Flag-Ank6ΔF-box were controls. Flag-Ank6-FboxAAAAA and Flag-Ank6-F-boxAAAAAF293M precipitated barely detectable levels of Cul1, while the remaining mutants failed to do so (Fig. 11). Flag-Ank6-F-boxAAAAA and its derivatives bearing additional mutations precipitated Rbx1 with relatively low efficiencies, all of which were considerably less than those of Flag-Ank6 but greater than those of Flag-Ank6ΔF-box. This suggests that additional F-box residues mediate direct interactions with Rbx1. Thus, conserved residues at positions 1, 2, 4, 9, and 17 are critically important for the F-boxes of O. tsutsugamushi Anks to interact with endogenous SCF proteins, but additional F-box amino acids of Ank6 and potentially other Anks participate in SCF binding.

FIG 11.

Assessment of the contributions of Ank6 F-box residues F293, D300, and D302 to SCF complex binding. Whole-cell lysates of HeLa cells expressing Flag-tagged BAP, Ank6, or the indicated mutants thereof were incubated with Flag antibody-conjugated agarose beads to immunoprecipitate (IP) Flag-tagged proteins and interacting proteins. Western-blotted input lysates were probed with Flag antibody to verify ectopic expression of the proteins of interest; antibodies targeting Skp1, Cul1, and Rbx1 to confirm their presence; and GAPDH antibody to validate that equal amounts of protein were present in each sample. Western-blotted eluates following immunoprecipitation were screened with indicated antibodies to verify pulldown of the Flag-tagged proteins and assess for Skp1, Cul1, or Rbx1 coimmunoprecipitation. Data are representative of seven experiments with similar results.

Ank1 and Ank6 competencies for binding Skp1, Cul1, and Rbx1 correlate with their abilities to inhibit NF-κB nuclear accumulation.

Truncated Ank1 and Ank6 proteins lacking the F-box cannot block TNF-α-induced NF-κB nuclear accumulation (Fig. 2 and 3) (21). We examined if this functional defect can be recapitulated by full-length F-boxAAA and F-boxAAAAA versions of each protein. Indeed, the percentages of TNF-α-treated HeLa cells exhibiting p65 in the nucleus were similar between cells expressing Flag-Ank1-F-boxAAAAA, Flag-Ank1ΔF-box, and Flag-BAP (Fig. 12) as well as cells expressing Flag-Ank6-F-boxAAAAA, Flag-Ank6ΔF-box, and Flag-BAP (Fig. 13). Flag-Ank1-F-boxAAA and Flag-Ank6-F-boxAAA were significantly less effective at impairing NF-κB than their wild-type counterparts but were significantly more effective at doing so compared to the F-boxAAAAA versions (Fig. 12 and 13). Therefore, Ank1 and Ank6 inhibit NF-κB nuclear accumulation when they can interact with Skp1, Cul1, and Rbx1 but are less capable of doing so when they can only bind Cul1 and Rbx1. Mutating F-box residues D14 and D16 to S and T, respectively, of Flag-Ank6-F-boxAAAAA did not further reduce its NF-κB modulatory ability (Fig. 13). Hence, the minor contributions of D14 and D16 of the Ank6 F-box to binding Cul1 and Rbx1 in the absence of SCF interactions mediated by L1, P2, E4, I9, and D17 are insufficient for Ank6 to functionally regulate NF-κB.

FIG 12.

L1, P2, E4, I9, and D17 of the F-box consensus are essential for Ank1 to inhibit NF-κB p65 nuclear accumulation. HeLa cells were transfected to express Flag-tagged BAP, Ank1, or the indicated mutants thereof. At 16 h, the cells were exposed to TNF-α or vehicle control for 30 min, after which they were fixed, screened with antibodies against the Flag epitope and p65, and examined by confocal microscopy. (A) Representative fluorescence images of cells viewed for Flag-tagged protein, p65, and merged images plus DAPI. (B) The mean ± SD percentage of transfected cells with p65 in the nucleus was determined. Triplicate samples of 100 cells each were counted per condition. Horizontal bars denote pairs for which statistical significance was assessed. Values that are statistically significantly different (****, P < 0.0001) are indicated. ns, not significant. Data are representative of three separate experiments with similar results.

