We have studied a previously identified protein (Plu1537, here renamed Pam) which in P. asymbiotica ATCC43949 is secreted in a temperature-dependent manner, suggestive of a host-specific role in insects. In the closely related insect-only pathogen P. luminescens TT01, Pam has been estimated to constitute more than 30% of total secreted protein , potentially indicating an important role in the lifestyle of this bacterium. Further sequence alignment and the inferred phylogeny of the pam genes from different Photorhabdus species suggest that pam is both ancestral and conserved throughout the genus. Where variable regions in amino acid sequence do exist, they could therefore be responsible for determining functional specificity of the protein within strains.
Given the characteristic dual lifecycle of Photorhabdus, with both a nematode-symbiotic and a insect-pathogenic stage, the limited similarity of Pam with B. thuringiensis Cry34 insecticidal protein, and the previous insecticidal studies with Pit , the first phenotypes tested with the pam mutant were toxicity to insects and symbiotic efficiency with the bacterium's partner nematode H. bacteriophora. Interestingly, the deletion of the pam gene did not affect the ability of P. luminescens TT01 to support nematode growth, the production of infective juveniles, re-association of the bacteria with the worm or their ability to re-infect an insect. Similarly, we were not able to demonstrate any difference in insect survival (measured by LT50) when G. mellonella were injected with wild-type or pam mutant strains, but this could result from the high redundancy of virulence factors in Photorhabdus . In the case of Pam recombinant protein, which did not cause toxicity either by injection or feeding assays, it is possible that Pam is not toxic by itself but requires a second, as yet unidentified, protein partner that operates in a binary toxin-type system. The closest known homolog of Pam is the 13.6 kDa Cry34 protein from B. thuringiensis, which only exerts effective mortality when coupled with its partner Cry35 [15, 16]. The precise mode of action of Cry34 toxins remains unclear, but susceptible insects show histopathological symptoms in the midgut epithelium, characterized by cell blebbing and vacuolation . We have not found any genes in Photorhabdus that are predicted to encode a component similar to Cry35. It should be noted that our findings are contrary to reports of toxicity of purified Pam protein by Li and co-workers . It is possible that the Pam variant they produced (Pit) as a GST-fusion from P. luminescens subsp. akhurstii YNd185, either has a much greater inherent toxicity to G. mellonella, or that the different method of purification used by these authors preserved Pam's toxic phenotype.
The fact that we did not find any toxic effect of Pam towards insects, or any decrease in the efficiency of interaction with the symbiotic nematode, led us to investigate whether it was expressed during insect infection at all. Western blots with anti-Pam antibody against proteins isolated from infected insects suggested that Pam was first produced at 48 h and not earlier during the infection process, and that it was continuously produced for at least 11 days after insect death. Although the possibility exists that earlier production was below the detection limit of our assay, we note that 48 h coincides with death of the insect as determined by LT50 assays. These results indicate that Pam may play a role in occupancy of the insect cadaver rather than killing of the host and are consistent with a previous study of P. luminescens genes upregulated upon insect infection, in which pam (plu1537) was not present among the identified genes encoding several toxins and metabolic enzymes . We have detected Pam both as secreted protein in the extracellular medium and bound to the EPS decorating the extracellular matrix surrounding cells. However, the observable structure of EPS/matrix is not significantly altered by the presence or absence of Pam. Although we observed no differences in mature biofilm, we found that Pam influences the early stages of bacterial attachment in hemolymph. SPR data from E. coli and P. luminescens cultures showed that membrane-bound Pam reduces the ability of cells to bind to the abiotic surface of the metallic gold of the probe, and that the secreted protein itself is able to bind to this surface. The observation that Pam expression increases binding to an abiotic surface in insect blood is in contrast to the findings from the SPR analysis which suggest Pam lowers the adhesive properties of the cell. However these observed differences in attachment between the wild type and pam mutant in the hemolymph are not directly comparable with the SPR data. In the first case the cells are grown in the media where attachment is assessed and the combination of secreted and cell-bound Pam contributes to the phenotype, while for SPR we analyzed washed cells and supernatant separately. Furthermore, insect blood is a far more complex environment than the PBS used to resuspend the cells in the SPR study, so potential interactions of Pam and the bacterium with components of the insect immune system must be considered. Together, these data indicate that Pam is a secreted adhesive factor that modifies the surface properties of the cell, affecting the attachment process, specifically cell-to-cell and cell-to-surface attachment. Although it is important to note that attachment to abiotic substrata is not the same as attachment to living or devitalized tissue, we believe that this modification of adhesion by Pam may be involved in one or several processes key to the biology of the bacterium. For instance, once Photorhabdus has been regurgitated by IJ nematodes, it must colonize and invade the midgut  and this establishment of a biofilm, following attachment, is recognized as an important step in many microbial infections . Since the effect of deleting Pam does not result in a complete gain or loss of attachment, the protein may allow some plasticity in colonization during the infection. Perhaps Pam allows adhesion to specific tissues or a transient attachment, or even to facilitate the release of cells from biofilms to colonize other tissues within the host, acting in an analogous manner to glycanases described other biofilm-forming bacteria .
