The high affinity of biotin to streptavidin has made it one of the most widely used affinity tags in proteomics. Early methods used biotin for enrichment alone and mostly ignored the biotin labeled peptide. Recent advances in labeling led to an increase in biotinylation efficiency and shifted the interest to detection of the site of biotinylation. This increased confidence in identification and provides additional structural information yet it requires efficient release of the biotinylated protein/peptide and sensitive separation and detection of biotinylated peptides by LC-MS/MS. Despite its long use in affinity proteomics the effect of biotinylation on the chromatographic, ionization, and fragmentation behaviour and ultimate detection of peptides is not well understood. To address this we compare two commercially-available biotin labels EZ-Link Sulfo-NHS-Biotin and Sulfo-NHS-SS-Biotin, the latter one containing a labile linker to efficiently release biotin to determine the effects of peptide modification on peptide detection. We describe an increase of hydrophobicity and charge reduction with increasing number of biotin labels attached. Based on our data we recommend gradient optimization to account for more hydrophobic biotinylated peptides and include singly charged precursors to account for charge reduction by biotin.
Biotin is an often-used affinity tag in many proteomic workflows. Its stable non-covalent interaction with its known binding partner, (strept)avidin, has been exploited to interrogate protein localization, synthesis and protein interaction. Traditionally most biotin-based labeling and enrichment approaches relied on indirect identification enrichment of biotinylated proteins and detection of enriched tryptic peptides. Direct detection of the biotinylated peptide has only recently become more popular for protein interaction and post-translational modification mapping as it provides direct evidence of the target protein and, more importantly, the site of biotinylation. With increased efficiency of the biotinylation reactions the number of biotin labels added to the protein of interest and their impact on the physicochemical properties of proteins and peptides increases. To date the potential effects of different biotin tags on mass spectrometric detection are only incompletely understood. Here we study two frequently used commercial biotin tags to guide optimal liquid chromatography and mass spectrometer configuration and enable accurate and complete identification of biotinylated peptides.
To enable detection of biotinylated peptides following enrichment, they first need to be released from their affinity matrix. While release from anti-biotin antibodies at low pH is straightforward, efficient reversal of the high-affinity interaction with streptavidin poses a challenge. One attractive solution is the introduction of a cleavable disulfide bridge linker between target protein and biotin moiety. This allows for release of the biotinylated protein or peptide in reducing conditions.
Typical acquisition by mass spectrometer relies on positive charges from the N terminus and C terminal lysine or arginine residue that resulted from trypsin digestion. However, because of its reactive amine group the same lysine is often the target for chemical modification including biotinylation. Modification of the lysine side chain can alter its ability to retain a positive charge and thus alter ionization, fragmentation and detection. Similarly, liquid chromatography is commonly optimized for primarily hydrophilic tryptic peptides and chemical adducts may alter hydrophobicity and chromatographic behavior of peptides. Here we hypothesize that the biotin tag significantly alters the physicochemical properties and detectability of peptides and that different biotin tags can fundamentally differ in their behaviour and suitability for MS detection.
We compare two commercially-available biotin labels: EZ-Link Sulfo-NHS-Biotin and EZ-Link Sulfo-NHS-SS-Biotin to determine the effects of peptide modification on peptide detection. Both reagents are water-soluble and possess the NHS ester moiety to facilitate biotinylation of primary amines. Biotin-NHS has a short linker, whereas biotin-SS-NHS has a longer linker that includes a disulfide bridge to release the captured protein from (strept)avidin in reducing conditions. We further characterize the effects of biotin modification compared to unlabeled peptides and suggest changes to chromatography as mass spectrometry methods to account for biotinylation induced changes.
10-15 µg prepared HeLa peptides were mixed with 100mM HEPES, pH=8.5 at a 1:2 ratio. The pH of the resulting solution was confirmed to be around pH=8.0-8.5. The commercially-available biotin was prepared in a stock solution of 20mM using 100mM HEPES, pH 8.5 as the diluant. The biotin was added to the HeLa peptides to a final concentration of 2mM biotin. For unlabeled peptides, an equivalent volume of 100mM HEPES, pH 8.5 was added. Samples were incubated on ice for 30 minutes. Labeling reaction was quenched using 50mM Tris-HCl, pH 8.0. For the reduction and alkylation of the biotin-NHS-SS label, a final concentration of 10mM DTT and 50mM CAA was added, respectively. During reduction, samples were incubated at 37°C for 30 minutes; alkylation was conducted at room temperature for 30 minutes.
Samples were prepared for LC-MS/MS analysis or other downstream assays using STaGE tips as prepared in Rappsilber et al.. Briefly, two small circular Empore SPE C18 disks were punched out using a flat-end needle. A straightened paper clip was used to position the disks into a P200 pipette tips. STaGE tips were conditioned with 40 µL methanol, 40 µL 0.1% Formic Acid, 60% Acetonitrile, and 40 µL 0.1% Trifluoroacetic Acid. Samples pH was adjusted to pH=2.0–3.0 using 10% TFA prior to loading. Samples were eluted with 40 µL 0.1% FA, 60% ACN. The ACN was eliminated from the samples via SpeedVac. Samples were re-suspended in 0.1% FA.
