Cover Page The handle http://hdl.handle.net/1887/32636 holds various files of this Leiden University dissertation.

(1)

Cover Page

The handle http://hdl.handle.net/1887/32636 holds various files of this Leiden University dissertation.

Author: Jongsma, Marlieke Lyrissa Maria

Title: A genome-wide cell biological analysis of genes involved in MHC class II antigen presentation

Issue Date: 2015-04-01

(2)

(3)

Chapter 6:

Summary & Discussion

M.L.M. Jongsma

(4)

(5)

Summary & Discussion 145

HLA-DQA1, HLA-DRB1 and HLA-

6

DRB1, inheritance of these alleles make a individual more susceptible to develop the disease [2]. Other examples of MHC class II related diseases are multiple sclerosis, linked to the HLA-DRB1 allele [3] and Celiac disease, linked to the alleles HLA-DQ2 and HLADQ8 [4]. Better insight into the regulation of MHC class II transcription, peptide loading and transport towards the cell surface is needed to understand MHC class II antigen presentation which thereafter can be manually regulated (for example by drugs) to prevent uncontrolled immune responses.

MHC class II molecules are specifically expressed in Antigen Presenting Cells (APCs), like B-cells, macrophages and dendritic cells.

Although APCs are the most representa- tive model systems to study the MHC class II antigen presentation pathway, they were not ideal cells to use for our flow cytometry- based RNAi screen. First, APCs are known to express Toll-like Receptors (TLRs) at their cell surface. siRNAs can activate these TLRs leading to interferon production that then increase MHC class II cell surface expression [5-8]. Since we used flow cytometry as a read out to detect changes in MHC class II levels at the cell surface, screening in APCs would have resulted in many off- target candidates. Second, it is hard to transfect immune cells with siRNAs. To overcome these problems we performed our screen in siRNA-based Genome wide screens have

been shown to be a powerful tool to investi- gate a wide variety of cell biological processes.

The strength of genome wide screening lies in the fact that it results in a huge unbiased dataset which includes known players of the studied process (positive internal controls), but also proteins and pathways which would otherwise not have been considered. Besides, the dataset can be used to cluster candidates with similar phenotypes as potential interaction partners that may function in the same network.

We performed a genome wide screen (21.245 genes) to identify new proteins involved in MHC class II antigen presentation. MHC class II molecules are known to present exogenously derived peptides to CD4+ T-cells to initiate and control immune responses. Although the pathway has been studied by many groups before (see Chap- ter 1), many questions remain on the regulation of antigen loading and the transport of peptide-loaded MHC class II towards the cell membrane. Proper understanding of the regulation behind MHC class II antigen presentation is important since MHC class II is strongly linked to auto-immunity [1].

Many autoimmune diseases are the result of improper antigen presentation leading to undesired immune responses. Type 1 Diabe- tes has been strongly linked to HLA-DQB1,

Biology is very complicated, since it includes many processes which are controlled by even more proteins and protein-networks. For a broad understanding of biology, these differ- ent proteins have to be defined and coupled to a biological process. My thesis is focused on the placement of new proteins in one specific biological process, namely antigen presenta- tion by MHC class II molecules. We are interested in MHC class II molecules because they are master regulators of many immune responses, for example antibody responses and CTL activation. Besides, MHC class II molecules are strongly related to almost all autoimmune diseases. Better understanding of the proteins and protein-networks con- trolling MHC class II expression and antigen presentation may provide new drug-targets allowing interference with processes controlling the immune response and autoimmune diseases. We decided to perform a siRNA based screen to identify new proteins involved in MHC class II expression and antigen presentation.

(6)

Chapter 6 146

MelJuSo cells, a melanoma cell line expressing many proteins involved in MHC class II antigen presentation that are also found on B cells, monocytes and dendritic cells [9]. Be- sides, MelJuSo cells lack TLRs and are very easy to transfect (>95% transfection efficiency) making them the perfect model system for our genome wide screen.

