Corrigendum: The human milk proteome and allergy of mother and child: exploring associations with protein abundances and protein network connectivity

Text Corrections

Corrections have been made to Materials and methods, Statistical methods, Missing data. This sentence previously stated:

In practice this resulted in a minimum of 66 and a median of 215 valid values.

The corrected sentence appears below:

In practice this resulted in a minimum of 49 and a median of 209 valid values.

Corrections have been made to Materials and methods, Statistical methods, Principal component analysis. This sentence previously stated:

For unsupervised data exploration, Principal Component Analysis (PCA) (40) was applied on the 300 × 647 data matrix (samples × proteins), using the FactoMineR package for R (41).

The corrected sentence appears below:

For unsupervised data exploration, Principal Component Analysis (PCA) (40) was applied on the 300 × 687 data matrix (samples × proteins), using the FactoMineR package for R (41).

Corrections have been made to Materials and methods, Statistical methods, Network inference and analysis - Covariance simultaneous component analysis (COVSCA). This sentence previously stated:

This fit was chosen as the best compromise between goodness of fit (68%) and the complexity of the COVSCA model (rank and number of the prototypical matrices).

The corrected sentence appears below:

This fit was chosen as the best compromise between goodness of fit (74%) and the complexity of the COVSCA model (rank and number of the prototypical matrices).

Corrections have been made to Results, paragraph one. This paragraph previously stated:

Proteomic analysis of all samples led to a total of 1629 identified proteins before filtering on missing values. After filtering these proteins on the requirement of being identified ≥ 25 times in at least one of the four mother-child allergy groups, 647 proteins remained for further data analysis. In this filtered dataset, the number of identified proteins per sample ranged between 256 and 586 (median = 458). The major milk proteins α-lactalbumin, albumin, lactoferrin, β-casein, and α_s1-casein, were in all analyzed samples among the 15 most abundant proteins. A complete overview of the 647 identified proteins can be found in Supplementary Table 1.

The corrected paragraph appears below:

Proteomic analysis of all samples led to a total of 1690 identified proteins before filtering on missing values. After filtering these proteins on the requirement of being identified ≥ 25 times in at least one of the four mother-child allergy groups, 687 proteins remained for further data analysis. In this filtered dataset, the number of identified proteins per sample ranged between 242 and 636 (median = 480). The major milk proteins α-lactalbumin, albumin, lactoferrin, β-casein, and α_s1-casein, were in all analyzed samples among the 15 most abundant proteins. A complete overview of the 687 identified proteins can be found in Supplementary Table 1.

Corrections have been made to Results, Univariate analysis. This section previously stated:

Differences in protein abundance between the different mother-child allergy groups were assessed with Kruskal-Wallis tests. After correction for multiple hypothesis testing, no significant differences were found among the four groups (Table 1). Kruskal-Wallis outcomes with uncorrected p < 0.05 were further assessed with Dunn’s post-hoc tests and subsequent correction for multiple testing, which resulted in 23 proteins that showed a difference between the groups with corrected p < 0.05 (Table 1). Most of these differences (n = 15) were found between the non-allergic group (M-C-) and the group where only the child ultimately developed an allergy (M-C+). Proteins that differed between these groups were primarily Ig chains (11 out of 15) and were mostly higher in abundance in the group where the mother was non-allergic and the child developed an allergy (Figure 3). Additionally, 4 of these Igs show also a higher abundance in milk from allergic mothers with children who did not develop an allergy. Further investigation of all identified Ig proteins showed that the mean abundance of these proteins is generally lower in the groups where mother, child or both are allergic, when compared to the non-allergic group (Figure 4). This effect is clearest in the comparison of the group where only the child developed an allergy with the group where both mother and child are nonallergic. Out of 81 Ig proteins, 75 have a mean abundance that is higher in the group where the child developed an allergy.

Figure 3

Figure 4

The corrected section appears below:

Differences in protein abundance between the different mother-child allergy groups were assessed with Kruskal-Wallis tests. After correction for multiple hypothesis testing, no significant differences were found among the four groups (Table 1). Kruskal-Wallis outcomes with uncorrected p < 0.05 were further assessed with Dunn’s post-hoc tests and subsequent correction for multiple testing, which resulted in 30 proteins that showed a difference between the groups with corrected p < 0.05 (Table 1). Most of these differences (n = 17) were found between the non-allergic group (M-C-) and the group where only the child ultimately developed an allergy (M-C+). Proteins that differed between these groups were primarily Ig chains (15 out of 17) and were mostly higher in abundance in the group where the mother was non-allergic and the child developed an allergy (Figure 3). Additionally, 3 of these Igs show also a higher abundance in milk from allergic mothers with children who did not develop an allergy. Further investigation of all identified Ig proteins showed that the mean abundance of these proteins is generally lower in the groups where mother, child or both are allergic, when compared to the non-allergic group (Figure 4). This effect is clearest in the comparison of the group where only the child developed an allergy with the group where both mother and child are nonallergic. Out of 83 Ig proteins, 77 have a mean abundance that is higher in the group where the child developed an allergy.

