SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Mannheimia haemolytica leukotoxin (LktA) is a member of the RTX toxin family that specifically kills ruminant leukocytes. Previous studies have shown that LktA induces apoptosis in susceptible cells via a caspase-9-dependent pathway that involves binding of LktA to mitochondria. In this study, using the bioinformatics tool MitoProt II we identified an N-terminal amino acid sequence of LktA that represents a mitochondrial targeting signal (MTS). We show that expression of this sequence, as a GFP fusion protein within mammalian cells, directs GFP to mitochondria. By immunoprecipitation we demonstrate that LktA interacts with the Tom22 and Tom40 components of the translocase of the outer mitochondrial membrane (TOM), which suggests that import of this toxin into mitochondria involves a classical import pathway for endogenous proteins. We also analysed the amino acid sequences of other RTX toxins and found a MTS in the N-terminal region of Actinobacillus pleuropneumoniae ApxII and enterohaemorrhagicEscherichia coli EhxA, but not in A. pleuropneumoniae ApxI, ApxIII, Aggregatibacter actinomycetemcomitans LtxA or the haemolysin (HlyA) from uropathogenic strains of E. coli. These findings provide a new evidence for the importance of the N-terminal region in addressing certain RTX toxins to mitochondria.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Mannheimia haemolytica is the primary bacterial pathogen associated with the bovine respiratory disease complex, a severe pleuropneumonia in cattle (Frank and Smith, 1983). One of the major virulence factors of M. haemolytica that contributes to the pathogenesis of bovine respiratory disease is a leukotoxin (LktA). LktA both activates and kills bovine leukocytes (Shewen and Wilkie, 1982; Clinkenbeard et al., 1989; Czuprynski et al., 1991; Maheswaran et al., 1992; Cudd et al., 2001).

Mannheimia haemolytica LktA is a member of the RTX (repeats in toxin) family of exotoxins produced by a wide variety of Gram-negative bacteria. The most studied RTX toxins include Escherichia coli haemolysin (HlyA), the Aggregatibacter actinomycetemcomitans leukotoxin (LtxA), the Bordetella pertussis adenylate cyclase-toxin (CyaA) and exotoxins from A. pleuropneumoniae (ApxI, ApxII, ApxIII and ApxIV) (Felmlee et al., 1985; Glaser et al., 1988; Chang et al., 1989; 1993; Lally et al., 1989; Frey et al., 1991; Jansen et al., 1993). New members recently added to the family include the RtxA toxin from epidemic strains of Vibrio cholerae, and plasmid-encoded enterohaemolysin (EhxA) from enterohaemorrhagic strains of E. coli (Bauer and Welch, 1996; Lin et al., 1999). The RTX toxins are characterized by the presence of glycine-rich non-apeptide repeats with a consensus sequence GGXGXDXUX, where U represents a large hydrophobic residue (Welch, 2001). These tandemly arrayed repeats are responsible for Ca2+-binding and are critical for toxin activity. Although the RTX toxins share significant homology in their amino acid sequence and structural features, individual members of the RTX family show differences in cell specificity and level of toxicity for target cells. For example, M. haemolytica LktA and A. actinomycetemcomitans LtxA are cytotoxic against a narrow range of target cells (ruminant and human leukocytes respectively). In contrast, E. coli HlyA, A. pleuropneumoniae ApxI and B. pertussis CyaA are cytotoxic for different cell types (erythrocytes, leukocytes and epithelial cells) from different species. The species and cell-type specificity of M. haemolytica LktA is mediated through the CD18 chain of the β2-integrin receptors (LFA-1 and Mac-1) (Ambagala et al., 1999; Lawrence et al., 2008).Specificity being determined by amino acids 5–17 of the signal sequence peptide, which is not cleaved from bovine CD18 (Shanthalingham and Srikumaran, 2009). At high concentrations, M. haemolytica LktA creates a pore in the cell membrane leading to cellular swelling and lysis (Shewen and Wilkie, 1982; Clinkenbeard et al., 1989; Cudd et al., 2001). At sublytic concentrations, target cells are activated or stimulated to undergo apoptosis (Czuprynski et al., 1991; Maheswaran et al., 1992). Previous studies in our laboratory have shown that LktA induces apoptosis in bovine lymphoblastoid (BL-3) cells via a caspase-9-dependent pathway (Atapattu and Czuprynski, 2005). It was further demonstrated that LktA is internalized by BL-3 cells and transported to mitochondria, where it exerts its cytotoxicity (Atapattu and Czuprynski, 2007; Atapattu et al., 2008). Mitochondria from LktA-treated BL-3 cells displayed large lesions in the outer mitochondrial membrane that correlated with collapse of the mitochondrial membrane potential (Δψ) and cytochrome c release. It has been suggested that mitochondrial damage and caspase-9 activation are due to direct binding of LktA to mitochondria (Atapattu et al., 2008).