FIG 13.

F-box residues L287, P288, E290, I295, and D303, but not D300 or D302, are critical for Ank6 to impair NF-κB p65 nuclear accumulation. HeLa cells were transfected to express Flag-tagged BAP, Ank6, or the indicated mutants thereof. At 16 h, the cells were exposed to TNF-α or vehicle control for 30 min, after which they were fixed, screened with antibodies against the Flag epitope and p65, and examined by confocal microscopy. (A) Representative fluorescence images of cells viewed for Flag-tagged protein, p65, and merged images plus DAPI. (B) The mean ± SD percentage of transfected cells with p65 in the nucleus was determined. Triplicate samples of 100 cells each were counted per condition. Horizontal bars denote pairs for which statistical significance was assessed. Values that are statistically significantly different (**, P < 0.01; ****, P < 0.0001) are indicated. ns, not significant. Data are representative of three separate experiments with similar results.

Dominant-negative Ank1 and Ank6 competitively antagonize the ability of O. tsutsugamushi to modulate NF-κB.

Given the genetic intractability of O. tsutsugamushi, we sought to investigate the roles of bacterially derived Ank1 and Ank6 in infected cells by leveraging our findings that both effectors inhibit NF-κB nuclear accumulation in ISR- and F-box-dependent manners (21). Presumably, ΔISR and F-boxAAAAA versions of Ank1 and Ank6 retain the ability to bind their cellular targets since their ankyrin repeat regions are unaltered. We rationalized that overexpressed Ank1ΔISR, Ank1-F-boxAAAAA, Ank6ΔISR, and Ank6-F-boxAAAAA would outcompete their O. tsutsugamushi counterparts to function as dominant-negative inhibitors. Transfected HeLa cells expressing GFP or GFP-tagged Ank1, Ank1ΔISR, Ank1-F-boxAAAAA, Ank6, Ank6ΔISR, or Ank6-F-boxAAAAA were infected with O. tsutsugamushi. At 24 h, TNF-α was added, followed by assessment of NF-κB p65 nuclear accumulation using immunofluorescence microscopy. Infected cells were identified by immunolabeling of intracellular O. tsutsugamushi organisms using antibody against outer membrane protein A (OmpA) (26). NF-κB was robustly detected in the nuclei of TNF-α-treated uninfected cells expressing GFP (Fig. 14). TNF-α-induced NF-κB nuclear accumulation was impaired in O. tsutsugamushi-infected cells expressing GFP and even more so in infected cells expressing GFP-Ank1 or GFP-Ank6 due to the additive effects of the endogenous and ectopically expressed effectors. Notably, however, TNF-α-driven NF-κB nuclear accumulation was almost completely restored in infected cells expressing GFP-Ank1ΔISR, GFP-Ank1-F-boxAAAAA, GFP-Ank6ΔISR, or GFP-Ank6-F-boxAAAAA, thereby substantiating our hypothesis and verifying that Ank1 and Ank6 each act in infection to modulate NF-κB in an ISR- and F-box-dependent manner.

FIG 14.

Overexpressed dominant-negative Ank1 and Ank6 proteins competitively inhibit the ability of O. tsutsugamushi to modulate NF-κB nuclear accumulation. HeLa cells were transfected to express GFP or GFP-tagged Ank1, Ank1ΔISR, Ank1-F-boxAAAAA, Ank6, Ank6ΔISR, or Ank6-F-boxAAAAA. At 16 h, each set of transfected cells was infected with O. tsutsugamushi (Ot). At 24 h postinfection, TNF-α was added for 30 min. The cells were fixed, immunolabeled for GFP, Ot OmpA, and NF-κB p65, stained with DAPI, and assessed for p65 nuclear accumulation using immunofluorescence microscopy. Uninfected cells expressing GFP and treated with TNF-α served as a positive control. OmpA immunosignal and DAPI staining were pseudocolored cyan to enhance their visibility. (A) Representative confocal images of cells viewed for GFP, p65, and OmpA as well as merged images. (B) The mean ± SD percentage of infected transfected cells exhibiting p65 in the nucleus was determined. Triplicate samples of at least 30 cells each were counted per condition. Values that are statistically significantly different (***, P < 0.001; ****, P < 0.0001) are denoted. Indicators of statistical significance relative to values obtained for GFP, GFP-Ank1, and GFP-Ank6 are colored black, red, and blue, respectively. Data are representative of three separate experiments with similar results.

