Imaging IGF-I uptake in growth plate cartilage using in vivo multiphoton microscopy

This paper describes and validates a novel method for imaging transport of biologically active, fluorescently labeled IGF-I into skeletal growth plates of live mice using multiphoton microscopy. Cellular patterns of fluorescence in the growth plate were completely distinct from our prior publications using biologically inert probes, demonstrating for the first time in vivo localization of IGF-I in chondrocytes and perichondrium. These results form important groundwork for future studies aimed at targeting therapeutics into growth plates.

Abstract

Bones elongate through endochondral ossification in cartilaginous growth plates located at ends of primary long bones. Linear growth ensues from a cascade of biochemical signals initiated by actions of systemic and local regulators on growth plate chondrocytes. Although cellular processes are well defined, there is a fundamental gap in understanding how growth regulators are physically transported from surrounding blood vessels into and through dense, avascular cartilage matrix. Intravital imaging using in vivo multiphoton microscopy is one promising strategy to overcome this barrier by quantitatively tracking molecular delivery to cartilage from the vasculature in real time. We previously used in vivo multiphoton imaging to show that hindlimb heating increases vascular access of large molecules to growth plates using 10-, 40-, and 70-kDa dextran tracers. To comparatively evaluate transport of similarly sized physiological regulators, we developed and validated methods for measuring uptake of biologically active IGF-I into proximal tibial growth plates of live 5-wk-old mice. We demonstrate that fluorescently labeled IGF-I (8.2 kDa) is readily taken up in the growth plate and localizes to chondrocytes. Bioactivity tests performed on cultured metatarsal bones confirmed that the labeled protein is functional, assessed by phosphorylation of its signaling kinase, Akt. This methodology, which can be broadly applied to many different proteins and tissues, is relevant for understanding factors that affect delivery of biologically relevant molecules to the skeleton in real time. Results may lead to the development of drug-targeting strategies to treat a wide range of bone and cartilage pathologies.

NEW & NOTEWORTHY This paper describes and validates a novel method for imaging transport of biologically active, fluorescently labeled IGF-I into skeletal growth plates of live mice using multiphoton microscopy. Cellular patterns of fluorescence in the growth plate were completely distinct from our prior publications using biologically inert probes, demonstrating for the first time in vivo localization of IGF-I in chondrocytes and perichondrium. These results form important groundwork for future studies aimed at targeting therapeutics into growth plates.

endochondral ossification is a well-delineated process in which a cartilage template is replaced by bone in a series of coordinated steps that cause bones to grow in length (37, 47, 48, 61). This cascade of events occurs in the growth plate, a band of cartilage located between the epiphysis and metaphysis at ends of the principal long bones. There is a complex interaction of systemic- and local-level inputs that regulate the rate and duration of linear growth (15, 47, 52, 61, 67, 83, 85). The vasculature is central to this process by delivering factors that are essential to support endochondral ossification (4, 11, 24, 25, 78, 79). Disruptions in vascular supply are common denominators of numerous skeletal pathologies, underscoring the importance of the vasculature in bone health and disease (4, 9, 10, 13, 24, 25, 59, 60, 72, 76, 78, 82). What makes the growth plate unique from other tissues, including adjacent bone, is that signaling molecules reach cartilage through an indirect route. Aside from cartilage canals (2), growth plates do not have a penetrating blood supply (9, 11, 72, 84). Instead, soluble factors diffuse into growth plates from three vascular routes: 1) epiphyseal vessels, 2) metaphyseal vessels, and 3) a ring vessel connected to a subperichondrial plexus surrounding the growth plate (11).

Promising therapeutic strategies for the treatment of growth plate disorders have emerged from research on the signaling pathways involved in linear growth regulation (27, 43, 56). However, one persistent and major challenge is a lack of adequate methods for targeting these molecules to avascular cartilage (63). Although some experimental techniques have been successful using localized injections, catheters, and drug-releasing implants (1, 5, 14, 45, 73, 89), we do not have a clear picture of how they are physically transported into cartilage. A better understanding of factors affecting molecular uptake in growth plates would enable us to better design drug delivery approaches to potentially treat a range of debilitating growth plate disorders in children.

