Practical Guide to Protein Composition Analysis
Most people approach protein sequencing by opening a textbook and reading about primary and secondary structure. That rarely helps when you're standing at a mass spectrometer with a sample that won't ionize. Protein composition analysis is less about theory and more about surviving the experimental details. I'll walk through what actually works in the lab, including the mistakes I've made and how to avoid them. Protein composition refers to the identity and proportion of amino acids making up a polypeptide chain. At its core, you are trying to determine which twenty standard amino acids are present, in what ratios, and how they are ordered. Modern labs mostly use mass spectrometry for this, but the field also relies on Edman degradation for smaller peptides and amino acid analysis by HPLC for bulk composition. Each method has trade-offs that matter in practice. Here is the sequence I follow when analyzing an unknown protein. It is not the most elegant, but it produces results consistently.
This is where most problems originate. If your sample is degraded or improperly prepared, no amount of downstream optimization will save it. For purified recombinant proteins, I typically reduce disulfide bonds with DTT at 56°C for thirty minutes, then alkylate with iodoacetamide in the dark for twenty minutes at room temperature. The alkylation step is critical because unreduced cysteines will scramble your fragmentation pattern entirely. I once spent three days troubleshooting a peptide identification that refused to make sense. The mass shifts were inconsistent across replicates. I eventually realized the DTT had oxidized during storage and wasn't actually reducing the disulfide bonds. Fresh DTT prep solved it immediately. Bad reagents are a silent problem because the data looks plausible until you check the chemistry.
Enzymatic Digestion
Trypsin is the standard enzyme because it cleaves at lysine and arginine, producing peptides in the ideal mass range for most instruments. I use a 1:50 enzyme-to-substrate ratio and digest overnight at 37°C. If the protein is heavily glycosylated or contains unusual modifications, trypsin sometimes stalls at certain sites. In those cases, I supplement with Lys-C, which cuts more aggressively at lysine residues and often recovers peptides that trypsin alone misses. For intact protein analysis, top-down MS gives you the full molecular weight and can detect modifications directly. For peptide-level identification, which is far more common, you run LC-MS/MS. The liquid chromatography separates the digest into individual peptides, and the mass spectrometer measures their masses and fragmentation patterns. You feed the data into a search engine like Sequest, Andromeda, or Mascot, which matches observed spectra against a theoretical database. The database choice matters enormously. If you are analyzing a non-model organism or a synthetic construct, you need to make sure your reference database includes the correct sequence. A missing gene entry means the search engine will either return low-confidence matches or nothing at all. I build custom databases for recombinant work to avoid this entirely.
Get the Full Details
Data Interpretation
Software generates candidate peptide identifications with associated score values, but automatic acceptance is a mistake. I manually inspect the spectrum for confident calls, particularly around modified peptides. The b-ion and y-ion series should cover the peptide sequence with minimal gaps. When they do not, the identification is likely incorrect or the modification is something the search engine was not configured to find. There are a few things that separate competent analysis from amateur results, and they are not covered in basic protocols. Chemical modifications are not anomalies. Oxidation of methionine, deamidation of asparagine and glutamine, and N-terminal pyroglutamate formation are common artifacts of sample handling. They shift peptide masses in predictable ways. If your search engine is not configured to account for these variable modifications, you will miss real identifications or misassign peptides. I typically allow methionine oxidation and N-terminal acetylation as variable modifications and keep the false discovery rate below one percent.
Partial digestion products are useful data. Not every peptide cuts cleanly at lysine or arginine. These partial products contain information about nearby modifications and can resolve ambiguous sequence regions. A beginner discards them as noise. An experienced analyst uses them to fill gaps.
Methods and When They Fail
No single technique handles every protein. Here is a blunt assessment of the main approaches. This method sequentially removes and identifies amino acids from the N-terminus. It is reliable for peptides under fifty residues and does not require a database. It fails completely when the N-terminus is blocked by acetylation or formyl groups, which is common in eukaryotic proteins. It also becomes unreliable after about thirty cycles due to cumulative yield loss. I use it when I need unambiguous N-terminal sequencing but rely on MS for everything else. This gives you the bulk amino acid composition after acid hydrolysis. It is accurate for ratios but destroys all sequence information. Tryptophan is destroyed during standard acid hydrolysis, so you need a separate measurement for that residue. This method is useful for quality control and verifying expression constructs, but it cannot tell you the order of residues.
This analyzes the intact protein and fragments it inside the mass spectrometer. It preserves information about combinatorial modifications that bottom-up approaches can miss. The downside is that it requires high-resolution instruments and works poorly for proteins above thirty kilodaltons. Most of the proteins I analyze are well above that threshold, so I rarely use this approach. This is the workhorse method. Digest the protein, separate the peptides, identify them by MS/MS, and reconstruct the sequence computationally. It handles large proteins well and benefits from decades of instrument and algorithm improvement. The main limitation is that it cannot resolve the exact order of identical peptides from different regions, and post-translational modifications can complicate database searches if they are unexpected. I analyzed a membrane protein expressed in insect cells a few years ago. The standard protocol produced almost no peptide identifications. The protein was hydrophobic and tended to aggregate during digestion. Trypsin could not access the cleavage sites effectively.
Instead of forcing the standard protocol, I switched to using 8M urea with one percent RAPIz surfactant in the digestion buffer. RAPIz is a zwitterionic detergent that is mass spectrometry compatible and does not suppress ionization the way SDS does. The digestion efficiency improved dramatically, and I went from fifteen identified peptides to over two hundred in a single run. The total analysis time dropped from three days to about four hours.
Common Pitfalls
Contamination is the most persistent problem. Keratin from skin and hair is everywhere in a lab environment. It shows up as a cluster of peptides in every spectrum and can dominate your results if you are not careful. I use clean bench practices, cover samples when not in use, and include a blank run at the start of every batch to identify contaminant peaks. Another issue is sample concentration. Too dilute and you get poor ionization. Too concentrated and you get aggregate formation that clogs the column. I aim for one microgram per microliter as a starting point and adjust based on instrument response. Database searches with inappropriate parameters waste significant time. If you search with too restrictive a mass tolerance, you miss real identifications. If you search with too loose a tolerance, you get false positives. I use a ten ppm precursor tolerance and a twenty ppm fragment tolerance for high-resolution instruments. Lower resolution instruments require wider tolerances, usually fifty ppm for precursors.
When to Consider Alternatives
If mass spectrometry is unavailable or the protein is extremely difficult to digest, consider N-terminal sequencing by Edman degradation combined with amino acid analysis for bulk composition. This older approach is slower and provides less detail, but it does not depend on database availability and can work for proteins from organisms with no genomic data. It is also useful for validating MS results when you suspect a misidentification. For large-scale compositional studies across many samples, targeted proteomics with multiple reaction monitoring can provide quantification that standard shotgun proteomics cannot. It requires knowing the peptides of interest in advance but delivers better reproducibility and lower limits of detection. The fundamentals remain the same regardless of the method: clean samples, proper controls, and careful interpretation of the data. The instruments and algorithms improve every year, but the basic chemistry does not change. Disulfide bonds still need reduction. Tryptic peptides still fragment along the backbone. Contaminants still show up where they are not wanted. Understanding these constants matters more than keeping up with every new software release.
