Reply to A human chemo-sensory modality to detect peptides in the nose? by A. Natsch

The comment by Natsch on our paper [1] addresses three main issues:

• (1) The first issue concerns the volatility of peptide stimuli. Natsch correctly notes that nonapeptides are not volatile and suggests that they may have been delivered to the olfactory mucosa in aerosolized form. Indeed, in preliminary tests, it was found that the air stream generated by the olfactometer contained the SLLPAIVEL (A*0201 ligand) test peptide. This was tested as follows: the air stream generated by the olfactometer was collected in a plastic bag, the contents washed out with copious amounts of distilled water, the eluate concentrated by freeze drying, redissolved in water and assayed by spectrophotometry; the absorption spectrum of the eluate was identical to that of the input peptide. Although the amounts of peptides delivered in such a way may vary depending on the amino acid composition, we concluded that it was possible to deliver peptides in this form. We note that the spatial activation patterns of sensory neurons of mice are invariant over a concentration range of several orders of magnitude [2].

• (2) The second issue concerns the design of psychometric tests. In our study, the subjects conducted the test at home; in the questionnaire, they were asked about their recent smoking habits and whether they had a cold on the test day. Hence, the study design was such that we could not a priori exclude subjects with a cold. The incidence of smoking was very low; thus, this parameter could not be used to establish a correlation with the extent of recent smoking. If the outcome of the psychometric tests had been, indeed, dependent only on chance, then no difference in the ability to distinguish between self- and non-self-stimuli between subjects with and without compromised smell perception would be expected. Hence, the inability to respond to peptides observed in subjects with compromised smell perception strengthens our positive finding in non-compromised subjects. In the psychometric test, stimuli were categorized as either ‘self’ or ‘non-self’; it is evident from the electronic supplementary material, table S1 in our paper [1], that in the majority of cases both types of stimuli consisted of peptides. For instance, when a volunteer lacking A*0201 and A*2402 alleles had to choose between A2 peptide in solvent and solvent alone or between A24 peptide in solvent and solvent alone, we expected that they would dislike both peptides, that is the ‘non-self’ stimuli. It is important to stress that the ‘solvent’ stimulus per se essentially represents a neutral stimulus, as the peptide was always delivered in the same solvent. There is thus no difference between any two options with regard to solvent. Any difference perceived by the study subjects must therefore be attributed to the peptide. In the example above, the solvent is equivalent to the ‘self’ stimulus for the purpose of categorizing the outcome of the test; a neutral stimulus should be preferred over a non-self one. Conversely, in a volunteer carrying, the two human leucocyte allele (HLA)-A alleles mentioned above, we would expect that both peptides be recognized as self-stimuli; according to the logic above, the scores recorded for only solvent compared with the peptide of one of these alleles would have to be interpreted as indicating a preference for ‘non-self’. Hence, in each of the three comparisons listed by Natsch, there is a clear prediction: there is always one option that is ‘self’ compared with the other, be it the solvent alone or peptide in solvent. Therefore, the results could be pooled for testing whether the prediction is fulfilled or not. The number of data points obtained from the psychometric tests precluded a meaningful comparison of preferences for self-peptide to solvent or non-self-peptide. Nonetheless, our present analysis allows us to make an assessment of whether a ‘self’ option is preferred to a ‘non-self’ option, at least one of these options being a peptide. As shown in fig. 1 [1], the results are conclusive. In future experiments, it might be desirable to increase the number of study participants to be able to consider only those situations both where two different peptides are compared as ‘self’ and ‘non-self’ stimuli and where one peptide, self or non-self, is compared with solvent.

• (3) The third issue concerns what Natsch refers to as the human ‘peptide’ sense. Our paper [1] demonstrates by way of psychometric tests and fMRI analyses that humans can perceive and evaluate major histocompatibility (MHC)-related peptides. The question as to what constitutes an individuality signal in behavioural paradigms has been extensively debated in the literature. While discussing the results of Wedekind et al. [3] and Roberts et al. [4], Natsch fails to make a distinction between inbred and outbred populations with respect to the outcome of preference tests [5]. Although an outbred population would be expected to aim at optimizing MHC diversity (too little and too much diversity is to be avoided), an inbred population would always benefit from higher diversity (i.e. it can only get better). The transition from optimizing MHC diversity in outbred populations to maximizing it in inbred populations through olfactory choice has been experimentally demonstrated in fish [6]. Hence, there is a trivial explanation for a seemingly irreconcilable result. In discussing the experiments of Spehr et al. [7], Natsch implicitly addresses a general point of experimental design. Interesting as it may be to use natural sources to achieve a certain effect, proof for identification of an active principle depends on replicating the observed effects with synthetic versions of the implicated substance(s). It is interesting to note that the paper by Kwak et al. [8], cited by Natsch as evidence against the conclusions of our paper [1], lacks such confirmatory evidence; moreover, nor is any neurophysiological correlate of individuality signalling reported. While it is of course obvious that rodents can be trained to recognize MHC-related odours, it is not clear whether volatile substances are the key elements for discrimination under natural conditions. Indeed, other groups have identified a different type of molecule, namely the structurally polymorphic major urinary proteins (MUPs), that function in individuality signalling [9–11], again a case of a very large and non-volatile active principle, examined in both natural and synthetic forms; it should be noted, however, that the principle of polymorphic MUPs only applies to rodents. Our previous studies in fish [12,13] and mice [7,14] have conclusively shown that peptides convey MHC-related signals that are predictably interpreted in diverse behavioural contexts. Moreover, in a recent study, MHC peptide ligands have been detected in the urine of mice [11]. Given that the MHC is an evolutionarily conserved facility of the adaptive immune system in vertebrates, our current data support the hypothesis that humans also evaluate body odour through the recognition of MHC peptide ligands. It is therefore likely that MHC-ligand peptides represent an important odour signal that influences MHC-based mate-choice decisions to optimize the immunogenetic status in the offspring.