Human leukocyte antigen class II quantification by targeted mass spectrometry in dendritic-like cell lines and monocyte-derived dendritic cells

The major histocompatibility complex II (HLA-II) facilitates the presentation of antigen-derived peptides to CD4+ T-cells. Antigen presentation is not only affected by peptide processing and intracellular trafficking, but also by mechanisms that govern HLA-II abundance such as gene expression, biosynthesis and degradation. Herein we describe a mass spectrometry (MS) based HLA-II-protein quantification method, applied to dendritic-like cells (KG-1 and MUTZ-3) and human monocyte-derived dendritic cells (DCs). This method monitors the proteotypic peptides VEHWGLDKPLLK, VEHWGLDQPLLK and VEHWGLDEPLLK, mapping to the α-chains HLA-DQA1, -DPA1 and -DRA1/DQA2, respectively. Total HLA-II was detected at 176 and 248 fmol per million unstimulated KG-1 and MUTZ-3 cells, respectively. In contrast, TNF- and LPS-induced MUTZ-3 cells showed a 50- and 200-fold increase, respectively, of total α-chain as measured by MS. HLA-II protein levels in unstimulated DCs varied significantly between donors ranging from ~ 4 to ~ 50 pmol per million DCs. Cell surface HLA-DR levels detected by flow cytometry increased 2- to 3-fold after DC activation with lipopolysaccharide (LPS), in contrast to a decrease or no change in total HLA α-chain as determined by MS. HLA-DRA1 was detected as the predominant variant, representing > 90% of total α-chain, followed by DPA1 and DQA1 at 3–7% and ≤ 1%, respectively.

Monocyte-derived DCs. To generate DCs, CD14+ monocytes were isolated from PBMCs collected with informed consent from six healthy donors (D1095, D1995, D18092, D1237, D1265, D1765, STEMCELL Technologies, Vancouver, BC, Canada). PBMCs from half-leukopaks were isolated by Ficoll (GE Health Sciences) density separation and CD14+ monocytes were isolated by magnetic bead-based cell separation (MACS, Miltenyi Biotec, MA, USA), using microbead-conjugated anti-CD14 antibodies and magnetic cell separation columns (Miltenyi Biotec) according to the manufacturer protocol. CD14+ monocytes were differentiated into DCs using a modification of a previously published protocol [42][43][44][45] . Briefly, monocytes were cultured at a density of 5 × 10 5 cells/mL in RPMI medium supplemented with 10% (v/v) FBS, GM-CSF (100 ng/mL) and IL-4 (17 ng/mL) for 6 days. To determine the LPS concentration for DC activation, independent sets of monocytic-DCs derived from four healthy human donors were activated with six increasing concentrations of LPS (0, 0.3, 1, 10, 32, 250 and 1000 ng/mL) on Day 5 as previously described 46 . This titration led to the selection of 32 ng/ml of LPS for stimulation of experimental monocytic DCs due to the consistent performance and activation response. On day 6 cells were harvested 22 h post LPS addition, washed with PBS three times, counted and stored at − 80 °C. Flow cytometry. DCs were harvested pre-and post-LPS stimulation and cell surface-associated HLA-DR (csHLADR) was detected by flow cytometry after staining with an anti-HLA-DR L243 antibody in non-permeabilized DCs. Briefly, cell pellets were washed with FACS staining buffer+ BSA (BD Pharmingen, 554,657) and stained for 30 min with the following conjugated antibodies: Alexa 488-HLA-DR (Clone L243; Biolegend, 307,620 B241328), PE-CD86 (Clone IT2.2; Biolegend, 305,406 B210795), BV421-CD11c (Clone 3.9; Biolegend, 301,628 B238001), APC-CD40 (Clone 5C3; Biolegend, 334,310 B219298) and live/dead fixable near-IR dead cell stain (Lifetech, L34976 1,868,118). Cells were washed three times with staining buffer and cell surface stain was analyzed with a BD Fortessa flow cytometer using BD FACS DIVA software. Further analyses were carried out using FlowJo software V. 10 www.nature.com/scientificreports/ Cell lysis, HLA-II-immunoprecipitation and proteolytic digestion. Cell lines and DCs were processed in an identical manner as described below. To obtain whole-cell lysates, pelleted cells were thawed and resuspended in 1 mL of cold lysis buffer containing 20 mM Tris, pH 8.0, 150 mM NaCl, 1% (v/v) CHAPS+ 1 mM PMSF (Sigma, A3428-10MG), 5 μg/mL Aprotinin (Sigma, A3428-10MG), 10 μg/mL Pepstatin A (Calbiochem, 516,481-5MG) and 10 μg/mL Leupeptin (Sigma, L5793-5MG). Lysates