DISCUSSION

This study validates the contributions of non-ankyrin repeat portions of O. tsutsugamushi Anks to interfacing with the endogenous SCF ubiquitin ligase complex and the effectors’ capacities to modulate eukaryotic processes. All 10 SMART-predicted F-boxes (Ank1, Ank2, Ank5, Ank6, Ank8, Ank9, Ank10, Ank12, Ank13, and Ank20) and the manually identified F-box of Ank17 were necessary and sufficient for Flag-tagged versions of these Anks to precipitate endogenous Skp1, Cul1, and Rbx1. Toxicity in yeast for 10 of these 11 was fully or partially F-box dependent. The mild toxicity of Ank4, which has a manually identified Ank4 F-box that weakly pulls down Skp1 but not Cul1 or Rbx1, is ablated when the F-box is removed. Ank4 potentially forms weak interactions with the SCF complex in eukaryotic cells that contribute to its effector function. Manually predicted Ank14, Ank15, and Ank16 F-boxes fail to precipitate SCF components. We conclude that Ank15 should no longer be considered an FBP. However, because Ank14 and Ank16 toxicity is F-box dependent, the Ank14 and Ank16 F-boxes, like that of Ank4, could form weak interactions with the SCF or another ubiquitin ligase complex that drives toxicity in cells but are not observable under the coimmunoprecipitation conditions used here. This possibility is supported by the AlphaFold models for Ank14 and Ank16. Hence, it cannot be concluded whether Ank14 or Ank16 is a bona fide FBP or not. Overall, O. tsutsugamushi Ank toxicity in yeast is largely F-box dependent. Exceptions are Ank1, the toxicity of which is F-box independent, and Ank6, which is nontoxic. Importantly, the toxicity screen establishes the likelihood that yeast suppressor screening can be used to identify eukaryotic cellular pathways that the 13 toxic Anks modulate, as has been performed for other intracellular bacterial effectors (37, 39–42).

The F-box residues that are essential for O. tsutsugamushi F-box-containing Anks to optimally interact with Skp1, Cul1, and Rbx1 are L1, P2, E4, I9, and D17 of the 21-amino-acid consensus sequence. These very residues correspond to those that are key for eukaryotic and other microbial F-boxes to nucleate the SCF complex through direct interactions with Skp1 and Cul1 (23, 24). Indeed, replacing L1, I9, and D17, which bind Skp1, with alanine (24, 26) was sufficient to eliminate Ank1, Ank5, and Ank6 interactions with Skp1 but not Cul1 or Rbx1. Further substituting P2, which interacts with both Skp1 and Cul1 (23, 24), and E4, which binds Cul1 (23), progressively reduced the abilities of Ank1, Ank5, and Ank6 bearing these mutations to precipitate Cul1 and Rbx1 to undetectable or barely detectable levels. The competencies of Flag-tagged Ank1 and Ank6 bearing LID-AAA substitutions to nucleate Cul1-Rbx1 versus the inabilities of LPEID-AAAAA versions to do so translated functionally. Flag-Ank1-F-boxAAA and Flag-Ank6-F-boxAAA were still able to partially inhibit NF-κB nuclear accumulation, while Flag-Ank1-F-boxAAAAA and Flag-Ank6-F-boxAAAAA were unable to do so. Other differences in the F-box sequence of Ank6 versus Ank1 and Ank5 that enabled residual coprecipitation of Cul1 and Rbx1, namely, D300 and D302 (positions 14 and 16 in the F-box consensus), are functionally irrelevant in terms of modulating NF-κB function. The role of consensus sequence P2 in recruiting the SCF complex is underscored when one considers that Ank4 weakly binds Skp1 and that Ank14, Ank15, and Ank16 fail to precipitate any SCF component. Each of these Anks lacks proline at position 2. Ank4, Ank14, and Ank16 have additional substitutions of L1, E4, or I9. The exception to this trend is Ank17, which has asparagine and tyrosine at positions 2 and 4 but still precipitates Skp1, Cul1, and Rbx1, presumably due to compensatory amino acid differences at other F-box positions. The poor capability of Ank4 and inabilities of Ank14, Ank15, and Ank16 to bind SCF complex proteins that were revealed by interrogating Ank-SCF complex interactions using antibodies against endogenous Skp1, Cul1, and Rbx1 were overlooked by our previous study that assessed interactions between Flag-tagged Anks and GST-Skp1 (26).