Targeting mitochondria has emerged as a common strategy by which pathogenic bacteria control host cell viability (Blanke, 2005; Kozjak-Pavlovic et al., 2008). Mitochondria have been reported to be the target for many bacterial virulence factors. These include exotoxins, type III secretion system (TTSS) effector proteins and outer membrane proteins (Galmiche et al., 2000; He et al., 2000; Kenny and Jepson, 2000; Muller et al., 2000; Nougayrede and Donnenberg, 2004; Choi et al., 2005; Braun et al., 2007). Nearly all of these proteins have been found to cause apoptosis in target cells, usually by inducing the release of pro-apoptotic factors like a cytochrome c and disrupting the mitochondrial membrane potential (Δψ) (Kozjak-Pavlovic et al., 2008). While the effect of these toxins on mitochondria has been investigated in detail, less is known about how these proteins access mitochondria. For some of them, classical N-terminal leader sequences characteristic for nucleus-encoded mitochondrial proteins have been identified (Saier et al., 1988; Kenny and Jepson, 2000; Lucattini et al., 2004; Nougayrede and Donnenberg, 2004; Nagai et al., 2005; Papatheodorou et al., 2006). These leader sequences generally contain a high number of arginine, alanine, leucine and serine in the absence of negatively charged (acidic) residues, and exhibit a tendency to form an amphiphilic α-helix (von Heijne, 1986; Reid and Walker, 1988; Claros and Vincens, 1996). Other proteins, like bacterial porins, possess a β-barrel structure similar to endogenous mitochondrial porins such as voltage-dependent anion-regulated channel (VDAC). It is believed that these proteins are incorporated into the mitochondrial membrane because of targeting information encoded in their β-barrel topology (Muller et al., 2000; Rapaport, 2003). Another group of proteins, consisting mostly of bacterial toxins, does not contain an N-terminal leader sequence or β-barrel structure. These proteins are sorted to mitochondria by an unknown route (Galmiche et al., 2000; Braun et al., 2007; Matarrese et al., 2007).

The purpose of this study was to examine the mechanism of mitochondrial targeting by M. haemolytica LktA. We used the bioinformatics tool MitoProt II to search for mitochondrial targeting motifs in LktA and constructed various LktA–GFP fusion proteins to study their intracellular localization. We also analysed amino acid sequences of other RTX toxins for mitochondrial targeting signals (MTSs) and toxin import into mitochondria.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

The N-terminal region of LktA represents a mitochondrial targeting signal

Recent studies in our laboratory demonstrated that M. haemolytica LktA binds to mitochondria in BL-3 cells and causes gross morphological changes in mitochondrial outer and inner membrane architecture (Atapattu et al., 2008). It has been shown that LktA-mediated damage of mitochondria is associated with cytochrome c release, activation of caspase-9 and cell death (Atapattu and Czuprynski, 2005; Atapattu et al., 2008). To better characterize mitochondrial targeting of LktA, we used MitoProt II software to analyse its amino acid sequence for mitochondrial targeting motifs. This analysis indicated that the first 54 N-terminal amino acids resemble a classical MTS, with a probability score for mitochondrial import of 0.84 (Table 1). Secondary structure prediction and helical wheel analysis revealed that this segment of LktA can form an amphiphilic helix, which is characteristic of mitochondrial targeting sequences (Fig. 1 and data not shown). To verify this experimentally, N-terminal (1–61 amino acids) and internal (59–115 amino acids) fragments of LktA were fused to green fluorescence protein (GFP) and transiently expressed in BL-3 cells. Cellular location of green fluorescence in these cells was analysed by confocal microscopy. As a positive control for mitochondrial targeting, we used a COX8[1–29]–GFP fusion protein that contains the 29-amino-acid MTS sequence from bovine cytochrome c oxidase subunit VIII (COX8) fused to GFP. As shown in Fig. 2, BL-3 cells transfected with pLktA[1–61]–GFP or pCOX8[1–29]–GFP constructs displayed green fluorescence in mitochondria, as confirmed by concomitant staining of mitochondria with MitoTracker Red. In contrast, cells transfected with the pLktA[59–115]–GFP plasmid that encodes the internal fragment of LktA fused to GFP, displayed diffused green fluorescence over the entire cell, similar to cells transfected with pAcGFP (GFP alone, Fig. 2). Mitochondrial transport of LktA[1–61]–GFP also was observed in other types of mammalian cells, including bovine bronchial epithelial cells (BBEC) (Fig. 3A), monkey fibroblasts (COS7) (Fig. 3B) and a mouse neuroblastoma cell line (N2a) (data not shown). These results indicate that the N-terminal region of LktA serves as a MTS and that import of M. haemolytica leukotoxin into mitochondria in mammalian cells is MTS-dependent. Importantly, we did not observe increased toxicity in mammalian cells transfected with pLktA[1–61]–GFP in comparison with pCOX8[1–29]–GFP plasmid, as determined by cell viability assay (Cell Titer 96 AQ assay, data not shown).

Table 1.  MitoProt II analysis of RTX toxins.
NameGenBank accesion no.Analysed region (AA)Probability of mitochondrial importCleavage site
LktAAAG40289540.84Not predicted
ApxIAAL5566660.09Not predicted
ApxIIAAU84700570.67Not predicted
ApxIIIAAK5005390.18Not predicted
HlyAP09983320.28Not predicted
HlyAYP_543860320.34Not predicted
LtxACAA34731210.23Not predicted
EhxACAA63849310.9331/32
image

Figure 1. Helical wheel diagram of the putative N-terminal α-helix of M. haemolytica LktA (amino acids 4–41; 3.6 residues per turn). The polar (including positively charged) residues are shown in gray and non-polar residues are shown in white.