were incubated for 1 h with end-to-end rotation at 4 °C then centrifuged at 2500 RCF for 5 min to recover the supernatant. In order to establish the linearity of signal detection in a titration experiment, serial dilutions of 200 μL whole lysate per cell type and condition were performed in duplicate in a 96-well plate format (LoBind deep-well plate, Eppendorf) to obtain samples corresponding to 100,000, 50,000 25,000, 12,500, 6250, and 3125 cells as well as blanks. Immunoprecipitation (IP) of total HLA-II protein was carried out in each sample (cell lysates equivalent to 100,000, 50,000 25,000, 12,500, 6250, and 3125 cells) by adding 10 μg of anti-panHLA-II antibody (ATCC HB-145 hybridoma, anti-HLA-DP/DQ/DR antibody), which was biotinylated using EZ-Link Sulfo-NHS-LC-Biotin (Thermo Fisher Scientific, 21,327). Antibody incubation proceeded overnight at 4 °C while shaking at 600 rpm. 50 μl of streptavidin magnetic beads (T1 MyOne, Thermo Fisher Scientific) were washed once with 1× PBS+ 0.1% Tween (PBST) and twice with 1× PBS prior to being added to each IP sample in a 96-well LoBind plate and incubated for 1.5 h while shaking at 1200 rpm at RT. To examine the completeness of HLA-II recovery during the immunoprecipitation step, a secondary IP was performed on the post-IP lysates, with selected samples, using the identical methodology described above in order to assess how much HLA-II α-chain, if any, remained in the lysate following IP.
Bead bound immunocomplexes (HLA-II-mAb-beads) were washed twice with PBST buffer followed by one wash with PBS before elution with 135 μL of 30 mM HCl/5% ACN using an automated bead handling system (KingFisher Flex, Thermo Fisher Scientific). Lyophilized stable isotope labeled (SIL) peptides (lysine 13 C 6 , 15 N 2 ) VEHWGLDKPLLK, VEHWGLDQPLLK and VEHWGLDEPLLK (New England Peptide, Massachusetts, USA ) were reconstituted in 30% ACN, 0.1% TFA at 69.3 pmol/μL. Peptides were diluted to 69.3 fmol/μL in an equimolar mix with 5%ACN, 0.1%TFA immediately prior to use. Samples were spiked with 4.6 μL of equimolar SIL peptide mix to a final concentration of 1.6 fmol/ μL (332 fmol of each SIL-peptide per sample). Reduction was performed by adding 2 μL of 500 mM TCEP (TCEP Bond Breaker, Thermo Fisher Scientific, 77,720) and incubation for 30 min at 60 °C. Alkylation was performed by the addition of 5 μL of 150 mM iodoacetamide to each sample followed by a 30 min incubation in the dark at RT. Sequencing grade lyophilized trypsin/Lys-C mix (Promega, V5071) was reconstituted in 20 mM Tris at 200 ng/μL immediately before use. Samples were then digested with 1 μg of trypsin/Lys-C overnight at 37 °C. Each sample digest was quenched by addition of 16 μL of 1% of TFA yielding a final volume of 200 µL.

Data processing.
To determine the total HLA-II levels, the peak area ratios of endogenous peptides to SIL reference peptides were calculated and multiplied by the total amount of reference SIL-peptide spiked per sample (332 fmol). Typically, lysates corresponding to 100,000, 50,000, 25,000, 12,500, 6250, 3125 and 0 cells were analyzed. Each quantification result was based on a total of six measurements as follows. Duplicate samples each with 100,000, 50,000 and 25,000 cells were used to derive the HLA-II α-chain level in each cell type or donor as well as to confirm linearity of HLA-II α-chain detection. Total HLA α-chain levels for each measurement were then normalized to 1 × 10 6 cells and averaged across the three different starting cell numbers to obtain the final amount of HLA-DRA1/DQA2, -DPA1 and -DQA1 (fmol/1 × 10 6 ). Of note, data from the lower cell numbers (12,500 6250 and 3125 cells) expectedly decreased in signal intensity and for some cell types resulted in low signal intensity unreliable for further data analysis. Thus, these lower cell numbers across all conditions were not considered in the final quantification result.     Fig. S2). Here, HLA-DR and csHLADR abundance did not change at concentrations > 1 ng/mL of LPS when detected by western blot and flow cytometry (Supplementary Fig. S2A,B), which was also consistent with the equivalent abundance of CD40 and CD86 ( Supplementary  Fig. S2C,D). Levels of csHLADR increased 1.5 to fourfold in CD11c+ singlets for all LPS-induced DCs ( Fig. 2A).