Using flow cytometry as a read-out, we de- tected peptide-loaded MHC class II at the cell surface using the Cy3-labelled antibody L243 and we measured the efficiency of peptide loading by staining for CLIP-loaded MHC class II using Cy5-labelled CerCLIP antibody. We performed over 100.000 two- color flow cytometry analyses which yielded an enormous data set containing information on all 21.245 human genes. The amount of candidates you eventually identify depends on a subjectively chosen threshold. Yet, the threshold is very important; If the threshold is set too high, it will result in only few remaining candidate genes and other potential canadates are lost, while a threshold set too low will yield many candidate genes including many false-positives that still have to be considered in further analyses. We choose a threshold of |z| > 3 (z-score), meaning that siRNA depletion of the selected candidates resulted in an increase or decrease in L243 or CerCLIP signal of at least 3x the standard deviation. This corresponded to a p value for the candidates of p<0.0027. We identified 276 candidate proteins potentially involved in MHC class II antigen presentation. 45 of these candidate proteins affected peptide loading since their depletion resulted in- creased or decreased levels of CLIP loaded MHC class II at the cell surface. Some candidates were already published to play a role in the MHC class II pathway, while others have not been linked to the MHC class II before. Remarkably, more than 60% of the identified proteins had a completely unknown function. These are in fact the most

interesting candidates and we decided to try to link these to functional networks. To determine the presence of currently known networks in our candidate list, all 276 candidates were analyzed by Ingenuity pathways Analysis and STRING [10]. However, network analyses did not provide any information about pathways involved in MHC class II antigen presentation as most candidates had an unknown function. Proteins with unknown functions are the hardest to place in a functional network, but probably most interesting to study.

Our initial screen was genome wide, but siR- NA screens can also be performed on smaller data sets. We performed two additional siRNA-based screens to place the identified candidates in networks. The concept is that genes with the same phenotype as other genes when depleted will likely be part of the same network. A quantitative PCR-based screen was performed to identify candidates involved in transcriptional control of the MHC class II pathway. We selected for candidates affecting MHC class II expression, the expression of its transcriptional master regulator and also looked at their effect on MHC class I expression to determine which candidates had a more general immune acti- vating effect (which is for example the case for IFNγ). The resulting nine candidates in- cluded various unknown proteins that (again) may act in transcriptional control networks.

To dissect these, we performed a ‘cross-cor- relative qPCR’ where first all candidates were individually silenced, where after the effect of their depletion on the transcription of the other eight candidates was assessed. The resulting information allowed us to build new unbiased transcriptional networks. One of the new transcription regulators identified was RMND5B. RMND5B is known to in- teract with SMADs, which are downstream signaling molecules of the TGFβ pathway [11]. TGFβ signaling leads to downregulat-

(7)

6

ed MHC class II levels [12] and additional RMND5B depletion further downregulated MHC class II levels at the cell surface. This suggests that RMND5B acts as an inhibitor of TGFβ activated MHC class II transcription.

Yet, most candidates did not affect MHC class II at the level of transcriptional control.

To place also these cadidates in networks, a second screen was performed. Here, we again depleted the candidate genes from the primary screen using siRNAs and we deter- mined their effect on the intracellular distribution of MHC class II containing vesicles, early endosomes and Golgi by confocal microscopy. Silencing occured in MelJuSo cells stably expressing HLA-DRβ1-GFP and mCherry-GalT2 (Golgi marker); early endosomes were visualized by endogenous EEA1 staining. Information on the intracellular distribution of MHC class II, the Golgi and early endosomes after depletion of the different candidates was used to cluster them according to their microscopy phenotypes.

One of these clusters contained proteins which all showed, after depletion, MHC class II at their membrane at the cost of intracellular MHC class II containing vesicles.

This phenotype resembled the MHC class II distribution in mature dendritic cells. To confirm this observation, shRNAs for these candidates were introduced in immature DCs using lenti-viral transduction. This indeed yielded immature dendritic cells with a mature dendritic cell phenotype and also showed that a phenotype strongly associated with mature DCs can be dissociated by inactivating the correct candidate protein as defined in our screen.

How do the identified candidate proteins regulate MHC class II vesicle transport?

This is an important point: how to translate a massive dataset into biological relevant new pathways? We managed to identify a

new mechanism which captures MHC class II vesicles intracellular (as in immature DCs) by combining various different high throughput screens (discussed above) com- bined with Yeast two hybrid and proteomics.