Measuring molecular transport into cartilage is a considerable technical challenge that was previously limited by a lack of available imaging techniques to visualize living tissue with adequate resolution (7, 12, 21, 23, 31, 88). In recent decades, multiphoton imaging has emerged as an exciting way to study live, intact tissues at cellular-level resolution (24, 29, 62, 65, 82). Multiphoton microscopy has advantages over other techniques for skeletal growth plate imaging because of its capacity to penetrate deep into connective tissues with minimal laser damage during long-term (3–4 h) imaging sessions (66, 69). Farnum, Williams, and colleagues developed a method to study tibial growth plates of live young mice using multiphoton microscopy (24, 82). Working with them, we optimized this platform to quantitatively assess molecular delivery to cartilage under controlled experimental conditions (66, 68, 69).

We previously used in vivo imaging to measure uptake of fluorescent molecules to mouse tibial growth plates using dextran tracers that approximate the size of various growth regulators (66, 69). Although we were able to track and even modulate molecular transport into the growth plate using mild limb heating, there are physical and physiological limitations in using biologically inert dextrans to study a process as dynamic as bone elongation [discussed by Serrat et al. (66)]. Insulin-like growth factor I (IGF-I) is a major hormone responsible for bone and body growth (3, 55). The purpose of this study was to develop methods for measuring uptake of biologically active IGF-I into growth plates of live mice using in vivo multiphoton microscopy. Here we describe techniques for labeling and testing the biological activity of fluorescently labeled IGF-I for in vivo imaging (Fig. 1). This methodology, which can be applied to many different proteins, may help improve our understanding of growth plate regulation by allowing us to quantitatively study factors that affect molecular uptake in the skeleton in real time.

Fig. 1.

Schematic illustrating key steps in fluorescent protein labeling using IGF-I. See text for details.

MATERIALS AND METHODS

Animals and tissue samples.

All procedures were submitted to and approved by the Institutional Animal Care and Use Committee at Marshall University (protocols 527 and 558). Male and female C57BL/6J mice (n = 10 total) were obtained at 5 wk of age from an in-house breeding colony. Half of the mice (n = 5) were used for terminal in vivo imaging of the proximal tibial growth plate. The remaining mice (n = 5) were euthanized without imaging, and the middle three metatarsal bones were removed for bioactivity testing, cleaned of skin and tendon, and stored at −80°C until use. Metatarsal bones were selected for the bioactivity test because they better represent native skeletal tissue compared with isolated cell lines. We found in a pilot test that metatarsals from 5-wk-old mice retain the ability to activate phosphorylation of the serine/threonine kinase Akt in response to treatment with IGF-I in culture.

Fluorescent protein labeling.

Fluorescently labeled IGF-I was used to develop methods for imaging physiologically relevant molecules in the growth plate. Of the many factors involved in linear growth regulation (22a, 46), IGF-I was selected as a starting point because it is the major circulating hormone of growth in humans and animals (32, 33, 86), it facilitates actions of other major hormones and growth factors (8, 55, 80, 87), and it has been implicated as a key regulator of differential growth plate activity (67) and chondrocyte hypertrophy (15) in elongating bones. Importantly, IGF-I is also within a size range (≤10 kDa) that we have previously shown to readily diffuse into cartilaginous growth plates (66).

Purified human IGF-I (7.6 kDa) was purchased commercially (100-11; PeproTech) and prepared at 1 mg/ml in 0.1 M sodium bicarbonate (36486-1L; Sigma). IGF-I was conjugated with Alexa Fluor 488 tetrafluorophenyl (TFP) ester (643 Da; A37570; Thermo Life Technologies), which is an amine-reactive dye that binds nonspecifically to the NH₂ terminus of proteins. It is therefore essential that the sample contains only the purified protein of interest and is free of serum and/or other binding proteins. Protein labeling was accomplished following the dye manufacturer’s protocol with modifications. The procedure is summarized below and in Fig. 1. Although only suggested by the manufacturer, we believe that the chromatography and bioactivity validation steps (detailed below) are essential parts of the labeling procedure.

