Functional proteomic strategies offer unique advantages over current molecular array approaches, as the epitopes identified can directly provide bioactive peptides for investigational and/or translational applications. The vascular endothelium is well suited to proteomic assessment by in vivo phage display, but extensive enrichment and sequencing steps limit its application for high throughput molecular profiling. To overcome these limitations we developed a quantitative PCR (Q-PCR) strategy to allow the rapid quantification of in vivo phage binding. Primers were designed for distinct clones selected from a defined phage pool to probe for age-associated changes in cardiac vascular epitopes. Sensitivity and specificity of the primer sets were tested and confirmed in vitro. Q-PCR quantification of phage in vivo confirmed the preferential homing of all phage clones to the young rather than old cardiac vasculature and demonstrated a close correlation with phage measurements previously determined using traditional bacterial-based titration methods. This Q-PCR approach provides quantification of phage within hours of phage injection and may therefore be used for rapid, high throughput analysis of binding of defined phage sequences both in vivo and in vitro, complementing nonbiased phage approaches for the proteomic mapping of vascular beds and other tissues.
- molecular profiling
- vascular epitopes
the vascular system represents an important target for distinguishing the molecular components of specific organs and stages of development and disease, since it is well established that vascular heterogeneity exists between organ beds as well as between developing and mature, and healthy and diseased tissues (1, 13). Analysis of the epitopes expressed by a particular vascular bed can reveal a molecular “fingerprint” of an organ or pathophysiological condition that may be useful for diagnostic purposes. Additionally, determination of epitopes expressed on the vascular surface provides a method for selective organ-specific targeting for drug delivery. Combined with the ease of access of the vascular system, molecular profiling of the vasculature represents a promising strategy for the design of tailor-made therapeutics that can be readily and specifically delivered to the target tissue of interest.
A number of techniques are used to study the distribution and levels of gene expression in tissues including differential display, subtractive hybridization, serial analysis of gene expression (SAGE), and microarrays (10). However, since gene expression is not necessarily indicative of the localization, expression level, or function of the corresponding protein product, proteomic approaches have been developed for analysis of the proteins and peptides themselves. Techniques such as comparative two-dimensional gel analysis are used to identify differentially expressed unknown peptides but require large amounts of material and subsequent mass spectrometry-based sequence analysis (14). To circumvent these limitations, antibody-binding arrays have been developed that can allow more rapid determination of protein expression but are limited to probing for known peptides for which suitable antibodies, with high affinity and low dissociation rates, can be generated (5).
A potentially powerful approach for the functional proteomic profiling of vascular beds is phage display biopanning. Similar to antibody array systems, phage display is routinely used for in vitro analysis of interactions between phage peptide sequences and targets including cell lines, extracellular matrix molecules, and monoclonal antibodies themselves (22, 25). The major strength of phage-based profiling lies in its utility for in vivo analysis of the molecular architecture of the endovascular surface in multiple organs or the entire host. Phage particles injected into the systemic circulation bind rapidly to cell surface ligands, thus providing a system well suited to the investigation of epitope expression on the vascular surface. We have previously used phage display biopans to functionally probe age-associated receptor distribution patterns in the murine heart, leading to the identification of novel cardiac pathways (3, 8, 9). For example, the increased titers of an integrin-like phage, ΨR3Y32, in the young compared with old murine heart led to the identification of an α8-integrin binding partner, tenascin-C, as a factor that is expressed in the cardiac vasculature. Subsequent analysis revealed the importance of tenascin-C for cardiac angiogenesis (3).
The identification of phage homing to the vasculature of specific organs and the determination of their endogenous binding partners also provide information about ligand-receptor interactions in vivo, and, as a consequence, these phage peptide sequences may provide valuable candidates for functional inhibition or organ targeting. Together, in vivo biopanning and molecular engineering of epitopes in bacterial hosts, directly employing recombinant phagemid technology, can be exploited to enhance epitope binding or for conjugation of therapeutic molecules. A number of groups have examined the use of compounds linked to tissue-specific homing peptides, demonstrating the ability of cytotoxic conjugates to inhibit tumor growth (2, 24). Arap and coworkers (2, 21), for example, have demonstrated that the delivery of the anticancer drug, doxorubicin, to tumor vessels via tagging with a tumor-homing phage peptide improved the efficacy of this drug, reducing tumor size and prolonging survival in murine studies.