We selected the GTPase Arl14/Arf7, which is one of the candidates affecting the intracellular localization of MHC class II molecules, as the candidate for follow up studies.

We showed that Arl14/Arf7 is localized on MHC class II positive vesicles in DCs. Since Arl14/Arf7 is a GTPase, which has to be activated by a GEF to perform its function, we used the data from the initial screen to identify its potential GEF protein. Two GEFs showed upregulation of MHC class II after their depletion, like Arl14/Arf7, and were tested in a GTPase assay. The GEF PSD4 was confirmed as a GEF for the GTPase Arl14/Arf7. Activated GTPases will bind to their effector proteins. Since Arl14/Arf7 effector proteins were unknown, we performed a Yeast-two-hybrid experiment and identified Arf7EP (Arf7 Effector Protein), a 29kD protein specifically expressed in immune cells, as a Arl14/Arf7 effector. Arl14/

Arf7 and Arf7EP did indeed co-localize in MelJuSo cells and their interaction was confirmed by co-immunoprecipitation experiments. Pulldown on Arf7EP identified the interacting actin motor protein Myosin 1E.

Interaction between Arf7EP and Myosin 1E was confirmed by a co-immunoprecipitation experiment. Together, PSD4, Arl14/Arf7, Arf7EP and Myosin 1E form a complex that anchors vesicles to actin (Model see Chapter 2, Figure 7). We expect that this anchoring mechanism gets inactivated by a GAP protein during dendritic cell maturation, allowing transport of MHC class II containing vesicles to the cell surface for antigen presentation.

Thus, by combining different high throughput screens, new pathways and new biology can be generated. The microscopy based im- ages and networks are used as a data reposi-

(8)

Chapter 6 148

tory to extract new candidates, as illustrated below.

A second candidate revealing an interesting phenotype in the microscopy screen is the E3 ligase RNF26. Depletion of RNF26 led to dispersion of MHC class II containing vesicles. We showed that RNF26 is an ER protein that not only affected the localization of MHC class II containing vesicles, but also that of other endocytic vesicles, suggest- ing a more general role for RNF26 and the ER in the control of endosomal transport.

Again, we decided to build a network that should explain how the E3 ligase RNF26 is able to control the endosomal pathway. We followed the same strategy as before. We produced the tail of RNF26 and performed a pulldown and proteomics to identify seven RNF26 interacting proteins including the actin-based motor protein Myosin VI. These proteins form ubiquitin dependent complex- es, regulated by RNF26, leading to Myosin VI-mediated vesicle tethering (Model see Chapter 4, Figure 6 and 8). RNF26 regulated Myosin VI-mediated vesicle localization plays an important role in the last step of the cell cycle since depletion of RNF26 shows, similar to Myosin VI depletion, many multinucleated cells and an elongated dura- tion of the mitosis. RNF26 is a RING-do- main containing E3-ligase and needs an E2 enzyme to ubiquitinate its target proteins. To identify the E2 of RNF26 we performed an additional microscopy-based RNAi screen on all 40 currently known E2 enzymes. This way we identified the E2-enzyme UBE2J1 as a potential candidate to pair with RNF26.

UBE2J1 is like RNF26 localized in the ER membrane [13] and interaction between the two proteins was identified by co-immunoprecipitation experiments.

Our follow-up studies on the candidates Arl14/Arf7 and RNF26 revealed two completely new mechanisms regulation vesicle

localization. So far we did follow up on only a few of the 276 identified proteins in our genome wide screen. The remaining candidates will contain many more interesting proteins with a function in the MHC class II pathway, but our data set will most likely also contain some false-positive candidates because of off-target effects of the used siR- NAs. Beside false-positive candidates our screen also contains false-negatives. We used MelJuSo cells as a model system, so we have to keep in mind that MelJuSo cells do have a different proteome compared to the pro- fessional APCs. Proteins not expressed in MelJuSo will not pop up as a candidate in our screen. Additionally we may miss candidates since some siRNAs do not lead to proper silencing or redundancy. Linking our networks to new networks that may enter the scientific literature, will improve the quality of the data set and allow testing for new ways that control MHC class II antigen presentation. These will yield new ways to further manipulate antigen presentation by MHC class II molecules as done by cytokines, in- fections and other mechanisms possibly not fully known at present.