First, 10 μl of deionized water were added to a single 100-μg tube of reactive dye (note that we did not use DMSO as recommended by the dye manufacturer because of the in vivo application). The dissolved dye was then immediately added to 100 μl of the purified protein solution, and the sample was incubated in the dark for 1 h at room temperature with gentle rocking. Nonreacted dyes were removed using commercially available fluorescent dye removal resin (suitable for proteins ≥6 kDa) and spin columns following the manufacturer’s instructions (22858; Thermo Pierce). For resin compatibility, the NaCl concentration of the reacted sample was first adjusted to 150 mM by adding 15 μl of 1 M NaCl to the labeling reaction. The labeling reaction was then added to a prepared spin column containing 175 μl of dye removal resin and centrifuged for 30 s at 1,000 g to collect the protein-dye conjugate (IGF-488). The degree of labeling (mol dye/mol protein) was determined in duplicate IGF-488 samples using a NanoDrop ND-1000 spectrophotometer (Nano-Drop Technologies, Rockland, DE) to measure protein concentration by absorbance at 280 nm (A₂₈₀) and 488 nm (A₄₈₈) using calculations provided by the dye manufacturer (55a). The labeled IGF-488 conjugate (8.2 kDa) was stored at 4°C for up to 1 wk, or at −20°C for up to 1 mo before use.

Thin-layer chromatography (TLC) was used to test for presence or absence of residual free dye in the IGF-488 sample. TLC was accomplished using chloroform (9175-33; JT Baker)-methanol (A412-4; Fisher Scientific) (1:1) as the eluent following published methods (40). The silica TLC plate (Z122777; Sigma) was spotted with 2 μl each of known free dye and protein-dye conjugate and allowed to develop in the chloroform-methanol solvent for 10 min. The plate was dried and spots were visualized at ×0.8 magnification using a ×0.4 reducing lens with a green fluorescent protein (GFP) fluorescence filter on a Leica MZ10F fluorescence stereomicroscope that was interfaced to a QImaging Micropublisher 5.0 camera (Surrey, BC, Canada).

In vitro bioactivity test.

Left and right metatarsals from a mixed-sex sample of n = 5 mice (10 samples) were fragmented into small pieces to facilitate diffusion in culture. Samples were incubated at 37°C in DMEM phenol-free media (SH30604.01; HyClone) and treated for 90 min with 100 ng/ml unlabeled IGF-I or fluorescently labeled IGF-488 or left untreated. Following treatment, tissues were homogenized at room temperature in the presence of radioimmunoprecipitation assay buffer (150 mM sodium chloride; 1% Triton X-100; 0.5% sodium deoxycholate; 0.1% SDS, and 50 mM Tris, pH = 8.0) containing protease inhibitor (PI78441; Thermo Scientific) and were sonicated for 4 s. After adding Laemmli buffer (161-0737; Bio-Rad), samples were boiled for 10 min (≥100°C) and then centrifuged for 1 min at 1,500 rpm. Supernatants were collected and stored at −80°C. Immunoblotting was subsequently done following standard Western-blotting methods.

Equal amounts of lysate from unlabeled, labeled, and untreated samples (25 μl each) were separated on SDS-10% polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (162-0177; Bio-Rad). Membranes were blocked using 5% powdered milk and 0.01% Tween 20 (P1379-100ML; Sigma) in PBS for 1 h at room temperature and thereafter probed with primary antibodies in blocking buffer (PBS containing 0.1% Tween 20) overnight at 4°C with gentle mixing. The following antibodies and dilutions were used: total Akt (4691S, 1:5,000; Cell Signaling Technology), phosphorylated Akt (Ser473, 4060, 1:2,000; Cell Signaling Technology), and phosphorylated Akt (Thr308, sc-16646-R, 1:500; Santa Cruz Biotechnology). Two Akt antibodies were used because phosphorylation on both residues (Ser473 and Thr308) is required for Akt activation (58). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH, sc-365062, 1:1,000; Santa Cruz Biotechnology) was used to normalize protein loading. The membrane was then rinsed and incubated with goat anti-rabbit poly-horseradish peroxidase (32260; Thermo Pierce) or goat anti-mouse poly-horseradish peroxidase (32230; Thermo Pierce) for 1 h at room temperature. An enhanced chemiluminescence system (WBKLS0500; EMD Millipore) was used for band detection. Normalized levels of total Akt (t-Akt) and phosphorylated Akt (p-Akt) were expressed as the ratios t-Akt/GAPDH and p-Akt/GAPDH. Densitometric analyses with background correction were performed using ImageJ software version 1.44 (https://imagej.nih.gov/ij/) following standardized methods (51).

In vivo multiphoton imaging.