Download figure to PowerPoint

image

Figure 2. Fluorescence confocal microscopy of BL-3 cells transfected with plasmids encoding different fragments of LktA fused to GFP (LktA[1–61]–GFP; LktA[59–115]–GFP), the MTS of cytochrome c oxidase (subunit VIII) fused to GFP (COX8[1–29]–GFP), or GFP alone. Cells were examined at excitation wavelength 488 nm to reveal the location of GFP (left panels). In the central panels, mitochondria were stained with MitoTracker Red 633 and visualized at excitation wavelength 633 nm. The right panels illustrate merged images with colocalized signals. Photos are representative of three independent experiments.

Download figure to PowerPoint

image

Figure 3. Fluorescence confocal microscopy of BBEC (A) and COS7 (B) cells transfected with plasmids encoding different fragments of LktA fused to GFP (LktA[1–61]–GFP; LktA[59–115]–GFP. The cells were examined at excitation wavelength 488 nm to reveal the location of GFP (left panels). To reveal mitochondria, cells were stained with MitoTracker Red 633 and examined at excitation wavelength 633 nm (central panels). Merged images with colocalized signals are shown in the right panels. Photos are representative of three independent experiments.

Download figure to PowerPoint

N-terminal 1–31-amino-acid sequence of LktA is critical for mitochondrial targeting of GFP

To map more precisely the sequence of LktA that is capable of directing GFP to mitochondria, we created various DNA constructs encoding several different LktA peptides fused to GFP. These included the 1–20, 1–31, 1–41, 1–55, 10–41, 10–55 and 20–61 amino acids peptides of LktA. The DNA constructs were expressed in COS7 cells, and intracellular signal from GFP was analysed by confocal microscopy. Mitochondrial localization of GFP was verified by MitoTracker Red staining and colocalization of signal was calculated using Image J software. As shown in Fig. 4, the minimal sequence with capacity to direct GFP to mitochondria contained N-terminal 1–31 amino acids of LktA. Removing the first 20 (LktA[20–61]–GFP) or 10 (LktA[10–41]–GFP and LktA[10–55]–GFP) amino acids from the N-terminal sequence of LktA resulted in the loss of mitochondrial localization of GFP. These results indicate that N-terminal 1–31-amino-acid segment of LktA is critical for mitochondrial targeting.

image

Figure 4. Mitochondrial localization of GFP fused to different N-terminal fragments of LktA in COS7 cells. Plasmids encoding different amino acid fragments (1–20, 1–31, 1–41, 1–55, 10–41, 10–55; 20–61 and 1–61 AA) of LktA fused to GFP were transfected into COS7 cells. MitoTracker Red 633 staining was performed 16 h post transfection. The cells were examined by fluorescence confocal microscopy at excitation wavelength 488 nm and 633 nm to reveal the location of GFP and mitochondria respectively. The percentages of GFP that colocalized with mitochondria were calculated using Image J software for three randomly selected cells. Data represents mean ± SD.

Download figure to PowerPoint

Mitochondrial targeting of LktA[1–61]–GFP is inhibited by cyclosporine A

It was previously demonstrated that pretreatment of BL-3 cells with the mitochondrial membrane stabilizing agent cyclosporine A (CyA) inhibits binding of LktA to mitochondria and significantly reduces LktA-mediated cytotoxicity (Atapattu et al., 2008). To determine if CyA has a similar effect on mitochondrial targeting of LktA[1–61]–GFP, transient expression of LktA[1–61]–GFP fusion protein in COS7 cells was performed in the presence or absence of 5 µM CyA. Confocal microscopy showed that, in the presence of CyA, green fluorescence was distributed through out the cytoplasm rather than localized to mitochondria (Fig. 5A). Interestingly, the same diffuse green fluorescent staining in the presence of CyA was observed in cells transfected with the control plasmid that encodes the MTS from bovine COX8 fused to GFP (Fig. 5A). The same inhibitory effect of CyA on the mitochondrial import of these two different GFP fusion proteins suggested a common mechanism for their mitochondrial translocation. Like other nucleus-encoded mitochondrial proteins, COX8 is transported into mitochondria in an MTS-dependent manner through the translocase of the outer mitochondrial membrane (TOM) complex (Khalimonchuk and Rodel, 2005). This led us to hypothesize that MTS-dependent association of LktA with mitochondria can involve the same protein machinery (i.e. TOM) as endogenous mitochondrial preproteins. It should be noted that CyA had no inhibitory effect on the level of GFP expression in COS7 cells, as shown by the control experiment presented in Fig. 5A (GFP panel). Quantification of green fluorescence signal for cells expressing GFP in the presence or absence of 5 µM CyA showed no significant difference in fluorescence intensity (Fig. 5B).

image

Figure 5. A. Fluorescence confocal microscopy of COS7 cells transfected with pLktA[1–61]–GFP; pCOX8[1–29]–GFP, or pAcGFP in the absence (left panels) or presence (right panels) of 5 µM cyclosporine A (CyA). Cells were visualized at excitation wavelength 488 nm to reveal the location of GFP. Photos are representative of three independent experiments. Bar = 50 µm. B. The panel illustrates the mean ± SD fluorescence intensity of three randomly selected images for COS7 cells transfected with pAcGFP (GFP alone) in the absence or presence of 5 µM CyA, as quantified using Image J software.

