Biomimetic Virulomics for Capture and Identification of Cell-Type Specific Effector Proteins John D. Lapek, Jr.,†,‡ Ronnie H. Fang,§,∥ Xiaoli Wei,§,∥ Pengyang Li,⊥ Bo Wang,*,⊥ Liangfang Zhang,*,§,∥ and David J. Gonzalez*,†,‡ †
Department of Pharmacology, ‡Skaggs School of Pharmacy and Pharmaceutical Sciences, §Department of NanoEngineering, and Moores Cancer Center, University of California, San Diego, La Jolla, California 92093, United States, ⊥ Department of Bioengineering, Stanford University, Stanford, California 94305, United States ∥
S Supporting Information *
ABSTRACT: An unmet challenge in the study of disease is to accurately streamline the identification of important virulence factors. Traditional, genetically driven approaches miss biologically relevant markers due to discordance between the genome and proteome. Here, we developed a nanotechnology-enabled affinity enrichment strategy coupled with multiplexed quantitative proteomics, namely Biomimetic Virulomics, for successful identification of cell-type specific effector proteins of both prokaryotic and eukaryotic pathogens. We highlight the power of Biomimetic Virulomics by capturing known virulence factors in a highthroughput, cell-type guided fashion. Additionally, a comprehensive characterization of the membrane protein component of biomimetics utilized in this strategy is provided. Interfacing cell-derived nanomaterials with multiplexed quantitative proteomics allow for a specific targeting strategy of virulence factors that can be utilized for drug discovery against prominent human diseases. KEYWORDS: biomimetics, affinity enrichment, quantitative proteomics, mass spectrometry, virulence factors
T
particularly at the protein-level, has proven to be challenging. Historically, direct detection of virulence-associated proteins by mass spectrometry has been hindered by high background from host proteins, making a proteomics approach difficult.3 Biomimetic nanosponges (NS), consisting of polymeric nanoparticle cores coated with endogenous cell membranes, have emerged as an effective means of neutralizing and retaining pore forming toxins,4 a classical virulence mechanism. NS are currently available for a variety of host cell types5−8 and have even been coated with bacterial membranes,9 thus demonstrating the potential for broad application of this technology. The endogenous membrane coatings camouflage the NS, allowing them to act as decoys for virulence factors that target specific host cell types. However, previous studies utilizing NS do not identify the virulence factors they neutralize a priori. Biomimetics have the added advantage of avoiding destruction by the host immune system.10,11 Intriguingly, biomimetics are known to bind and neutralize toxins while
he term virulence is commonly used regarding infectious disease. By its definition, however, virulence can be used to describe the severity of any disease. As such, virulence factors can be defined as molecules that add to the effectiveness of a disease and thus allow the disease to thrive in a host. Diseases that can be encompassed by this definition not only include those of bacterial or viral origin but also a plethora of diseases including cancer.1 Conversely, host-derived defense molecules can also be included under the umbrella of virulence factors, as they are molecules that aid in the prosperity of the parent organism in its environment while eliciting a virulent effect on their target. Under a classical definition of virulence, infectious diseases remain a major global health concern with associated high morbidity and mortality rates.2 Within host−pathogen interactions, virulence is governed by biomolecules produced by the pathogen that target different tissues during infection. Thus, the identification of pathogenic effector proteins with cell-type specificity is paramount in understanding pathophysiology associated with infection and for future development of effective intervention strategies. However, the identification of relevant virulence factors, © 2017 American Chemical Society
Received: April 17, 2017 Accepted: September 11, 2017 Published: September 11, 2017 11831
DOI: 10.1021/acsnano.7b02650 ACS Nano 2017, 11, 11831−11838
Article
www.acsnano.org
Article
ACS Nano maintaining natural conformations of toxin proteins, allowing the immune system to mount a response and increase survival.6 Through this study, we sought to overcome the challenges associated with direct identification of virulence factors, biochemically understand the constituents that comprise a NS, and identify proteins bound and neutralized by the NS. Toward this goal, multiplexed quantitative proteomics12 was combined with cell-type specific NS affinity capture to yield an affinity enrichment workflow, Biomimetic Virulomics (Figure 1). Multiplexed quantitative proteomics gives a two-fold
Figure 1. Biomimetic Virulomics affinity enrichment workflow. Coated NS are incubated with pathogen-derived, protein containing supernatants. Bound NS are separated from supernatants. Both fractions are digested with trypsin, peptides labeled with TMT reagents, and peptides identified and quantified by mass spectrometry.