Experiments were performed in the Marshall University Molecular and Biological Imaging Center using an upright Leica TCS SP5 II Multiphoton Microscope with a Ti:sapphire pulsed femtosecond laser with dispersion compensation (Chameleon Vision II; Coherent, Santa Clara, CA). Methods followed our published protocols for imaging proximal tibial growth plates of live 5-wk-old mice (66, 69). Farnum et al. (24) provide a diagram of the in vivo platform, along with details of the growth plate selection and surgical approach. In brief, mice were injected with oxytetracycline (OTC, 50 mg/kg; Norbrook; 200 mg/ml diluted in sterile water), which labels mineralizing bone and facilitates growth plate localization on the multiphoton microscope. Mice were then anesthetized with isoflurane (3.5% induction, graded to 1.5% maintenance) and positioned in dorsal recumbency atop a heating pad on a custom-built imaging stage. The left hindlimb was extended into a perfusion chamber containing lactated Ringer solution for water immersion imaging. Surgical procedures were performed under a Leica MZ10F fluorescence stereomicroscope. The proximal tibial growth plate was exposed by making a small surgical incision through the superficial fascia of the biceps femoris and gastrocnemius muscles between the medial collateral ligament and saphenous vessels as originally described by Farnum, Williams, and colleagues (24, 82). The surgical procedure was completed in <10 min and left the perichondrium, its vascular network, and joint capsule intact. The growth plate was warmed to a physiological temperature (32–34°C; 6, 28) using a perfusion pump and in-line solution heater as previously detailed (66, 69).

The limb was oriented under bright-field illumination on the multiphoton microscope using a Leica HC PL Fluotar ×5/0.15-numerical aperture (NA) air objective. The growth plate was imaged using a Leica HCX Irapo L ×25/0.95-NA water immersion dipping objective. Excitation wavelength was tuned to 880 nm, and emissions were collected in reflected-light mode using external detectors with emission filters for the separation of second-harmonic generation (SHG) from collagen (407–507 nm) and OTC/IGF-488 fluorescence (475–575 nm) using a 495-nm beam splitter. SHG from collagen in the perichondrium and OTC fluorescence in epiphyseal and metaphyseal bone were used to identify the growth plate and define imaging depth. Imaging depth was standardized at 50 µm deep to the deepest edge of the perichondrium, which was defined by collecting sequential optical sections in a z-series at 5-µm intervals from superficial to deep as previously detailed (66, 69).

After the growth plate was localized, mice were given a single 100-µl intraperitoneal injection of fluorescently labeled IGF-488 (~90 µg). The rationale for the dose is that it is within a range that stimulates bone growth in mice (70) and pharmacokinetic data are available (at 10 µg/g body wt; 35). Our preliminary data demonstrated that a single injection of 90-µg fluorescently labeled IGF-I could be visualized and quantified in vivo, reaching peak values in the growth plate within ~90 min. For a 15–16-g 5-wk-old mouse, this dose represents ~6 µg/g body wt. The z-series images were collected immediately before the injection and at 15-min intervals spanning 90 min after the injection using simultaneous collection channels for SHG and OTC/IGF-488. Growth plate location was verified at each time point by matching morphology of SHG collagen in the z-series. All image processing was done in ImageJ software version 1.44 (https://imagej.nih.gov/ij/) as described previously (66).

RESULTS

Fluorescent protein labeling.

Degree of labeling determined by NanoDrop spectrophotometry was between 0.93 and 0.95 in four separate labeling batches, yielding a nearly 1:1 dye-to-protein ratio. Thin-layer chromatography (TLC) confirmed that there was no detectable residual free dye in the IGF-488 samples (Fig. 2), assuring that fluorescence in the growth plate was solely dye-conjugated IGF-I. Figure 2, A and B, shows two representative labeling batches of IGF-488 run on TLC plates with a free dye standard. Samples were blotted at the bottom of the plates. Free dye appears as a streak toward the top of the plates, while dye-conjugated IGF-488 remains at the bottom. The absence of a streak in the IGF-488 sample indicates that all dye is bound to the protein. Figure 2C shows TLC plates from a preliminary labeling reaction in which IGF-I was also conjugated with Alexa Fluor 555 dye. TLC again verified the absence of unbound dye in the IGF-488 sample. By comparison, a large streak of residual free dye is evident in the IGF-555 plate, indicating that that particular sample was overlabeled and could not be reliably used for protein tracking in vivo.