Download figure to PowerPoint

M. haemolytica LktA interacts with components of the TOM complex

The TOM is a multi-subunit complex that consists of import receptor proteins (Tom20, Tom22 and Tom70), and the general import pore for protein translocation (Tom40) (Nakamura et al., 2004; Becker et al., 2008). Tom22 is an essential component of the import receptor complex that functions as a receptor for preproteins and serves as a docking point for both Tom20 and Tom40. To understand the involvement of TOM machinery in the mitochondrial import of LktA, we analysed interaction of LktA with Tom22 and Tom40 by immunoprecipitation in a cell-free assay. Mitochondrial lysates from BL-3 or COS7 cells were incubated with native LktA and then immunoprecipitated using anti-Tom22 or anti-Tom40 antibodies. Immunoprecipitated proteins were then blotted onto nitrocellulose and probed for LktA. As shown in Fig. 6, LktA was immunoprecipitated from BL-3 and COS7 mitochondrial lysates by both anti-Tom22 and anti-Tom40 antibodies but not by irrelevant antibody (Fig. 6) or naked agarose beads (data not shown).

image

Figure 6. Immunoblot analysis of BL-3 (A) or COS7 (B) mitochondrial lysates incubated with LktA and immunoprecipitated with anti-Tom40 and anti-Tom22 antibodies or anti-mouse IgG antibody (irrelevant Ab control). The blots were probed for LktA using anti-LktA antibody (MM601).

Download figure to PowerPoint

Mitochondrial targeting of other RTX toxins

LktA shares significant amino acid sequence homology with other RTX toxins. However, the N-terminal region of RTX toxins shows high sequence divergence (Welch et al., 1995; Welch, 2001). We tested the amino acid sequences of several RTX toxins for their mitochondrial targeting using MitoProt II software (Table 1). These analyses revealed that the 57 N-terminal amino acids of A. pleuropneumoniae ApxII, represents a MTS, with a 0.67 probability score for mitochondrial import. This is the RTX toxin with the highest homology to LktA. In contrast, the N-terminal sequences of HlyA from two different uropathogenic strains of E. coli (J96 and UT189) had very low probability scores for mitochondrial import (0.28 and 0.34 respectively). The probability scores of < 0.5 for these toxins indicate that HlyA might not be imported into mitochondria. Interestingly, the N-terminal sequence of plasmid-encoded haemolysin, EhxA from enterohaemorrhagic E. coli, had a probability score of 0.93. Moreover, EhxA was the only RTX toxin for which the cleavage site of MTS was defined by MitoProt II (Table 1). The presence of a cleavage site in MTS indicates that EhxA should be targeted to the mitochondrial matrix. The lowest probabilities for mitochondrial targeting in the group of RTX toxins we analysed were for A. actinomycetemcomitans LtxA (0.23), A. pleuropneumoniae ApxIII (0.18) and A. pleuropneumoniae ApxI (0.08). To confirm the results obtained in silico, we chose three RTX toxins (ApxI, ApxII from A. pleuropneumoniae and HlyA from E. coli strain J96), for which GFP fusion proteins were constructed and analysed for cellular localization. Fluorescence analysis of the RTX–GFP fusion proteins transiently expressed in COS7 cells confirmed the predictions of mitochondrial targeting calculated by MitoProt II. As shown in Fig. 7, ApxII[1–61]–GFP (probability score 0.67) localized to mitochondria, although some green fluorescence staining was also observed in the nucleus and cytoplasm. As expected, cells transfected with pHlyA[1–61]–GFP (probability score 0.28) and pApxI[1–61]–GFP (probability score 0.08) exhibited diffuse green fluorescence through the cells, suggesting that these RTX toxins are not transported to mitochondria. To confirm these results we performed experiments in which the intracellular trafficking of the native RTX toxins, ApxI, ApxII and HlyA, within BL-3 cells was analysed by confocal microscopy. BL-3 cells were treated with 0.5 U of ApxI, ApxII, HlyA or LktA for 60 min and then stained with an appropriate anti-RTX antibody and with MitoTracker Red. Three randomly selected BL-3 cells (for each RTX toxin) were spatially scanned to determine the positions of RTX toxin (green) and mitochondria (red). The degree of colocalization of signal was calculated using Image J software. As shown in Fig. 8, ApxII (58% ± 6.1%) and LktA (82% ± 5.6%), but not ApxI (6% ± 3.2%) or HlyA (11 ± 3.1%), colocalized with mitochondria in BL-3 cells. These observations are in agreement with the results obtained using RTX–GFP fusion proteins, and from in silico analysis. Together these findings demonstrate for the first time that the N-terminal region of certain RTX toxins carries the signal for mitochondrial targeting, and that sequence differences in this region are responsible for differing abilities of various RTX toxins to traffic to mitochondria.

image

Figure 7. Fluorescence confocal microscopy of COS7 cells transfected with plasmids encoding the N-terminal (1–61) sequence from several RTX toxins fused to GFP (ApxII and ApxI from A. pleuropneumoniae and HlyA from uropathogenic E. coli). The cells were examined at excitation wavelength 488 nm to reveal the location of GFP (left panels). To reveal mitochondria, cells were stained with MitoTracker Red 633 and examined at excitation wavelength 633 nm (central panels). Merged images with colocalized signals are shown in the right panels. Photos are representative of three independent experiments. Bar = 30 µm.

