Impact of Leishmania donovani infection on the HLA I self peptide repertoire of human macrophages

Macrophages are specialized antigen-presenting cells that process and present self-antigens for induction of tolerance, and foreign antigens to initiate T cell-mediated immunity. Despite this, Leishmania donovani (LD) are able to parasitize the macrophages and persist. The impact of this parasitizing and persistence on antigen processing and presentation by macrophages remains poorly defined. To gain insight into this, we analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) and compared the HLA-I self-peptidomes, proteasome compositions, HLA expression and activation states of non-infected and LD-infected THP1-derived macrophages. We found that, though both HLA-I peptidomes were dominated by nonapeptides, they were heterogeneous and individualized, with differences in HLA binding affinities and anchor residues. Non-infected and LD-infected THP1-derived macrophages were able to sample peptides from source proteins of almost all subcellular locations and involved in various cellular functions, but in different proportions. In the infected macrophages, there was increased sampling of plasma membrane and extracellular proteins, and those involved in immune responses, cell communication/signal transduction and metabolism/energy pathways, and decreased sampling of nuclear and cytoplasmic proteins and those involved in protein metabolism, RNA binding and cell growth and/or maintenance. Though the activation state of infected macrophages was unchanged, their proteasome composition was altered.


Introduction
Visceral leishmaniasis (VL) is a vector-borne neglected tropical and subtropical disease caused by Leishmania donovani (LD) and other Leishmania species that are transmitted by infected female phlebotomine sandflies and obligate intracellular protozoan parasites in vertebrates [1]. It is the second largest parasitic killer disease after malaria, with 200 000 to 400 000 estimated new cases each year. The infections result in high fatality [1][2][3] that is prone to worsen with HIV/VL and other co-infection [4][5][6][7]. PLOS  Subsequently, THP1MФ and YFP-BHU5 infected THP1MФ (THP1MFiy) were used immediately to access the uptake of YFP-BHU5 by THP1MФ and its effects on HLA-ABC and HLA-DR expression. THP1MФ and BHU5 infected THP1MФ (THP1MFi) were used to access viability, HLA-ABC, HLA-A Ã 02:01 and CD83 expression, and were harvested by 10 min centrifugation at 800 x g, shock-frozen in liquid nitrogen and stored as pellets at -80˚C for isolation of HLA.

Parasite uptake, viability, activation and HLA expression
To determine the uptake of LD by THP1MF and effects on hosts viability, CD83, HLA-ABC, HLA-A Ã 02:01 and HLA-DR expression flow cytometry was used. THP1MФ and THP1MFiy or THP1MFi were stained with fluorochrome-labeled monoclonal antibodies against CD11b, CD83, HLA-ABC (BD Bioscience, Heidelberg, Germany), HLA-A Ã 02:01 and HLA-DR (BioLegend, Eching, Germany), Calcein-AM (Invitrogen, Carlsbad, CA, USA) and Propidium Iodide (PI) (Sigma-Aldrich, Germany). The expression of these markers on the cell surface and of calcein and PI fluorescence was determined with a FACSCalibur flow cytometer (Becton Dickinson, Heidelberg, Germany) on 20,000 for forward versus sideward scatter-gated events. CellQuest (Becton Dickinson, Heidelberg, Germany) and WinMDi 2.9 (Purdue University, USA) software were used to process and analyze the data, respectively. The uptake was assessed by CD11b expression against YFP fluorescence, and cell viability assessed using Calcein-AM (Invitrogen, Carlsbad, CA, USA) and propidium iodide (PI) (Sigma-Aldrich, Germany). CD83 was used as a marker of macrophage activation.