Although csHLADP, csHLADQ and other cell surface markers were not analyzed by flow cytometry in this panel, DC-maturation upon LPS was also confirmed in all cases with the robust increase in detection of the costimulatory molecules CD40 and CD86 at the cell membrane (Fig. 2B).

HLA-II α-chain abundance in cell lines by western blot.
We detected HLA-DR protein by western blot (WB, method details provided in Supporting Information, Supplementary Fig. S3) to evaluate if the increase of csHLADR in LPS-induced MUTZ-3 cells observed by flow cytometry was a consequence of an increased translocation to the plasma membrane or an increase in the total HLA-II protein amount. Total HLA-DR protein signal normalized by β-actin control increased 2-to fourfold in LPS and TNF-stimulated cells compared with undifferentiated MUTZ-3 (Fig. 3). While this data supports the increase in csHLADR in LPS-induced MUTZ-3 cells observed by flow cytometry, WB methods not only lack the ability to differentiate between HLA-DR, -DP and -DQ variants and but are also unsuitable for reliable quantification, especially in the absence of a protein standard. Further confounding WB normalization and interpretation was that actin, used as a loading control,

LC-MS/MS method development.
We next developed an HLA-II LC-MS/MS assay based on HLA-IIimmunopurification using an anti-panHLA-II antibody (HB-145), which is frequently used in HLA-II immunopeptidomics studies 50,51 . At the amino acid-sequence level, HLA-II α-chains are less variable relative to HLA-II β-chains. We therefore targeted the HLA-II α-chain polypeptides from the three major groups of HLA variants:   (Fig. 4). As expected, the relative intensity of fragment ions was similar for both endogenous (light) and reference SIL-peptide (heavy) for the three target sequences. The averaged peak area ratio from duplicates of each sample corresponding to 25,000, 50,000 and 100,000 cells showed the expected increase of HLA-II peptide signal approximately proportional to the input amount analyzed ( Supplementary  Fig. S4). No saturation effect was observed at any of the higher cell number samples, which provided evidence of linear response of the assay within this cell range. Furthermore, analysis of a second IP of the supernatant following the first IP confirmed that the HLA alpha chain recovery was typically > 95% and ≥ 90% in all conditions. This included cell lysate equivalents up to 100,000 stimulated DCs with increased HLA-II levels for at least three sets of DCs ( Supplementary Fig. S5). Therefore, and in the absence of HLAII α-chain protein reference standards, the concentrations determined from this procedure were considered a suitable quantitative measure of HLAII alpha chain abundance.
Because the fundamental antigen presentation pathways are misregulated in KG-1, we then examined HLA-II levels in both unstimulated and stimulated MUTZ-3 cells. While HLA-DRA1 was detected at 268 fmol per million unstimulated MUTZ-3 cells, like in KG-1 cells, HLA-DPA1 and -DQA1 proteotypic peptides were also undetectable in unstimulated MUTZ-3 (Fig. 5). MUTZ-3 cells induced with TNF and LPS showed a significant elevation of HLA-DRA1/DQA2-levels corresponding to 12 and 60 pmol per million cells, respectively, representing a 50-and 200-fold increase over the unstimulated condition (Fig. 5). However, HLA-DPA1 peptide was also upregulated in MUTZ-3 cells and became detectable at 2.2 and 6.2 pmol per million cells with TNF-and LPS-stimuli, respectively, representing between 7.5 and 9.4% of the total HLA-II amount (Fig. 5). Notably, HLA-DQA1 peptide was detected only after LPS-maturation signal at 288 fmol per million cells, being the lowest HLA α-chain detected in MUTZ-3 (Fig. 5), constituting < 1% of the total HLAII detected. The quantitative HLA-II α-chain LC-MS/MS method, in comparison to the WB method, allowed for a more reliable and specific assessment of total HLA-DRA1 expression, which increased 50-and 200-fold in MUTZ-3 in response to LPS and TNF stimulation, respectively. In contrast, csHLADR levels measured by flow cytometry increased only 3-to 4-fold, suggesting that only a fraction of the newly synthesized HLA-II protein is translocated to the cell surface upon LPS and TNF stimuli.