The ankyrin repeat-F-box/PRANC domain arrangement does not occur in mammalian F-box or ankyrin repeat-containing adaptors, suggesting that these proteins have arisen from gene shuffling during host-endosymbiont or host-pathogen interactions (29). In support of this notion, the PRANC domain has only been identified in Anks of the Chordopoxvirinae, which consist of all vertebrate poxviruses, O. tsutsugamushi, Wolbachia species, and the parasitic wasps Nasonia vitripennis and Cotesia congregata that Wolbachia species infect (12, 15, 16, 20, 26, 46). Genomic and phylogenetic analyses concluded that N. vitripennis most likely laterally acquired a PRANC-containing Ank gene from Wolbachia and expanded it over time (46). Moreover, a general theme observed for chordopoxviruses is that the larger number of PRANC-containing Anks present in these genomes correlates with broader host range (12), a trend that is echoed by the versatility of O. tsutsugamushi to survive in mites and multiple mammalian hosts. While there is no documented association of chordopoxviruses with O. tsutsugamushi (12), it is noteworthy that cowpox, monkeypox, ectromelia virus, and O. tsutsugamushi all infect numerous cell types of small rodents (2, 27). Given that the PRANC-containing Anks are more closely related to each other than to those of Wolbachia, N. vitripennis, and C. congregata and genes for these proteins are lacking in Leptotrombidium spp. (46–48), it is tempting to speculate that lateral exchange of these genes could have occurred between the ancestors of chordopoxvirus and O. tsutsugamushi due to a shared host environment. Whether the non-F-box portion of the PRANC domain is functional aside from linking the Ank-ISR region to the F-box appears to be unlikely based on data here. Deleting the PRANC domain nullifies the abilities of PRANC/F-box-containing Anks to bind Skp1, Cul1, and Rbx1 and prevents Ank1 and Ank6 from modulating NF-κB. However, this is likely due to changes in tertiary structure of PRANC proteins that compromise accessibility of the F-box residues that mediate SCF complex binding. Indeed, the PRANC domain in the absence of the F-box is insufficient to facilitate SCF binding while the F-box alone is sufficient. The non-F-box portion of the PRANC domain could simply be an ancestral sequence that was retained by O. tsutsugamushi but performs no direct functional role.

