Is the major outer membrane lipoprotein LipL32 really exposed on the surface of Leptospira?

Here's a study that may come as a surprise to those in the leptospirosis field.  The outer membrane lipoprotein LipL32 is believed to be the dominant protein on the cell surface of pathogenic species of Leptospira.  However, according to a new PLoS One article written by Pinne and Haake at UCLA, LipL32 may not be present on the surface at all.  This is an important issue to get right because the function proposed for LipL32, attachment to the extracellular matrix during infection, assumes that the lipoprotein is exposed on the surface of the spirochete.  More importantly, a number of research groups have already committed a lot of time and resources towards generating LipL32-based vaccines, which in current formulations confer (at best) weak protection against leptospirosis in rodent models (see this review for a critical analysis of the vaccine studies).

Pinne and Haake assessed surface exposure of LipL32 by two methods.  The first involved adding proteinase K to suspensions of Leptospira to digest proteins exposed on the surface of the spirochete.  When they did this, they saw that the known surface-exposed proteins OmpL37 and OmpL47 were degraded.  On the other hand, LipL32 didn't break down at all unless the spirochetes were first lysed by boiling them in a detergent (see Western blot below).

Figure 1B from Pinne and Haake (2013).  Increasing concentrations of proteinase K (up to 150 ug/ml) were added to suspensions (first five lanes) or lysates (last five lanes) of L. interrogans.  Following incubation, LipL32 was examined in a Western blot.  Source.

They next added LipL32 antibodies to Leptospira to see if they bound to the surface of the spirochete.  They did not, providing additional evidence that LipL32 was not exposed on the cell surface.  The authors tested LipL32 antibodies from different sources in an attempt to rule out the  possibility that failure of antibody binding was due to the surface-exposed portion of LipL32 not being antigenic.  LipL32 antiserum raised in rabbits, monoclonal LipL32 antibodies raised in mice, and LipL32 antibodies purified from the sera of leptospirosis patients all failed to bind the surface of Leptospira unless the outer membrane was chemically (with methanol or EDTA) or physically disrupted.  Note that antibodies raised against OmpL54, a known suface-exposed protein, reacted strongly with intact Leptospira (last pair of images below).

Figure 3A from Pinne and Haake (2013).  Bound antibody was detected with a secondary fluorescent antibody (green).  The spirochetes were also stained with DAPI, a penetrating dye that stains DNA (blue).  Left column, intact Leptospira, right column, methanol-treated Leptospira. Source.

In light of these results, the authors took another look at the 2005 study by Cullen and coauthors, who claimed LipL32 was surface exposed.  In contrast to Pinne and Haake, Cullen and colleagues detected binding of LipL32-specific antibodies to intact Leptospira in three different assays.  However, Pinne and Haake point out that antibody binding in their assays was extremely weak when  the abundance of LipL32 is considered.  The most striking example was the immunoelectron microscopy image of Leptospira treated with gold-labeled LipL32 antibody (see next image).  Yes, the surface ended up labeled, with a mean of 10.8 gold particles per spirochete cell.  However, we now know that there are 38,000 copies of LipL32 in each bacterial cell, making LipL32 the most abundant protein of L. interrogans (see this blog post about the Leptospira protein census).  If LipL32 were really surface exposed, the surface of the spirochete should have been packed with gold particles.

Figure 5 from Cullen et al. (2005).

Cullen and colleagues also mixed suspensions of Leptospira with a biotin probe that reacts with primary amines (mostly on lysine side chains).  The probe should have reacted solely with surface-exposed proteins since it's unable to penetrate the outer lipid bilayer and is assumed to be too large to diffuse through outer membrane porins.  Biotinylated proteins were separated by two-dimensional electrophoresis and identified by mass spectrometry (see next figure).  Although LipL32 was one of the few proteins labeled with biotin, it's hard to make a firm conclusion about its surface exposure because proteins known to be located underneath the outer membrane, FlaB1 (a flagellar protein) and GroEL (a cytoplasmic heat shock protein), were also labeled with biotin. It's possible that LipL32 was labeled despite being located underneath the surface because the membrane was damaged while the spirochetes were being harvested for the experiment.

Modified from Figure 2 of Cullen et al., 2005.  Biotinylated proteins were separated by two-dimnesional electrophoresis.  Spots were removed and analyzed by mass spectrometry to identify proteins.  Multiple spots for each protein in the result of members of the population of each protein reacting with different numbers of biotin molecules.  "LipL32.16" was generated by proteolysis of LipL32.