Constitutive and immunoproteasome expression in THP1MФ and THP1MФi
The impact of the infection by LD on the constitutive and immunoproteasome expression of THP1MF was determined by semi-quantitative RT-PCR. Total RNA was extracted from THP1MФ, THP1MФi, HeLa cell line, and HeLa clone 33/2 (A2+/IP) using Nucleospin RNA II Purification Kit (Macherey-Nagel, Duren, Germany). cDNAs were prepared from 500ng of DNase-treated RNA using superscript III reverse transcriptase (Invitrogen, CA, USA). RT-PCR was carried out with 500ng of cDNA using the following constitutive (β1, β2 and β5) and immunoproteasome (β1i, β2i and β5i) subunit-, and GADPH-specific forward and reverse primers [24]

Processing and analysis of data
MS and MS/MS spectra were processed using Data Analysis 3.4 and Biotools 3.1 (Bruker Daltonics). Peptides were identified against Swissprot human protein sequence database version 56.3 (20,408 reviewed non-redundant protein sequences) integrated in a local MASCOT server (version 2.2). Precursor and fragment mass tolerances of 100 ppm and 0.5 Da were used respectively, oxidation of methionine was allowed as a possible modification. Peptide-spectrum matches were validated using a statistical evaluation -10logP, where logP is the logarithm to the base 10 of P (P<0.05). De novo sequencing using Sequit software [27] and manual inspection were used to further validate the identified peptides. Peptides source proteins were annotated using Uniprot [28] and classified according to subcellular locations and biological functions using human protein reference database [29]. Peptides were assigned to their respective HLA using netMHCpan in the Immune Epitope Database IEDB [30,31] and SYFPEITHI [32], and their predicted HLA binding affinities were determined using netMHCpan in IEDB with a binding affinity IC 50 threshold of 500nM. Binding motifs for the nonapeptides were visualized using sequence logos [33,34].

Statistical analyses
Macrophage activation, HLA expression and proteasome subunit expression between THP1MФ and THP1MФi/or THP1MФiy were compared by paired 1-tailed Student's t-test and differences indicated as significant when Ã p < 0.05. Data are presented as the mean ± standard deviation from three independent experiments.

Parasite uptake and effects on viability, activation and HLA expression
To assess the uptake of LD by THP1MF and to determine its effects on viability and expression of HLA-ABC, HLA-A Ã 02:01, HLA-DR and CD83 by the host, flow cytometry was used as detailed in Materials and Methods. The uptake of YFP-BHU5 by THP1MФ based on CD11b expression versus the YFP fluorescence was 70.25±5.59%. The viability of THP1MФi was slightly lower compared to THP1MФ based on calcein and PI fluorescence (Fig 1A). The levels of HLA-ABC ( Ã p<0.05), HLA-DR and HLA-A Ã 02:01 were lower on the infected compared to the non-infected cells, while those of the macrophage activation marker CD83 were unchanged (Fig 1B and 1C). The decrease in MHC class I and II expression on infected cells has also been observed previously, albeit in murine studies [14].

Impact of parasite uptake on constitutive and immunoproteasome mRNA expression
To determine the impact of LD on the hosts constitutive and immunoproteasome subunit expression by THP1MF, semi-quantitative RT-PCR was carried out as detailed in Materials and Methods. THP1MФ and THP1MФi expressed both the constitutive proteasome subunits (β1, β2 and β5) and the immunoproteasome subunits (βi1, βi2 and βi5) (Fig 2A). Semi-quantitative analysis of the RT-PCR bands showed a reduction of the mRNA (subunit/GADPH) of β1 ( Ã p<0.05), β2 ( Ã p<0.05), and no significant change for the β2i (p = 0.28) and β5i (p = 0.15) subunits in THP1MФi compared to THP1MФ with 14%, 25%, 9% and 25%, respectively (Fig 2B).

Self-ligands presented by HLA I of THP1MФ and THP1MFi
2.3 x 10 9 THP1MFi cells were lysed, and affinity chromatography and LC-MS/MS was used to isolate the HLA class I molecules and analyze the HLA-bound peptides. A total of 86 nonredundant HLA class I self-ligands were identified from 82 source proteins of THP1MФi (S1 Table) compared to 347 non-redundant HLA-I self-ligands derived from 282 source proteins identified for 2.8 x 10 9 THP1MF cells at the same time and reported earlier [26]. Only 17 HLA-I self-peptide sequences and 18 source proteins were found to be shared between THP1MФ and THP1MФi.
With 82%, nonapeptides were also the most dominant among the peptides shared between THP1MФ and THP1MФi.