The findings here reinforce those of our previous report that the ISR and F-box are essential for Ank1- and Ank6-mediated immunomodulation (21) and demonstrate for the first time the role of these or any Ank during O. tsutsugamushi infection. Both effectors’ ISRs are dispensable for interacting with the SCF complex but are required for inhibition of NF-κB nuclear accumulation. This could be due to their influence on the structure-function relationship of both effectors. Additionally, the Ank1 and Ank6 ISR domains are more closely related to each other than to those of any other O. tsutsugamushi Ank and exhibit the greatest degree of amino acid identity to each other, 84.1%, than any other domain (21). Thus, the Ank1 and Ank6 ISRs potentially drive specificity for the canonical NF-κB pathway, possibly by coopting the importin β1 and/or exportin 1 proteins that regulate NF-κB nuclear cycling (21). However, their ankyrin repeat domains exhibit only 37.2% identity to each other (21), which presumably leads to Ank1 and Ank6 impairing NF-κB nuclear accumulation by binding distinct host proteins. This rationale is supported by the effectors’ differential toxicity in yeast. The disparate abilities of full-length, F-boxAAA, and F-boxAAAAA versions of Ank1 and Ank6 to modulate NF-κB validate that both effectors must interact with Skp1 and Cul1, thereby recruiting Rbx1 and nucleating the functional SCF E3 ubiquitin ligase complex to do so. Combined with the yeast toxicity and dominant-negative assay results, these data indicate that each effector orchestrates the ubiquitination and proteasomal degradation of one or more proteins upstream of NF-κB nuclear translocation that prevents this event from occurring. O. tsutsugamushi counters TNF-α-induced degradation of p105 (NFKB1), which functions to retain NF-κB in the cytoplasm (49). A cowpox PRANC/F-box-containing Ank called vIRD (viral inducer of receptor-interacting protein of kinase 3 [RIPK3] degradation; CPXV006) binds RIPK3 via its ankyrin repeats and directs its K48-linked polyubiquitination and proteasomal degradation to block necroptosis (29). vIRD and its orthologs in other orthopoxviruses also inhibit TNF-α-induced p105 degradation to retain NF-κB in the cytoplasm (50, 51). Whether Ank1, Ank6, or any other O. tsutsugamushi Ank functions analogously to vIRD is unknown.

Although genetic tools are being developed for O. tsutsugamushi (52), no such system is available yet. Moreover, knockout complementation approaches would be practically impossible to apply to genes encoding Ank1, Ank6, and several other Anks because they are multicopy. Consequently, most experiments in this study were limited to using ectopically expressed Anks. Nonetheless, employing this very approach with ΔISR and F-boxAAAAA versions of Ank1 and Ank6 as competitive inhibitors of their endogenous counterparts enabled us to functionally interrogate them during infection for the first time and thereby circumvent O. tsutsugamushi genetic intractability. This model could be extended to other Anks whose functionality is ISR or F-box dependent once their target host proteins/pathways are identified. The dominant-negative GFP-tagged Ank1 and Ank6 proteins did not fully restore NF-κB accumulation relative to uninfected cells expressing GFP, which could be due to incomplete inhibition of bacterially derived Ank1 and Ank6 or other unidentified O. tsutsugamushi proteins that contribute to the negative regulation of the NF-κB pathway.

In summary, this report further establishes the importance of the Ank repertoire to O. tsutsugamushi pathogenesis by confirming that the modular ankyrin repeat domain-ISR-PRANC/F-box arrangement of most of these effectors is critical for their ability to co-opt the SCF E3 ubiquitin ligase complex and modulate host cell processes. Specifically, five F-box residues that are functionally relevant in eukaryotic and other microbial FBPs are responsible for nucleating the SCF complex. The ISR, at least of Ank1 and Ank6, either contributes to specificity for the host pathway that is targeted or is functionally critical because of its tertiary structural importance. Also for Ank1 and Ank6, the PRANC domain potentially maintains accessibility of the F-box for it to bind the SCF complex. While most Anks’ toxicity in yeast is F-box dependent, some exhibit F-box-independent toxicity and others are nontoxic. Future studies will build on the foundation established here by combining yeast suppressor screening, coimmunoprecipitation coupled with proteomics, and polyubiquitination analyses to identify the array of host proteins/pathways that the Anks modulate and utilize overexpressed dominant-negative versions of Anks to confirm the effectors’ functional roles during O. tsutsugamushi infection.

MATERIALS AND METHODS

HeLa cell cultivation and O. tsutsugamushi infection.

HeLa human cervical epithelial cells (CCL-2; American Type Culture Collection [ATCC], Manassas, VA) were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Gemini Bio-Products, Sacramento, CA, USA) at 37°C in a humidified incubator with 5% CO₂. O. tsutsugamushi strain Ikeda was propagated every 3 to 4 days in HeLa cells in RPMI 1640 medium supplemented with 1% FBS and 1× Anti-Anti (ThermoFisher Scientific, Waltham, MA) at 35°C in a humidified incubator with 5% CO₂. Synchronous infections were performed at a multiplicity of infection (MOI) of 10, which was verified by assessing coverslips using antiserum specific for OmpA (26) and immunofluorescence microscopy as described below.

