Binding BSA to silver surface via large difference in isoelectric points of the two materials

Binding BSA to silver surface via large difference in isoelectric points of the two materials

We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

I would like to bind BSA to a silver surface so that I can utilize plasmonic sensing to detect the BSA. There seems to be two methods of doing this: 1) to rely on electrostatic forces or 2) to form a covalent bond. The former seems to be the easier route, especially for someone like me who does not have a background in chemistry/biology.

So I found a paper (, which has been cited 54 times, that determined a silver (oxide) surface has an isoelectric point (iP) of ~10 pH. So this should mean that it has a positive charge when a neutral pH solution covers the surface. In other literature BSA has been shown to have an iP of pH 4-5, so it should become negatively charged in a neutral pH solution. So the BSA should bond to the silver surface in a neutral solution, correct?

The reason I ask is two-fold. First, I found a more recent paper ( that has a conflicting value for the iP of a silver film (~3 pH). Second, I found a paper ( were the authors adsorbed BSA onto silver films, but with the intermediate step of applying 2-mercaptoethanesulphonate to the silver surface. Why the need for this if the silver should attract the BSA electrostatically?

In addition, if BSA bonding to silver were to work in this way, am I correct in thinking that a neutral solvent such as phosphate buffered saline would hinder the electrostatic bonding process because the positive and negative ions from the salt solution would bond to the respectively charged molecules and neutralize the charge?

Concerning the apparent discrepancy between your first and second citations, the difference in values is addressed by the authors in the more recent work:

Perusal of the literature yields few IEP values for silver metal. Chau and Porter reported an IEP of 10.4 for an evaporated silver film as determined by contact angle titration.(6) The method employed in their investigation used only neutral and basic liquids, and the maximum contact angle achieved on the silver films was ∼35°, in contrast to the maximum we observe of 52° at pH 3.2 (Figure 1d). This dichotomy may be indicative of the amphoteric behavior of the native oxide at the surface.

Concerning your third citation where the authors bound BSA to silver via a 2-mercaptoethanesulphonate linkage monolayer, the authors give their reasoning in the abstract:

The direct adsorption of proteins on metal surfaces usually leads to loss of their enzymatic activity

So, to answer one of your questions

Why the need for this if the silver should attract the BSA electrostatically?

BSA here is a proxy for other proteins. The purpose of the research was to examine how the structure of the linkage monolayer changed in response to protein adsorption, not how BSA specifically binds a silver substrate.

Deciphering molecular mechanism of silver by integrated omic approaches enables enhancing its antimicrobial efficacy in E. coli

Affiliation CAS Key Laboratory of Magnetic Resonance in Biological Systems, State Key Laboratory of Magnetic Resonance and Atomic and Molecular Physics, National Centre for Magnetic Resonance in Wuhan, Wuhan Institute of Physics and Mathematics, Chinese Academy of Sciences, Wuhan, P. R. China

Roles Data curation, Investigation, Methodology

Affiliation Department of Chemistry, The University of Hong Kong, Hong Kong, P. R. China

Roles Data curation, Methodology

Affiliation School of Biological Sciences, The University of Hong Kong, Hong Kong, P. R. China

Roles Investigation, Methodology

Affiliation School of Chemistry, Sun Yat-sen University, Guangzhou, P. R. China

Affiliation Department of Chemistry, The University of Hong Kong, Hong Kong, P. R. China

Roles Investigation, Software

Affiliation Department of Chemistry, The University of Hong Kong, Hong Kong, P. R. China

Roles Conceptualization, Methodology

Affiliation State Key Laboratory of Environmental Chemistry and Ecotoxicology, Research Centre for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing, P. R. China

Roles Writing – review & editing

Affiliation School of Chemistry, Sun Yat-sen University, Guangzhou, P. R. China

Roles Methodology, Resources

Affiliation State Key Laboratory of Genetic Engineering, Zhongshan Hospital and School of Life Sciences, Fudan University, Shanghai International Centre for Molecular Phenomics, Collaborative Innovation Centre for Genetics and Development, Shanghai, P. R. China

Roles Methodology, Resources

Affiliation Singapore Phenome Center, Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore

Roles Conceptualization, Methodology, Writing – original draft, Writing – review & editing

Affiliation Department of Chemistry, The University of Hong Kong, Hong Kong, P. R. China

Roles Conceptualization, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Software, Supervision, Writing – review & editing

Affiliation Department of Chemistry, The University of Hong Kong, Hong Kong, P. R. China


The adsorption characteristics of three proteins [bovine serum albumin (BSA), myoglobin (Mb), and cytochrome c (CytC)] onto self-assembled monolayers of mercaptoundecanoic acid (MUA) on both gold nanoparticles (AuNP) and gold surfaces (Au) are described. The combination of quartz crystal microbalance measurements with dissipation (QCM-D) and pH titrations of the ζ-potential provide information on layer structure, surface coverage, and potential. All three proteins formed adsorption layers consisting of an irreversibly adsorbed fraction and a reversibly adsorbed fraction. BSA showed the highest affinity for the MUA/Au, forming an irreversibly adsorbed rigid monolayer with a side-down orientation and packing close to that expected in the jamming limit. In addition, BSA showed a large change in the adsorbed mass due to reversibly bound protein. The data indicate that the irreversibly adsorbed fraction of CytC is a monolayer structure, whereas the irreversibly adsorbed Mb is present in form of a bilayer. The observation of stable BSA complexes on MUA/AuNPs at the isoelectric point by ζ-potential measurements demonstrates that BSA can sterically stabilize MUA/AuNP. On the other hand, MUA/AuNP coated with either Mb or CytC formed a reversible flocculated state at the isoelectric point. The colloidal stability differences may be correlated with weaker binding in the reversibly bound overlayer in the case of Mb and CytC as compared to BSA.

North Carolina State University.

Colorado State University.

Royal Institute of Technology.

Institute for Surface Chemistry.

Author to whom correspondences should be addressed. Phone: (919)515-8915. Fax: (919)515-8909. E-mail: [email protected]

Materials and Methods


DOPC (1,2-dioleoyl-sn-glycero-3-phosphocholine) was purchased from Avanti Polar Lipids as a solution in chloroform. Cholesterol was obtained in powder form from Sigma Life Science. Pentavalent MVL5, PEG2000-lipid, RGD-, iRGD-, cRGD-, and RPARPAR-PEG2000-lipid were synthesized as previously described [40,57,58]. The pGFP plasmid encoding the GFP gene was purchased from Promega, propagated in Escherichia coli, and purified using Qiagen Giga or Mega Prep kits. Stock solutions of pGFP were prepared in deionized water (dH2O). For in vitro studies, the pGFP plasmid was labeled using YOYO-1 dye (Molecular Probes). For in vivo studies, the pGFP plasmid was labeled using the Mirus Bio Label IT Nucleic Acid Labeling Kit with Cy5 (excitation/emission maximum: 649 nm/670 nm) (see details below).

Liposome and DNA Preparation

Stock solutions of MVL5, cholesterol, and PEG2000-lipid were prepared by dissolving them in a 3:1 (v/v) chloroform/methanol mixture. RGD-, iRGD-, and cRGD-PEG2000-lipid were dissolved in a 65:25:4 (v/v/v) chloroform/methanol/dH2O (dH2O, deionized water) mixture. RPARPAR-PEG2000-lipid was dissolved in methanol. Lipid solutions were combined volumetrically at the desired molar ratio, and the solvent was evaporated by a stream of nitrogen followed by incubation in a vacuum overnight (12� h). The appropriate amount of high resistivity water (18.2 M㪜m) was added to the dried lipid film to achieve the desired lipid concentration (1 mM). Hydrated films were incubated overnight (12� h) at 37 ଌ to form liposomes. The liposome suspension was then sonicated for 7 minutes using a tip sonicator to promote the formation of small unilamellar vesicles. Plasmid purification was performed according the manufacturers protocol. For in vitro experiments, pGFP was labeled with YOYO-1, using a dye/basepair ratio of 1:30 by incubating the appropriate amounts of dye and pGFP at 37 ଌ overnight. For in vivo experiments, pGFP was labeled using Cy5 according to the manufacturer’s protocol with one modification: the incubation time at 37 ଌ was increased from 1 to 2 h to improve labeling efficiency. Characterization of the nanoparticles (size by dynamic light scattering and zeta potential) is described in the Supplementary Material (Tables ST1 and ST2).

Membrane Charge Density and Charge Ratio

For the lipid formulations used in this study (10/70/10/10, molar ratio of MVL5/DOPC/Cholesterol/x, with x=PEG-lipid and/or peptide-PEG-lipid), the membrane charge density was low at σM𢒀.0061 e/Å 2 . The membrane charge density can be calculated from the equation σM=[1–Φnl/(Φnl+rΦcl)]σcl [25]. Here, r=Acl/Anl is the ratio of the headgroup areas of the cationic and the neutral lipid σcl=eZ/Acl is the charge density of the cationic lipid with valence Z Φnl and Φcl are the molar fractions of the neutral and cationic lipids, respectively. In our nanoparticles, the neutral lipid component is a mixture of DOPC, cholesterol, and the PEG-lipid (with and without peptide). For simplicity, we assigned an estimated average headgroup area (that of DOPC) to this lipid mixture. (Compared to DOPC, cholesterol’s headgroup is much smaller, while those of the PEG-lipids are larger.) Thus, the membrane charge density was calculated using Anl=72 Å 2 , rMVL5=2.3, and ZMVL5=5.0 [25,26].

Cell Culture

PC-3 cells (ATCC number: CRL-1435 human prostate cancer) and M-21 human melanoma cells (a gift from David Cheresh) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) (Invitrogen) supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin/streptomycin (Invitrogen). Cells were passaged every 72 h to maintain subconfluency and cultured in an incubator at 37 ଌ in a humidified atmosphere containing 5% CO2. MKN-45P human gastric cancer cells were originally isolated from parental MKN-45 cells (a gift from Joji Kitayama) as described [59]. The MKN-45P cells were cultivated in DMEM (Lonza) containing 100 IU/mL of penicillin and streptomycin, and 10% of heat-inactivated fetal bovine serum (GE Healthcare).