The binding motifs for HLA I in THP1MF and THP1MFi
To determine whether there was a difference in binding motifs of the nonapeptides in THP1MF and THP1MFi, we used sequence logos [33,34]. In these sequence logos, the height of each column of amino acids is equal to the number of peptide sequences (in bits), and the relative height of each amino acid within each column is proportional to the frequency of the amino acid at that position. The most frequent primary anchor amino acids at position 2 of infected and non-infected THP1MФ HLA-A Ã 02:01-bound peptides were L, and with about equal but lower representation I, Y and M; for the C-terminus these were L and V. For HLA-B Ã 15:11-bound peptides from infected and non-infected cells, P was most prominent at position 2, and Y and F at the C-terminus followed by M in case of the infected cells. For HLA-C Ã 03:03, A and Y were dominant at position 2 of peptides derived from non-infected cells whereas no prominence was found at this position for peptides from infected cells. At the C-terminus of HLA-C Ã 03:03-bound peptides, L was most frequent followed by F in noninfected and M in infected THP1MФ (Fig 4).

Discussion
The total number of HLA class I-restricted self-peptides and source proteins identified from THP1MFi was four-fold lower compared to those identified in THP1MF, and were heterogeneous and individualized. Only a few peptides were found to be shared between the two despite expressing the same HLA alleles.
The strong decrease in the number of HLA class I-restricted peptides from LD-infected THP1MФ has been reproduced in independent experiments and thus not a technical issue, but may rather be due to the following. Firstly, the overall MHC-I expression at the cell surface of THP1MФi was lower compared to THP1MФ including, though not significantly, that of HLA-A Ã 02:01 (Fig 1B and 1C). Reduction of MHC I-restricted antigen presentation upon infection with LD parasites through reduction of MHC I present at the cell surface has also been observed in murine studies albeit no comparative peptidome studies had been done [14,39]. Though our focus was on HLA I-restricted self-peptides, we also observed lower expression of HLA II by THP1MФi compared to THP1MФ. This observation concurred with murine studies on MHC II, and showed that Leishmania inhibits antigen presentation by repressing MHC II expression [14,39,40]. Secondly, the infection of THP1MФ by LD resulted in a slight decrease in host cell viability (Fig 1A), which might be due to the fact that naturally Leishmania promastigotes, upon uptake by macrophages, transform to amastigotes and multiply to eventually rupture the macrophages [2]. Thirdly, although CD83 expression, a marker of macrophage activation, was unchanged in THP1MФi compared to THP1MФ indicating a lack of activation, LD infection resulted in decreased expression of β1 and β2 constitutive proteasome subunits, which could translate into decreased antigen processing efficiency. The impact of proteasome on the quality and quantity of MHC class I ligands had been studied using wild type and proteasome subunits deficient murine dendritic cells [41,42]: expression of proteasome subunits correlated with increased generation of peptides that are suitable for binding to MHC I molecules.
The heterogeneity and individuality in the HLA I self-peptides and source proteins identified in THP1MФ and THP1MФi depicts differences in protein expression, processing and presentation, as was in other cells and tumor samples [25,26,35,36]. The nonapeptides are the optimum lengths for MHC class I binding [26] and though the infection of THP1MФ by LD did not change the nonapeptides dominance in the identified HLA I-bound peptides, profound differences in antigen processing and presentation were evident, firstly, in the HLA restriction of identified peptides. For THP1MF the percentage of the peptides identified for the different HLA-restrictions ranked in the order HLA-A Ã 02:01 > HLA-B Ã 15:11 > HLA-C Ã 03:03 while in THP1MФi they were in the order HLA-B Ã 15:11 > HLA-C Ã 03:03 > HLA-A Ã 02:01. Though HLA-B Ã 15:11 peptides were dominant after infection with LD, only 26% of them were within the IC50 threshold of 500nM. In general, after infection there was a shift of peptides towards