Allele-Independent Turnover of Human Leukocyte Antigen (HLA) Class Ia Molecules

Abstract

Major histocompatibility complex class I (MHCI) glycoproteins present cytosolic peptides to CD8+ T cells and regulate NK cell activity. Their heavy chains (HC) are expressed from up to three MHC gene loci (human leukocyte antigen [HLA]-A, -B, and -C in humans), whose extensive polymorphism maps predominantly to the antigen-binding groove, diversifying the bound peptide repertoire. Codominant expression of MHCI alleles is thus functionally critical, but how it is regulated is not fully understood. Here, we have examined the effect of polymorphism on the turnover rates of MHCI molecules in cell lines with functional MHCI peptide loading pathways and in monocyte-derived dendritic cells (MoDCs). Proteins were labeled biosynthetically with heavy water (2H2O), folded MHCI molecules immunoprecipitated, and tryptic digests analysed by mass spectrometry. MHCI-derived peptides were assigned to specific alleles and isotypes, and turnover rates quantified by 2H incorporation, after correcting for cell growth. MHCI turnover half-lives ranged from undetectable to a few hours, depending on cell type, activation state, donor, and MHCI isotype. However, in all settings, the turnover half-lives of alleles of the same isotype were similar. Thus, MHCI protein turnover rates appear to be allele-independent in normal human cells. We propose that this is an important feature enabling the normal function and codominant expression of MHCI alleles.

Fig 1. Measuring MHCI protein turnover by ²H₂O labeling.

(A) SINEW work flow. See text for details. (B) New protein synthesis may support cell growth, increase net protein levels per cell, or replace protein lost to turnover. These processes contribute additively to protein synthesis. (C) Assignment of peptides to particular MHCI alleles or isotypes. First, LC-MS/MS data (Step 4 in panel A) are screened against sequence databases to identify tryptic fragments derived from any MHCI molecules (top, color-coded boxes). MHCI alleles present in each donor are identified by HLA genotyping, and their predicted amino acid sequences are subjected to tryptic digestion in silico. These virtual digests are compared with each other to identify a subset of peptides that are specific to particular alleles or isotypes (symbolised by boxes with text labels).

Fig 2. Effect of ²H₂O labeling on peptide mass isotopomer distributions.

(A) Mass isotopomer distributions of a MHCI-derived, B isotype-specific tryptic peptide from KG-1 cells after labeling with ²H₂O for various times. Lines connect data at each time point. Within error, each mass isotopomer changed from its initial (unlabeled) to final plateau (fully-labeled) value at the same rate, which is identical to the rate of protein fractional synthesis. (B) For the same peptide, MIDA models for the unlabeled and fully-labeled mass isotopomer distributions (dashed and solid lines, respectively) were compared with experimental data (symbols). RMSD values were 0.20% and 0.25%, respectively, for unlabeled and fully-labeled samples).

Fig 3. MHCI turnover in KG-1 cells.

Proliferating KG-1 cells were labeled with ≈ 5% ²H₂O in media and MHCI molecules immunoprecipitated with W6/32. ²H incorporation into selected peptides (S1 Table; identified here by the four N-terminal amino acids and assigned to isotypes and alleles as shown) was quantified by LC-MS. One of two independent experiments is shown here; S1 Table summarizes results for both. (A-C) Fractional synthesis was calculated for different peptides derived from HLA-A (A), HLA-B (B), and HLA-C molecules (C) (mean ± SD of the informative mass isotopomers) and plotted against time. In (B) and (C), allele- and isotype-specific peptides exhibited no significant differences in fractional protein synthesis (p = 0.24 and p = 0.55, respectively, by F test). Single-exponential curve fits (with 95% confidence intervals) are based on a pooled analysis of all peptides from each isotype. (D) Exponential growth of KG-1 cells during ²H₂O labeling. The corresponding time course of the fraction of new cells is shown in panels (A-C) for comparison with protein synthesis. (E) Fractional synthesis rates (per hour, mean ± SEM) of MHCI isotypes (from (A-C)), compared with cell growth (from (D)). The differences between HLA-C and the other isotypes, and those between the MHCI fractional synthesis rates and cell growth, were significant (each p < 0.0001, F test). (F) Turnover half-lives of different MHCI isotypes, calculated from the excess of mean fractional protein synthesis rates over the cell growth rate.

Fig 4. MHCI turnover in LCL721 cells.

LCL721 cells were labeled with ≈ 5% ²H₂O in media and folded MHCI molecules immunoprecipitated with W6/32. ²H incorporation into tryptic peptides was quantified by LC-MS. (A-C) Fractional synthesis was calculated for different peptides derived from HLA-A (A), HLA-B (B), and HLA-C molecules (C) (mean ± SD of different mass isotopomers) and plotted against labeling time. Full sequences and analytical metrics for all informative peptides (identified by four amino acids in single-letter code or by charge) are in S2 Table. In (A) and (B), allele- and isotype-specific peptides exhibited no significant differences in fractional protein synthesis (p = 0.32 and p = 0.29, respectively, by F test); in (C), only C1-specific peptides were identified. Single-exponential curve fits (with 95% confidence intervals) are based on a pooled analysis of all peptides from each isotype. (D) Exponential growth of LCL721 cells during ²H₂O labeling. The corresponding time course of the fraction of new cells is shown in panels (A-C), for comparison with protein synthesis. Panels (A-D) were from the same experiment. (E) Fractional synthesis rates (per hour, mean ± SEM) of MHCI isotypes (from (A-C)), compared to fractional cell growth rates (from (D)). Two independent experiments are shown. Fractional synthesis rates for individual peptides are in S2 Table. (F) Turnover rates of different MHCI isotypes, calculated from the excess of mean fractional protein synthesis rates over the cell growth rate.

Fig 5. Summary of fractional synthesis rates of MHCI isotypes in MoDCs.

(A) Scheme illustrating MoDCs differentiation, followed by mock or LPS (100 ng/ml) stimulation, with subsequent ²H₂O labeling for 72 hours, beginning 24 hours after LPS treatment. (B) Comparison between turnover rates of MHCI isotypes in individual unstimulated MoDC cultures from 8 HDs and one B27-negative AS patient. In this setting, fractional synthesis was taken to equal turnover (see text). (C) Effect of LPS on fractional synthesis rates of MHCI isotypes in two HDs.

Fig 6. HLA-A fractional synthesis in unstimulated MoDCs.

Each panel shows isotype-specific (grey symbols) and allele-specific (black or white symbols) peptides from an individual HD or AS patient. SDs of individual data points are shown, as well best single-exponential curve fits to pooled data, with 95% confidence intervals (dashed lines). Fractional synthesis rate constants (k) are shown (means ± SEM).

Fig 7. HLA-B fractional synthesis in unstimulated MoDCs.

Analysis as in Fig 6, except that separate curve fits are shown for HD2 (p = 0.04, F test). The significance of this result is doubtful, as explained in the text. B*27+ donors are identified; note that B27 allele-specific peptides proved suitable for analysis in HD6 and 7, but not in HD4 and 5.

Fig 8. HLA-C fractional synthesis in unstimulated MoDCs.

Analysis as in Fig 6.