Figure 1. Flowchart of the BioID2 application to detect in situ
interactions supporting therapeutic proteins secretion.
Figure 2. Expression of bait-BirA proteins results in a
substantial increase in biotinylated proteins. a) Successful secretion
of the bait-BirA proteins into the culture supernatant was evaluated by
Western blot using HRP-anti-flag antibody. b) The immunoblotting
biotinylation profiling of the model proteins and WT control in HEK293
cells with HRP-streptavidin. When the BirA domain was fused to the model
proteins, biotin addition led to the biotinylation of a subset of
proteins (B+) which are not seen in WT or absence of biotin. This
demonstrates that the BioID labeling system tags interactions as
secreted proteins are synthesized and trafficked through the secretory
pathway. A few endogenously biotinylated proteins appear in the absence
of biotin and in the WT.
Figure 3. Bait-BirA fusion proteins are colocalized with
biotin-staining. Co-Immunofluorescence demonstrated the intracellular
colocalization of the biotin-labeled proteins (stained with
Streptavidin-Dylight 594 and illustrated in green color) and bait-BirA
(stained with anti-flag monoclonal antibody-Dylight 650 and illustrated
in red color), while WT did not show increased biotinylation under the
same experimental conditions.
Figure 4. Dozens of proteins show significantly increased
biotinylation after expression of bait-BirA proteins. Volcano plot
showing the distribution of the quantified biotinylated proteins by MS
according to p-value and fold change. As depicted the bait-protein
significantly showed the highest fold change compared to WT almost in
all cases, indicating the capability of the BioID labeling system to tag
the in vivo interactors within the live cells. SeAP is a truncated form
of Alkaline phosphatase, placental type (ALPP).
Figure 5. Interacting proteins are enriched for secretory
pathway machinery. (a) To determine if significant interactions enrich
for secretory pathway-related genes, we performed an iterative
enrichment analysis in which we included the most significant
interactions first and iteratively added interactors with lower fold
changes. The y-axis indicates the overall coverage of 3 secretory
pathway-related gene sets and the x-axis the significance cutoffs (rank
ordered by fold change). The coverage of the gene set (top) along with
their corresponding hypergeometric enrichment p-values (bottom) are
shown. The top 300 hits for each secretable BirA sample (Fig. S3 for all
hits) showed significant enrichment of the secretory pathway components
and co-secreted proteins among the most significant hits for all samples
except Signal-BirA (which is a lone secreted BirA and not a mammalian
secreted protein). (b) Quantified interactions between interactors
(x-axis) and the model proteins (y-axis), where the shadowed entries
indicate significant interactions. The features of the model proteins,
detailed in Fig. 6, are summarized for each model protein on the left
and the interactors are labeled on top based on their secretory pathway
attributes.
Figure 6. The bait proteins show diversity in their PTM and
structural content. Dots and lines represent known PTM sites and
structural motifs in panel (a) and (b) respectively. The hills and
valleys indicate protein tertiary features. Note that solvent
accessibility and structural motifs are only available for regions
covered by the PDB structure, whereas predicted features such as protein
hydrophobicity and disorder are available for the entire protein.
Figure 7. Detected interactors correlate with protein features.Interactors were associated with specific (a) PTMs and (b) structural
features of model proteins. The heatmap shows the standardized
interaction affinities estimated between certain interactors and PTMs or
structural features across all model proteins (see methods). Only
interactors having significant associations with model protein features
are shown.