The Potential Consequences for Cell Signaling by a Class of NOD-Like Receptor Proteins (NLRs) Bearing an N-terminal Signal Sequence Cell Signaling

The authors gratefully acknowledge the support of the UK Biotechnology and Biological Sciences Research Council (BBSRC) for the work on cellular 2A-like sequences.


Commentary
Toll-like and NOD-like receptors (TLRs/NLRs) recognise pathogenor damage-associated molecular patterns (PAMPs/DAMPs) and are key initiators of cell signalling pathways by which the cell mounts an appropriate defence response. Whilst TLRs are type I transmembrane proteins, NLRs are cytosolic proteins. Determination of the genome sequence of the purple sea urchin (Strongylocentrotus purpuratus) revealed this echinoderm had a greatly expanded repertoire of TLR/NLR innate immunity genes in comparison with vertebrates possessing an acquired immune system [1]. The open reading frame (ORF) of a number of NLR genes in S. purpuratus (and organisms from other phyla) commence with a leader sequence with sequence similarity to '2A/2A-like' oligopeptide sequences found in many virus genomes, which mediate an unusual translational 'recoding' event in which translation arrests at the C-terminus of 2A (mid-ORF; no stop codon present), but then may recommence the synthesis of the downstream protein as a separate, entirely discrete, translation product [2][3][4][5]. Bioinformatic analyses also suggested, however, that these S. purpuratus NLR 2A-like sequences might also function as cotranslational signal sequences. We showed this type of 2A-like leader sequence possessed both activities [6], and in this commentary we speculate upon the possibility that such a single gene could produce both cytoplasmic and membrane-bound/secreted forms of such NLRs. '2A' and '2A-like' oligopeptide sequences were first characterised from a group of single-stranded positive (mRNA) sense RNA viruses called picornaviruses. These short (~20aa) sequences were shown to mediate a co-translational 'cleavage' of a polyprotein at their own Ctermini. This apparent 'cleavage' was, however, not a product of proteolysis, but a newly-characterised form of translational 'recoding' in which the synthesis of a specific peptide bond at the C-terminus of 2A was 'skipped' . Briefly, the model of this mechanism states that when the elongating ribosome translates the 2A sequence, the nascent polypeptide sequence within the exit tunnel of the ribosome (2A) interacts with the tunnel to pause elongation and inhibit peptide bond formation. This is proposed to bring about the release of the nascent peptide synthesised to that point, but then translation of the sequences downstream of 2A may recommence. In this manner the translation products are actually synthesised as discrete products rather than as a single product which is then cleaved apart [2][3][4][5]. Many 2As possess only partial recoding activity: if, for example, a particular 2A sequence recodes translation only 50% of the time, then in 50% of the translation products the peptide bond is formed and the resulting translation profile is a mixture of the recoded ('cleaved') products (50%) plus the fusion protein (50%).

Cellular 2A-like Sequences
Database probing with a motif characteristic of 2A ([-D(V/ I)ExNPGP-]) revealed the presence of '2A-like' sequences in the genomes of many other RNA viruses but, surprisingly, also multiple occurrences within the genome of S. purpuratus. Here, 2A-like sequences were observed within two main types of cellular sequence: non-LTR retrotransposons [7,8] and certain NLRs [6]. In the latter case our surprise was compounded in that these particular 2A-like sequences formed the N-termini of certain NLR proteins: until that point all of the (virus) 2A/2A-like sequences we had studied were internal features used for the co-expression of protein domains up-and down-stream of 2A (Figure 1). These cellular NLR-2As were of approximately the same length as virus 2As (~25-30aa), but comprised a higher proportion of hydrophobic residues and were predominantly flanked by charged residues -suggestive of a signal sequence and, indeed, bioinformatic analyses confirmed these NLR-2As could be signal sequences. But how to show these NLR-2As could (i) mediate translational recoding and (ii) could function as N-terminal signal sequences? Given the lack of immortalised S. purpuratus cell-lines, the limited availability of primary cell-lines, the lack of anti-NLR antibody probes etc., we chose to recapitulate the system using fluorescent reporter proteins (to study the effects upon protein localisation) and, since the function of N-terminal sequences is conserved across kingdoms, we used both immortalised, readily transfectable, mammalian (HeLa) cells and plant (Nicotiana benthamiana) leaf epidermal cells. Our data showed that these NLR-2As could both mediate (partial) translational recoding and act as N-terminal signal sequences [6]. When the NLR-2A recoded itself from the fluorescent reporter protein (mCherryFP), this protein localised to the cytoplasm. In the proportion of translation products where the NLR-2A did not recode -remained as an N-terminal feature of mCherryFP -then mCherryFP was secreted from the cell. Our data showed that in mammalian cells the NLR-2A was removed from the secreted mCherryFP, presumably by signal peptide peptidase.