Flow Cytometry

Cells were detached using enzyme-free cell dissociation buffer (Gibco), seeded in 24-well plates at a density of 45 000 cells/well, and incubated for 18 h. A total of 1 μg of pGFP (10% YOYO-1–labeled) was used for each sample (i.e., two wells). This DNA was diluted to 250 μL with DMEM. The appropriate volume of liposomes (to reach the desired lipid/DNA charge ratio) was also diluted to 250 μL with DMEM. The diluted liposome and DNA solutions were mixed and incubated at room temperature for 20 min to allow nanoparticle formation. After washing the cells with PBS, 200 μL of NP solution was added to each well. Control wells received only DMEM or only (labeled) DNA. Cells were incubated with nanoparticles for 5 h, rinsed with PBS, detached with enzyme-free cell dissociation buffer (Gibco), and suspended in 200 μL of DMEM. Cells were maintained on ice after harvesting to inhibit further uptake of NPs during the measurement. Fluorescence was measured using a Guava EasyCyte Plus Flow Cytometry System (Millipore). Cell solutions were passed through a 100 μm filter to disperse aggregates prior to measurement. The filtered cell solution was divided in two. One half was mixed with a Trypan Blue (Gibco) solution (0.4% in water, w/v) at a 4:1 (cell:Trypan Blue) v/v ratio and incubated for 5� min before the measurement, quenching extracellular fluorescence. The other half of the cell solution was mixed with PBS at the same 4:1 v/v ratio and measured immediately. The software parameters were set such that 10,000 events constituted a single measurement, though some samples with significant cell detachment only reached 𢒂,000 events before time expired on the measurement while also showing an increased ratio of debris to cells. The flow cytometry results were analyzed using the Cyflogic software (CyFlo). Events were sorted using forward and side scattering to separate cells from debris. A single acceptance window was used for each plate of cells, taking care to account for any shifting of the scattering due to high NP binding. The green (YOYO-1) fluorescence distribution of the accepted events (cells) was log-normal, making the geometric mean a more accurate measure of the distribution than the arithmetic mean. The data plots ( Figures 2B and ​ and3) 3 ) show the normalized geometric means which were obtained by subtracting the geometric mean of the control (DMEM only) cells’ autofluorescence. The error bars show the uncertainty in the geometric mean which was calculated from the coefficient of variation (CV) of the fluorescence distribution using the following equation: σERROR=log(CV 2 𠄱)×I/N. Here, I is the geometric mean and N is the number of counted cells. The propagated uncertainty of the mean NP fluorescence included the error of both the total fluorescence and control autofluorescence distributions. Control samples (DMEM only, DNA only, PEG-lipid only) were repeated across multiple experiments and are reported as the average of those experiments (with a clear outlier removed in one case for the PEG-lipid only control).

Flow cytometry measurements of binding and internalization of peptide-tagged CL𠄽NA NPs as a function of peptide density. PC-3 and M-21 cells were incubated with NPs containing fluorescently labeled DNA. The NPs were formulated at ρch=1.5 with lipid mixtures of MVL5/DOPC/Cholesterol/PEG2000-lipid/peptide-PEG2000-lipid at a 10/70/10/10–x/x molar ratio (x as indicated in the plots peptide=RGD, iRGD, cRGD, or RPARPAR). Control NPs contained only nontargeted PEG2000-lipid (“PEG” x=0). Naked DNA (without lipid) and cell culture medium (DMEM, with autofluoresence) only were the negative controls. (A) Example distributions of fluorescence intensity for M-21 cells incubated with NPs containing 5 mol% peptide. The area under each curve (total cell count) differs between samples due to variations in cell detachment and cellular debris. The inset shows the distributions with their maximum height normalized to 1. (B) Plots of the average DNA fluorescence per cell, obtained by calculating the geometric mean of each fluorescence distribution and subtracting the mean autofluorescence of cells that received no DNA (𠇍MEM”). Green bars show the mean of the average DNA fluorescence, representing combined NP binding and internalization. Blue bars show the average DNA fluorescence measured in the presence of Trypan Blue (which quenches fluorescence from extracellular label), representing NP internalization only. For NPs containing ϡ mol% cRGD-PEG2000-lipid, a large number of cells detached from the growth substrate the fluorescence of these cells was measured independently (𠇌RGD (det.)”). At 5 mol% cRGD-PEG2000-lipid, too few attached PC-3 cells remained to obtain data (*).

Flow cytometry measurements of binding and internalization of peptide-tagged CL𠄽NA NPs (2.5 mol% peptide-PEG-lipid) as a function of lipid/DNA charge ratio (ρch). PC-3 and M-21 cells were incubated with NPs containing labeled DNA. The NPs were formulated at ρch=0.5, 1.0, 1.2, 1.5, and 5 using lipid mixtures of 10/70/10/7.5/2.5 MVL5/DOPC/Cholesterol/PEG2000-lipid/peptide-PEG-lipid (peptide=none (control), RGD, iRGD, cRGD, or RPARPAR). (A,C) Plots of the mean of the cell-associated fluorescence, representing combined NP binding and internalization. The inset shows the data for the detached population of cells treated with cRGD-tagged NPs. (B,D) Plots of the mean fluorescence measured in the presence of Trypan Blue (which quenches fluorescence from extracellular label), representing NP internalization only. All data was normalized by subtracting the mean fluorescence (autofluorescence) of control cells which received no DNA.

In vivo biodistribution studies

Athymic nude mice were purchased from Harlan Sprague Dawley. All the animal experimentation protocols were approved by Estonian Ministry of Agriculture, Committee of Animal Experimentation (Project #42). Mice were injected intraperitoneally with 2휐 6 MKN-45P cells, and the MKN-45P tumors were allowed to grow for two weeks. CL–NA NPs containing Cy5-labeled pGFP were injected intraperitoneally (0.1 mL of 5 mg/mL solution was diluted in 0.5 mL of PBS). In vivo imaging was performed 4 h and 24 h after injection (See Figure S3 in the Supplementary Material). After 24 h, the animals were perfused with 10 mL of PBS. The tumors and organs were excised for fluorescence visualization using an Optix MX3 (Advanced Research Technologies). Fluorescence quantification was done using the OptiView analysis software. The average fluorescent signal (n=3) was normalized by dividing by the control mouse fluorescence (receiving no injection) and the total grams of tissue. Tumor nodules were separated from membranous tissue connecting the nodules and pooled (for each set of NPs n=3). Tissues were snap-frozen in liquid nitrogen and stored at �ଌ for further analysis.

Immunofluorescence and microscopic imaging

The snap-frozen tumors and organs were cryosectioned at a thickness of 10 μm and fixed with 4% of paraformaldehyde in PBS. The nuclei of cells were counterstained with 1 μg/ml DAPI. Tissue sections were imaged on a Zeiss LSM 510 (for lower magnification images) and an Olympus FV1200MPE (for higher magnification images), and the confocal images were analyzed with the ZEN lite 2012 and Olympus FluoView image software, respectively.

Samples: sampling, bulking and multiplexing

Target traits and phenotyping for sampling

Bulked segregant analysis was originally designed to target the traits controlled by major genes with large effect and less confounded by environments. Recent developments in BSA have increased the power of bulked segregant analysis in identifying minor causal alleles (Bernier et al., 2007 Sun et al., 2013a Tuberosa et al., 2010 Venuprasad et al., 2011 Vikram et al., 2012 Xu and Crouch, 2008 Xu et al., 2008 Figures 2 and 3). A simulation study indicated that BSA can be used for mapping QTL with relatively small effects, as well as linked and interacting QTL. With the original population size of 3000, selection of 10% of the extreme individuals from each tail and marker density of 5 cM, we could have the power of 95% to detect a QTL that explains only 1% of the phenotypic variation (Sun et al., 2013a ). This has been supported by a study in yeast with identification of several genes with minor effects for chemical resistance traits and mitochondrial function (Ehrenreich et al., 2006 ).

The power of BSA largely depends on the feasibility of classifying individuals into groups with extreme phenotypes, which in turn depends largely on precision phenotyping under well-managed environments, particularly for the traits with low heritability and largely affected by environments. To improve phenotyping precision in field conditions, it is important to reduce ‘signal-to-noise’ ratio, by selection of research plots with low spatial variability in soil properties, uniform application of inputs with good weed, pest and disease control, use of adequate plot borders, use of experimental designs to control within replicate variability and data analysis to reduce or remove spatial trends (Xu, 2016 Xu et al., 2008 , 2012 ). Precision phenotyping also depends on the utilization of new field-based techniques (precision fertilization, water management and weed control remote sensing techniques for accurate evaluation of secondary traits) and correct selection, calibration and application of phenotyping instruments (such as neutron probes, radiation sensors and chlorophyll and photosynthesis meters).

For biotic and abiotic stresses, phenotyping needs to be performed simultaneously in two contrasting environments, or near iso-environments (NIEs) (Xu, 2015 , 2016 ), with one imposing much less stress on plants than the other. The effect of the stress environment can be measured using the much-less-stress or normal environment as a control. A relative trait value is then derived from two direct trait values to ascertain the sensitivity of plants to the stress. Traits suitable for measurement under NIEs include all abiotic/biotic stresses (e.g. disease resistance and drought tolerance) and agronomic practices (e.g. weed control). A relative trait value can be also derived by measuring of the same trait under the NIEs with one neutral factor significantly different such as plant responses to photoperiod or day-length.


Two contrasting sampling methods, trait-based sampling and marker-based sampling, have been used in BSA (Figure 3). The former is based on the phenotypic extreme plants for a trait of interest, and the plants are selected from the high and low tails of the phenotypic distribution (Lander and Botstein, 1987 Lebowitz et al., 2014a ). The second approach is based on molecular markers evenly covering the genome of the entire germplasm collection or segregating populations (Edwards et al., 2010 Soller and Beckmann, 2013 ), and individuals are selected by genotyping in the target region. In genetics, the former sampling method tends to be used for rough mapping (Vikram et al., 2012 ), while the latter mainly applies to fine mapping (Boopathi et al., 2013 Frouin et al., 2009 Yang et al., 2014 ). However, BSA is mainly based on trait-based sampling.

The power of BSA largely depends on sampling-related factors, particularly sample sizes including entire population size and tail size (the number of individuals selected for bulking) (Figure 3). In segregant-based BSA, the population size required mainly depends on population type, distance between markers, recombination frequency in the target region and genetic architecture of the target trait (Xu et al., 2008 ). As recombination frequency and relative information of genotypes usually vary across population types, the population size required in constructing the linkage map might also vary.

For the complex trait controlled by minor genes, other factors associated with the target genes, such as gene number, gene effect, gene interactions and the relative positions on chromosomes, should be taken into account to determine the required population size (Yan et al., 2009 ). To effectively identify marker–trait association, the population size should also consider marker availability and genotyping cost (Xu et al., 2008 ). At the same time, with the reduction in genotyping cost, the increase in population size becomes more feasible.