Download figure to PowerPoint

image

Figure 8. Fluorescence confocal microscopy of BL-3 cells treated with native RTX toxins. BL-3 cells were incubated with 0.5 U of ApxI, ApxII, HlyA or LktA for 1 h and then stained with MitoTracker Red 633. The cells were fixed, permeabilized and incubated first with an appropriate anti-RTX antibody followed by a second antibody conjugated to Alexa Fluor 488. The slides were examined at 488 nm (left panels) and excitation wavelength 633 nm (central panels). Merged images with colocalized signals are shown in the right panels. Photos are representative of three independent experiments. Bar = 10 µm.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Previous studies in our laboratory showed that M. haemolytica LktA binds to mitochondria in bovine lymphoblastoid (BL-3) cells, resulting in mitochondrial outer membrane damage and cell death (Atapattu and Czuprynski, 2005; Atapattu et al., 2008). In this study, we identified the N-terminal sequence (54 amino acids) of LktA as an MTS. We showed that ectopic expression of this sequence as a GFP fusion protein within the BL-3 cells directs GFP to mitochondria. Further peptide mapping of this N-terminal sequence indicated that the first 31 amino acids of LktA are critical for mitochondrial targeting. These results suggest that LktA is transported to mitochondria in an MTS-dependent manner. MTS-dependent mitochondrial import has been reported for two other bacterial proteins, EspF and Map from enteropathogenic E. coli (EPEC) (Kenny and Jepson, 2000; Nougayrede and Donnenberg, 2004; Papatheodorou et al., 2006). These multifunctional effector proteins are injected into host cells via the type III secretion system (Kenny, 2002; Zaharik et al., 2002). EspF triggers apoptosis by dissipating Δψm, causing cytochrome c release and activating caspase-9 and 3. Map causes dissipation of Δψm and changes in mitochondrial morphology. Both of these proteins possess N-terminal mitochondrial presequences with features typical for MTS of nuclear-encoded mitochondrial proteins (Nougayrede and Donnenberg, 2004; Papatheodorou et al., 2006). In addition, cleavage sites for these mitochondrial presequences have been predicted, which suggest that EspF and Map should be transported into the mitochondrial matrix. Indeed, the expression of Map in yeast cells, followed by mitochondria isolation and fractionation, confirmed its mitochondrial matrix localization (Papatheodorou et al., 2006). Our in silico analysis did not predict a cleavage site for the MTS of LktA. Therefore, we assume that LktA will associate with mitochondrial membranes rather than be imported into the mitochondrial matrix.

Using several different mammalian cell lines, we demonstrate that the mitochondrial targeting sequence of LktA is recognized by several cell types from different species. LktA[1–61]–GFP was transported to mitochondria in bovine epithelial (BBEC), monkey fibroblasts (COS7) and mouse neuronal (N2a) cells. These observations are in agreement with the previous report of Atapattu et al. (2008), in which protein transfection of LktA resulted in its transport to mitochondria in a murine macrophage cell line (RAW 264.7) that is normally resistant to LktA. Together these findings indicate that LktA potentially can be toxic to non-ruminant cells if it is delivered across the cell membrane into the cytoplasm.

Mitochondrial targeting signal-dependent trafficking of LktA[1–61]–GFP to mitochondria raises questions about the route and mechanism for mitochondrial import of LktA. Virtually all endogenous mitochondrial proteins, synthesized in the cytoplasm, possess an N-terminal MTS that is recognized by the TOM (Becker et al., 2008). These precursor proteins initially interact with the receptor proteins, Tom70, Tom20 and Tom22 on the mitochondrial surface and then are delivered to Tom40, which provides a channel for protein translocation (Rapaport et al., 1997; Kunkele et al., 1998; Meisinger et al., 1999). Identification of an MTS in LktA led us to hypothesize that transport of LktA into mitochondria might also involve the TOM machinery. This hypothesis was supported by the observation that mitochondrial import of LktA[1–61]–GFP and COX8[1–29]–GFP in COS7 cells was equally inhibited by CyA. These two GFP constructs contain MTS from LktA and COX8, respectively, the latter of which is known to be recognized by and transported through the TOM complex (Khalimonchuk and Rodel, 2005). Our finding that LktA was immunoprecipitated with Tom22 and Tom40 from BL-3 and COS7 mitochondrial lysates confirms the interaction of LktA with TOM and suggests that translocation of LktA into mitochondria involves a classical mitochondrial import pathway for endogenous proteins. Components of the TOM complex previously have been shown to participate in mitochondrial import of Map from E. coli and the bacterial porin PorB from Neisseria gonorrhoeae (Muller et al., 2000; Papatheodorou et al., 2006). Although PorB, in contrast to Map, does not carry an N-terminal mitochondrial signal, it still follows the classical import pathway and is recognized by Tom20 and transported through the Tom40 channel (Muller et al., 2000). Interestingly, association with Tom22 and Tom40 also has been demonstrated for the pro-apoptotic endogenous protein Bax (Bellot et al., 2007; Cartron et al., 2008). This interaction seems to be critical for its insertion into the mitochondrial outer membrane, which leads to the release of apoptogenic proteins such as cytochrome c, AIF and Smac into the cytosol (Danial and Korsmeyer, 2004). Peptide mapping indicated that Bax interacts with TOM22 via its first α-helix, and possibly two central α-helices, which are homologous to the pore forming domains found in pore forming toxins such as a colicin and diphtheria toxin (Bellot et al., 2007; Lalier et al., 2007).