HLA-II α-chain abundance in DCs by LC-MS/MS.
We next analyzed HLA-α chain levels in human monocyte-derived DCs, where HLA-DRA1 was the most abundant variant detected in the six sets of donorderived DCs, corresponding to ≥ 90% of total α-chain (Fig. 5). Quantification of HLA-DRA1 showed some variability between donors, resulting in 20.2, 16.0, 3.7, 44.1, 17.5 and 17.7 pmol/1 × 10 6 unstimulated DCs for D1095, D1995, D18092, D1237, D1265, D1761, respectively. HLA-DPA1 was the second most abundant α-chain in all cases, representing 5 to 10% of total HLA-α chain at 1.3, 0.9, 0.3, 3.8, 1.0 and 0.9 pmol/1 × 10 6 donor-derived cells from of D1095, D1995, D18092, D1237, D1265, D1761, respectively (Fig. 5). In contrast, HLA-DQA1 was detected only in three donors (D1095, D1265, D1761) at very low levels (~ 0.1 pmol/1 × 10 6 of DCs), constitut- www.nature.com/scientificreports/ ing < 1% of total α-chain when detected (Fig. 5). Of note, loss of DQA1 expression in D1095 and D1265 was observed after LPS stimulation. While morphologic and immunophenotypic changes were observed in all LPS-treated DCs, the ratio of abundance of HLA-DRA1, -DPA1 and -DQA1 proteins remained relatively consistent before and after LPS induction (> 90%, 5-10% and < 1%, respectively). Three sets of LPS-stimulated DCs (D1995, D1809 and D1237) showed a significant decrease in total amount of HLA α-chain detected compared with unstimulated cells (30-70% reduction). In contrast, HLA-II levels in the remaining sets of DCs (D1905, 1237 and 1761) did not change significantly after LPS treatment. These results contrast with the increase in HLA-II detected in the cell line model MUTZ-3, where total HLAII increased upon stimulation. The variable response to LPS activation in donor-derived DCs may be due to factors such as HLA background of donors, as well as multifactorial regulation of the molecular players in the biosynthesis of the HLA polypeptides and its translocation mechanisms from the intracellular compartment to the cell surface in response to maturation signals. While acute myeloid leukemia-derived cell lines KG-1 and MUTZ-3 exhibited low levels of cell surface HLA-DR, suggesting the low abundance of HLA-II and resembling their lineage origin as CD34+ DC-precursors, HLA-II quantification in primary donor-derived DCs by MS demonstrated a decrease or no change in total levels of HLA-DRA1, -DPA1 and -DQA1 after LPS induction, which contrasts with the increase of HLA-DR at the cell membrane observed in all cases (Fig. 6). Altogether these results indicate a differential response of HLA-II levels to the activator LPS, which may respond to the individual variability associated with haplotype and other factors involved in the biosynthesis, translocation and stabilization of HLA-II at the cell membrane.