Taking genetic mapping as an example, how many individuals should be sampled from phenotypic extremes usually matters with the entire population size and gene effect. For small- to moderate-sized populations (each with 200–500 individuals), optimum tail size would be 20%–30% of the entire population (Gallais et al., 2011 Navabi et al., 2011 ). With the increase in the population size, selected proportion (SP) required for a given power of QTL detection will decrease. For a QTL of large effect (with phenotypic variation explained (PVE) = 10%–15% or larger), each tail should contain at least 20 individuals (or SP > 10%) selected from an entire population of around 200 (Sun et al., 2013a ). For a QTL of medium effect (PVE = 3%–10%), each tail should contain 50 individuals (SP = 5%–10%) from an entire population of 500–1000. For QTL of small effect (PVE = 0.2%–3%), each tail should contain 100 individuals (or SP < 5%) from an entire population of 3000–5000 (Sun et al., 2013a ). In terms of the optimum SP, it should consider the cost balance between genotyping and phenotyping for the selected samples (Darvasi and Soller, 2013 Gallais et al., 2011 ).


Selected samples of phenotypic extremes may come from single or bidirectional selection, which results in uni-, bi- and multibulks for the target traits, providing comparative analyses with different options (Figure 2). There are four types of BSA. For qualitative traits such as disease resistance with two distinct phenotypes (R, resistance S, susceptible), two bulked samples with qualitative difference can be generated (Figure 2a). For most quantitative traits with normal distribution, two bulked samples can be selected from two tails with extremely low and high phenotypic values, respectively (Figure 2b). To increase statistical power and reduce the false positives, multiple bulks can be selected independently from each of the two tails (Figure 2c). In many cases, where only one bulk is available for the target trait from one tail while the other tail was killed by lethal genes or due to severe stresses, BSA can be performed by comparing the bulk with a group of individuals randomly selected from a control population under no stress with normal allele frequencies for the target trait (Figure 2d).

There are two ways to bulk sampled individuals. Tissues sampled from the phenotypic extremes are pooled first, and then, a single DNA/RNA/protein isolation is performed or DNA/RNA/protein is isolated first from each extreme individual, and then, an equal amount of the extraction from each individual is bulked. As the two bulking methods for DNA analysis do not give significantly different results (Liu et al., 2014 ), bulking before extraction is more cost-effective.

When only one extreme (most resistant individuals under a severe abiotic and biotic stress condition) is available or reliable estimation of allele frequencies is not possible, BSA using a single bulk can be performed by comparing the available bulk with a phenotypic control that is randomly selected from the individuals under a normal environment (Xu and Crouch, 2008 ), or by comparing with the theoretical expectation. A similar situation is that the target trait is associated with a lethal gene so that only survivors from one tail are available for being used as single bulk (Figure 2d).

To increase the power of BSA, multiple parallel bulks have been proposed to form from the same population (Ghazvini et al., 1991 Xu et al., 2008 ). Only the positive genetic signal will show up consistently between parallel bulks (Figure 2), which provides confirmation with each other to reveal the true genetic difference because the probability for false positives showing up simultaneously in different bulks becomes much lower as the number of bulks increases. For the traits controlled by lethal genes or severely selected under stress conditions, we may only get the extreme phenotypic data for one tail. In this case, we may just do single bulk analysis to see whether observed genetic signal (e.g. allele frequencies in DNA analysis) in the bulk deviates significantly from the expected, from the individuals under normal condition or from nonlethal case (Figure 2d).


Multiplexing can be performed for samples and markers, both of which perform multiple assays in one reaction. Sample multiplexing is usually used along with individual-based selective genotyping, which makes it possible to achieve the same low cost as BSA by multiplexing many samples, while marker multiplexing is used along with BSA. Multiplexing will increase the total number of samples or markers without drastic increase in cost and time. Bulked samples can be also multiplexed as individual samples, resulting in a further cost reduction and throughput increase.

As an example for sample multiplexing, a unique sequence (barcode) can be attached to each sample so that multiple samples can be pooled in one sequencing run but can be distinguished and sorted during data analysis ( The barcode sequences can be designed follow the instructions ( The total number of available barcodes is determined by barcode length and the number of indices. Dual indexing, namely two indices used in multiplexing, further increases the total number of samples that can be pooled. With multiplex sequencing, a large number of samples can be simultaneously sequenced during a single experiment, while multisample pooling improves productivity by reducing time and reagent use. Illumina now provides a 384-sample kit which allows as many as 96 samples to be analysed in one run. For single nucleotide polymorphism (SNP) genotyping, up to tens or even hundreds of samples can be labelled individually but mixed and analysed as one sample (Livaja et al., 2005 Takagi et al., 2013b ). As RNA can be analysed by its cDNA form in BSA procedure, the protocols for DNA multiplexing can be generally used for multiplexing RNA samples.

Early marker multiplexing efforts started with mixing several pairs of primers in PCR analysis (Henegariu et al., 2014 ). Marker multiplexing has been used for assays at DNA, RNA and protein levels. At DNA or cDNA level, SNP data can be obtained using one of the numerous multiplex SNP genotyping platforms that combine a variety of chemistries, detection methods and reaction formats. Putting thousands of markers onto a single chip is one of the best ways to multiplex markers. In maize, several SNP chips have been developed (Ganal et al., 1996 Unterseer et al., 2009 Yan et al., 2013 ), which can genotype 1536–600 K SNPs per run.

Multiplex sequencing has been accomplished by random DNA shearing followed by barcode tagging with short DNA sequences (barcodes) and pooling samples into a single sequencing channel (Craig et al., 2011 ), or using an inexpensive barcoding system to sequence restriction site-associated genomic DNA (i.e. RAD tags) (Baird et al., 2008 ). The former has been used to rapidly determine the complete organellar and microbial genome sequences (Cronn et al., 1994 ) and also for discovery and mapping of genomic SNPs (Huang et al., 2010 , 2012 ). The latter has been used for high-density SNP discovery and genotyping.

To multiplex proteins or parallel protein interaction profiling, a single-molecular interaction sequencing (SMI-seq) technology has been developed. First, DNA barcodes are attached to proteins collectively via ribosome display or individually via enzymatic conjugation. To construct a random single-molecule array, the barcoded proteins are then assayed en masse in aqueous solution and subsequently immobilized in a polyacrylamide thin film for amplification and sequencing (Gu et al., 1999 ).

Protein multiplexing can be also performed by mass spectrometry. To eliminate the intrinsic bias towards detection of high-abundance proteins, significant progress has been made in a large-scale study to detect a limit of

2 μg/mL (Addona et al., 2009 ), and a biomarker validation pipelines established to detect proteins in the ng/mL range in plasma (Addona et al., 2011 Whiteaker et al., 2011 ).

Antibody colocalization microarray (ACM) as a novel concept for protein multiplexing without mixing has been used to quantify proteins in the serum of patients with breast cancer and healthy controls, with six candidate biomarkers identified (Pla-Roca et al., 2012b ). ACM involves a physical colocalization of both capture and detection antibodies, spotting of the capture antibodies, and sample incubation, followed by spotting of the detection antibodies. Up to 50 targets and their binding curves can be produced. By comparing with enzyme-linked immunosorbent assay or conventional multiplex sandwich assay, the ACM can be validated.

1 Introduction

The field of nanomaterials is growing rapidly. In medicine, tremendous efforts are being made to achieve a successful translation of nanomaterials as carrier systems for diagnostic as well as therapeutic purposes. 1 Regardless of this purpose, the nanomaterials come into contact with biological fluids and therefore with proteins. These fluids can be either extracellular fluid, such saliva, mucus, or blood, or intracellular fluid, the cytoplasm. To understand and predict the behavior of nanomaterials in biological systems, it is essential to characterize the interactions between naturally occurring proteins and nanomaterial surfaces. 2 This is equally relevant for materials outside the medical field where contact with the natural environment cannot be avoided. These are usually planar surfaces such as membranes, 3 coatings, 4 or even air–water interfaces, 5 which naturally also interact with proteins when they are present.

Independent of the material interface and application goal, controlling the interaction of materials with proteins is increasingly seen as a design tool for producing surfaces with a specific function that allows modulation of their properties to improve their efficacy in a biological context. Such control presupposes knowledge of the interaction mechanisms and principles to enable control over the relevant processes, as well as to finally engineer the impact on the desired application. Thus the combination of different techniques to characterize protein–surface interactions goes hand in hand with applying the obtained knowledge to create specific functional systems. In this context, we aim to give a detailed overview of the techniques that are currently applied to characterize and understand the underlying mechanisms of protein interactions at the nanomaterial interface, and show some current directions for engineered protein-repellent surfaces as well as contemporary strategies to obtain functional protein coatings to overcome current challenges for medicinal applications.


Compared with the majority of nanomaterials, QDs have reached an advanced stage of technological development, driven in part by successful applications as labels for biomedical imaging and molecular detection. The research literature is rich with reports of QDs with broadly tunable physicochemical and photophysical properties deriving from combinations of diverse materials and coatings. However, it is still not clear which properties are required for specific labeling applications and to what degree analytical outcomes compare with those from standardized methods based on conventional noncolloidal reagents. Uncertainties are especially prevalent for recent QDs based on multidentate polymer coatings that balance trade-offs between size and stability, which may elicit complicated interactions with endogenous biological materials. In this report, we focused on isolating single biophysical parameters by maintaining fixed dimensions of the core and the coating. The results are complex, as physicochemical characterization showed that most of these QDs were similar in size, charge, and monodispersity, but performance results in protein solutions and cells showed widely divergent behaviors in some cases (BSA solutions and permeabilized cells) but not others (live cells). For intracellular molecular labeling, differences between well-performing materials (OEG and ZW-OEG) were subtle but very important, with substantial interplay between the QD properties and the bioaffinity molecules (antibodies and protein A adaptors). These results show that there is still a major need for advanced analytical methods and sensitive characterization tools that can pinpoint underlying differences between these materials and how they interact with biological components and biospecimens. Based on the synergy observed between OEG and ZW groups within a single molecular entity, it is also clear that there is substantial room for further development of these materials, particularly if advances can be informed by insights into the fundamental nature of intermolecular forces mediating nonspecific binding in complex biological specimens.


RGMs and NET1 bind simultaneously to their receptor NEO1 at the cell surface

To dissect the molecular mechanisms that control NEO1 signaling via interactions with RGM and NET1, we identified the minimal binding regions for the NET1-NEO1 interactions using surface plasmon resonance (SPR) and cell surface binding experiments ( Figures S1 A–S1D). Our data show that the minimal complex is composed of the three membrane proximal FN domains of NEO1 (NEO1FN456) and a NET1 construct lacking the C-terminal NTR domain (NET1ΔNTR) ( Figureਁ A), in line with previous biochemical studies (Geisbrecht etਊl., 2003 Kruger etਊl., 2004 Mille etਊl., 2009 Xu etਊl., 2014). We previously mapped the RGM-binding region on NEO1 to the two membrane-proximal FN domains (NEO1FN56) (Bell etਊl., 2013).