The vast complexity inherent in phage display biopanning, while providing an unbiased approach to epitope profiling, limits its quantitative power in proteomic assessment. Phage-based protocols currently require multiple rounds of phage transformation, amplification, and injection, followed by extensive sequencing of phage DNA to compare the frequency of individual phage clone sequences (22) (Fig. 1). Although this approach can provide important information, it may nevertheless allow analysis of only a fraction of the total phage population, with a diminished power to detect lower frequency clones. Moreover, this approach is laborious and requires further phage injections to confirm the titers of individual phage clones.
To overcome these limitations, we have developed a quantitative PCR (Q-PCR)-based approach to allow the analysis of multiple, known phage sequences simultaneously (Fig. 1). The generation of phage clone-specific PCR primers not only increases the throughput of samples that can be analyzed but also allows for precise quantitative analysis of the binding of specific phage epitopes. Although such a strategy does not replace the nonbiased approach of large-scale sequencing to identify novel epitopes, the development of a faster means of phage binding analysis of a more limited pool of known epitopes can complement traditional approaches and broaden the applications of phage display to provide a rapid system for determining the molecular profile of specific vascular beds. In this study, we demonstrate the use of the Q-PCR-based approach to simultaneously and rapidly evaluate the binding capacity of multiple phage sequences that are thought to bind preferentially to the young (3-mo-old), rather than old (18-mo-old), cardiac vasculature.
MATERIALS AND METHODS
Phage Clone-Specific PCR Primer Design
A phagemid pool consisting of four phage clones encoding a cyclically constrained seven-amino acid variable region was selected based on a previous biopan involving injection of a pSKAN phagemid library (MoBiTec) into young (3-mo-old) and old (18-mo-old) mice (9). Phagemid clones from this biopan were sequenced and translated amino acid motifs analyzed for homology to known mammalian proteins. Over 100 clones were sequenced from both the young and old pools. Of the sequenced clones, four were selected that were present in the pool of phage that bound to the vasculature of the young murine heart and were not present in the aging pool.
To amplify phage DNA from all four phage clones, universal forward and reverse primers (forward primer, #2897, 5′-GGAGGTCTAGATAACGAGG-3′; reverse primer, #1255, 5′-AGTTGTTTAGCAAAATCCC-3′; MoBiTec) were used. To amplify individual phage clones, the universal forward primer (#2897) was used, and reverse primers were designed using Oligo 4.0 software (Molecular Biology Insights) to select 19-mer primers that were complementary to >10 base pairs of the 21-base pair variable region of each phage clone (Table 1).
Phage Primer Sensitivity and Specificity
To test the sensitivity of each set of phage clone-specific primers, phage DNA was isolated from minipreps of WK6λmutS Escherichia coli cultures infected with the individual phage clones using the QIAprep spin miniprep kit (Qiagen). PCR was performed using 200 ng of phage DNA, 5 pmol of both forward and reverse primers, and 2X Hotstar PCR mastermix (Qiagen) per 25-μl reaction. The following cycling program was used: 95°C for 10 min followed by 30 cycles of 95°C for 45 s, 55°C for 45 s, 72°C for 45 s, and finally 72°C for 10 min. Reactions were performed in duplicate. For negative controls, DNA was substituted with water.
To test the specificity of the designed primers for their corresponding phage DNA, the primer pair for a particular phage clone was used in a PCR reaction containing DNA encoding for the three other phage clones. A lack of a PCR product after 30 cycles (using the above PCR program) was deemed to demonstrate lack of cross-reactivity between phage primers and DNA from other phage clones.
