LAUSR

The correspondence from Lauc et al.1 argues that a high level of H5N4S2 (abbreviated as Hex5HexNAc4NeuAc2 in our original article2, which has a combined relative abundance of 5.4 %) indicates contamination of other plasma glycoproteins in our immunoglobulin G (IgG) samples. Therefore, they speculate that many trace glycans identified in our study originated from other plasma glycoproteins, but not IgG. We are thankful for the interest in our work; however, we do not agree with this conclusion as it is inferred from a generalization of a “relative abundance-based indicator”, incorrect interpretation of our data, and more importantly, incomplete consideration of the intrinsic high sensitivity and specially improved detection sensitivity for acidic glycans of our method. Herein, we address the query point by point and provide additional experimental results in support of our published article. Firstly, Lauc et al.3–10 claim that the relative abundance of H5N4S2 on IgG is below 1% in their >30,000 samples, and therefore a high level of H5N4S2 can be considered as an indicator of contamination. However, because relative abundance is merely calculated from the signal response without using standard, therefore can vary with the analytical method and instrument conditions, it is not suitable to generalize a relative abundance-based indicator to the studies of another lab (especially labs that use different analytical approaches). It is common to observe varied relative levels of glycans across studies of different labs. Taking H5N4S2 as an example, relative abundance varying from 0.57 to 10% has been reported11,12. As aforementioned, the relative level of H5N4S2 cannot be simply used to indicate contamination across studies. Another issue is that we captured IgG from serum by using protein A. Therefore, it is not applicable to speculate non-specific binding of denatured protein to protein G (a phenomenon often occurred during the purification of IgG from dried bloodstrains as mentioned by Lauc et al.) based on the observation of >5% H5N4S2 to our study. Of note, our method differs from that of Lauc’s lab in almost every step3–10,12, and was developed specifically to improve the detection of acidic glycans by: (1) on-chip enrichment (afford up to 25-fold signal gain); (2) a separated and optimized mobile phase for acidic glycans (enhanced the signal intensity of acidic glycans by approximately fivefold); and (3) optimized multiple reaction monitoring (MRM) parameters for quantification of each glycan species (further enhanced the detection of lowabundance species by ~1000-fold). Because of these improvements, the relative level of acidic glycans, especially those lowabundance acidic glycans, are commonly higher than the values reported by Lauc et al., while neutral glycans’ level are generally lower3–10,12. Hence, the relative high level of H5N4S2 in our study is mostly a result of the improved detection, not from other plasma glycoproteins. We would also clarify that the particularly high standard deviation mentioned by Lauc et al. refers to the deviation of the levels of ten healthy individuals (see Supplementary Data 4 of the original article2), but not the deviation of the calculated levels of the same sample. So the “particularly high standard deviation” indicated great individual difference, which is quite common and reasonable. It is not a correct judgement of DOI: 10.1038/s41467-018-05082-y OPEN