Identification of the minimal NEO1-NET1 interaction region, related to Figureਁ

(A, B) SPR equilibrium binding experiments of different NET1 and NEO1 constructs. Graphs show a plot of the equilibrium binding response against used NEO1 construct concentrations (left panels: full-length NEO1 ectodomain (eNEO1), right panels: NEO1 FN type III domains 4 to 6 (NEO1FN456). Ligands immobilized on SPR sensor chip: A, full-length NET1 B, NET1ΔNTR. (C) Immunofluorescence staining of FLAG-tagged full-length human DCC (DCCFL) and mouse NEO1 (NEO1FL) overexpressed in COS-7 cells (green). Left panel: bound NET1ΔNTR is stained via a Rho ID4 tag (red) right panel: transfected cells were incubated with buffer only as a negative control and stained as in the left panel. (D) Western blot of COS-7 cells transfected with the indicated plasmids used in C. α-tubulin serves as a loading control. (E, F) Proximity ligation assay (PLA) to test for simultaneous binding of NET1 and RGMB to NEO1. (E) COS-7 cells were transfected with a NEO1-mVenus fusion protein or the corresponding empty vector, and with full-length RGMB (wild type or RGMB-A186R). Transfected cells were incubated with NET1ΔNTR before performing the PLA assay. PLA signals are shown in red and NEO1-mVenus transfected cells in green with nuclei in blue. (F) PLA signals were quantified and values from 3 individual experiments were plotted. A two-tailed, unpaired t test showed the statistical significance as p = 0.0107.

NET1 and RGMB can simultaneously bind NEO1 and form a ternary complex

(A) Schematics of NEO1, NET1, and RGMB. SP, signal peptide TM, transmembrane helix IG, immunoglobulin-like domain FN, fibronectin type III domain CD, cytoplasmic domain LN, laminin domain LE, laminin epidermal growth factor-like repeats LC, netrin (NTR) domain N-RGM, RGM N-terminal domain identified to bind to BMP ligands (Healey etਊl., 2015) vWfD, von Willebrand factor D-like domain GPI, glycosylphosphatidylinositol anchor.

(B and C) Proximity ligation assays (PLA) were performed to test for simultaneous binding of NET1 and RGMB to NEO1.

(B) Cos-7 cells were either transfected with a NEO1-mVenus fusion protein or empty vector. Cells were incubated with NET1ΔNTR and RGMBECD (wild type or RGMBECD-A186R). NEO1-mVenus positive cells are shown in green, nuclei are stained with DAPI and PLA signals in red.

(C) Number of PLA signals per NEO1-mVenus positive cells. Individual values are plotted from 4 independent experiments. Statistical significance was determined using a two-tailed, unpaired t test with p < 0.0001.

(D) Ribbon representation of the NEO1-NET1-RGMB protomer observed in the 3.25 Å resolution crystal structure, with NEO1FN456 in red, NET1ΔNTR in਋lue and RGMBCORE in yellow. A schematic is shown. See also Figure S1 .

Since cells can simultaneously encounter RGMs and NET1 in a NEO1-dependent manner in vivo (Kee etਊl., 2008 Moon etਊl., 2011 Muramatsu etਊl., 2011 O’Leary etਊl., 2013 Wilson and Key, 2006), we asked whether a stable ternary NEO1-NET1-RGM complex can exist at the cell surface, or whether NET1 and RGM would compete for NEO1 binding. To distinguish between these two options, we carried out in situ proximity ligation assays (PLAs) (Srberg etਊl., 2006). We mutually incubated full-length NEO1 expressing cells with purified NET1ΔNTR and the full-length ectodomain of RGMB (RGMBECD) and observed numerous PLA foci, which can only be generated when NET1 and RGMB come into close proximity (㱀nm) ( Figures 1 B and 1C). The number of foci was diminished when wild type RGMB was replaced with a RGMB mutant (RGMB-A186R [Healey etਊl., 2015]) that cannot efficiently bind to NEO1 ( Figures 1 B and 1C). This result suggests that NEO1 is specifically required for bridging NET1 and RGMB and that a complex of RGMB, NET1, and NEO1 forms on the cell surface. This scenario describes signaling in “trans,” whereby RGM and NEO1 are expressed in different cells, for example in RGM-mediated axon guidance. We also observed a high number of foci from NET1-RGMB interactions when full-length RGMB (RGMBFL) and NEO1 are co-transfected in the presence of soluble NET1ΔNTR ( Figures S1 E and S1F). This shows that the ternary complex can also form in “cis,” a situation occurring in hepatocytes or chondrocytes (Zhang etਊl., 2009 Zhou etਊl., 2010) where the RGM-NEO1 complex regulates BMP signaling. It remains to be seen what the effect of NET1 in these signaling scenarios is.

The structure of the ternary NEO1-NET1-RGM complex reveals a “trimer-of-trimers”

To understand how NEO1, NET1, and RGMB assemble into a ternary complex, we determined the crystal structure of the minimal ternary NEO1-NET1-RGMB complex, composed of NEO1FN456, NET1ΔNTR and the RGMB NEO1-binding region (RGMBCORE) ( Figureਁ D Methods S1 Table S1) to 3.25 Å resolution. Strikingly, this ternary complex is assembled into a 3:3:3 stoichiometry, composed of three 1:1:1 “protomer” complexes arranged around a 3-fold symmetry axis ( Figureਁ D and Figureਂ A). To test whether such a “trimer-of-trimers” arrangement can exist in solution, we carried out analytical ultracentrifugation (AUC) ( Figureਂ B) and small angle X-ray scattering (SAXS) ( Figureਂ C) analyses. Both methods suggest the assembly of a 3:3:3 NEO1-NET1-RGMB complex as the major species.

Structure of the NEO1-NET1-RGMB ternary complex

(A) Two 90°-rotated ribbon representations of the NEO1-NET1-RGMB trimer-of-trimers complex. The relative location of the plasma membrane is depicted. The solvent accessible surface is shown in the right panel in addition to the ribbons. The inset shows an outline complex architecture and symmetry. Disordered regions are depicted as dotted lines. Color-coding is as in Figureਁ D.

(B) Sedimentation velocity AUC experiment of the NEO1FN456-NET1ΔNTR-RGMBECD complex at different concentrations. The major species corresponds to a 3:3:3 complex.

(C) Guinier region analysis of the NEO1-NET1-RGMB complex from SEC-SAXS experiment suggests a molecular weight of 410 kDa, corresponding to the 3:3:3 NEO1FN456:NET1ΔNTR:RGMBECD stoichiometry. The inset shows the SAXS intensity plot for the final merged data.

(D) Selected 2D class averages used for cryo-EM map reconstruction of the NEO1-NET1-RGMB ternary complex.

(E) Ribbon representation of the 3:3:3 NEO1-NET1-RGMB cryo-EM complex. View and coloring as in (A). The crystallographic 3:3:3 NEO1-NET1-RGMB complex fitted into the cryo-EM map as a single rigid body (depicted in light cyan) is shown for comparison. The cryo-EM electron potential (grey mesh) is calculated to 6.0 Å resolution.

(F) Close-up view of the NEO1-RGMB interface highlighted in (E). The model of the ternary 3:3:3 complex fits the cryo-EM map better when refined as six rigid bodies (see also STAR Methods). Distances (Å) between selected Cα atoms are indicated. The curved arrow highlights the movement of the NEO1FN56-RGMB segment by up to 10 Å relative to the NEO1FN4-NET1 segment. See also Figures S2 and ​ andS3 S3 and Methods S1 and S2.

We further interrogated stoichiometry and architecture of the ternary complex using size-exclusion chromatography coupled to multiangle light scattering (SEC-MALS), and cryo-electron microscopy (cryo-EM). The ternary complex composed of NEO1FN456, NET1ΔNTR and RGMBECD could be readily formed on SEC ( Figures S2 A–S2D). The peak fraction contained all three proteins and was further analyzed using SEC-MALS. The experimentally determined molecular weight (MW) (422.7 ± 2.8 kDa) matched the calculated MW of the glycosylated complex (433 kDa) confirming the presence of the 3:3:3 assembly in solution ( Figure S2 B). We further analyzed the same sample with cryo-EM (Methods S2). Single particle analysis of the NET1-NEO1-RGMB complex revealed their trimeric symmetry ( Figureਂ D), consistent with the complex determined by crystallography ( Figureਂ A). We reconstructed a cryo-EM map to 6.0 Å resolution in which the crystallographic 3:3:3 complex could be readily fitted ( Figures 2 E and 2F). Using a similar SEC-MALS experiment as for RGMB, we could show that the full-length extracellular domains of the other two human RGM family members RGMA and RGMC form ternary complexes with NEO1FN456 and NET1 ( Figures S2 E–S2H) suggestive of ternary 3:3:3 RGM-NEO1-NET1 architecture.

SEC, MALS and SDS-PAGE analysis of the ternary NEO1-NET1-RGM complexes, related to Figureਂ

(A) SEC of the ternary NEO1FN456-NET1ΔNTR-RGMBECD complex. The SEC fraction (elution volume

9.8-10.1 ml) indicated with a red line was analyzed using MALS (panel B) and cryo-EM. SEC fractions indicated with a blue line (elution volume