Real-Time PCR Analysis of Phage Titers
Q-PCR reactions were performed using serial dilutions of phage DNA, 5 pmol of both forward and reverse primers, and 2X SYBR green (Applied Biosystems, California) per 25-μl reaction. Q-PCR reactions were performed in a Cepheid SmartCycler. The following cycling program was used: 95°C for 10 min followed by 40 cycles of 95°C for 45 s, 55°C for 45 s, and 72°C for 45 s. Reactions were performed in duplicate. Copy numbers of phage DNA were determined based on spectrophotometric analysis and serial dilution to produce standard curves for each phage clone. Cycle threshold (Ct) values were defined, using the SmartCycler software, as the cycle number at which fluorescence emission is 5 SD above background levels.
In Vivo Phage Peptide Cardiac Biopanning
Studies employing 3- and 18-mo-old C57BL/6 mice were approved by and performed in compliance with the Institutional Animal Care and Use Committee of Weill Medical College of Cornell University.
Phage injection and elution.
Female C57BL/6 mice 3- and 18-mo-old (n = 3 per group) were anesthetized with 2.5% Avertin and injected with the phage pool (3 × 1010 colony-forming units per phage clone/100 μl PBS for injections of the pool of four phages, or >1011 colony-forming units of ΨY12 alone per 100 μl PBS), via the tail vein. Four minutes after injection, mice were killed, and hearts were explanted. Hearts from mice injected with the pool of four phages were then incubated for 10 min at room temperature in SM buffer [5.8 g/l NaCl, 2 g/l MgSO4·7H2O, 1 M Tris·HCl (pH 7.5), and 2% gelatin], a high-salt buffer that disrupts noncovalent bonds between the phage and their cellular receptors, and thus releases phage from the vascular surface. Eluates were used for titration via both standard methods (i.e., serial 10-fold dilution, infection of WK6λmutS E. coli and culture on selective media containing the antibiotics tetracycline and ampicillin) and Q-PCR analysis, using both universal primer pairs and clone-specific primers, without any bacterial amplification step. For all samples, 0.2 μl of eluate were used per 25-μl Q-PCR reaction. This ensured that for all phage clones, Ct values were greater than background levels and fell within the linear range of the corresponding dilution curve (i.e., 18 < Ct < 30). Hearts from mice that were injected with ΨY12 alone were explanted and embedded in paraffin according to standard techniques.
Antibodies to the phage coat protein, pIII (MoBiTec), and tumor necrosis factor receptor (TNFR)-1 (Santa Cruz) were used on 10-μm paraffin sections of cardiac tissue. For consistency, heart sections for all analyses were selected from the midpapillary region of the ventricles. Primary antibodies were labeled with biotin (Innogenex) and visualized with Texas red- and FITC-conjugated avidin (Vector Labs). For costaining, antibody incubations were performed sequentially. Before addition of the second biotinylated antibody, a blocking step involving incubation of the sections with three changes of avidin was performed.
Statistical analysis was performed using the unpaired Student's t-test. ΔCt values and phage titers are expressed as means ± SD. A P value of <0.05 was considered statistically significant.
Selection of Young Cardiac-Homing Phage and Phage-Specific Primer Design
In vivo phage biopanning was previously performed in 3- and 18-mo-old mice to identify phage preferentially homing to the young cardiac vasculature (9). From this biopan, four phage clones were selected for further investigation on the basis of their presence in the pool of phage recovered from young, but not old mice and homology to known mammalian proteins (Table 1). These four clones were previously designated ψY12, ψY32, ψY58, and ψY79. Notably, ψY12, a phage clone with homology to TNF-α, was selected for comprehensive analysis based on the distribution of a number of TNF-α-like phage in the young, but not old, pool, as previously reported (9). Phages ψY32, ψY58, and ψY79 were selected based on their identification in the young, but not old, phage pool.