lower affinity binders. A previous systematic mapping and characterizing of peptide ligands derived from B Ã 1508, B Ã 1501, B Ã 1503, and B Ã 1510 showed endogenous peptide loaded into B15 to be flexible both in the location of and amino acids at the N-proximal anchors [43]. In addition to this, additional preference of aliphatic amino acids was observed at the C-Terminus after infection, which would, though unconfirmed in Prilliman et al. [43], result in lower binding affinity of the peptides. The differences in antigen processing and presentation were evident in the peptide anchor motifs. For the HLA-A Ã 02:01-bound peptides there were no dominant accessory anchor amino acids at position 6 in THP1MФi compared to the dominant hydrophobic anchor in THP1MФ. For HLA-C Ã 03:03 there was no anchor motifs at position 2 in THP1MФi but a strong preference for A and Y in THP1MФ but a higher percentage of peptides within the IC50 threshold of 500nM in THP1MФi compared to THP1MФ.
In both THP1MF and THP1MFi the peptide source proteins were derived from almost all subcellular locations and were involved in almost all molecular functions of the cells. But differences were observed. Firstly, in THP1MFi compared to THP1MF, there was an increase of source proteins from plasma membrane and extracellular proteins and a decrease in source proteins from nucleus and cytoplasm (Fig 5A), and no peptides were identified from ribosomes, cytoskeleton and centrosomes unlike in THP1MF. Secondly, in THP1MFi compared to THP1MF, there was an increase of source proteins involved in immune responses, cell communication/signal transduction and metabolism/energy pathways and a decrease in those involved in protein metabolism, RNA binding, cell growth and/or maintenance (Fig 5B). The differences in source protein peptide sampling in THP1MF and THP1MFi, would imply differences in protein turnover, as protein turnover correlates with source protein presentation [44,45]. LD has been shown previously in proteomic studies to globally alter protein expression in THP1 cells [46].
In summary, the infection of macrophages with LD has profound effects on the self-peptide repertoire presented by MHC I molecules, which in parts can be explained with changes in antigen processing including the composition of the proteasomes, and altered protein expression and turn-over in different cellular compartments. In conclusion, the self-displayed by infected macrophages is very different from the self of uninfected cells. This difference may relate to T cell-mediated autoimmune reactions which may explain some of the immune pathology observed in LD patients, as changes in self-peptidome have been shown previously to impact T cell-mediated immune responses [47,48]. Furthermore, though our focus was on the self-antigens, also the processing of LD antigens may be affected but likely not in the same manner as the self-antigens. LD antigens as exogenous antigensare processed via the MHC class I cross-presentation pathway whereas the self-antigens as endogenous antigens are processed via the classical MHC class I antigen processing and presentation pathway [49,50]. To check for alterations, a comparison of LD MHC I peptidomes from THP1 derived macrophages incubated with dead LD versus infected with live LD would be required. The results of such studies would however be difficult to compare because dead parasite are expected to be processed through the MHC class II antigen-processing pathway in endolysosomes and the epitopes primarily be presented by MHC class II molecules. Lastly, LD and HIV infect the same host cells, macrophages, but differently, and persist in different subcellular compartments and use different survival mechanisms that affect different key players in the MHC class I antigen processing and presentation pathways [8,14,39,[51][52][53][54]. LD/HIV co-infection is expected to result in mutual impact on processing and presentation of antigens of both agents. Given the increased fatality of HIV/VL co-infection cases, determination of the effect of LD/ HIV co-infection on the self-peptidomes as well as HLA peptidomes of the pathogens, which are yet to be determined, would be vital.
Supporting information S1 Table. HLA I ligands and source proteins identified from THP1MФi. (PDF)