8-12 ml) were analyzed on SDS PAGE (panels C and D). (B) SEC-MALS analysis of the NEO1FN456-NET1ΔNTR-RGMBECD complex. Calculated MW of 1:1:1 mol:mol:mol complex is 144.4 kDa (129.35 kDa of protein plus 15.06 kDa of seven Asn-linked Man9GIcNAc2 glycans). Calculated MW of 3:3:3 complex is 433.24 kDa. The NEO1-NET1-RGMB complex eluted as two peaks with corresponding MW of 422.7 kDa and 117.9 kDa (indicated with red lines). (C, D) SDS PAGE analysis of SEC fractions. Fractions were heated (100 ଌ, 10 minutes) in the presence or absence of 2-mercaptoethanol (panels C and D, respectively). (E) NEO1FN456 co-elutes with extracellular domain of RGMA (RGMAECD) on SEC, suggesting that NEO1 and RGMA form a binary complex. SEC fractions were analyzed using SDS-PAGE under non-reducing and reducing conditions. Under reducing conditions, the RGMAECD dissociates into two fragments (labelled N-term. and C-term.) due to an autocatalytic cleavage mechanism. SEC fractions containing the binary NEO1-RGMA complex used to form the ternary NEO1-NET1-RGMA complex are indicated. SEC running buffer: 150 mM NaCl, 10 mM HEPES pH 7.5, 2 mM CaCl2, 0.02% NaN3 (flow rate 0.3 ml/min Superose 6 Increase 10/300 GL column 21 ଌ). (F) SDS-PAGE analysis (non-reducing and reducing conditions) of NET1 and NEO1-RGMA used to assemble the ternary NEO1-NET1-RGMA complex for SEC-MALS analysis. Traces corresponding to absorbance at 280 nm, light scattering and molecular masses derived from SEC-MALS are shown in black, blue and red, respectively. Calculated molecular masses based on protein amino acid sequences: NET1ΔNTR, 49.2 kDa plus 3 Asn-linked glycans, 5.6 kDa FN domains 4𠄶 of NEO1, 39.2 kDa plus 2 Asn-linked glycans, 3.8 kDa RGMA, 42.2 kDa plus 3 Asn-linked glycans, 5.6 kDa. Thus, calculated mass of the glycosylated NEO1-NET1-RGMA ternary 3:3:3 complex is 437.0 kDa. The ternary complex dissociated on SEC-MALS as suggested by a major peak corresponding to 79.97 kDa. However, an additional peak corresponding to 444.4 kDa, which is consistent with the NET1:NEO1:RGMA 3:3:3 mol:mol:mol complex, was also observed. (G) FN domains 4𠄶 of NEO1 co-elute with the full-length extracellular domain of RGMC (RGMCECD) on SEC, suggesting that NEO1 and RGMC form a binary complex. SEC fractions were analyzed using SDS-PAGE under non-reducing and reducing conditions. Under reducing conditions, a fraction of RGMCECD dissociates into two fragments (labelled N-term. and C-term.) as observed for RGMAECD (E). (H) SEC and SDS-PAGE analysis of the ternary NET1–NEO1–RGMC complex. The ternary NEO1-NET1-RGMC complex elutes as two major peaks (12.5 and 13.9 ml peaks) at lower elution volume compared to the binary NEO1-RGMC complex (16.3 ml, G) or NET1 in isolation, suggesting that the NEO1-NET1-RGMC ternary complex forms in solution. SEC running buffer: 150 mM NaCl, 10 mM HEPES pH 7.5, 2 mM CaCl2, 1 mM sucrose octasulfate, 0.02% NaN3 (flow rate 0.3 ml/min Superose 6 Increase 10/300 GL column 21 ଌ). SEC input was 0.6 ml of the ternary complex at 2.6 mg/ml.

The backbone of the “trimer-of-trimers” super-complex is essentially formed by interactions between NEO1 and NET1. NET1 is an elongated and rigid molecule, making contacts with two neighboring NEO1 molecules in order to bridge individual NEO1-NET1-RGM protomers within the super-complex ( Figures 2 A and ​ and3A 3 A and Figures S3 A and S3B). The NET1 LN domain (NET1LN) forms the major interaction site with NEO1, binding to NEO1FN4 (“Interface-1,” Figureਃ A, right panel). The combined buried surface area between NET1LN and NEO1FN4 is 686 Å 2 and involves predominantly hydrophobic interactions, with NET1 residue F55 in the center. The second interface is formed by the NET1LE3 of the same NET1 molecule and a neighboring NEO1FN5 domain. In contrast to “Interface-1,” this interaction is of a predominantly hydrophilic nature with a buried surface area of 572 Å 2 (“Interface-2,” Figureਃ A, left panel). Both interfaces are highly conserved amongst NEO1 orthologues ( Figures S3 A and S3B). The NEO1 FN5 and FN6 domains interact with RGMB, as observed in our previous structural analysis of binary NEO1-RGM complexes, via the high-affinity “site 1” interface (Bell etਊl., 2013). A total of 4 N-linked sugars and 4 non-covalently linked sucrose-octasulfate (SOS) molecules are bound at the complex surface ( Figures S3 C–S3G). The NET1ΔNTR and RGMBCORE molecules form a very minor interaction ( Figure S3 H), but no binding was observed between NET1ΔNTR and the ectodomains of RGMA and RGMB in solution ( Figures S3 I–S3L). This suggests that NET1-NEO1 interactions are the driving force for the formation of the “trimer-of-trimers” super-complex.

Interface analysis of the ternary NEO1-NET1-RGMB super-complex

(A) Close-up views of the observed NET1-NEO1 interfaces (right: interface 1, left: interface 2). Residues are displayed in stick representation and labelled according to domain color-coding. A Ca 2+ ion bound to NET1 LN (grey sphere) and hydrogen bonds (dashed black lines) are displayed. Mutated residues are in bold and underlined.

(B) SPR equilibrium binding curves for the NET1-NEO1 interaction. A schematic of the experiment and the calculated Kd values are shown.

(C) AUC analysis of the NEO1FN456-NET1ΔNTR-RGMBECD complex, using NET1ΔNTR WT and mutants. Both NET1 interface-1 and -2 mutants abolish the 3:3:3 stoichiometry of the NEO1-NET1-RGMB super-complex.

(D) Overlapping expression of NET1 RNA (in situ hybridization), and NEO1 and RGMB protein (immunohistochemistry) in consecutive coronal sections of E16 mouse striatum. Boxed area is shown at higher magnification for NEO1 and RGMB. Scale bar, 100 μm.

(E) RGMB immunoprecipitation (IP) from adult mouse cortex was followed by immunoblotting. Input samples (lane 1), IP using control non-specific IgGs (cntrl) (lane 2), and anti-RGMB IP (lane 3). NEO1 and NET1 co-IP with RGMB from adult mouse brain lysates.

(F and G) Functional analysis of the effect of NET1 on RGMA-mediated growth cone collapse.

(F) Representative examples of growth cones from mouse P0 cortical neurons. Neurons were stained with the microtubule marker Tuj1 (green) and F-actin marker phalloidin (red). Scale bar, 10 μm.

(G) Quantification of growth cone collapse. Growth cones were treated with control or RGMA alone and in combination with different NET1 variants. Proportions of collapsed growth cones relative to control are displayed. n = 3 experiments, one-way ANOVA followed by Tukey’s multiple comparison test. ∗ p < 0.05. Data are shown as means ± SEM.

(H–J) Comparison of binary NEO1-RGM (PDB ID 4BQ6 [Bell etਊl., 2013]) and the ternary NEO1-NET1-RGMB complexes shown as ribbons. The ternary NEO1-NET1-RGMB protomer complex architecture (I) clashes with the NEO1-RGM dimer-of-dimers signaling conformation (H) when superimposed on NEO1 (marked with an asterisk) (J). See also Figure S3 , Figure S4 , Figure S5 .

Structural and functional analysis of the ternary NEO1-NET1-RGM complex, related to Figures 2 , ​ ,3, 3 , and ​ and4 4

(A, B) Surface representations of NET1-NEO1 interactions The NEO1-NET1 Interface-1, formed by the NEO1 FN4-NET1 LN interaction is shown in A. Interface residues are mapped onto solvent accessible surfaces displayed in open-book view (blue, left panel in A). Residue conservation calculated with ConSurf server ( is mapped onto the protein surfaces according to a white-to-black gradient (right panel in A). Surfaces are highlighted with a line. The NEO1-NET1 Interface-2, formed by the NEO1 FN5-NET1 LE3 interaction is shown in B. Presentation is as in A. (C-G) Sugar sites identified on the ternary NEO1-NET1-RGMB crystal structure. (C) Ribbon presentation of the NEO1-NET1-RGMB protomer with the 4 N-linked N-acetylglucosamine (NAG yellow) and 4 sucrose-octasulfate (SOS light blue) molecules depicted as sticks. (D-G), Close-up views of the 4 SOS-binding sites with residue side chains within hydrogen-bonding distance shown in stick representation and labelled. Potential hydrogen bonds are displayed as dashed black lines. (H) NET1-RGM interaction analysis in the ternary trimer-of-trimers complex determined by X-ray crystallography. Overall 1:1:1 trimer architecture is displayed on the left. The close-up shows the interface between NET1 and RGMB. The sigmaA-weighted 2Fo-Fc map of the final refinement in AUTOBUSTER is displayed and contoured at 1σ. RGMB is ordered to residue D323 and a dashed line denotes disordered residues linking to a putative helical stretch of Ala residues, which were built into this density as the sequence could not be unambiguously assigned. (I) Non-reducing SDS-PAGE of purified RGMAECD and RGMBECD used as analytes for SPR injections. (J) Schematics of the experimental SPR set up. NET1ΔNTR (ligand) was attached to a streptavidin-coupled sensor chip via a biotinylated C-terminal Avi-tag. RGMECD and NEO1FN456 (analytes) were injected to probe interactions. (K, L), SPR equilibrium binding curves for NET1ΔNTR binding experiments with NEO1FN456 (K and L same measurement for comparison), RGMBECD (K) and RGMAECD (L). (M, N) SPR equilibrium binding curves for the NEO1-NET1 interaction. A schematic of the experiment (NEO1: red, NET1: blue) and the calculated Kd values are shown. The maximal response for the wild type NEO1FN456:NET1ΔNTR interaction represents 100% binding. Sensorgrams for NEO1:NET1ΔNTR interactions, corresponding to Figureਃ B and Figure S6 J are shown in (B).

The ternary NEO1-NET1-RGMA complex is essential for NET1 inhibition of RGMA-mediated growth cone collapse

In order to verify our observed NET1-NEO1 interfaces, we designed NET1 mutants to ablate specific interaction sites and tested these for NEO1 binding in SPR ( Figure 3 B, Figures S3 M, and ​ and3N). 3 N). When compared to wild-type NET1ΔNTR, NET1ΔNTR F55R (“NET1-Interface-1” mutant) designed to disrupt NET1-NEO1 “Interface-1” ( Figureਃ A, right panel), bound to NEO1FN456 with an approximately ten-fold lower affinity but did not change NET1 binding to NEO1FN56. On the other hand, NET1ΔNTR Q443N/R445T (“NET1-Interface-2”), designed to introduce an Asn443-linked glycan to disrupt the NET1LE3-NEO1FN5 interface, abolished binding to NEO1FN56 ( Figureਃ B) and reduced NEO1FN456 binding approximately two-fold ( Figures S3 M and S3N). This supports the model of two independent NET1-NEO1 binding interfaces, as observed in our complex structure. We also tested the effects of the NET1ΔNTR Interface-1 and -2 mutants using AUC ( Figureਃ C). Both NET1 mutant complexes with NEO1FN456 and RGMBECD abolished the “trimer-of-trimers” super-complex, in agreement with our observation that both NET1-NEO1 interface 1 and 2 are necessary to form the super-complex.