As mentioned above, MHC class II molecules are linked to auto-immune diseases.

Candidates from our screen may be used as drug-targets to treat certain diseases. Since protein depletion is comparable to chemi- cally inhibiting the same candidate, we can integrate chemical and genetic screens to identify new target-drug combinations in an unbiased manner, coupled to a defined biological effect; manipulation of MHC class II antigen presentation. To identify potential drugs affecting MHC class II antigen presentation, we performed a compound screen using a LOPAC compound library. Screen- ing of these compounds on MelJuSo cells did not lead to the discovery of a potential drug regulating MHC class II antigen presentation. Nowadays many more compound

(9)

6

libraries are available and also these can be tested for their effects on MHC class II antigen presentation. Identification of drugs regulating the MHC class II pathway, will be a great step forward in the treatment of autoimmune diseases related to MHC class II.

In summary, we performed a highly success- ful siRNA-based genome wide screen which identified new players involved in the MHC class II antigen presentation pathway. We identified a transcriptional network involved in MHC class II and CIITA transcription, an Arl14/Arf7 and Myosin1E regulated mechanism of MHC class II vesicle transport important for dendritic cell maturation and a more general mechanism controlling endosomal vesicle localization regulated by the E3 ligase RNF26 and Myosin VI which plays an important role in mitosis. This was only pos- sible by establishing new technologies that allow placing new and unknown proteins in networks in an unbiased manners. This work did not only result in new technologies but also new biology, relevant in the control of MHC class II antigen presentation.

Reference List

1 Chaplin,D.D. and Kemp,M.E. (1988) The major histocompatibility complex and autoimmunity. Year Immunol. 3, 179-198 2 Roark,C.L. et al. (2014) Multiple HLA epitopes contribute to type 1 diabetes susceptibility. Diabetes 63, 323-331

3 Abolfazli,R. et al. (2014) Relationship between HLA-DRB1* 11/15 genotype and susceptibility to multiple sclerosis in IRAN.

J. Neurol. Sci.

4 Denham,J.M. and Hill,I.D. (2013) Celiac disease and autoimmunity: review and con- troversies. Curr. Allergy Asthma Rep. 13, 347-353

5 Reynolds,A. et al. (2006) Induction of the interferon response by siRNA is cell type- and duplex length-dependent. RNA. 12, 988-993

6 Hornung,V. et al. (2005) Sequence-specific potent induction of IFN-alpha by short in- terfering RNA in plasmacytoid dendritic cells through TLR7. Nat. Med. 11, 263-270 7 Judge,A.D. et al. (2005) Sequence-dependent stimulation of the mammalian innate immune response by synthetic siRNA. Nat.

Biotechnol. 23, 457-462

8 Sioud,M. (2005) Induction of inflamma- tory cytokines and interferon responses by double-stranded and single-stranded siR- NAs is sequence-dependent and requires endosomal localization. J. Mol. Biol. 348, 1079-1090

9 Wubbolts,R. et al. (1996) Direct vesicular transport of MHC class II molecules from lysosomal structures to the cell surface. J. Cell Biol. 135, 611-622

10 Snel,B. et al. (2000) STRING: a web- server to retrieve and display the repeatedly occurring neighbourhood of a gene. Nucleic Acids Res. 28, 3442-3444

11 Colland,F. et al. (2004) Functional proteomics mapping of a human signaling pathway. Genome Res. 14, 1324-1332

12 Dong,Y. et al. (2001) The Smad3 protein is involved in TGF-beta inhibition of class II transactivator and class II MHC expression.

J. Immunol. 167, 311-319

13 Tiwari,S. and Weissman,A.M. (2001) Endoplasmic reticulum (ER)-associated degradation of T cell receptor subunits. In- volvement of ER-associated ubiquitin-con- jugating enzymes (E2s). J. Biol. Chem. 276, 16193-16200