Premature ovarian failure (POF) is a frequent pathology leading to female infertility. Clinically, this disorder is characterized by amenorrhoea under 40 years and high plasma gonadotrophins levels. The high incidence of cases considered as idiopathic is mainly due to a poor understanding of the complex interactions between genetic and environmental factors underlying this condition. BMP15, as well as its close paralog GDF9, have attracted much interest in the field of human reproduction. Since the first report of a pathogenic BMP15 mutation (p.Tyr235Cys) in a POF case,1 genetic screenings of large panels of women presenting with nonsyndromic POF have revealed further potentially pathogenic variants.2 BMP15 and GDF9, as other members of the TGF-β family, are synthesized as precursors, involving a signal peptide, a pro-region and a C-terminal region corresponding to the mature peptide. Post-translational processing includes signal peptide removal, dimerization and a further cleavage in order to release the bioactive mature peptide. Up to now, some BMP15 mutations located in the propeptide (i.e. p.Tyr235Cys, p.Leu148Pro, p.Arg68Trp, p.Glu211X) have been considered as the best candidates to have pathogenic effects. However, the involvement of some BMP15 variants in the aetiology of POF is still unclear.2, 3 A recent publication describes two novel BMP15 heterozygous mutations (p.Ser5Arg, p.Arg138His) in women with non-syndromic POF as well as four previously known substitutions.4 This work, using an elegant functional reporter assay of BMP15 activity, helps assess whether these variants are involved in POF or not. Their results provide strong evidence for a functional impact of previously known and novel variants (i.e. p.Leu148Pro, p.Arg68Trp and p.Arg138His). They confirm that BMP15 mutations, involving rather drastic amino acid changes at conserved positions, interfere with normal protein function. This is the case of p.Leu148Pro and p.Arg68Trp that we and others have previously noticed but whose functional effects were to be shown. Interestingly, co-expression of the mutant and wild-type (WT) proteins leads to a decreased bioactivity. This points to a mature-protein concentration insufficiency. However, it is not clear whether this is linked to a dominant negative effect (DNE) or not. Such a possibility would not be unprecedented since BMP15 acts as a dimer, and abnormal dimers containing processed and unprocessed chains have been reported for the mutant p.Tyr235Cys.1 Very recently, we have described in Clinical Endocrinology, the potentially damaging variant p.Ser5Arg in a POF patient of Tunisian origin,5 inherited from her mother who had POF at 36 years old. Our in silico analyses predicted a quantitative alteration (but not an abolition) of the processing of the signal peptide. In agreement with this prediction, Rossetti et al.4 found a significantly decreased bioactivity of the mutant protein when expressed alone (−15% with respect to the normal protein) even though a decrease in protein processing was not detectable in the Western-blot experiments. However, in their experimental conditions, when BMP15-Ser5Arg was co-expressed with the WT protein the global activity was not different from the WT level. As already said, this was not the case for seemingly more damaging variants such as p.Arg138His, p.Leu148Pro, p.Arg68Trp. This has led the authors to consider p.Ser5Arg as weakly pathogenic (if at all). However, the case of the p.Ser5Arg and of other mutations can be revisited. Let us first consider the case of p.Leu148Pro, which seems to be one of the worst mutations since the pre-proprotein is hardly processed (according to figure 1 of Rossetti et al.4). Intriguingly, the virtual absence of mature peptide does not correlate with the retention of 2/3 of the WT activity in the functional assay. This can be explained by assuming that the reporter system, in their experimental conditions, is extremely sensitive to BMP15 and therefore working near saturation. Thus, even a low amount of BMP15-Leu148Pro would be enough to elicit a rather strong signal transduction. In such circumstances, even a DNE induced by this mutation might be overlooked. To confirm or infirm the existence of such a DNE it would be interesting to test this mutation (as well as p.Arg138His, p.Arg68Trp) with the biological assays previously used for the analysis of p.Tyr235Cys. In further studies, before reaching a firm conclusion on the pathogenicity of a BPM15 ‘variant’ a dose-response assessment of mutant BMP15 activity using the reporter system described in ref. Rossetti et al.4, complemented as much as possible by other bioassays, should be recommended and even necessary. The dosage-response analysis seems also necessary for mutations in the mature peptide, because it can be shown that DNEs in such cases are also protein-concentration dependent. The explanation runs as follows. Let us consider that the interaction between normal (A) and abnormal (A′) BMP15 monomers leads to impaired dimers (AA′ and A′A′). This is because AA has two ‘normal arms’ to interact with the receptor while AA′ and A′A′ have either one or none respectively. Accordingly, when the BMP15 receptor is far from being saturated only normal dimers (i.e. AA) will effectively recognize it (∼25% of productive complexes, which is normally not enough to ensure a normal phenotype). However, as dimer concentration increases, the receptor becomes saturated by normal dimers, which easily outcompete the abnormal ones (more details in ref. 6). This can completely mask the DNE. Careful analyses can also be critical in the case of mild alterations, which are, in fact, the most likely to be encountered. Indeed, they might act as predisposing factors requiring the co-occurrence of alterations at other loci and can therefore persist in the population when ‘travelling’ alone or through the males. Premature ovarian failure can be considered as a rather quantitative character (i.e. at least its age of onset), because any interruption of menses for more than 6 months before 40 may lead to a diagnosis of POF. If the almost-certainly damaging p.Leu148Pro (−33% of activity with respect to the normal protein) is associated with a secondary amenorrhoea, then it is conceivable that, for instance, that p.Ser5Arg (−15% of activity) might also lead to POF (presumably with a later-onset?). Finally, it is worth keeping in mind that BMP15 mutations and variants might also be interfering with GDF9 through the formation of impaired heterodimers. A potential functional impact of these dimers can now be easily tested with the useful assays described by Di Pasquale et al.1 and Rossetti et al.4 We thank two anonymous referees for their helpful comments on this MS. R.A.V. and his laboratory is funded by Institut Universitaire de France (IUF), Association pour la Recherche contre le Cancer (ARC), the CNRS and Université Paris VII.