The HARPER Lab

Department of Cell Biology

Research Projects

BioPlex Large Scale Interaction Proteomics

BioPlex: an ORFeome-based, Mass Spectrometry-driven, Human Protein Interaction Network

Because a cell's phenotype reflects its underlying proteome, mass-spectrometry-enabled proteomic studies can provide essential biological insights. Whereas surveys of protein expression and select post-translational modifications have achieved near-comprehensive scope, mass-spectrometry-based protein interaction profiles have typically examined small protein families while leaving vast swaths of the interaction landscape unexplored. To achieve more comprehensive examination, we have established a high-throughput pipeline capable of elucidating interactions for several hundred baits per month and are now systematically mapping human protein interactions at unparalleled depth and breadth. The emerging interaction network is providing unique insights into both normal and pathological biological processes by revealing novel functions for familiar proteins and providing tantalizing clues into roles of unknown proteins.

Our high-throughput platform for protein interaction profiling relies upon expressing HA-tagged versions of human proteins within HEK293T cells; following immunoprecipitation, the baits and their interacting partners are identified using Q-Exactive mass spectrometers aided by customized data analysis techniques. To date we have completed analysis of over 3,200 bait proteins and their interacting partners, defining an interaction network that spans 33,000 interactions among over 8,000 proteins. The resulting network reveals clusters matching known subcellular structures such as the proteasome, signalsome, and mediator complexes, while suggesting biological functions for unknown proteins and highlighting shared functional and regulatory clusters across the interactome. Whether viewed individually or in aggregate, these interaction profiles offer unique insights into both known and unknown proteins while also illuminating larger patterns of proteomic regulation.

Proteomic and cell biological dissection of mitochondrial networks

Interaction map of Complex I from mitochondria determined by mass spectrometry.

A new area of research in the lab involves the analysis mitochondrial disease genes. Our long-term goal is to employ quantitative proteomi approaches to understand how mitochondrial disease genes alter protein assemblies within the mitochondria that are required for energy production, metabolism, and mitochondrial quality control. In our initial foray into this area, we employed interaction proteomics to partners of 15 Complex I components previously found to be mutated in humans with mitochondrial disease. Complex I is the largest (44 subunits) and arguably the least understood complex in the electron transport chain (ETC). It is known to be assembled through a very complex process involving multiple intermediate complexes, and a number of assembly factors. Using our interaction proteomics platform, we identified a previously unrecognized member of the ESCIT-ACAD9-TMEM126B-NDUFAF1 (refered to as MCIA) assembly factor complex, which is known to play a role in integrating Complex I components within the mitochondrial inner membrane. This new protein - TIMMDC1/C3orf1 - is a multipass transmembrane protein which we demonstrated to reside in the mitochondrial inner membrane and to be required for assembly of Complex I. Our studies demonstrated that TIMMDC1 is required for Complex I activity. To explore the biological functions of TIMMDC1, we developed a quantitative proteomic approach that allows us to identify Complex I intermediates that accumulate when TIMMDC1 is depleted by RNAi. This method allows the identification of individual sub-complexes that accumulate in response to loss of a particualr assembly factor. These studies allowed us to put the function of TIMMDC1 into the context of existing models for Complex I assembly. Additional ongooing studies seek to employ analogous approaches to dissect the functions of other mitochondrial proteins that have been linked to human disease.

Targets and Mechanisms of CRLs

SCF targets and regulatory kinases we have worked on over the years.
Structures of phosphodegrons bound to 2 F-box proteins (collaboration with Nikola Pavletich)
Systematic analysis of substrates for the FBXL family of CRLs

The Uniquitin-Proteasome Pathway (UPP) controls the abundance and activity of a large number of proteins in the cell. Through the work of this and other labs, the Cullin-RING ubiquitin ligase system has emerged as a major system controlling the stability and regulated proteolysis of hundreds of cellular proteins in key signaling and disease systems. CRLs are composed of an E3 ligase module containing a neddylated cullin (CUL1-7) and a RING finger protein (RBX1 or 2) and a substrate adaptor module which interacts with substrate. For example, CUL1’s adaptor module is composed of SKP1 and a member of the F-box family of proteins, of which 68 exist in mammals. Thus, this super-family of E3s which is referred to as the SCF (SKP1-CUL1-F-box protein) encodes a large variety of E3s involved in a plethora of signaling pathways, ranging from Wnt, NFkB, Hedgehog, and numerous circuits that control cell division. Our research efforts on the CRL system, including: 1) identification of new substrates for CRLs, with a focus on CUL1 and CUL4-based CRLs linked with disease, 2) architecture of the CRL system in mammalian cells analyzed using quantitative protoemics, 3) mechanisms of SCF regulation.

Recently, we have developed methods for systematic identification of CRL substrates [referred to as PAC (Parallel Adaptor Capture) Proteomics] and have applied this approach to a class of F-box proteins - The leucine rich repeat FBXL class. This led to the identification of numerous candidte substrates for each of the 20 FBXL proteins analyzed. We examined a candidate substrate of FBXL17 in detail. This substrate - BACH1 - is a transcriptional repressor of NRF2 dependent genes. We found that FBXL17 promotes turnover of BACH2 in order to "license" NRF2 promoters for activation by NRF2. The use of this approach has the potential to identify substrates broadly across all CRL type ubiquitin ligases. See: Tan et al, Molecular Cell, 2013.