Validation of Primer Design for Phage Titer Analysis
To specifically amplify individual phage clones using Q-PCR, a universal forward primer (primer #2897) was used for all phage clones in combination with phage clone-specific reverse primers, designed to span the variable region of the phagemid DNA sequence (Fig. 1, Table 1). To confirm the capacity of phage-specific primer pairs to amplify their corresponding phagemid DNA, 200 ng of DNA from individual phage clones were amplified using the respective phage-specific primer pair. Each set of primers was able to amplify its corresponding phage DNA, as determined by detection of PCR products on agarose gels (Fig. 2A). Furthermore, primer specificity was tested by combining PCR primers for a specific phage clone with DNA from the three other clones. In all four cases, no product was detected on the gels after 30 PCR cycles, illustrating that each set of primers was selective for its respective phage clone (data not shown).
Specificity of the phage-specific primer pairs was then tested using Q-PCR. Each set of primers was used to amplify its corresponding phage DNA either alone or in a reaction mixture containing DNA of the three other phage clones at 100-fold excess concentrations. In both cases the threshold cycle number values, Ct (i.e., the cycle during the early exponential phase of DNA amplification at which fluorescence emission, representing the amount of DNA product produced, first becomes statistically significant above background levels), were comparable. This demonstrates that each set of phage-specific primer pairs is able to amplify its own phage DNA but does not cross-react and amplify other phage sequences (Fig. 2B). Furthermore, primers specific for each phage clone were employed in negative control tests, where reaction mixtures contained either no phage DNA, or DNA from all clones other than that corresponding to the phage primers. Under both conditions, Ct values were at background levels (i.e., >30), further confirming that each set of primers is specific for its corresponding phage DNA and does not amplify other phage DNA sequences (Fig. 2B). All four clones were tested in this way, and the specificity of each primer pair was confirmed.
Sensitivity of the phage-specific primer pairs for DNA amplification was tested using Q-PCR. Serial dilutions of phage DNA, from 200 ng to 20 pg, or ∼1011–107 copies of each phage plasmid (as determined by spectrophotometric analysis), were prepared and amplified with their respective primer pairs. For all four clones, increasing DNA concentrations correlated with decreasing Ct, demonstrating a linear correlation between the two parameters (R2 > 0.9 for all serial dilution curves) within this range (Fig. 2C). Importantly, all dilutions tested gave Ct readings above background levels, demonstrating that this approach is sensitive enough to detect as little as 20 pg of phage DNA.
In Vivo Analysis of Phage Clone Binding Using Q-PCR
Once the sensitivity and specificity of the phage-specific primer pairs were established, the application of these primers for in vivo analysis of phage binding was performed. For analysis of phage binding to the cardiac vasculature, the four phage clones were pooled and injected via the tail vein into both young and old mice (3 × 1010 phage/clone/mouse). After 4 min, animals were killed, and phage was eluted from the cardiac tissue. Both standard titration via bacterial infection and Q-PCR reactions were performed directly using the eluate.
Total phage eluted from each mouse was determined via Q-PCR using universal forward (#2897) and reverse (#1255) primers. (Fig. 3, A and B). Analysis of ΔCt values (i.e., difference between Ct of sample and Ct of negative control) for total phage recovered indicated that phage binding was ∼25 times higher in young mice (ΔCt in young mice = 6.09 ± 1.76; ΔCt in old mice = 1.43 ± 1.28; a difference in ΔCt of 4.7 corresponds to an order of magnitude difference between young and old titers; Fig. 3B). Titers determined by bacterial-based methods also showed a difference in phage binding of one order of magnitude (Fig. 3C), confirming that Q-PCR analysis of relative titers is a valid alternative to standard 10-fold dilution titration techniques. Individual phage clone binding was then analyzed from the cardiac eluates of the pooled phage using Q-PCR and clone-specific primers, confirming that each of the four clones preferentially bound in the young rather than old mouse heart (Fig. 3, D and E). As expected, given the lower cardiac binding capacity, there was generally more variability in phage titers in the old mice compared with the young mice (Fig. 3D). On the basis of these data, the age-associated fold difference in individual phage clones was calculated, ranging from ∼7-fold (ΨY58) to 17-fold (ΨY32, Fig. 3E).