Our observation that mitochondrial import of LktA[1–61]–GFP in COS-7 cells is inhibited by CyA is consistent with the previous report of Atapattu et al. (2008). In that study, pretreatment of BL-3 cells with CyA significantly inhibited binding of LktA to mitochondria and LktA-mediated toxicity. Two mechanisms were proposed to explain the protective effect of CyA in that study. The first mechanism is based on the ability of CyA to deplete dynamin-2, the vesicle scission protein required for clathrin-mediated internalization of LktA. The second is based on the observation that CyA inhibits cyclophilin D, a mitochondrial matrix peptidilpropyl cis-trans isomerase involved in the regulation of the mitochondrial permeability transition (Tsujimoto and Shimizu, 2007; Leung and Halestrap, 2008). Oxidative or other cellular stresses promote translocation of cyclophilin D to the inner membrane. This translocation appear to be a key factor that triggers opening of the non-specific pore known as the mitochondrial permeability transition pore (mPTP). The opening of mPTP results in loss of mitochondrial membrane potential, swelling of mitochondria and activation of apoptotic or necrotic mechanisms leading to cell death. Atapattu et al. (2008) proposed that interaction of CyA with cyclophilin D can reduce the mitochondrial membrane permeability in LktA-mediated apoptosis. Our observations that CyA inhibits MTS-dependent translocation of LktA[1–61]–GFP and COX8[1–29]–GFP to mitochondria suggest that CyA might also interfere with MTS-dependent import of mitochondrial proteins. It is possible that this observation also results from the inhibition of cyclophilin D. Beside its role in opening the mPTP, cyclophilin D has been found to be involved in folding of proteins newly imported into mitochondria (Matouschek et al., 1995; Rassow et al., 1995). Perhaps inactivation of cyclophilin D by CyA might impair folding of translocated proteins and thus slow their mitochondrial import. It was previously demonstrated that LktA interacts with cyclophilin D, and this interaction is important for mitochondrial translocation of the former (Atapattu et al., 2008).

Identification of an MTS in the N-terminal region of LktA led us to search for MTSs in other RTX toxins. It is especially interesting that the N-terminus is the most divergent region among members of the RTX family (Welch, et al., 1995; Welch, 2001). Although no functional significance for this portion of RTX toxins was previously found, it has been speculated that it is important for initiating cell contact through non-specific (electrostatic) interactions with the lipid membrane of target cells (Welch, 2001). Our study reveals for the first time that the N-terminal region of RTX toxins carries a signal important for intracellular targeting of certain RTX toxins to mitochondria. Using MitoProt II we identified an MTS in the N-terminal region of A. pleuropneumoniae ApxII and enterohaemorrhagic E. coli EhxA, but not in A. pleuropneumoniae ApxI, ApxIII, A. actinomycetemcomitans LtxA or haemolysin HlyA from uropathogenic strains of E. coli. Further studies of the cellular localization of RTX–GFP fusion proteins performed with three selected RTX toxins (ApxI, ApxII and HlyA), as well as cellular localization study of the native RTX toxin ApxI, ApxII and HlyA, confirmed mitochondrial targeting predicted by MitoProt II. Our results indicate that differences in the N-terminal amino acid sequence of RTX toxins specify their intracellular trafficking. We suggest that different intracellular trafficking patterns contribute to differences in toxicity of RTX toxins for target cells.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Bacterial strains and growth conditions

Mannheimia haemolytica A1 (isolated from a pneumonic bovine lung) was kindly provided by R.E. Briggs (Ames, IA). Bacteria were incubated without shaking in brain-heart infusion (BHI) broth (Difco Laboratories) at 37°C. A. pleuropneumoniae (strain UWP36N) producing ApxI, and A. pleuropneumoniae (strain UWP34) producing ApxII, were kindly provided by S.E. West (Madison, WI). Bacteria were cultured in BHI broth supplemented with 10 µg ml−1 NAD, in a 5% CO2 atmosphere at 37°C. E. coli WAM1824 carrying pSF4000 plasmid encoding HlyA from clinically pathogenic E. coli J96 was kindly provided by R.A. Welch (Madison, WI). Bacteria were incubated in Luria–Bertani (LB) broth supplemented with 20 µg ml−1 chloramphenicol at 37°C. E. coli DH5α (BioPionier), used for all cloning experiments, were grown in LB medium supplemented with 100 µg ml−1 kanamycin.

Eukaryotic cell cultures

The BL-3 bovine lymphoblastoid cell line (ATCC CRL-8037) was cultured in RPMI medium (Mediatech, Inc.) supplemented with 10% fetal bovine serum (Gibco, BRL) at 37°C in the presence of 5% CO2. Primary BBEC were generously provided by D.S. Allen-Gipson (Omaha, NE). BBEC were immortalized by transfection with SV40-large T antigen by the method of Stins et al. (1997). Immortalized BBEC were maintained in DMEM/F-12 medium (Mediatech, Inc.) supplemented with 10% fetal bovine serum, 2 mM glutamine and Pen/Strep (Sigma), for up to 50 passages. COS7 cells (a monkey kidney cell line) obtained from L.A. Shuler (Madison, WI) were cultured in DMEM/F12 medium (Mediatech, Inc.) supplemented with 10% fetal bovine serum, 2 mM glutamine and Pen/Strep (Sigma). The mouse neuroblastoma N2a cell line (ATCC CCl-131) was grown in EMEM medium (Mediatech, Inc.) supplemented with 10% fetal bovine serum, 2 mM glutamine and Pen/Strep (Sigma).