To assess the functional impact of NET1 on the NEO1-RGM interaction, we confirmed that RGMA, RGMB, NET1, and NEO1 expression patterns in the nervous system allow interactions between these molecules in a “trimer-of-trimers” super-complex. We found widespread overlap in the distribution and expression of RGMs, NET1, and NEO1 in various brain regions of the adult mouse brain, including in the thalamus, cortex, and cerebellum ( Figures S4 A–S4C). For example, in line with previous data, these molecules were localized to cell types in the ventricular-subventricular zone, such as astrocytes and neurons, that reside in close proximity and that functionally interact ( Figure S4 D). Furthermore, RGMB, NET1, and NEO1 also displayed overlapping expression in different regions of the embryonic brain, including the striatum ( Figureਃ D). Next, we assessed whether RGMs, NET1, and NEO1 are present in the same protein complex in brain tissue. We performed immunoprecipitation (IP) of RGMB from adult mouse cortex followed by detection of RGMB, NET1, and NEO1. As predicted by the PLA experiments ( Figures 1 B and 1C) and expression data ( Figures 3 D and ​ andS4A), S4 A), NEO1 and NET1 were co-immunoprecipitated with RGMB ( Figureਃ E). These observations, together with our data showing lack of RGMB-NET1 and RGMB-DCC binding ( Figures S3 K and S3L) (Bell etਊl., 2013), suggest that RGMs, NET1, and NEO1 co-exist in multimeric protein complexes on neural cells.

Expression of NEO1, NET1, RGMA and RGMB, related to Figureਃ

(A) Protein expression of NET1, NEO1 and RGMB in the adult mouse brain detected by western blot analysis. (B) Sagittal overview of the adult mouse brain. (C)In situ hybridization for NEO1, NET1, RGMA and RGMB in sagittal sections from the adult mouse brain (obtained from the Allen Brain Atlas ( Regions of interest are indicated in boxed regions in B: (i) anterodorsal nucleus of the thalamus, (ii) cerebellum and (iii) olfactory nucleus. Images are obtained from the Allen Brain Atlas. Olf bulb, olfactory bulb ctx ant, anterior half of the cortex ctx post, posterior half of the cortex hip, hippocampus Th, thalamus AON, anterior olfactory nucleus ACB, nucleus accumbens. Scale bar = 500 μm. (D) scRNAseq dataset analysis (Mizrak etਊl 2019) for co-expression of RGMA/B, NEO1 and NET1 in adult V-SVZ. Single-cell expression levels of cluster-specific marker genes in adult ventricular-subventricular zone (V-SVZ) cells plotted on UMAP embedding (Cldn10, Mog, Ccnd2, Tmem119, Meg3, Egfl7, Ccdc153, Vtn, Pdgfra, Fyn). In addition, expression levels of Neogenin (NEO1), Netrin-1 (NET1), RGMA and RGMB are shown. Clusters marked as [clusterID]. OPC, oligodendrocyte precursor COP, committed oligodendrocyte precursors.

To investigate the functional role of these protein complexes, we first used an NEO1-dependent growth collapse assay in which dissociated cortical neurons (CNs) were exposed to RGMA in the absence or presence of NET1 proteins (van Erp etਊl., 2015) ( Figures 3 F and 3G). CNs express NEO1 ( Figure S5 A), and addition of RGMA induced CN growth cone collapse (van Erp etਊl., 2015). This inhibitory effect of RGMA was diminished when CNs were simultaneously exposed to RGMA and wild type NET1ΔNTR, suggesting that NET1 can inhibit RGMA-mediated growth cone collapse. In contrast, when CNs were incubated with RGMA and either NET1 interface-1 or interface-2 mutant, RGMA-mediated growth cone collapse was unaffected ( Figures 3 F and 3G). Since embryonic CNs express NEO1 and DCC ( Figure S5 A), we confirmed that the observed effect of NET1 was independent of the NET1 receptor DCC (that shares some 80% sequence similarity to NEO1). We crossed Emx1-IRES-cre mice with Dcc fl/fl mice (Krimpenfort etਊl., 2012), to delete Dcc from CNs (Gorski etਊl., 2002 Liang etਊl., 2012). We observed no difference in RGMA-induced growth cone collapse and NET1 rescue in Dcc lox/lox Emx1 cre/wt compared to Dcc lox/wt Emx1 wt/wt CNs ( Figure S5 B). Taken together, our functional and structural analyses suggest a mechanism for NET1-mediated inhibition of NEO1-RGM signaling, whereby the formation of the NEO1-NET1-RGM “trimer-of-trimers” super-complex is the driving force as it is incompatible with NEO1 dimerization and subsequent downstream signal activation ( Figureਃ H𠄳J).

Silencing of RGMA-mediated growth cone collapse by NET1 is DCC-independent, related to Figures 3 , ​ ,5, 5 , ​ ,6, 6 , and ​ and7 7

(A) Immunocytochemistry of NEO1, deleted in colorectal cancer (DCC) and TUJ1 in P0 mouse cortical neurons at DIV3. Scale bar = 50 μm. (B) Mean ± S.E.M. of the percentage of collapsed growth cones following exposure to RGMA or RGMA + NET1FL in cortical neurons from Emx1-Cre -/- Dcc fl/+ (control) or Emx1-Cre +/- Dcc fl/fl (knockout) mice. Emx1-Cre -/- Dcc fl/+ (mean ± S.E.M.): vehicle = 18.83 ± 2.17, RGMA = 49.40 ± 3.61, RGMA + NET1FL = 19.47 ± 1.47. Emx1-Cre +/- Dcc fl/fl (mean ± S.E.M.): vehicle = 22.35 ± 1.48, RGMA = 45.28 ± 3.15, RGMA + NET1FL = 27.88 ± 2.53. n = 6 experiments, two-way ANOVA with Tukey’s multiple comparisons test, ∗∗∗ p < 0.0001. (C-G) Quantification of migration distance in SVZ-NSC assays and analysis of GFP + neurons following IUE. Migration distance (per 50 μm bin) of SVZ-neuroblasts related to (C) Figures 5 C and 5D, (D) Figures 5 E and 5F, (E) Figures 6 A and 6B, (F) Figures 6 I and 6J, and (G) Figures 6 G and 6H. (H-J) Cortical migration of GFP + electroporated neurons. (H) At E14 embryos were in utero electroporated (IUE) with expression vectors for GFP, RGMA, and/or NET1 (each condition has GFP). Embryos were harvested two days later at E16. Migration distance from the VZ to the MZ was measured per bin (1-8) (i.e. the number of GFP + cells per bin/total GFP + cells). (I-J) Electroporation of RGMA or NET1 caused an increase in the number of GFP + in bins near the VZ, indicating reduced migration towards the MZ. Simultaneous overexpression of RGMA and NET1 in part rescued this inhibitory effect. The reduction in the number of GFP + cells in more upper layers was visible in the images but did not reach statistical significance due to the low numbers of these more superficially located neurons. One-way ANOVA followed by Sidaks multiple comparisons test: RGMA vs. GFP bin 1 p < 0.0001, RGMA vs. GFP bin 2 p = 0.0305, NET1 vs. GFP bin 1 p = 0.379, NET1 vs. GFP bin 2 p = 0.362. GFP: n = 6 animals, RGMA: n = 6 animals, NET1: n = 4 animals, NET1+RGMA: n = 5 animals. Marker: 100 μm. VZ, ventricular zone SVZ, subventricular zone IZ, intermediate zone CP, cortical plate MZ, marginal zone.

The structure of the binary NET1-NEO1 complex suggests NET1-mediated NEO1 clustering

We next determined the crystal structure of the binary complex between NEO1FN456 and NET1ΔNTR ( Figure਄ A Table S1). The observed NEO1FN456:NET1ΔNTR interfaces are equivalent to the ternary NEO1-NET1-RGMB complex, with NET1 linking two separate NEO1 molecules via their FN4 and FN5 domains respectively ( Figure਄ A). The NEO1FN56 structural unit is positioned differently when compared to the ternary NEO1-NET1-RGM structure, undergoing a rotation of some 130 o around the FN4-5 interdomain linker ( Figure S6 A). Interface-1 and -2 determinants resemble that of our ternary complex and are also observed in a previously reported crystal structure of the chick NET1ΔNTR with NEO1FN45 (Xu etਊl., 2014) ( Figure਄ B). However, we did not observe a 2:2 hetero-tetrameric arrangement as in the chick NET1-NEO1 structure, mediated by an antiparallel NET1 dimer. NET1ΔNTR exists as a monomer at concentrations up to 81 μM when analyzed by SAXS ( Figures S6 B–S6G), in agreement with previous NET analyses (Finci etਊl., 2014 Grandin etਊl., 2016 Reuten etਊl., 2016). Crystal packing analysis of our binary NET1-NEO1 complex suggests a continuous NET1-NEO1 arrangement generated by a crystallographic two-fold axis, mediated by NET1-NEO1 interface 1 and 2 ( Fig.਄ C). This suggests a ligand-induced receptor clustering mechanism, resulting in a similar arrangement to that proposed for the NET1-DCC complex (Xu etਊl., 2014) ( Figure਄ D). In both complexes, interfaces 1 and 2 are highly conserved ( Figure S6 H). Our SPR analysis using NET1 interface-1 and-2 mutants is in agreement with this model, with equivalent NET1-binding properties observed for DCC ( Figures S6 I and S6J) as for NEO1 ( Figureਃ B).

Structure and functional characterization of the binary NET1-NEO1 complex

(A) Cartoon representation of the binary NET1ΔNTR-NEO1FN456 complex. NET1ΔNTR contacts two NEO1FN456 chains, using the same interfaces observed in the “trimer-of-trimers” NEO1-NET1-RGMB super-complex structure. Interfaces 1 and 2 are labelled.

(B) Comparison of NET1-NEO1 interfaces (interface 1: top panel, interface 2: lower panel). Superpositions were calculated using NEO1 FN4 (top panel, “interface-1”) and FN5 (lower panel, “interface-2”) as template. The binary (light blue/blue) and ternary (light red/red) NET1ΔNTR-NEO1FN456 complexes from this study and the previously determined NET1ΔNTR-NEO1FN45 complex (orange/light orange, PDB Id. 4PLN [Xu etਊl., 2014]) are shown as ribbons.

(C) Overall arrangement of the NET1-NEO1 complex, which forms a continuous array in the crystal. The relative orientation of the plasma membrane is depicted. The region marked corresponds to the protomer in (A).

(D) Cartoon representation of the previously published DCCFN45-NET1ΔNTR complex (PDB Id 4PLO [Xu etਊl., 2014]) shown in the same orientation as the NET1ΔNTR-NEO1FN456 complex from C. Both complexes form a similar, continuous array in the crystal. The DCC FN6 domain missing in the DCC-NET1 complex is depicted schematically.

(E and F) Sedimentation velocity AUC experiments of the binary NET1ΔNTR-NEO1FN456 complex. The complex reveals concentration-dependent increase of oligomerization, characterized by a shift to higher s(S) values (E). This can be inhibited by NET1 interface-1 and -2 mutants that both result in a reduction of the apparent molecular weight (F), corresponding to a 1:1 NET1-NEO1 complex stoichiometry.