Peptide concentration and total were determined using the Nanodrop. A total of 1 µg peptide per sample was injected for analysis. Mass spectrometric analyses were performed on Q Exactive HF Orbitrap mass spectrometer coupled with an Easy-nLC 1200 liquid chromatography system. Buffer A was 2% Acetonitrile and 0.1% Formic acid. Buffer B was 95% ACN and 0.1% FA.
For comparison between biotin-NHS and biotin-SS-NHS samples, a 35-cm homemade analytical column with pre-column was used. Liquid chromatography gradient was at a flow of 300 nL/min using the following 67-minute gradient profile: 0:3, 3:8, 40:27, 52:42, 53:90, 60:90, 67:100. Top 12 method with a full-scan MS spectrum with mass range of 350-1660 m/z was collected at a resolution of 120000, maximum injection time of 30 ms, and an AGC target of 2e5, MS/MS scan was acquired at 15000 resolution, maximum injection time of 60 ms, and an AGC target of 2e5. Normalized collision energy was set to 28. Dynamic exclusion was set to 30 s. Charge state exclusion was set to ignore unassigned, +1, +5 and greater charges. Comparison between biotin-NHS and unlabeled samples, and comparison between inclusion and exclusion of singly-charged peptides were done on a 50-cm µPAC column, with and without pre-column, respectively. Liquid chromatography gradient was at a flow of 300 nL/min using the following 85-minute gradient profile: 0:4, 5:9, 10:10, 15:12, 20:14, 25:15, 30:17, 35:18, 40:19, 45:21, 50:24, 55:27, 60:80, 85:80. Top 12 method with a full-scan MS spectrum with mass range of 400-1800 m/z was collected at a resolution of 60000, maximum injection time of 75 ms, and an AGC target of 3e6. MS/MS scan was acquired at 15000 resolution, maximum injection time of 50 ms, and an AGC target of 5e4. NCE was set to 28. Dynamic exclusion was set to 20 s. Charge state exclusion was set to ignore unassigned, +1, +5 and greater than +8 charges, but +1 was removed from the list during the inclusion of singly-charged peptides.
Fluorescence signal was measured from biotinylated peptides using Quantitative Fluorometric Peptide Assay and following the manufacturer’s protocol. An unlabeled peptide sample and 0.1% FA in water were used as controls for minimum and maximum labeling, respectively. To confirm peptide amounts from the fluorometric assay were comparable, peptide totals were measured using Quantitative Colorimetric Peptide Assay.
Raw MS DDA data acquired from the Q Exactive HF were searched with MaxQuant using the built-in Andromeda search engine, and embedded standard Orbitrap settings which included first search peptide tolerance at 20 ppm and main search peptide tolerance at 4.5 ppm. The false discovery rate for protein, peptide, and PSM were set at 1%. Trypsin/P specific digestion mode was used. Carbamidomethyl was set as fixed modification, but was set as variable modification during the assessment of carbamidomethyl cysteines. Oxidation and Acetyl were set as dynamic modification. Additional dynamic modifications, as described in Supplementary Figure 1, were assigned for lysine and N terminus, as needed. For the comparison between biotin-NHS and biotin-NHS-SS, raw files for each biotinylated samples were searched separately but grouped with unlabeled controls. The human protein database was downloaded from Uniprot. For the comparison between biotin-NHS and unlabeled controls, and comparison between inclusion and exclusion of singly-charged peptides, the human protein database was downloaded from Uniprot. Common contaminants were embedded from MaxQuant. MaxQuant peptide identifications were obtained from the evidence.txt file output. Peptide sequences that were labeled as Potential contaminant or Reverse were excluded from further analysis. Determination of labeling efficiency and other comparison of features were done using R. All statistical tests were done using t-tests. For the statistical testing of the amino acid frequencies at the N terminus, an adjusted p-value was calculated using Benjamini and Hochberg method. All box plots prepared were in the style of Tukey: the box boundaries represent Q1 and Q3 while the line within the box represents the median; whiskers extend from the upper and lower quartiles to the maximum and minimum values, respectively, with the outliers excluded. Outliers are defined as 1.5 × IQR of Q1 or Q3 and are displayed as points beyond the minimum and maximum values, respectively.
Tryptic peptides were prepared from HeLa lysates and were labeled with either biotin-NHS or biotin-SS-NHS. During biotinylation, primary amines at the N terminus or lysine residues makes a nucleophilic attack on the NHS ester moiety present in the biotin labels. This yields in a stable amide bond linking biotin and the peptide. To confirm biotinylation, we assessed the relative number of free amino groups in the peptides by conducting an assay using an amine-reactive fluorescent agent. In this assay, biotinylated amino groups are blocked from reacting with the fluorescent agent. Therefore, a fully biotinylated sample would yield a lower fluorescence signal. We found that both biotinylating reagents effectively labeled amine groups in the samples, with labeling efficiencies observed to be greater than 90%.