To confirm that phage titers assessed by Q-PCR reflect phage binding in vivo, ΨY12 alone was injected into 3- and 18-mo-old mice, which were killed 4 min later, followed by dissection and paraffin embedding of hearts. Immunostaining of sections from these hearts for the phage coat protein pIII demonstrated expression within the cardiac vasculature, confirming the binding of ΨY12 to vessels within the heart. Importantly, comparison of immunostains in young and old hearts revealed higher levels of ΨY12 binding to the young, rather than old, cardiac vasculature, in agreement with phage titers deduced by Q-PCR analysis (Fig. 3F). Moreover, these immunostains showed that the ΨY12 clone, which has homology to TNF-α, colocalized with TNFR1 in the hearts, with greater staining and colocalization in the young hearts compared with the old (Fig. 3G). In agreement with ΨY12 binding, TNFR1 protein expression was also found at higher levels in the young vs. old cardiac vasculature, as we have previously described (9). These data thus demonstrate that phage titers assessed by Q-PCR analysis correlate with the expression of putative endogenous receptors for phage motifs in vivo, confirming the efficacy of phage display combined with Q-PCR for the identification of differentially expressed vascular cell receptors.
The present study has demonstrated the utility of Q-PCR to enhance the throughput and comparative assessment of in vivo phage binding for functional proteomic studies of the vasculature. By employing the combination of a universal primer with designed phage clone-specific primers we have successfully determined binding titers for individual phage clones eluted from the cardiac vasculature as part of a defined pool of phage. This approach is both sensitive and specific for each phage clone and the results are comparable to previous phage titer studies. Furthermore, it is extremely rapid compared with standard techniques, taking advantage of the parallel throughput approach of Q-PCR analysis rather than serial processing via phage enrichment and sequencing.
Our data demonstrate that the combination of phage-specific primer design and Q-PCR constitutes a powerful system that is ideally suited to in vivo molecular profiling and high throughput analysis of vascular epitope quantification. In the development of this approach, initial analysis of primer specificity and sensitivity is essential, since cross-reaction of primers with other phage sequences (either from variable or consensus regions) would result in false positives and inaccurate titers. We have demonstrated in vitro that each primer pair was indeed specific for its corresponding phage DNA, even in the presence of a 100-fold excess of the other phage clones. Moreover, the generation of standard curves demonstrated that primers were able to maintain linearity in phage DNA amplification over a range of concentrations spanning five orders of magnitude, allowing for the specific detection of as little as 20 pg of phage DNA.
In vivo testing of our novel approach demonstrated its capacity to detect the differential, age-associated distribution patterns of select phage clones. Indeed, comparison with ΨY12 binding titers that we have previously determined by traditional bacteria-based titration approaches (9) confirms the accuracy of the Q-PCR-based approach described here: we previously found that the ΨY12 phage had an ∼10-fold higher concentration in young hearts compared with old hearts [106 in the young heart and 105 in the old heart (9)]. This is in good agreement with the 12-fold difference we calculated by Q-PCR from the pool analysis described here. Together, these data demonstrate that the Q-PCR-based approach is a valid and powerful method for phage titration of single phage clones and can replace traditional techniques to determine individual phage titers. It should be noted that phage titers are generally considered to be indicative of the level of phage binding, and in vivo biopanning cannot distinguish between increases in phage affinity and increased numbers of binding sites. However, we can nevertheless make direct comparisons between interactions of a particular phage clone in one host or vascular bed vs. another.