Plasmid constructs

The eukaryotic expression vector pAcGFP (Clontech), encoding GFP under control of the early CMV promoter, was used for the construction of various LktA–GFP and other RTX–GFP fusion proteins. Standard techniques were used for DNA manipulation (Sambrook and Russell, 2001) unless specified by the manufacturer.

The sequences of lktA gene encoding different N-terminal (1–61, 1–20, 1–31, 1–41, 1–55, 10–41, 10–55 and 20–61 amino acids) and internal (59–115 amino acids) fragments of LktA were amplified by PCR from total M. haemolytica genomic DNA using primers listed in Table 2. The sequences encoding N-terminal (1–61-amino-acid) fragments of other RTX toxins (ApxI, ApxII and HlyA) were amplified by PCR from genomic DNA isolated from A. pleuropneumoniae (UWP36N), A. pleuropneumoniae (UWP34) and plasmid DNA isolated from E. coli WAM1824, respectively, along with their primers, presented in Table 2. A DNA fragment encoding the 1–29-amino-acid MTS of bovine cytochrome c oxidase subunit VIII (COX8) was amplified by PCR from cDNA prepared from BBEC cells. The primers used for COX8 MTS amplification are presented in Table 2. The PCR products were inserted into the pAcGFP vector (Clontech) using XhoI/SalI sites. The resulting plasmids, designated pLktA[1–61]–GFP, pLktA[1–20]–GFP, pLktA[1–31]–GFP, pLktA[1–41]–GFP, pLktA[1–55]–GFP, pLktA[10–41]–GFP, pLktA[10–55]–GFP, pLktA[20–61]–GFP, pLktA[59–115]–GFP, pApxI[1–61]–GFP; pApxII[1–61]–GFP; pHlyA[1–61]–GFP and pCOX8[1–29]–GFP were verified by DNA sequencing with appropriate vector-specific primers.

Table 2.  Oligonucleotides used in this study.
Primer name5′-3′ primer sequence
  1. F, forward primer; R, reverse primer.

lktA1-FCCGCTCGAGATGGGAACTAGACTTACAACCC
lktA10-FGAGCTCGAGATGGGGCTAAAAAACACTTTAACGG
lktA20-FGGTCTCGAGATGAGTGGCTTACATAAAGCCGGTC
lktA59-FCCGCTCGAGATGGGTATTGGTTTACAGGATTTAG
lktA20-RGTGGTCGACCTTTTGGTTGCCGTTAAAGTG
lktA31-RGAGGTCGACGTTAATGATTGACCGG
lktA41-RGAGGTCGACGCCCCAGTTTTTAAAG
lktA55-RGAGGTCGACTCATATTGGTAATTTTGG
lktA61-RCCCGTCGACCCATTACCTTGTTCAGTATC
lktA115-RGCCGTCGACAATTTATCAATTTGTGGAGCGG
apxI1-FCCGCTCGAGATGGCTAACTCTCAGCTCGATAG
apxI61-RCACGTCGACAGACTTGAGCCGGTACTAGC
apxII-1 FCTACTCGAGATGTCAAAAATCACTTTGTCATCATTAAAATCG
apxII-61 RCACGTCGACTGACCCGAATCATAGCC
hlyA-1 FGGCCTCGAGATGCCAACAATAACCACTGC
hlyA-61 RGAGGTCGACCCTGTCCTTTATAATCTTTAGG
cox8-1 FCCGCTCGAGATGTCTGTGCTGACTCC
cox8-29 RCCTGTCGACTTGGGATGTATCTGGGC

GFP fusion proteins expression

For GFP expression experiments, the adherent cells (BBEC, COS7 and N2a) were seeded on glass cover slips in 24-well tissue culture plates and incubated for 16–24 h until the monolayers reached 60–70% confluence. Non-adherent BL-3 cells were seeded in 24-well tissue culture plates at 105 cells per well. Transfection of BL-3, BBEC and COS7 cells was performed using Fugen-HD (Roche) according to the manufacturer's protocol. N2a cells were transfected using Lipofectamine 2000 (Invitrogen).

Confocal microscopy

GFP expressing cells were incubated for 30 min with 25 nM MitoTracker Red 633 (Molecular Probes) in RPMI at 37°C in the presence of 5% CO2. After several washes with RPMI (DMEM/F-12 or EMEM) medium the cells were fixed on glass cover slips with 4% paraformaldehyde for 10 min at room temperature, and then washed twice with PBS. The slides were mounted in Vectashield mounting medium (Vector), and fluorescence was visualized by laser scanning confocal microscopy at an excitation wavelength of 488 (green) and 633 nm (red) using an Eclipse TE200-U microscope (Nikon). Fluorescence intensity in images was quantified using Image J software (http://rsbweb.nih.gov/ij/).