Plasmid constructs.

pFlag-BAP was purchased from Sigma-Aldrich (St. Louis, MO). Constructs encoding mammalian codon-optimized full-length Anks, Ank1ΔISR, Ank1ΔF-box, Ank4ΔF-box, Ank6ΔISR, Ank6ΔF-box, Ank9ΔF-box, and Ank13ΔF-box N-terminally fused to the Flag tag and Ank1 and Ank6 N-terminally fused to GFP were described previously (16, 18, 21, 26). Constructs encoding mammalian codon-optimized Ank2ΔF-box, Ank5ΔF-box, Ank8ΔF-box, Ank10ΔF-box, Ank12ΔF-box, Ank14ΔF-box, Ank15ΔF-box, Ank16ΔF-box, Ank17ΔF-box, and Ank20ΔF-box N-terminally fused to the Flag tag were made by PCR amplification of the nucleotide sequences encoding the first amino acid through the end of the F-box using primers listed in Table S1 in the supplemental material and pBMH constructs carrying the coding sequence for each respective Ank using PCR conditions described previously (16). The PCR products were restriction digested and ligated into p3XFlag-CMV-14 (Sigma-Aldrich) as described previously (16). Mammalian codon-optimized genes encoding Ank1 with F-box residues L285, I293, and D301 converted to alanines (Ank1-F-boxAAA); Ank5 with F-box residues L301, I309, and D317 converted to alanines (Ank5-F-boxAAA); and Ank6 with F-box residues L287, I295, and D303 converted to alanines (Ank1-F-boxAAA) with flanking BamHI and SalI restriction cloning sites were provided by Genewiz (South Plainfield, NJ). The inserts were subcloned by restriction digestion followed by ligation into p3XFlag-CMV-7.1 as described previously (16) to generate pFlag-Ank1-F-boxAAA, pFlag-Ank5-F-boxAAA, and pFlag-Ank6-F-boxAAA. These constructs were used as the template, primers listed in Table S1, and the TaKaRa Bio USA (San Francisco, CA) In-Fusion mutagenesis protocol to generate pFlag-Ank1-F-boxAAAA, pFlag-Ank5-F-boxAAAA, and pFlag-Ank6-F-boxAAAA. pFlag-Ank1-F-boxAAAA, pFlag-Ank5-F-boxAAAA, and pFlag-Ank6-F-boxAAAA were used as the template, primers listed in Table S1, and In-Fusion Mutagenesis to yield pFlag-Ank1-F-boxAAAAA, pFlag-Ank5-F-boxAAAAA, and pFlag-Ank6-F-boxAAAAA, respectively. pFlag-Ank6-F-boxAAAAA was used as the template in conjunction with primers listed in Table S1 and In-Fusion mutagenesis to generate constructs encoding Flag-tagged Ank6-F-box-AAAAA proteins having additional substitutions of F293, D300, and/or D302. Inserts encoding Ank1ΔF-box, Ank1-F-box-AAAAA, Ank6ΔF-box, and Ank6-F-boxAAAAA were subcloned from their respective p3XFlag-CMV-7.1 into pEGFP-C1 (53) using EcoRI, BamHI, and restriction-ligation cloning as described previously (16) to generate constructs encoding each protein N-terminally fused to GFP. All generated constructs were sequenced to ensure nucleotide fidelity (Genewiz).

Immunofluorescence microscopy.