(G) Guinier region analysis of SEC-SAXS data collected for the NEO1FN456:NET1ΔNTR complex at 2.4 mg mL 𢄡 (blue) and 5.9 mg mL 𢄡 (yellow) gives larger Rg and MWVC values at higher concentrations. See also Figure S6 .

Structural and functional analysis of binary NET1-NEO1 and NET1-DCC complexes, related to Figure਄

(A) Flexibility between NEO1 FN4 and FN5-6 domains. Superposition of the binary NEO1-NET1 (gold) and NEO1-NET1-RGMB (red) complex structures. Superimpositions were calculated using NET1 as template. NET1 and RGMB are colored as in Figureਁ A. Due to flexibility in the interdomain linker region between FN domains 4 and 5, the position of the NEO1 FN5-6 region varies greatly in relation to the FN4 domain. NEO1FN56 forms a structural unit. (B, C) Fit of an ensemble of NET1ΔNTR models to experimental scattering data. Experimental (black) and calculated (red) scattering curves are displayed to a maximal momentum transfer of q = 0.37 Å -1 , with fit value (χ 2 ) displayed (B). A distribution of NET1ΔNTR models as calculated by MultiFOXS and MES is displayed, color-coded as per model (C) (D) Guinier region for experimental and calculated scattering, with radius of gyration (Rg) calculated from experimental data annotated. (E) Normalized pair-distance distribution function, with the derived maximum intra-particle diameter (Dmax). This suggests that NET1ΔNTR behaves as a monomer in solution. (F-G) Fitting between experimental (black) and calculated (green) scattering data (F) from a proposed X-linked NET1ΔNTR dimer (G) (PDB ID. 4PLN). (H) Comparison of the binary NEO1-NET1 and DCC-NET1 interfaces (‘Interface-1’: left panel, ‘Interface-2’: right panel). Superpositions were calculated using NEO1 FN4 (for ‘Interface-1’) and FN5 (for ‘Interface-2’) as template, respectively. The binary NEO1FN456-NET1ΔNTR complex from this study and the previously determined DCCFN45-NET1ΔNTR (PDB ID 4PLO) and DCCFN56-NET1ΔNTR (PDB ID 4URT) complexes are shown as ribbons. (I, J) SPR binding analysis to characterize the NET1 interaction with the NEO1 paralogue DCC. SPR equilibrium binding curves for the DCC-NET1 interaction (B) and corresponding sensorgrams (C) are presented. A schematic of the experiment (DCC: grey, NET1: blue) and the calculated Kd values are shown. The maximal response for the wild type DCCFN456:NET1ΔNTR interaction represents 100% binding.

To validate the NET1-NEO1 clustering model, we performed AUC analysis of the NEO1FN456-NET1ΔNTR complex. We observed a concentration-dependent shift of the complex towards higher molecular weights (10� S), indicative of NET1ΔNTR-induced NEO1 clustering ( Figure਄ E). This was inhibited by mutations in both NET1-NEO1 interface 1 and 2 ( Figure਄ F). In support of this finding, analysis of the NEO1FN456-NET1ΔNTR complex using SAXS gave a notable difference in scattering profile and calculated molecular weight at different concentrations ( Figure਄ G). Our data support a cell surface receptor oligomerization model in which a single NET1 molecule links two NEO1/DCC receptors together, leading to NEO1/DCC clustering in a concentration-dependent manner.

RGM inhibition of NET1-mediated neuronal migration

RGMA is a repulsive guidance cue for cortical interneurons migrating out of the MGE and it induces growth cone collapse in CNs. In line with our discovery of the “trimer-of-trimers” silencing complex, these effects can be silenced by NET1 (O’Leary etਊl., 2013) ( Figures 3 F and 3G). To functionally examine the NET1-NEO1 interaction and to test whether RGMs can inhibit NET1-mediated NEO1 signaling, we cultured mouse subventricular zone neurospheres (SVZ-NSCs) on different NET1 variants. NET1 and RGMA are expressed along the migratory route of mouse SVZ-neuroblasts en route to the olfactory bulb in the rostral migratory stream (RMS) and NET1 promotes SVZ-neuroblast migration in vitro (Bradford etਊl., 2010 O’Leary etਊl., 2015). This migration assay allows testing of NET1-NEO1 signaling output, because SVZ-neuroblasts express NEO1 and rely on this receptor for NET1-mediated migration ( Figures 5 A and 5B) (O’Leary etਊl., 2015).

NET1 mediates SVZ-neuroblast migration via NEO1

(A) Schematic of the neurosphere migration assay. Neurospheres were generated from the adult mouse subventricular zone (SVZ) subsequently plated on control or NET1 proteins.

(B) Immunocytochemisty for NEO1, DCC, and TUJ1 (to label SVZ-neuroblasts) in DIV5 SVZ-NSC cultures. SVZ-neurospheres (NSCs) and neuroblasts (arrowheads) express NEO1 and DCC. Boxed areas are shown at higher magnification on the right. Scale bar, 50 μm.

(C and D) Analysis of migrating neurons from SVZ-NSCs grown on full-length NET1 constructs. Ablation of either NET1-NEO1 interface-1 or -2 interactions causes loss of NET1-mediated neuron migration. Mean ± SEM of the relative number of Tuj1/DCX positive migrating neurons per neurosphere: vehicle = 100.00, NET1FL-WT = 223.25 ± 27.51, NET1FL-Interface-1 = 114.20 ± 6.07, NET1FL-Interface-2 = 85.06 ± 13.02, n = 3-4 experiments. Brown-Forsythe to test significant difference between SDs (p < 0.05): ns. One-way ANOVA followed by Tukey’s multiple comparisons test: vehicle vs. NET1FL-WT P = 0.0003, NET1FL-WT vs. NET1FL-Interface-1 P = 0.0007, NET1FL vs. NET1FL-Interface-2 p < 0.0001. Representative samples of mouse SVZ-NSCs grown on coverslips coated with indicated proteins are shown in (D). Boxed areas are shown at higher magnification on the right of each panel. Migrating neurons (white arrowheads) were identified via labelling with the microtubule markers TUJ1 (green) and DCX (red) as well as the nuclear marker DAPI (blue).

(E and F) Analysis and representative samples of migrating SVZ neuroblasts (white arrowheads) grown on NET1ΔNTR. NET1 lacking the C-terminal NTR domain fails to increase neuron migration. Mean ± S.E.M of the relative number of TUJ1/DCX-positive migrating neurons per neurosphere: control (vehicle) = 100.00, NET1ΔNTR = 86.18 ± 2.215, n = 2 individual experiments. Unpaired t test: p = 0.0247. See also Figure S5 .

SVZ-NSCs grown on full-length NET1 (NET1FL) showed a significant increase in the number of migrating neurons compared to control substrate ( Figures 5 C and 5D), in line with previous work (O’Leary etਊl., 2015). Interestingly, NET1ΔNTR (that inhibits RGM-mediated growth cone collapse [ Figureਃ G]) was not sufficient to activate NEO1-depenent cell migration ( Figures 5 E and 5F), suggesting a crucial role of the NET1 C-terminal NTR domain, potentially because it interacts with heparan sulphate proteoglycans (Kappler etਊl., 2000). When NET1FL interface-1 or -2 mutants were used no enhanced migration was observed, in agreement with our structural and biophysical analyses and supporting that both NET1-NEO1 interfaces are essential for function ( Figures 5 C and 5D).

Next, we examined whether RGMA can counteract the positive effect of NET1 on neuron migration and found that RGMA blocked the promotion of SVZ-neuroblast migration by NET1FL in a concentration-dependent manner ( Figures 6 A and 6B, Figures S5 C–S5G). Addition of RGMA had no effect on neuroblast proliferation or differentiation ( Figures 6 C�). Furthermore, we showed that NET1 could induce neuroblast migration following DCC ablation while RGMA could still inhibit this effect ( Figures 6 G and 6H, Figures S5 C–S5G data not shown). RGMB also blocked the promotion of SVZ-neuroblast migration by NET1 ( Figures 6 I and 6J) suggesting a general effect of RGMs.

RGMs inhibit NET1-mediated SVZ-neuroblast migration

(A and B) RGMA inhibits SVZ-neuroblast migration mediated by NET1-NEO1 signaling in a concentration-dependent manner. Analysis (A) and representative samples (B) of SVZ-NSCs grown on full-length NET1 and different concentrations of mouse RGMA. 2x RGMA = 1.2 μg/ml, 10x RGMA = 6.0 μg/ml. Mean ± SEM of the relative number of TUJ1/DCX-positive migrating neurons per neurosphere: NET1FL-WT = 100.00, NET1FL-WT + 2x RGMA = 68.52 ± 7.17, NET1FL-WT + 10x RGMA = 52.04 ± 10.27, n = 6 experiments. Bartlett’s test to test significant difference between SDs (p < 0.05): p π.0001. Kruskal-Wallis followed by Dunn’s multiple comparisons test: NET1FL-WT vs. NET1FL-WT + 2x RGMA p = 0.0289, NET1FL-WT vs. NET1FL-WT + 10x RGMA p = 0.0023. Arrowheads indicate neuroblasts. Scale bar, 50 μm.

(CF) RGMA does not influence SVZ neurosphere proliferation and differentiation.

(C) Overview of the proliferation assay.

(D) Ratio of EdU-positive over DAPI-positive cells. Mean ± SD vehicle = 0.381 ± 0.091, 2x RGMA = 0.277 ± 0.027, 10x RGMA = 0.296 ± 0.059. n = 3 experiments, one-way ANOVA with Tukey’s multiple comparisons test, vehicle vs. 2x RGMA p = 0.5238, vehicle vs. 10x RGMA p = 0.6369. (E) Overview of the differentiation assay. (F) Ratio of TUJ1-positive over DAPI-positive cells. Mean ± SD vehicle = 0.447 ± 0.016, 2x RGMA = 0.433 ± 0.036, 10x RGMA = 0.482 ± 0.041. n = 3 experiments, one-way ANOVA with Tukey’s multiple comparisons test, vehicle vs. 2x RGMA p = 0.9389, vehicle vs. 10x RGMa p = 0.6897.

(G and H) Silencing of NET1-mediated neuronal migration in neurospheres by RGMA is DCC-independent. Analysis (G) and representative samples (H) of SVZ-NSCs derived from Emx1 wt/wt Dcc lox/wt (control) and Emx1 cre/wt Dcc lox/lox (DCC knockout) mice grown on full-length NET1 with and without addition of 2x RGMA. Mean ± SEM of the relative number of TUJ1/DCX-positive migrating neurons per neurosphere: Emx1 wt/wt Dcc lox/wt mean ± SEM NET1FL = 100.00, NET1FL + 2x RGMA = 48.263 ± 9.535, Emx1 cre/wt Dcc lox/lox mean ± SEM NET1FL = 100.00, NET1FL + 2x RGMA = 43.977 ± 6.099. n = 3 experiments, two-way ANOVA with Sidak’s multiple comparisons test, NET1FL vs NET1FL + 2x RGMA p < 0.0004 for both genotypes. Arrowheads indicate neuroblasts. Scale bar, 50 μm.