Corrections have been made to Results, Non-human proteins, paragraph one. This sentence previously stated:

In the current study, several non-human proteins were identified (n = 9), including albumin from dog, horse, and cat, as well as bovine α_s1-casein and BLG (Table 2).

The corrected sentence appears below:

In the current study, several non-human proteins were identified (n = 11), including albumin from dog, horse, and cat, as well as bovine α_s1-casein and BLG (Table 2).

Corrections have been made to Results, Network analysis, Network inference, paragraph 2. This paragraph previously stated:

To investigate this pattern further, proteins with differential connectivity > 50 were selected for further investigation (Supplementary Table 1). These proteins had the largest differences in connectivity among the four different groups and were selected for further functional analysis, to determine possible functional consequences of the differences between the networks. The selection resulted in 173, 171, and 153 proteins for the comparison of the group with non-allergic mother and child with respectively (i) allergic mother and non-allergic child, (ii) non-allergic mother and allergic child, and (iii) allergic mother and child groups. From these proteins, 95 proteins occurred in all three selections, showing a similarity in differential connectivity. Interestingly, GO overrepresentation analysis of these proteins showed a significant overrepresentation of proteins involved in translation initiation (p = 1.08×10⁻¹⁵). This overrepresentation is due to 24 ribosomal proteins and translation initiation factors (EIF3A, EIF4A1, EIF5A).

The corrected paragraph appears below:

To investigate this pattern further, proteins with differential connectivity > 50 were selected for further investigation (Supplementary Table 1). These proteins had the largest differences in connectivity among the four different groups and were selected for further functional analysis, to determine possible functional consequences of the differences between the networks. The selection resulted in 160, 168, and 144 proteins for the comparison of the group with non-allergic mother and child with respectively (i) allergic mother and non-allergic child, (ii) non-allergic mother and allergic child, and (iii) allergic mother and child groups. From these proteins, 79 proteins occurred in all three selections, showing a similarity in differential connectivity. Interestingly, GO overrepresentation analysis of these proteins showed a significant overrepresentation of proteins involved in translation (p = 9.13×10⁻⁹). This overrepresentation is due to 23 ribosomal proteins and translation elongation factor EEF1A1P5.

Corrections have been made to Results, Network analysis, Network modeling, paragraph 3. This paragraph previously stated:

The loadings for component 1 (see Figure 8) are overrepresented by proteins involved in gluconeogenesis (p = 0.0003), the synthesis of glucose. This component accounts for separation between the non-allergic group and the other three groups. The second component, which drives the separation of the groups on allergy status of the child, shows a significant overrepresentation of proteins involved in the positive regulation of DNA biosynthetic processes (p = 0.0013). This overrepresentation is mainly due to 5 members of the tailless complex polypeptide 1 ring complex (TRiC or CCT). In addition, several proteins involved in translation processes show differences in correlation patterns on this component.

The corrected paragraph appears below:

The loadings for component 1 (see Figure 8) are overrepresented by proteins involved in gluconeogenesis (p = 0.003), the synthesis of glucose. This component accounts for separation between the non-allergic group and the other three groups. The second component, which drives the separation of the groups on allergy status of the child, does not show a significant overrepresentation of gene ontology terms. However, among these proteins are 11 proteins involved in translation as well as 3 members of the tailless complex polypeptide 1 ring complex (TRiC or CCT).

Corrections have been made to Discussion, Differences in immunoglobulin abundances between groups with different allergy statuses, paragraph 6. This sentence previously stated:

Notably, soluble CD14, a protein in human milk that may be protective against the development of food allergies (56, 57), was not different between the allergy groups in our study (uncorrected p = 0.53).

The corrected sentence appears below:

Notably, soluble CD14, a protein in human milk that may be protective against the development of food allergies (56, 57), was not different between the allergy groups in our study (uncorrected p = 0.43).

Error in Figure/Table

For the same reason as explained above, corrections have been made to Tables 1–3, and Figures 3–8.

Figure 5

Figure 6

Figure 7

Figure 8

The corrected Tables and Figures appear below.

Incorrect Supplementary Material

For the same reason as explained above, corrections have been made to Supplementary Figure 1 and Table 1.

The correct Supplementary Material has been added to the original article.

The authors apologize for these errors and state that this does not change the scientific conclusions of the article in any way.