Based on the intense labeling of LipL32 with biotin, Cullen and coauthors declared LipL32 the most abundant protein on the cell surface.  They speculated that LipL32 is poorly accessible to large molecules such as antibodies and proteases (which in their study failed to digest any protein when added to intact Leptospira) because the LPS side chains act as a "rainforest canopy" that can be penetrated only by smaller molecules such as biotin. This is a reasonable supposition because LipL32 is up to 60Å in length, whereas the distance between the outer membrane and the surface of the LPS layer is 92Å, according to a cryoelectron microscopy study of L. interrogans.  On the other hand, Pinne and Haake concluded that LipL32 is entirely or almost entirely subsurface since their assays failed to detect even a hint of the lipoprotein on the surface of Leptospira.  They maintain that the reactivity of surface probes with LipL32 observed by Cullen and colleagues was an artifact generated by the presence of damaged spirochetes in their assays and the massive copy number of LipL32. 

The results from Pinne and Haake's study do not rule out the "rainforest canopy" model since they did not test smaller surface probes that could penetrate into the LPS side chain layer.  Additional studies are needed to pin down the location of LipL32 relative to the surface of Leptospira.

References

Pinne, M., & Haake, D.A. (2013). LipL32 is a subsurface lipoprotein of Leptospira interrogans: Presentation of new data and reevaluation of previous studies. PLoS ONE, 8 (1) DOI: 10.1371/journal.pone.0051025

Cullen, P.A., Xu, X., Matsunaga, J., Sanchez, Y., Ko, A.I., Haake, D.A., & Adler, B. (2005). Surfaceome of Leptospira spp. Infection and Immunity, 73 (8), 4853-4863 DOI: 10.1128/IAI.73.8.4853-4863.2005

Related posts

• Protein census of Leptospira interrogans
• The major outer membrane protein of Leptospira interrogans: Not essential for infection?

An autoantigen targeted during Lyme arthritis

Infection by the spirochete Borrelia burgdorferi, if left untreated, can lead to a form of Lyme disease called Lyme arthritis.  About 10% of Lyme arthritis patients end up with a chronic form that doesn't go away with antibiotic treatment.  Allen Steere's group has long suspected that the antibiotic-refractory form of Lyme arthritis involves an autoimmune process.  This notion seems reasonable since those with antibiotic-refractory Lyme arthritis tend to have certain forms of the HLA-DR gene that are also common among those afflicted with the autoimmune disease rheumatoid arthritis.

Several groups have been searching for autoantigens (self antigens) that could drive the joint inflammation seen in Lyme arthritis patients.  Several candidate protein autoantigens were identified based on their short sequence similarities (molecular mimicry) to a T-cell or antibody epitopes in the B. burgdorferi OspA protein, which is targeted by the immune system in many Lyme arthritis patients, especially those with the antibiotic-refractory form.  However, further studies demonstrated that none of these autoantigens were likely to stimulate a sufficiently robust T-cell or antibody response that could account for the prolonged joint swelling experienced by patients with antibiotic-refractory Lyme disease (see this excellent review article for the complete story).  Therefore, an additional approach is needed to identify additional autoantigen candidates, an approach that does not assume that molecular mimicry underlies antibiotic-refractory Lyme arthritis.

An unbiased approach for finding autoantigens is to gather all of the different self-peptides being displayed by the HLA-DR molecules in the synovial tissue of the swollen joint and then figure out which of these peptides are capable of stimulating T cells.  At one time this approach wasn't possible since the individual peptides presented by HLA-DR molecules are found in such tiny amounts in human tissues, but the sensitivity of today's liquid chromatography/tandem mass spectrometry systems have improved to the point where many of the peptides can now be sorted and sequenced.

Steere's new study, which appeared in January's print issue of Arthritis and Rheumatism, was conducted in collaboration with Catherine Costello's group in Boston University.  The study was a follow-up to an earlier one published two years ago.  The authors extracted the inflamed synovial tissue from the swollen knee of a 12 year old boy suffering from antibiotic-refractory Lyme arthritis (see picture below).  He had gone through three months of antibiotic therapy a year prior to the procedure.  The tissue was culture and PCR negative.  When the patient's HLA-DR genes were examined, he turned out to have a copy of the DRB1*0101 allele, one of the HLA-DRB gene variants that places individuals at a higher risk for antibiotic-refractory Lyme arthritis.

Figure 1A from Drouin et al., 2013

The boy's synovial tissue was ground up, and HLA-DR-specific antibodies were used to capture the HLA-DR molecules with their bound peptides.  The peptides were then analyzed by liquid chromatography/tandem mass spectrometry.  The authors identified 120 different self-peptides from this analysis.  When each peptide was chemically synthesized and mixed with the boy's blood mononuclear cells, one peptide turned out to stimulate proliferation of his T cells.  This peptide came from a human protein called endothelial cell growth factor, or ECGF.