Fig. 2.

A and B: thin-layer chromatography (TLC) results from two representative labeling batches of IGF-488 using chloroform-methanol (1:1) to separate free dye from the labeled protein. The silica sheet was viewed on a fluorescence stereomicroscope using a green fluorescent protein filter. Free dye and dye-conjugated protein samples were blotted at the bottom of the plates. Free dye appears as a streak toward the top of the plates, while dye-conjugated IGF-488 remains at the bottom. C: TLC plates from a separate labeling reaction in which IGF-I was also conjugated with Alexa Fluor 555 dye. TLC again verified the absence of unbound dye in the IGF-488 sample. By comparison, a large streak of residual free dye is evident in the IGF-555 plate, indicating that that particular sample was overlabeled and not suitable for in vivo protein tracking. Note that the images shown in C were taken without a reducing lens and are shown as a two-frame montage.

In vitro bioactivity test.

We next conducted a bioactivity test in which we treated left-right pairs of 5-wk-old mouse metatarsal bones in vitro with IGF-488 and measured phosphorylated Akt (p-Akt) as a biomarker of IGF-I signal activation. Figure 3 shows results from left and right metatarsal samples from n = 3 mice (n = 6 left-right samples) run on the same blot. Western blot (Fig. 3, top) shows that expression of p-Akt (Thr308) was increased in metatarsal bones treated with either labeled (lanes 2, 4, and 6) or unlabeled IGF-I (lanes 3 and 5) compared with an untreated sample (lane 1). Figure 3, bottom, shows normalized values of p-Akt (Thr308) expressed as the ratio p-Akt/GAPDH for each sample. IGF-488 treatment increased expression of p-Akt in bones from the same animals similar to the native unlabeled protein (compare lanes 3 and 4 with lanes 5 and 6). Densitometry analyses revealed that bones treated with IGF-488 had over fourfold greater p-Akt levels relative to untreated bones from the same mouse (left-right pair in lanes 1 and 2). Bones from two different mice treated with either labeled (lanes 4 and 6) or unlabeled IGF-I (lanes 3 and 5) showed over threefold greater p-Akt levels relative to the untreated control (lane 1). Results for p-Akt at the Ser473 phosphorylation site were similar (data not shown). Experiments were repeated with a total of n = 5 mice using left-right pairs (n = 10 left-right samples). Although we did not have enough samples to conduct paired statistical testing for each set of comparisons, independent-samples t-testing revealed that there was no significant difference between p-Akt levels in bones treated with labeled and unlabeled IGF-I (t = 0.81, P = 0.48); additionally, both treatments (samples pooled) were significantly higher than the untreated control bones (t = 8.96, P < 0.001). These data show that labeled IGF-I retains comparable biological activity of the native protein as assessed by these methods.

Fig. 3.

Western blot (top) for phosphorylated Akt (p-Akt, Thr308) shows IGF-I activation in 5-wk-old mouse metatarsal bones treated ex vivo with labeled and unlabeled IGF-I. Left and right metatarsal samples from n = 3 mice (n = 6 left-right samples) run on the same blot are shown. Adjusting for loading control (GAPDH), densitometry analyses of each sample (bottom) revealed that bones treated with IGF-488 had over fourfold greater p-Akt levels relative to untreated bones from the same mouse (left-right pair in lanes 1 and 2). Bones from two different mice treated with either labeled (lanes 4 and 6) or unlabeled IGF-I (lanes 3 and 5) showed over threefold greater p-Akt levels relative to the untreated control (lane 1). IGF-488 treatment increased expression of p-Akt in bones from the same animals similar to the native unlabeled protein (compare lanes 3 and 4 with lanes 5 and 6). Experiments were repeated using left-right pairs from n = 5 mice (n = 10 left-right samples) total. A.U., arbitrary units.

In vivo multiphoton imaging.