Quantitative and systematic analysis of the ubiquitin modified proteome

Work flow for ubiquitin capture proteomics

Together with the Gygi lab, we have used proteomic approaches to identify ~19,000 ubiquitination sites in ~5000 human proteins in multiple cell lines. The approach involves enrichment of “diGly renmants” using a monoclonal antibody that reacts with lysine residues carrying this modification, followed by mass spectrometry analysis. Using this approach, we have examined the temporal control of accumulation of proteins in response to proteasome inhibition, allowing us to classify various types of substrates. We have also examined the role of protein synthesis in defining the ubiquitin modified proteome, and have found that identification of a significant fraction of sites requires ongoing protein synthesis. Finally, we have used this approach to identify candidate substrates of cullin-RING ubiquitin ligases. These studies provide a resource for the identification and analysis of ubiquitinated proteins.

The database can be searched online at: https://gygi.med.harvard.edu/ggbase/

Applying ubiquitin proteomics to diseases such as Parkinson’s Disease

The parkin-dependent ubiquitylome.

PARKIN is a Ubiquitin ligase that is mutated in early onset Parkinson’s Disease. In response to mitochondrial depolarization, PARKIN is recruited to the mitochondrial outer membrane (MOM) in a PINK1 kinase dependent manner. There, PARKIN ubiquitylates proteins on the MOM to regulate mitochondrial dynamics and the process of mitophagy, wherein mitochondria are targeted for autophagy. We applied quantitative diGLY proteomics to the problem of PARKIN substrate identification, thereby revealing the landscape of the PARKIN modified ubiquitylome - the collection of ubiquitylation sites in PARKIN-dependent ubiquitylation targets. This work provides a resource for the field and will enable an understanding of how PARKIN regulates mitochondrial function.

We are currently extending these studies to other quality control pathways including ALS.

Interaction Networks

  CompPASS
Hierarchical Clustering of the Dub Proteome

A second major goal of our work is to develop proteomic and informatic approaches that allow for the rapid identification of protein interactions within signaling networks and within particular classes of proteins - for example specific classes of ubiquitin pathway genes. We have established a platform for proteomic and informatic analysis of genesets. The centerpiece of this platform is a suite of software called CompPASS which greatly simplifies parallell processing of large proteomic datasets and facilitates the identification of interaction networks. We have applied this methodology to deubiquitinating enzymes, the human autophagy system, cullin RING Ubiquitin Ligases, and many other pathways and genesets.

Macro and Selective Autophagy

The human autophagy interaction network examined by mass spectrometry.
Ferritin delivery to lysosomes via the NCOA4 cargo adaptor protein

We have used proteomic tools in conjunction with the CompPASS system to analyze the interaction landscape of the human autophagy system. Autophagy is a process by which cellular proteins and organelles are sequestered into a double-lipid bilayer structure called the autophagosome and are then delivered to the lysosome for degradation. The turnover of cellular proteins through this pathway provides both recycled building blocks such as amino acids, and also is responsible for the turnover of most of the highly stable proteins in the cell. The autophagy system is critical for the ability of the cell to respond to stress, particularly in the context of cancer. Our proteomic analysis has identified an interaction network containing more than 400 proteins, derived from 32 primary autophagy related proteins and 33 new secondary bait proteins. Extensive validation and functional studies have revealed numerous aspects of how the machinery works together to build autophagosomes and deliver them to the lysosome. RNAi experiments examining a sub-set of genes in the pathway have revealed dozens of genes that affect the production of autophagosomes or flux through the autophagy pathway. See Behrends et al, Nature, 2010.

A major question concerns the identity and functions of cargo adaptors, in particular those that are involved in selective autophagy. Moreover, the identity of specific proteins that are targeted to the lysosome via autophagy for selective degradation is largely unknown. We have purified autophagosomes and employed a quantitative proteomic approach to identify resident autophagosomal proteins, identifying more than 200 proteins that are intimately associated with autophagosomes. Exploration of one of these - NCOA4 - led to the discovery of a pathway that targets ferritiin for lysosomal degradation when cells need iron. Ferritin is a critical protein for ior homeostasis. 24 copies of ferriton form a cage which binds ~4000 iron atoms and when iron levels are too high, ferritin is induced and captures excess iron, thereby protecting cells from the Fenton reaction which produces toxic oxidative species. However, when iron levels are reduced, ferritin is targeted to the lysosome, where it is degraded, thereby releasing the iron for its incorportation into important iron-dependent proteins. NCOA4 functions as an essential component by linking ferritin to the autophagosome, which is then targeted to the lysosome for degradation. In cells which lack NCOA4, ferritin isnt collected in the autophagosome and therefore isn't targeted to the lysosome. Mechanistically, NCOA4 interacts with the autophagosomal protein ATG8, which is a key step in the process. These studies suggest a new framework inwhich to understand the links between autophagy and iron control, a process that is linked to numerous diseases, including neurodegeneration. See: Mancias et al, Nature, 2014.

Quantitative Proteomics of Signaling Networks

Quantitative analysis of CRL complexes by AQUA proteomics

Proteome networks undergo reorganization in response to cellular signaling. This often results in either protein turnover, or changes in the modification state of proteins. However, understanding how networks are reorganized in response to perturbations such as small molecule inhibitors of a pathway are often difficult, and require quantitative approaches. We are applying multiple quantitative proteomic approaches to discern how various signaling networks are altered in response to pathway perturbation. We have performed a systematic analysis of the cullin-RING E3 ligase network in response to inhibition of the neddylation system using a small molecule inhibitor. This work has revealed that the abundance of cullin adaptor proteins in the cell control the assembly state of the CRL network, and there is little evidence for the involvement of CAND1 in global sequestration of cullins in response to acute dennedylation. These studies employ multiplex AQUA (absolute quantification) as a tool to survey the concentrations for more than 14 proteins in the CRL network simultaneously. We are currently developing this approach further with the intent to characterize dynamic changes in tumor suppressor and oncogenic signaling systems.