BL-3 cells, treated with 0.5 U of ApxI, ApxII, HlyA or LktA for 1 h, were washed three times with PBS and incubated with 1 µM Mitotracker Red 633 for 30 min. The cells were fixed on the glass slides with 4% paraformaldehyde, washed three times with PBS and permeabilized with cold acetone (−20°C) for 10 min. For ApxI and ApxII detection, cells were incubated with 1:500 dilution of mouse anti-RTX antibody (List Biological Laboratories Inc., Campbell, CA) in 1% bovine serum albumin in PBS, followed by 1 h incubation with anti-mouse antibody conjugated with Alexa Flour 488 (Invitrogen). For HlyA detection cells were incubated with rabbit anti-HlyA serum (kind gift from RA Welch, Madison, WI) followed by 1 h incubation with anti-rabbit antibody conjugated with Alexa Flour 488 (Invitrogen). For LktA detection, cells were incubated with mouse anti-LktA antibody (MM601) followed by 1 h incubation with anti-mouse antibody conjugated with Alexa Flour 488 (Invitrogen). The slides were mounted in Vectashield mounting medium (Vector), and fluorescence was visualized by laser scanning confocal microscopy at an excitation wavelength of 488 (green) and 633 nm (red) using an Eclipse TE200-U microscope (Nikon).

Cell viability assay

The viability of COS7 cells transfected with pLktA[1–61]–GFP or pCOX8[1–29]–GFP was determined using the Cell Titer 96 AQ assay (Promega, Madison, WI) according to the manufacturer's protocol. Briefly, plasmid transfected COS7 cells were incubated for 16–24 h at 37°C in the presence of 5% CO2. The cells were washed and DMEM containing [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, MTS] was added to the wells. After 1–2 h further incubation absorbance was measured at 490 nm. Controls included untreated cells and cells transfected with pAcGFP plasmid.

Cyclosporine A treatment assay

COS7 cells were seeded on glass cover slips in 24-well tissue culture plates and incubated 16–24 h until the monolayers reached 60–70% confluence. The cells were washed with HBSS buffer (Mediatech, Inc.), and 500 µl of Opti-MEM medium (Gibco) containing 5 µM CyA was added to the wells. The cells were then transfected with pLktA[1–61]–GFP, pCOX8[1–29]–GFP or pAcGFP plasmids using Fugene HD (Roche) according to the manufacturer's protocol. After 18–24 h incubation at 37°C, green fluorescence were examined by confocal microscopy.

Isolation of mitochondria

BL-3 or COS7 cells were resuspended in mitochondrial isolation buffer (250 mM sucrose, 20 mM Hepes pH 7.4, 1.5 mM MgCl2 10 mM KCl, 1× protease inhibitors cocktail) and incubated for 10 min on ice. Cells were lysed by sonication (20 s, on ice), and remaining intact cells were removed by centrifugation at 1000 g for 10 min at 4°C. Supernatants were collected and centrifuged again at 1000 g for 10 min. Mitochondria from supernatants were then pelleted at 20 000 g for 20 min at 4°C. Finally mitochondrial pellets were resuspended in PBS with 1× protease inhibitor cocktail.

Co-immunoprecipitation of LktA

Protein A/G agarose beads (Calbiochem) were washed three times with PBS and then incubated for 24 h at 4°C with anti-Tom22 anti-Tom40 antibodies (Santa Cruz). Mitochondria isolated from BL-3 or COS7 cells were lysed by sonication (2 min, on ice) in the presence of a protease inhibitors cocktail (Thermo Scientific). The mitochondrial lysates (300 µg) were incubated overnight with LktA (90 µg) at 4°C. To preclear samples, mixtures of mitochondrial proteins and LktA were incubated with 100 µl of protein A/G agarose beads (Calbiochem) for 1 h at room temperature. The beads were removed by centrifugation and supernatants were incubated with protein A/G agarose beads conjugated to anti-Tom22 or anti-Tom40 antibodies overnight at 4°C. As a control, mitochondrial lysates were incubated with protein A/G agarose beads conjugated to anti-mouse IgG antibodies (irrelevant antibody control). After washing, the proteins were eluted from beads using 30 µl of 10% SDS. The eluted fractions were separated by SDS-PAGE, blotted to polyvinylidene difluoride and probed with anti-LktA antibody (MM601).

RTX toxins purification and activity control

Crude LKT was prepared and purified as described previously and stored at −80°C until it was used in experiments (Atapattu and Czuprynski, 2005). ApxI, ApxII and HlyA were prepared in a similar manner. To control the activity of purified RTX toxin, BL-3 cells (1 × 106 cells per ml) were incubated with 2 µl of ApxI, ApxII, HlyA or LktA for 120 min at 37°C, washed and resuspended in antibiotic-free RPMI medium. Cytotoxicity was quantified using the Cell Titre 96 AQ assay (Promega).

Bioinformatics tools

MitoProt II (http://ihg2.helmholtz-muenchen.de/ihg/mitoprot.html; Claros and Vincens, 1996) was used to predict subcellular location of RTX toxins. MitoProt II predicts protein as a mitochondrial based on a probability cut-off of > 0.5. Prediction of secondary structure of LktA was performed by the PSIRED server (http://bioinfadmin.cs.ucl.ac.uk/psipred/psiform.html, Jones, 1999; Bryson et al., 2005). Helical wheel analysis of LktA N-terminal sequence was performed by BioEdit (http://www.mbio.ncsu.edu/bioedit/bioedit.html).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

We are grateful to Dr Evgeni Sokurenko (University of Washington, Seattle) for support and helpful advice. We thank Dr Yoshihiro Kawaoka (University of Wisconsin-Madison) for access to his laboratory's confocal microscope for this study. We also thank Kati Hellenbrand for technical assistance. This work was funded by grants 2004-14841 and 2006-17522 from the USDA National Research Initiative, by grant WIS04884 from the Wisconsin Agricultural Experiment Station and by the Walter and Martha Renk Endowed Laboratory for Food Safety.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References