Once we determined that no excess dye remained in the IGF-488 sample and that the labeled conjugate retained its biological activity, our next step was to demonstrate feasibility of imaging its uptake in the growth plate in vivo. We collected time-lapse images of IGF-488 infiltrating the proximal tibial growth plates of live, 5-wk-old mice (n = 5) for 90 min after a single intraperitoneal injection. A representative timed series (Fig. 4) showed that IGF-488 is evident in the growth plate within 30 min from the injection. IGF-488 infiltration is depicted by the increasing fluorescence intensity (white) relative to background oxytetracycline (OTC) fluorescence in metaphyseal bone. Figure 5 shows optical sections of the perichondrium (Fig. 5A) and growth plate (Fig. 5B) of a live mouse 90 min after the IGF-488 injection. Collagen in the perichondrium (Fig. 5, panels at left) was visualized using second-harmonic generation (SHG) to document imaging depth. Fluorescence in the growth plate resembles round chondrocytes, suggesting receptor-mediated cellular localization. In contrast, the diffuse and punctate fluorescence in the more superficial, collagenous perichondrium lacks a cellular pattern. Optical sections of the same growth plate at higher magnification (×4 zoom) show cellular detail in the growth plate, demonstrating IGF-488 localization to chondrocytes (Fig. 6). All five of the mice examined showed the same pattern of fluorescence in the growth plate, with little variation in the cellular localization of IGF-488. Differences observed between mice were due largely to variation in imaging conditions, such as overlying fat or fascia obscuring the growth plate or slight differences in limb tilt that caused differential tissue scattering on the microscope (69). The representative images shown here are those in which such tissue scattering was minimal. The overall patterns among all mice, however, were the same.

Fig. 4.

Time-lapse images of fluorescently labeled IGF-I transport into the growth plate. Multiphoton images of a 5-wk-old mouse tibial growth plate from 10 to 90 min after an intraperitoneal injection of IGF-I (90 µg) labeled with Alexa Fluor 488 tetrafluorophenyl (TFP) ester (IGF-488) to validate visualization in vivo. The 0-min image shows oxytetracycline (OTC) fluorescence in metaphyseal bone at the bottom of the image. Growth plate is between the arrowheads shown at the left of each frame. IGF-I uptake, evident within 30 min from the injection, is shown by the increasing fluorescence intensity (white). Peak values were attained in the growth plate within ~90 min in a sample of n = 5 mice.

Fig. 5.

Multiphoton images of labeled IGF-I in the perichondrium (A) and growth plate (B). Fluorescence in the growth plate (B) resembles round chondrocytes, suggesting receptor-mediated localization. IGF-488 in the perichondrium (A) is diffuse and punctate, suggesting IGF-binding protein entrapment. Z-stack images from superficial to deep were taken 90 min after injection of fluorescently labeled IGF-I (IGF-488). Collagen, viewed by second-harmonic generation (SHG), and IGF-488 are in separate imaging channels. The region of the growth plate is shown by the white arrowheads.

Fig. 6.

Optical sections of the same growth plate in Fig. 5 taken at a higher magnification (×4 zoom) show cellular detail in the cartilage. The fluorescence is evident in a distinct pattern that appears localized to chondrocytes (red arrowheads). Sections were imaged at 5-µm increments spanning 80–90 µm below to the superficial edge of the perichondrium. See Fig. 5 for orientation.

DISCUSSION

Advances in understanding bone biology at the molecular level have led to the identification of many novel drug targets for a wide range of skeletal pathologies. Despite these potential therapies, a major obstacle in their practical implementation is a lack of methods for studying their delivery in the intact organism. More whole animal approaches are needed in skeletal physiology research (38). Pioneering technological developments such as dynamic in vivo imaging and high-resolution computerized tomography scanning now enable the integrated study of macroscale and microscale processes as a functional unit. Protein-imaging modalities are one way to optimize these technologies to further our understanding of the interplay between cells, tissues, and organs in development and disease of the postnatal skeleton. Fluorescent proteins are being increasingly used in diverse research applications ranging from living cells (16, 18–20, 36, 53, 81) to tissues and whole organisms (22, 50). One key advantage of protein labeling over genetically encoded protein tags such as GFP, which are not as bright or photostable (17), is that the approach can be readily applied across many different proteins, tissue types, and animal strains. Labeling methods (54, 77, 90) are particularly useful for tracking molecules through time and space after a systemic injection. Because of the wide range of fluorescent dyes and probes available, multicolor imaging can even be employed to track different proteins in the same live cells (30, 81). Here we developed methods for imaging the transport of fluorescently labeled, biologically active IGF-I (IGF-488) into skeletal growth plates of live, 5-wk-old mice. This approach is important for understanding mechanisms of bone elongation by enabling us to track delivery of physiologically relevant drugs and growth factors to the skeleton in real time.