The predicted amino acid sequence of Pam gives little clue to its role or of the potential structure that mediates its adhesive properties. To get an insight into the structure of Pam, we analyzed the protein with circular dichroism spectroscopy. Our far-UV CD data strongly indicate that Pam is a helical protein, with 5.5 helix segments per 100 residues and an average helix length of 10.5 residues. By contrast, only 8% of residues are expected to form β-strands. We obtained only very weak spectra for Pam in the near-UV wavelengths, but 1D 1H and 2D 1H-15N HSQC NMR spectra (data not shown) and high melting temperature from differential scanning calorimetry experiments confirm that the protein has well defined tertiary structure. A degree of tertiary structural prediction is available from the far-UV spectra, specifically the position of the spectral cross-over from positive to negative, and the magnitude of the negative maximum at 208 nm . These both suggest that Pam is a α+β protein. Rather than having intermixed segments, such proteins have separate α-helix and β-sheet-rich regions . Interestingly, although Pam is not secreted at 37°C in P. asymbiotica, it shows thermal stability far beyond this. Differential scanning calorimetry revealed that the protein does not begin to thermally denature until heated to temperatures above 60°C. The transition midpoint is 77.4°C, suggesting that Pam is particularly thermostable for a protein produced by an organism considered to be psychrophilic . In fact, this midpoint is approaching that seen in thermophilic bacteria and archaea [23–25]. Without high resolution structural analyses we are unable to explore precise contributions to the thermal stability of Pam, but the high α-helix content is likely to be significant; thermostable proteins are richer in α-helices than mesophilic proteins . The observed ability of Pam to refold to its native conformation following denaturation may be biologically significant; this folding indicates that the protein can form its native structure in the absence of molecular chaperones, outside of the cell if it is secreted as an unfolded polypeptide. It is as yet not clear how Pam is secreted from the cell as we can detect no recognizable signal motifs, neither were found in Pit .
Finally, although the role of this highly secreted protein in Photorhabdus biology has not yet been completely elucidated, we have shown its possible relevance in cell attachment. Our findings indicate that Pam is a secreted adhesive factor of Photorhabdus that modifies attachment of cells to surfaces in biotic (hemolymp) and abiotic (SPR) conditions. Thus, it might be involved at different cell-cell and cell-surface adhesion stages during the insect host colonization, such as the production of a biofilm-like matrix on the insect gut, the spread of bacteria within the insect cadaver and potentially a resource protection role, binding to the insect tissues and preventing other saprophites from taking advantage of the biomass. The high levels of secretion and the degree of conservation within the genus are congruent with Pam modulating these important activities. Very little is known about Photorhabdus infections in humans, but a recent study has found that, unlike the extracellular growth of P. luminescens in insects , a clinical isolate of P. asymbiotica is a facultative intracellular pathogen when incubated with human macrophage-like cells . Future studies may investigate what role if any Pam has in the infection of mammalian cells.