Unlike conventional large-scale biopans that employ bacterial-based phage titration methods and typically analyze the distribution of >100 phage epitopes, the biopan used in this study involved the analysis of only a small subset of identified phage clones. This pilot study, however, now paves the way for the expansion of this approach to analyze much larger pools of clones. Indeed, one can envisage the development of proteomic arrays consisting of multiwell plates coated with different primer sequences for large-scale analysis of epitope expression. To identify potentially important epitopes that would be incorporated into such an array system, initial unbiased biopans would need to be performed, as was done to identify the phage epitopes examined in this study. Primers sensitive and specific for different variable regions will be essential in the expansion of this technology and may necessitate the reengineering of specific variable regions to reduce nonspecific PCR amplification. Alternatively, increasing the size of the variable region could be employed to improve primer specificity.
The main utility of phage display approaches is the ability to examine epitope binding to vascular targets in an in vivo system. The vascular system is rapidly accessible both for binding and elution of phage, a feature that can be exploited for the targeted delivery of therapeutic agents to damaged organs. Indeed, in the current study, animals were killed just 4 min after systemic phage injection to select clones binding the vascular endothelial surface. This is in contrast to biopanning strategies targeting nonvascular cells/tissues, such as the myocardium, in which it has been shown that intravenous phage injection requires much longer of in vivo circulation (i.e., hours) to ensure passage of phage particles across the endothelial barrier, and subsequent complicated elution protocols (19, 28). Although phage binding in the cardiac vasculature alone was analyzed in the current study, a further advantage of phage display biopanning is that the vasculature of multiple target organs in a single host can be analyzed simultaneously, as we and others have previously demonstrated (3, 16, 18, 19). Thus the investigation of phage binding in multiple vascular beds may be greatly enhanced by the combination of phage display with Q-PCR analysis.
The validation of our novel approach to phage titer determination after in vivo biopanning now paves the way for the expansion of phage display technologies to new applications. The combination of Q-PCR-based phage titration with laser capture microdissection from histological sections of tissues, for example, would permit the analysis of epitope levels and distribution in specific blood vessels and vascular cell types, as has been used by Yao et al. (27) to demonstrate vascular heterogeneity of the pancreatic islet. With this approach, individual vessels could be identified and the molecular profile of different vessel types, or the same vessel type in different tissue samples, analyzed. Furthermore, the added step of immunohistochemical staining to identify specific cell populations may also extend the applications of phage display to identify epitopes expressed on individual cells or cell types, as has been demonstrated in studies combining laser capture with two-dimensional gel electrophoresis and cDNA arrays (4, 23). Such an in vivo approach may have advantages over, or may potentially complement, in vitro approaches to examine the molecular profile of cultured vascular cells or cell membranes in vitro (6, 20, 26).
Another advantage of PCR-based phage analysis approaches is that binding of specific phage epitopes to target sequences represents a receptor-ligand interaction that can be analyzed by modification of the phage sequence and/or exploited for use as a competitive inhibitor or agonist for downstream signaling. Such approaches could improve on traditional phage biopanning techniques that have led to the identification of peptides that formed the basis for the development of a wide range of therapies, including anticancer vaccines (15), anticoagulants (11, 12), anti- and proangiogenic agents (17, 25), and antitumor therapies (7). In addition to its uses in experimental models, PCR-based biopanning could have direct clinical applications. Given that the vasculature of an organ contains a unique molecular profile, the targeting of specific vascular beds via organ-specific markers represents a promising approach for drug delivery.
In conclusion, this study demonstrates that the combination of phage clone-specific primer design and real-time Q-PCR analysis is a viable and advantageous alternative to current phage titration methods. Although this technique does not replace standard phage titration approaches for examining distributions of unknown epitopes, our data suggest that the use of phage-specific primer pairs may act as a valuable complement to full-scale epitope profiling and will provide an efficient system for analysis of known phage clone sequences.
This work was supported in part by National Institute on Aging Grants AG-19738, AG-20320, and AG-20918.
We are grateful to Inga Duignan and Ruby Choi for technical assistance.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: J. M. Edelberg, Dept. of Medicine, Div. of Cardiology, Weill Medical College of Cornell Univ., 520 E. 70th St., New York, NY, 10021 (e-mail:).
- Copyright © 2006 the American Physiological Society