HeLa cells that had been seeded onto glass coverslips in 24-well plates were transfected with 0.4 μg of plasmid DNA. At 16 h, the cells were incubated with medium containing 25 ng mL⁻¹ TNF-α (Life Technologies, Grand Island, NY) or vehicle control (molecular-grade H₂O) for 30 min at 37°C in a humidified incubator with 5% CO₂. Coverslips were washed with phosphate-buffered saline (PBS) followed by fixation and permeabilization with ice-cold methanol. Samples were blocked and immunolabeled as described above, after which they were incubated with 0.1 mg mL⁻¹ 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen) in PBS for 1 min, washed three times with PBS, and mounted with ProLong Gold antifade mounting medium (Invitrogen). Primary antibodies used were rabbit anti-p65 (1:1,000; ThermoFisher Scientific, Waltham, MA [catalog number PA1-186]) and mouse anti-Flag (1:1,000; Sigma-Aldrich [F1804]) in PBS containing 5% (vol/vol) bovine serum albumin (BSA). Secondary antibodies used were Alexa Fluor 488-conjugated goat anti-mouse IgG and Alexa Fluor 594-conjugated goat anti-rabbit IgG (1:1,000; Invitrogen [A11001 and A11037]) in 5% BSA. For cells expressing GFP or GFP-tagged proteins, the primary antibodies used were rabbit anti-GFP (1:1,000; Invitrogen [A6455]), mouse anti-p65 (1:250; Santa Cruz, Dallas, TX [sc-8008]), and rat anti-O. tsutsugamushi outer membrane protein A (OmpA) (1:500) (26), which were detected using Alexa Fluor 488-conjugated goat anti-rabbit IgG (1:1,000; Invitrogen [A11034]), Alexa Fluor 594-conjugated goat anti-mouse IgG (1:1,000; Invitrogen [A11032]), and Alexa Fluor 350-conjugated goat anti-rat IgG (1:1,000; Invitrogen [A21093]). Coverslips were imaged with an Olympus BX51 spinning disc confocal microscope (Olympus, Shinjuku City, Tokyo, Japan) or via laser scanning confocal microscopy as described previously (54). For images generated using laser scanning confocal microscopy, brightness and contrast were normalized across all conditions based on visual signal intensity in ImageJ-Fiji (NIH, Bethesda, MD) (55). For these images, ImageJ was also used to pseudocolor Alexa Fluor 350 and DAPI signal to cyan in order to enhance visibility. The mean ± standard deviation (SD) number of transfected cells exhibiting NF-κB p65 in the nucleus was determined by counting 100 cells per coverslip per sample in triplicate.

Immunoprecipitation and Western blotting.

HeLa cells grown in 25-cm² flasks to 80% confluence were transfected with 1 to 6 μg plasmid DNA using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) per the manufacturer’s instructions. At 16 h, cells were harvested and lysed in high-saline Tris buffer (50 mM Tris-HCl, 400 mM NaCl, 1 mM EDTA [pH 7.4]) with 1.0% Triton X-100 (TBHS-T) containing Halt protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific). To facilitate lysis, tubes were kept on ice for 2 h and vortexed for 5 s every 5 min. Lysate concentration was determined using a Bradford assay (Bio-Rad, Hercules, CA). Aliquots of each were saved for input lysate. The remaining lysate, which was ≤400 μg of protein, was saved for the subsequent immunoprecipitation. Protein A/G agarose beads (ThermoFisher Scientific) were washed with TBHS-T buffer twice, centrifuged at 8,400 × g for 30 s, and added to normalized cell lysates in a final volume of 400 μL. The lysates were rotated with beads at 4°C for 4 h, followed by centrifugation at 8,400 × g for 30 s. Supernatants were mixed with anti-Flag M2 affinity gel (MilliporeSigma, St. Louis, MO) that had been washed twice with TBHS-T buffer. Lysates were rotated with beads at 4°C overnight followed by centrifugation at 8,400 × g for 30 s. The beads were subjected to eight 5-min washes with TBHS-T at room temperature. Beads were resuspended in 40 μL of 2× Laemmli buffer and incubated at 100°C for 5 min to elute bound proteins. Input lysates (20 to 30 μg) or the eluted proteins (40 μL) were resolved by SDS-PAGE, screened with antibodies, exposed to chemiluminescent substrate, and imaged as described previously (18). Primary antibodies used were rabbit anti-p65, rabbit or mouse anti-Flag (1:1,000; Sigma-Aldrich [F7425 or F1804]), rabbit anti-Skp1 (1:1,000; Cell Signaling [2156S]), rabbit anti-Cul1 (1:1,000; Abcam, Waltham, MA [ab75817]), rabbit anti-Rbx1 (1:1,000; Abcam [ab133565]), mouse anti-GAPDH (1:750; Santa Cruz [sc-365062]). Bound primary antibodies were detected with horseradish peroxidase-conjugated horse anti-mouse, anti-rabbit, or anti-rat IgG (1:10,000; Cell Signaling Technology).