What's the function of ECGF?  The protein stimulates angiogenesis, the sprouting of new blood vessels from pre-existing ones.  Angiogenesis is a general feature of inflammatory arthritis, including Lyme arthritis and rheumatoid arthritis.

The authors went on to examine the T- and B-cell responses to ECGF in other Lyme arthritis patients.  The T-cell response was determined by measuring the amount of interferon-γ secreted by the patients' blood mononuclear cells upon exposure to ECGF in vitro.  In antibiotic-refractory patients, the T-cell response was observed in 38% (14/37) subjects against 30% (8/27) among Lyme arthritis patients who responded to antibiotics. The difference between the two groups was not statistically significant.  The B-cell (antibody) response was examined by ELISA in a larger group of patients.  17% (19/109) of antibiotic-refractory patients and 8% (6/77) of antibiotic-responsive patients had an IgG antibody response against ECGF that was higher than among healthy controls, yet the difference between the antibiotic-refractory and -responsive groups again was not statistically significant (P = 0.09).  So a link between an autoimmune response to ECGF and antibiotic-refractory arthritis was not clear-cut.  However, in support of a link, the authors mentioned that almost all of the Lyme arthritis patients with a T-cell response to ECGF (20/21, 98%) had one of the HLA-DR alleles known to be a risk factor for antibiotic-refractory arthritis.

The authors also looked at the levels of ECGF in the swollen joints of Lyme arthritis patients.  Those with antibiotic-refractory Lyme arthritis had much higher levels of ECGF in their joint fluid (mean 448 ng/ml, 37 subjects) than those whose arthritis responded to antibiotic treatment (mean 154 ng/ml, 19 subjects, P < 0.0001)

Further evidence for a link between an immune response to ECGF and chronic Lyme arthritis came from a group of untreated Lyme disease patients who were followed in the late 1970s, before the cause of Lyme disease was known.  Sera from sequential bleeds were still available from many of these patients.  If an autoimmune process involving ECGF was responsible for the disease, then the immune response to the autoantigen should have appeared before the disease symptoms.  This turned out to be the case.  Six of the seven Lyme arthritis patients who had antibodies against ECGF developed the antibody response before their joints swelled up.  The duration of the arthritis attack was longer in Lyme arthritis patients with an immune response to ECGF, lasting a median of 67 weeks in the seven patients with an ECGF antibody response and only 17 weeks in the 20 Lyme arthritis lacking the response (P = 0.004).

Steere's paper proposes that the immune response to ECGF leads to a persisting, autoimmune form of arthritis in those who have a high level of ECGF in their joint fluid.  In those patients, T cells that recognize ECGF would be activated more easily because of the high levels of ECGF available for phagocytes to engulf, process, and display to the T cells.  These events would lead to a chronic form of arthritis that would persist even when the spirochetes were cleared from the joints by the immune system or antibiotics.  These patients also have a lot of ECGF in their synovial tissue.  Antibody against ECGF could bind to the tissue and trigger attack by complement, contributing to the tissue damage.

Molecular mimicry doesn't appear to be involved in triggering an immune response to ECGF.  The authors were unable to identify any B. burgdorferi proteins that could cross-react with ECGF.

The immune response to ECGF can't be the whole story since most patients with antibiotic-refractory Lyme arthritis don't generate a T-cell or antibody response to the protein.  An autoimmune process in these other patients may involve other self-antigens waiting to be discovered.  Other host and spirochete factors also influence the course of Lyme arthritis (see this post, which gives the spirochete's point of view).


References

Drouin, E.E., Seward, R.J., Strle, K., McHugh, G., Katchar, K., Londoño, D., Yao, C., Costello, C.E., & Steere, A.C. (2013). A novel human autoantigen, endothelial cell growth factor, is a target of T and B cell responses in patients with Lyme disease. Arthritis & Rheumatism, 65 (1), 186-196 DOI: 10.1002/art.37732

Seward, R.J., Drouin, E.E., Steere, A.C., & Costello, C.E. (2010). Peptides presented by HLA-DR molecules in synovia of patients with rheumatoid arthritis or antibiotic-refractory Lyme arthritis. Molecular & Cellular Proteomics, 10 (3) DOI: 10.1074/mcp.M110.002477

A helpful review

Steere, A.C., Drouin, E.E., & Glickstein, L.J. (2011). Relationship between immunity to Borrelia burgdorferi Outer-surface protein A (OspA) and Lyme arthritis. Clinical Infectious Diseases, 52 (Supplement 3) DOI: 10.1093/cid/ciq117

Related post

• The genetics of both host and pathogen matter in antibiotic-refractory Lyme arthritis