Conclusions
The elucidation of total HLA-II levels of represents a valuable opportunity to build our understanding of antigen presentation and subsequent T-cell activation. Here we present the first study that elucidates the abundance of the total levels of HLA-II protein by targeting specifically α-chain polypeptides, which ranged from 100 to 200 fmol per million dendritic-like cells and 3-30 pmol per million unstimulated and stimulated DCs. These amounts correspond to 6-12 × 10 4 and 1.8 × 10 6 -1.8 × 10 7 HLA-II molecules per cell for cell lines and DCs, respectively. Consistent with a previous published results [34][35][36][37][38][39][40][41] , we confirmed an increase of csHLADR following TNF and LPS treatment for the monocytic-like cell line MUTZ-3, where both stimulants elicited a response of similar magnitude. This is also consistent with the increase in allogenic T cell-activation potential by induced-MUTZ-3 found by T cell proliferation assays 38 . However, our assessment of total HLA-II abundance by mass spectrometry showed a larger effect in stimulated MUTZ-3 cells, suggesting that an accumulation of HLA-α chain may occur at the intracellular compartments. Moreover, the varied response to both proinflammatory stimuli could be explained by the high heterogeneity in MUTZ-3 cells despite the clonal origin [53][54][55] , . Furthermore, signal transduction upon different maturation signals may be also affected by cell cycle phase and receptor-mediated responses at the single-cell level. Overall, these findings are consistent with the proinflammatory response by MUTZ-3 differentiated into DCs in cytokine production and chemokines as well as the immunophenotypic profile reported, which maturation more closely recapitulates that of DCs [34][35][36]38,41 .We have shown that HLA-DRA1 protein is the most abundant α-chain variant (> 90%), followed by -DPA1 (3-7%) and -DQA1 (< 1%) when detected, irrespective of the HLA-II haplotype in dendritic-cell lines KG-1 and MUTZ-3. This was also observed in monocyte-derived DCs, where HLA-DRA1 levels were consistently higher than -DPA1, and -DPA1 was consistently higher than DQA1. These findings are consistent with the predominance of HLA-DRA1 followed by -DPA1 and -DQA1 recently reported at the transcriptional level by RNA-seq 49 . In half the donor samples DQA1 was not detected. In the case of KG-1 cells DPA1 and DQA1 were not detected by MS despite the haplotype (DRB1*11:01:02, Figure 6. DCs display increased levels of csHLA-DR upon LPS-activation in contrast with total levels HLA-DRA1 α-chain. csHLA-DR and totalHLA-DRA1 fold induction between unstimulated and LPS-treated DCs, statistical significance was analyzed with a two-tailed Student's t Test (*p < 0.05). Each point for csHLA-DR represents a replicate (n = 3). For totalHLA-DRA1 each point represents the mean for two injections corresponding to 25,000, 50,000 and 100,000 cells per donor (n = 3) and the SD (error bars). www.nature.com/scientificreports/ *14:01:01G, DRB302:02:01G DQA1*01:01,*05:01; DPA1*02:01 and *03:01). Furthermore, because the equivalent relative abundance of HLA-DRA1 and -DQA1 at the transcriptional and translational level have been previously reported 10,11 , it is possible to hypothesize that differential mechanisms of the protein synthesis and half-life underline HLA-II abundance. Noteworthy, HLA-DQA1 levels were downregulated after LPS induction in two sets of DCs (D1095 and D176), suggesting differential stability mechanisms for this variant. For example, while MARCH1 and MARCH8 E3 ligases are known to ubiquitinate HLA-II molecules in order to maintain them in the endosomal and lysosomal compartments for storage and degradation [23][24][25][56][57][58][59] , HLA-DQ has been found ubiquitinated by an alternative ligase (MARCH9) 58,59 . Furthermore, monocytic differentiation into DCs and maturation induced by GM-CFS+ IL-4 and LPS, respectively led to a signaling cascade of upstream activators in a rapid manner 7,60-64 , which leads to a transient increase in de novo MHCII synthesis at the transcriptional level, followed by a subsequent decrease resulting from silencing from its transcriptional activator CIITA 14,29,30 . Therefore, potential differences in the rates of CIITA silencing and transcriptional shutdown in DCs at the time of harvest and analysis could impact the quantification depending on the individual variability. For example, the detection of HLA-DRA1 and -DQA1 transcripts at the same levels after 6 h of stimulation with LPS changed significantly for -DQA1, where its expression dropped to baseline levels after 24 h 10 . Thus, differences in the developmental state and differentiation intermediaries when monocyte-derived dendritic cells were stimulated with LPS and harvested at 22 h cannot be ruled out. Furthermore, the low abundance of HLA-DQA1 highlights a fundamental question regarding its association with a wide range of human disorders such as celiac disease and the potential role of HLA-DQA1 protein levels change in disease processes.
While the focus of this work is on method development for MS quantification of HLA-II, further application of this method will complement characterization studies of DC maturation and other processes relevant to DC biology. Thus, quantification of total HLA α-chains is expected to contribute to key research fields in the context of adaptative immunity such as immunopeptidomics, transplantation and vaccine studies. Finally, because HLA abundance differences appear to be individual-, cell-and haplotype-specific markers in cancer [65][66][67][68][69] , MS quantification of HLA α-chains may contribute to use of HLA-II as a potential clinical biomarker.

Data availability
The datasets generated and analyzed during the current study are available from the corresponding author upon request.