Yeast toxicity assays.

pYes-Ank13 and pYes-Ank13ΔF-box were previously described (18). Inserts encoding the remaining O. tsutsugamushi Anks and their respective F-box deletion mutant versions listed in Table 2 were cloned into a modified pYesNTA-Kan vector (38) in-frame with the Gal promoter and His tag coding sequence. ank ORFs were PCR amplified using primers listed in Table S1, Platinum Taq DNA polymerase high fidelity (Invitrogen), and the appropriate pFlag-Ank or pFlag-AnkΔF-box plasmid as the template. Thermal cycling conditions used were 98°C for 30 s, followed by 25 cycles of 98°C for 10 s, 55°C to 57°C for 30 s, 72°C for 2 min, and a final extension at 72°C for 10 min. Gel-purified amplicons were doubly digested with 20 U each of KpnI-HF and XbaI for 30 min at 37°C, gel purified again, and ligated into pYesNTA-Kan to generate pYes-Ank and pYes-AnkΔF-box constructs. S. cerevisiae was transformed with pYesNTA-Kan constructs for expressing each Ank, ΔF-box versions thereof, C. trachomatis CT694 (43), or empty pYesNTA-Kan. Yeast transformants were plated on uracil dropout medium containing glucose (noninducing conditions). Single colonies were picked and expanded in uracil dropout broth containing glucose. Toxicity was assessed by diluting each culture to an optical density at 600 nm of 0.2 and spotting 10-fold serial dilutions onto 2% glucose or 2% galactose agar plates, the latter of which induced expression of the fusion protein. Plates were incubated at 30°C for 48 h. Images were captured using a UVP GelDoc-It (Analytik Jena, Jena, Germany). Toxicity due to expression of each Ank and ΔF-box mutants thereof were scored as being very strong, strong, mild, or nontoxic based on the amount of growth observed for each 10-fold serially diluted transformant relative to that observed for yeast transformed with vector only or with positive control CT694 or Ank13, which were strongly toxic and very strongly toxic, respectively.

Bioinformatic analysis.

AlphaFold (31, 32) was used to predict tertiary structures for Ank1 (predicted local distance differentiation test [pLDDT] [56] score of 84.2), Ank1ΔF-box (pLDDT, 88.2), Ank1ΔISR (pLDDT, 68.8), Ank1ΔPRANC (pLDDT, 82.9), Ank6 (pLDDT, 83.8), Ank6ΔF-box (pLDDT, 88.9), Ank6ΔISR (pLDDT, 65.5), Ank6ΔPRANC (pLDDT, 71.1), Ank14 (pLDDT, 75.2), Ank16 (pLDDT, 78.2), and Ank17 (pLDDT, 78.4). Specific regions of the resulting program database files were color-coded using the PyMOL Molecular Graphic System, version 1.2r3pre, Schrödinger, LLC. The F-box and ISR regions of each Ank listed in Table 1, the sequence coordinates of which were identified previously (26), were analyzed by CLUSTAL Ω (https://www.ebi.ac.uk/Tools/msa/clustalo/) (57) to generate multiple sequence alignments. The alignments were subsequently analyzed using Interactive Tree of Life (iTOL) v4 (https://itol.embl.de/) (58) to construct phylogenetic trees.

Statistical analyses.

Statistical analyses were performed using the Prism 8.0 software package (GraphPad, San Diego, CA). One-way analysis of variance (ANOVA) with Tukey's post hoc test was used to test for a significant difference among groups. Statistical significance was set at P values of <0.05.