Next-generation phage display: integrating and comparing available molecular tools to enable cost-effective high-throughput analysis

Abstract

Background: Combinatorial phage display has been used in the last 20 years in the identification of protein-ligands and protein-protein interactions, uncovering relevant molecular recognition events. Rate-limiting steps of combinatorial phage display library selection are (i) the counting of transducing units and (ii) the sequencing of the encoded displayed ligands. Here, we adapted emerging genomic technologies to minimize such challenges.

Methodology/principal findings: We gained efficiency by applying in tandem real-time PCR for rapid quantification to enable bacteria-free phage display library screening, and added phage DNA next-generation sequencing for large-scale ligand analysis, reporting a fully integrated set of high-throughput quantitative and analytical tools. The approach is far less labor-intensive and allows rigorous quantification; for medical applications, including selections in patients, it also represents an advance for quantitative distribution analysis and ligand identification of hundreds of thousands of targeted particles from patient-derived biopsy or autopsy in a longer timeframe post library administration. Additional advantages over current methods include increased sensitivity, less variability, enhanced linearity, scalability, and accuracy at much lower cost. Sequences obtained by qPhage plus pyrosequencing were similar to a dataset produced from conventional Sanger-sequenced transducing-units (TU), with no biases due to GC content, codon usage, and amino acid or peptide frequency. These tools allow phage display selection and ligand analysis at >1,000-fold faster rate, and reduce costs approximately 250-fold for generating 10(6) ligand sequences.

Conclusions/significance: Our analyses demonstrates that whereas this approach correlates with the traditional colony-counting, it is also capable of a much larger sampling, allowing a faster, less expensive, more accurate and consistent analysis of phage enrichment. Overall, qPhage plus pyrosequencing is superior to TU-counting plus Sanger sequencing and is proposed as the method of choice over a broad range of phage display applications in vitro, in cells, and in vivo.

Figure 1. Comparison of TU-counting and qPhage.

(A) Representation of phage genome and relative location of the cloning site and two sets of primers used. Primer set #1 targets the Tet^R gene and was used for quantification with real-time PCR; primer set #2 flanks the insert coding for the peptide displayed in pIII, and served for large-scale sequencing (B and C, respectively). TU-counting and qPhage titration output [determined by cycle-thresholds (Ct)]; note the limited quantification range for TU-counting relative to qPhage. Comparative results of TU-counting and qPhage titration of Fd-tet (D) and RGD-4C phage (E) are shown. Asterisk indicates that the high bacterial density prevented accurate TU determination.

Figure 2. Phage binding and internalization assays.

Binding of αv integrin-binding ligand phage (displaying RGD-4C) or insertless phage to endothelial cells. Quantification by conventional TU-counting (A) and qPhage (B) are shown. For internalization, endothelial cells were incubated with RGD-4C phage or insertless phage for short (5 minutes) or long (ON) incubation. Internalized phage particles were detected by immunostaining (C) or qPhage (D).

Figure 3. Overlap of sequences revealed by Sanger sequencing and next-generation pyrosequencing.

Venn diagrams represent the peptides revealed by Sanger sequencing (purple) and large-scale next-generation pyrosequencing (salmon) in tissues such as bone marrow, fat, muscle, skin (A) or the non-selected CX7C library (B). The dark-pink area represents sequences found by both approaches. Numbers indicate the total peptides in each group and their percentage relative to TU-counting (purple and overlap areas) or relative to next-generation sequencing (salmon-colored area). No overlap was seen for the non-administered phage library sequences, produced by both methods. Circle sizes are proportional to the number of sequences revealed by each strategy.

Figure 4. Saturation plots of peptide diversity coverage after next-generation sequencing.

The plot shows the number of distinct peptides observed in bone marrow (A), fat (B), muscle (C), skin (D) or the non-selected library (E), as a function of the total number of sequences evaluated for each tissue after filtering. All tissues investigated attained or nearly attained saturation (as determined by the predicted number of distinct peptides in each tissue), whereas nothing approaching saturation was observed for the unselected library (straight line).

Figure 5. Analysis of cost and time required to generate phage sequences using Sanger- or 454-pyrosequencing methods.

Cost (A) and time (B) to generate sequences with Sanger-sequencing of individual TU (red) versus DNA amplification (qPhage) followed by next-generation sequencing (blue). Data and estimates used for this analysis are presented in Table S6 online.