Innate Immune Signalling in S. purpuratus
The systems outlined above allowed us to demonstrate that these NLR-2As were a new type of bi-functional translational recoding/ The presence of a functional N-terminal signal sequence (functional at least in mammalian and plant cells) together with putative transmembrane/internal signal-anchor/stop-transfer domains between the NACHT and LLR domains raises a number of questions with regards the sub-cellular localisation, potential membrane association (and, if this is the case, the protein orientation in the membrane) of these S. purpuratus NLR proteins. As shown in (Figure 3) (left panels), if the 2A-like signal sequence does not mediate a translational recoding event (remaining fused to the NLR), then the protein could either be secreted from the cell (no effect of putative TMD/internal signalanchor/stop-transfer sequences), or, become a type 1 transmembrane protein (putative TMD/internal signal-anchor/stop-transfer sequence functional), although it is difficult to envisage a signalling function of such a molecule either being completely secreted from the cell or with this particular orientation within the membrane. As shown in ( Figure  3) (right panels) shows the possibilities if the 2A-like signal sequence does mediate a translational recoding event (2A-like signal sequence not present at the N-terminus of the NLR): the NLR could become cytoplasmic (no effect of putative TMD/internal signal-anchor/stoptransfer sequences), or, perhaps, become either a type I or a type II transmembrane protein (internal signal-anchor/stop-transfer sequence functional). The orientation of the protein in the membrane is determined by the difference in the charges of the 15 residues flanking the first internal signal-anchor/stop-transfer sequence, with the more positive portion facing the cytosol [10]. Translation of the mRNA, however, can produce 2, distinct, translation products. In the first case, the NLR 2A-like signal sequence does not mediate translational recoding (remains fused to the NLR protein), the NLR could either be secreted from the cell, or, if the putative transmembrane domain (TMD) is recognised by the exocytic pathway, adopt a type I transmembrane orientation with the LRR cytoplasmic and the NACHT oligomerisation and pyrin (PYR) domains oriented to the exterior of the cell (left panels). In this illustration we have assumed that the 2A-like signal sequence is removed by signal peptide peptidase (as is the case in mammalian cells). In the second case, where the 2A-like S-seq does mediate translational recoding, the NLR could be localised to the cytoplasm (putative TMD not recognised by the exocytic pathway), or, if the putative TMD is recognised by the exocytic pathway then adopts the orientation of either a type I or type II transmembrane protein -depending upon the charge distribution surrounding the putative TMD (right panels).
Again, it is difficult to envisage the function of such an NLR oriented as a type I transmembrane protein, but if it became a type II trans membrane protein, then could it fulfil the same type of function as a TLR? Interestingly, the 2A-like signal sequence could be acquired by these S. purpuratus NLR genes by a process of 'exon-shuffling' since it is encoded by its own mini-exon preceding the bulk of the NLR gene: in the case of the NLR described above (XP011670951.1), encoding the entire signal plus six amino acids downstream MDGFCLLYLLLILLMRSGDVETNPG ↓ PNTAVS; 2A-like signal sequence underlined, site of translational recoding indicated by vertical arrow).
Our paper presented analyses of the function of these N-terminal 2A-like signal sequences of NLR genes in S. purpuratus [6]. It should be noted, however, that we have detected a similar 2A-NLR gene organisation in other phyla; the cephalochordate Branchiostoma floridae, the sponge Amphimedon queenslandica, the mollusc Lottia gigantea and the arthropod Ixodes scapularis [11].
These data, plus subsequent bioinformatic analyses presented here, raise a number of important questions as to the function of these 'NLR-like' genes in these organisms. It may prove to be the case that they may have not only expanded the number of innate immunity genes, but also expanded the functionality of some of these NLR proteins -potentially involving novel signalling pathways.