Figure 2. Confirmation of RBCNS and MNS membrane presence through coating process. (a) Z-average size of bare PLGA cores and RBCNS (n = 3, mean ± SD). (b) Zeta potential of bare PLGA cores and RBCNS (n = 3, mean ± SD). (c) Transmission electron microscopy image of RBCNS. Scale bar = 100 nm. (d) Z-average size of bare PLGA cores and MNS (n = 3, mean ± SD). (e) Zeta potential of bare PLGA cores and MNS (n = 3, mean ± SD). (f) Transmission electron microscopy image of MNS. Scale bar = 100 nm.
advantage over normal proteomics. First, the issue of missing values between replicates and samples can be addressed, while increasing throughput.13 In a traditional proteomics experiment, missing values can lead to false positives and false negatives, where proteins may be considered true or false binders based solely on their presence or absence in a list of identified proteins. Second, the relative quantitation enabled by multiplexed quantitative proteomics allows us to determine fold-enrichment, a measure of specificity, when comparing samples.
allows us to leverage this information to guide future mechanistic studies of virulence factor action. Quantitative proteomics was used to characterize two hostderived RBCNS, human and mouse, and MNS from mice (Figure 3). In total, 981 proteins were identified specifically associated with the human RBCNS. Of these, 937 (>95%) were statistically specific, assessed by π value,17 to the RBCNS (Figure 3a). The π value was used as a statistical measure to account for variance in measurement due to the heterogeneity of the RBCNS population. The remaining proteins were likely not statistically associated due to degradation of a small population of RBCNS or from residual membrane component in the supernatant from the coating process. Though these results are not a comprehensive catalogue of all RBC proteins, they represent an advance in the number of proteins derived from RBCs,18 where previous studies required a compendium of techniques to define only 592 RBC proteins. Additionally, there is well-documented difficulty in proteomics relating to the identification of membrane proteins, especially multipass transmembrane proteins.19−21 Taken together with previous in vivo and in vitro findings,4−11 RBCNS demonstrate an ability to accurately reflect interactions between host cells and pathogens. Despite this, it is possible that not every host cell−pathogen interaction will be captured or identified, regardless of the cell type. Failure to identify interactions could arise from technical limitations of the mass spectrometer, the length and nature of the effector molecule, and whether or not the effector molecule embeds in the membrane with high affinity.
RESULTS/DISCUSSION Prior to applying the workflow, we first characterized the NS used in subsequent experiments. NS with red blood cell (RBC) membrane (denoted RBCNS) and macrophage membrane (denoted MNS) were prepared for host cell-type specific enrichment of pathogen-derived effector proteins (Figure 2). For both RBCNS and MNS, an increase in diameter was observed once the NS underwent the membrane cloaking process (Figure 2a and 2d). Furthermore, a reduction in zeta potential, indicating the presence of the membrane and its insulating properties (Figure 2b and 2e), was noted. Membrane coating, known to exist in a right-side-out orientation,14 was confirmed by electron microscopy, and the particles exhibited a characteristic core−shell structure (Figure 2c and 2f). NS made from both RBC and nucleated cell membranes have been previously shown to possess narrow size distributions and exceptional stability under physiological conditions.5,11 Having confirmed the presence of membranes on the NS, we sought to define the host cell-type derived protein component of the membranes cloaking the NS. The rationale for this is that virulence factors can bind, inhibit, and even degrade specific host membrane proteins through a variety of mechanisms.15,16 Understanding which host membrane proteins are present 11832
DOI: 10.1021/acsnano.7b02650 ACS Nano 2017, 11, 11831−11838
Article
ACS Nano
cells would identify upward of 10,000 proteins.24 We hypothesized that this reduction in complexity of host-derived sample matrix would allow a deeper characterization of proteins derived from pathogens with full implementation of the Biomimetic Virulomics workflow. To benchmark the Biomimetic Virulomics platform, we utilized human-derived RBCNS to selectively enrich for hemolytic proteins from Group A streptococcus (GAS), a human specific pathogen. As hemolysis is a characteristic virulence mechanism of GAS and a clinical diagnostic phenotype, we first confirmed the ability of the RBCNS to bind and neutralize the hemolytic activity of secreted proteins collected from GAS cultures, recapitulating previous work.4 Cell-free supernatants from GAS were fractionated to identify hemolytic fractions (HF),25 and lytic ability was confirmed by heme release from blood. After incubation of the HF with the RBCNS, no quantifiable lysis was observed, indicating neutralization of the hemolytic proteins (Figure 4a). Upon confirming neutralization of lysis, RBCNS were incubated with cell-free GAS supernatants to identify GAS proteins specifically bound to the RBCNS. In parallel, the effect of blocking RBCNS was assessed with bovine serum albumin prior to incubation with GAS supernatants. A statistical difference in specific binding proteins was not observed (Supplementary Figures 2 and 3). As such, blocked NS were utilized for subsequent characterization with the assumption that blocking more accurately mimicked in vivo conditions. Prior to full implementation, we ensured blocking did not affect the stability of proteins associated with the NS (Supplementary Figures 4 and 5). In total, 602 GAS proteins were identified (35% of the GAS proteome), with only 58 (