(I and J) RGMB inhibits neuroblast migration mediated by NET1. Analysis (I) and representative samples (J) of SVZ-NSC cultures grown on full-length NET1 with and without addition of 2x RGMB. Mean ± SEM of the relative number of TUJ1/DCX-positive migrating neurons per neurosphere: NET1FL = 100.00, NET1FL + 2x RGMB = 24.55 ± 5.253. n = 3 experiments, paired two-tailed t test p = 0.0048. Scale bar, 50 μm. See also Fig. S5.

Finally, we assessed whether interactions between RGM and NET1 can lead to silencing of their individual biological effects in vivo. In the embryonic cortex, RGMA acts as a repulsive cue for migrating CNs via NEO1 (van Erp etਊl., 2015). NET1 promotes the migration of various types of neurons and displays very low expression in the cortex (Brignani etਊl., 2020 O’Leary etਊl., 2015 Yung etਊl., 2018). We designed an in utero electroporation (IUE) study to examine whether NET1 could silence the repulsive effect of RGMA on cortical neuron migration in vivo. Expression constructs for RGMA or NET1 (in combination with a GFP plasmid for visualization) were electroporated at E14 following which the pregnant mothers received a pulse of EdU at E15. EdU labelling enabled analysis of neurons that would migrate into a region of strong RGMA expression (±NET1 expression) that was generated by IUE at E14. Expression vectors were targeted to neuronal progenitors in the VZ at E14, followed by immunostaining at E16 or E17 ( Figures 7 A and 7B). Three days after GFP electroporation, EdU + neurons were found throughout the cortex, including in the upper CP ( Figures 7 C and 7D). Electroporation of RGMA created a non-permissive zone for subsequent EdU + neurons, resulting in a reduction of the number of EdU + neurons in upper cortical areas (quantified in the CP) ( Figures 7 C and 7D). Electroporation of NET1 also reduced the migration of EdU + neurons, most likely because neurons got trapped in deeper regions exogenously expressing this attractive cue ( Figures 7 C and 7D). Knockdown of NEO1 partially rescued the reduced migration of EdU + neurons, indicating that the NET1-mediated effect requires NEO1 ( Figures 7 C and 7D). Finally, we tested co-electroporation of RGMA and NET1 and failed to detect a reduction in CN migration, both following analysis of EdU + and GFP + neurons ( Figures 7 C and 7D, Figures S5 H–S5J). These data together with our observations from growth cone collapse and SVZ-NSC experiments ( Figures 3 , ​ ,5, 5 , and ​ and6) 6 ) and work by others (O’Leary etਊl., 2013) suggest that simultaneous binding of the functionally competing ligands NET1 and RGM blocks NEO1 receptor signaling.

In vivo inhibitory interactions between RGMA and NET1

(AD) In vivo inhibitory effects of RGMA and NET1 on embryonic mouse cortical neuron migration are silenced in the presence of both cues.

(A) Graphical overview of the in utero electroporation (IUE) experiment. Embryos were electroporated at E14 with a GFP construct in addition to (combinations of) different expression vectors (RGMA, NET1, or shRNA). At E15, pregnant mothers were injected with EdU to label the population of cortical neurons born at E15. At E17, migration of Edu + neurons was quantified in the cortical plate (CP) in 4 different bins (1𠄴).

(B) Immunohistochemistry showing NET1 expression in the deep part of the E16 cortex following co-electroporation of GFP and NET1-mCherry.

(C) EdU staining on E17 coronal sections of the mouse cortex to visualize migrating neurons born at E15, one day after IUE of the VZ at E14. Scale bar, 100 μm.

(D) Quantification of Edu + neuron migration using the bins shown in (C). Upper graph, IUE of RGMA and NET1 constructs reduced migration of EdU + neurons, an effect silenced when RGMA and NET1 are co-electroporated. Lower graph, reduced migration of neurons following NET1 electroporation is partly rescued by knockdown of NEO1 (shNEO1). One-way ANOVA followed by Sidak’s multiple comparisons: RGMA vs. GFP bin 4 p = 0.0094, NET1 vs. GFP bin 4 p < 0.0001, NET1 vs. RGMA+NET1 bin 1 p = 0.0231, NET1 vs. RGMA+NET1 bin 4 p < 0.0001, NET1+shSCR vs. GFP bin 2 = 0.0366, NET1-shSCR vs. GFP bin 4 p < 0.0001, NET1+shNEO1 vs. GFP bin 4 p = 0.0108. GFP, RGMA, and NET1+ RGMA: n = 6 embryos, NET1 and NET1+shSCR: n = 4 embryos, NET1+shNEO1: n = 7 embryos. i.p., intraperitoneally E, embryonic day VZ, ventricular zone SVZ, subventricular zone IZ, intermediate zone MZ, marginal zone.

(E and F) Model for NEO1 signaling via the NET1 and RGM guidance molecules in trans.

(E) NET1-induced clustering of NEO1 at the cell surface via Interface-1 and -2 interactions can lead to NEO1 intracellular interactions, inducing e.g. attractive guidance and outgrowth (left panel). In contrast, RGM binding to potentially pre-clustered NEO1 results in NEO1 dimerization in a signaling compatible conformation (Bell etਊl., 2013) (right panel). This architecture leads to activation of downstream signaling resulting in repulsive guidance (e.g., growth cone collapse), a process that can be potentiated by BMP morphogens (Healey etਊl., 2015).

(F) Combined binding of RGM and NET1 to NEO1 results in “trimer-of-trimers” super-complexes, preventing cell surface clustering, thereby inhibiting both RGM-mediated repulsive but also NET1-mediated attractive signaling. See also Figure S5 , Figure S7 , Figure S8 .

Binding BSA to silver surface via large difference in isoelectric points of the two materials - Biology

a Medway School of Pharmacy, Universities of Kent and Greenwich, Chatham, Kent, UK
E-mail: [email protected]

b Departament de Biologia Ceŀlular, Fisiologia i Immunologia, Universitat Autònoma de Barcelona, Bellaterra, Spain

c National Centre for Sensor Research, Biomedical Diagnostics Institute, Dublin City University, Dublin 9, Ireland

d School of Biotechnology, Dublin City University, Dublin 9, Ireland

e Department of Chemistry, MacDiarmid Institute for Advanced Materials and Nanotechnology, University of Auckland, Private Bag 92019, Auckland 1142, New Zealand

f Department of Chemistry, University of Surrey, Guildford, Surrey, UK


Despite the potential of antibody-coated nanoparticles (Ab-NPs) in many biological applications, there are very few successful, commercially available examples in which the carefully engineered nanomaterial has made it beyond the laboratory bench. Herein we explore the robustness and cost of protein-nanoparticle conjugation. Using multivalent polyamidoamine (PAMAM) dendrimers and dextran as crosslinkers, it was possible to retain colloidal stability during (i) NP-linker binding and (ii) the subsequent conjugation reaction between linker-coated NPs and proteins to generate monodisperse Ab-NPs. This was attributed to the physicochemical properties of the linkers, which were inherited by the NPs and thus benefited colloidal stability. Attaching negatively charged, EDC/sulfo-NHS-activated PAMAM to the NPs contributed to overall negative charge of particles, and in turn led to high electrostatic attraction between the protein and PAMAM-coated NPs during the reaction conditions. In contrast, using an uncharged, EDC/NHS-activated PAMAM dendrimer led to NP aggregation and lower protein binding efficiency. Dextran as a cost-effective, uncharged macromolecule allowed for steric repulsions between neighbouring particles during protein binding, thus inducing NP stability in solution, and also produced monodisperse Ab-NPs. By freeze-drying Ab-NPs from a 1% BSA solution it is possible to reconstitute the solid-form colloid back to a stable state by adding solvent and simply shaking the sample vial by hand. The consequences of the different surface chemistries and freeze-drying stabilizers on the colloidal stability of the NPs were probed by dynamic light scattering. The performance of Ab-NPs was compared in a simple fluorescence linked immunoassay in whole serum. Interestingly, the signal-to-noise ratios were similar for Ab-NPs using PAMAM and dextran, despite dextran binding fewer Abs per NP. We believe this work provides researchers with the tools and strategies for reliably generating Ab-NPs that can be used for a variety of biological applications.

Gold NanoUrchins

Figure 4. TEM of 100 nm Gold NanoUrchins

Gold NanoUrchins have unique optical properties compared to spherical gold nanoparticles of the same core diameter. The spiky uneven surface causes a red shift in the surface plasmon peak and a larger enhancement of the electromagnetic field at the tips of the Gold NanoUrchin spikes compared to spherical particles. As an example, 100nm spherical gold nanoparticles have an SPR peak at 570nm while 100nm Gold NanoUrchins have a SPR peak at around 680nm, figure 4.

Figure 5. Left - UV-VIS spectra of 100 nm Gold NanoUrchins (blue) and 100 nm standard gold nanoparticles (green). Note the red-shift in the SPR-peak. Right - UV-VIS spectra of Gold NanoUrchins ranging in size from 50 nm to 100 nm in diameter.


In our previous studies, we developed different types of scaffolds for tissue regeneration 8,9,22,70 . In the current study, a novel type of skin scaffold capable of drug delivery, based on hybrid utilization of PCL microspheres embedded into the PVA/SA composite hydrogel, was introduced to accelerate cell-induced tissue regeneration. The appropriate blending of SA and PVA polymers used in the proposed hydrogel caused to develop an optimized porous structure, as an imitator of skin tissue ECM, which possessed the adequate mechanical properties, gel-like physics, sufficient degradation rate, high swelling ability and an appropriate WVTR. The addition of PCL microspheres encapsulating bFGF, as a key stimulator of fibroblast migration and proliferation, to the hydrogel substrate created a burst-free and sustained release kinetics of this GF. The fabricated scaffold inhibited S. aureus and E. coli growth as well. Furthermore, the hybrid scaffold showed no cytotoxic effect in-vitro regarding the MTT assay and the cells were able to proliferate within the structure. The in-vivo evaluations also indicated that the constructed hybrid scaffold effectively accelerated the burn-wound healing process through promoting epithelialization and collagen deposition.

Watch the video: Dynafit Low Tech Race Binding (July 2022).


  1. Amery

    I apologize, but in my opinion you admit the mistake. I can defend my position. Write to me in PM, we will discuss.

  2. Kigagor

    Rather than criticize better write their options.

  3. Mut

    Well done, this sentence was just about

  4. Severin

    Exam +5

Write a message