Fluorescent protein labeling and validation.

Following a standard labeling protocol, we show that there are two critical additional steps needed before using the labeled protein conjugate in an experimental application (Fig. 1): 1) verify that free dye is removed from the sample, and 2) validate biological activity of the labeled conjugate. For example, we used identical protocols and reagents to conjugate IGF-I with Alexa Fluor 488 and Alexa Fluor 555 dyes in a pilot study. Although we had good results with the Alexa Fluor 488 dye, we obtained a very low degree of labeling with the Alexa Fluor 555 dye. Figure 2C demonstrates the importance of thin-layer chromatography to verify dye removal, as there was a large streak of unbound free dye remaining in the IGF-555 sample. If we had used that sample for in vivo imaging without the validation step, we would not be able to discriminate the fluorescence of the protein from that of excess free dye in the growth plate.

Likewise, it is essential to validate the biological activity of the protein conjugate since the labeling process can affect interactions of the protein with other binding proteins and/or receptors. For example, residues near the NH₂ terminus of the growth hormone (GH) molecule are important for binding to its receptor (41, 64). In a proof-of-concept experiment, we previously demonstrated that fluorescently labeled GH could be visualized in the growth plate in vivo (66); however, we do not know whether the fluorescent tag interfered with its ability to activate the GH receptor since we did not conduct bioactivity testing in that study. It would be essential to test its biological activity before conducting further experiments with labeled GH.

In vivo multiphoton imaging of IGF-488 in the growth plate.

In our experiments using labeled IGF-488, we observed patterns of fluorescence in the growth plate that were completely distinct from any of our other prior results using biologically inert dextrans. In all of our experiments using dextrans and small fluorescein dye, fluorescence in the growth plate was observed primarily in the matrix, and chondrocytes were evident only as dark, nonfluorescent shadows (66, 69). The opposite is seen here using biologically active IGF-488, where fluorescence is mostly restricted to the chondrocytes with little in the surrounding matrix (Figs. 4–6), suggesting receptor-mediated cellular localization. Interestingly, the pattern of IGF-488 fluorescence in the perichondrium was diffuse and punctate (Fig. 5), unlike anything we observed with biologically inert dyes. Previous studies have shown that IGF-I receptor expression is relatively low in perichondrium compared with growth plate, but IGF-binding proteins (IGFBPs) are up to 50-fold higher in the perichondrium (57). We hypothesize that the diffuse pattern of IGF-488 fluorescence in perichondrium may reflect IGFBP entrapment. These results are significant because they demonstrate, for the first time, IGF-I localization in perichondrium and growth plate chondrocytes in a live, intact animal.

These experiments open exciting possibilities for subsequent work using fluorescent labeling to track biomolecule transport into and around the growth plate under normal and abnormal conditions. Specifically, these methods could be useful in developing effective treatments for skeletal pathologies that are caused by defects in molecular transport. For example, the human skeletal disorder hereditary multiple exostoses syndrome is caused by dysregulated transport of Indian hedgehog (Ihh), a signaling molecule that helps regulate chondrocyte proliferation and differentiation in the growth plate (42). Another example involves the transport of myokines (muscle-derived growth factors) across the periosteum during bone healing (44). Tracking protein transport in vivo could be a major advancement toward understanding and treating a range of different bone diseases. Importantly, because the pattern and distribution of fluorescent IGF-I in the growth plate differed substantially from our previous studies using inert dextrans, this type of imaging approach will be needed to test and validate in vivo delivery platforms that utilize biologically active molecules.

Caveats and technical considerations.

Although the labeling approach described here can yield valuable information about growth plate dynamics in vivo and is broadly applicable to the study of many proteins and tissue types, there are limitations that should be considered. First, the approach must start with a purified protein sample. TFP esters used here (e.g., Alexa Fluor 488) bind nonselectively to the NH₂ terminus of any protein (45a). It is therefore critical that the sample does not contain serum and/or other binding proteins, which would also be labeled. As noted above, the NH₂ terminus of some proteins is involved in receptor binding (e.g., GH), so validating biological activity is essential after coupling it with a fluorescent tag. Also, the labeled protein will be slightly larger because of the addition of the fluorescent dye (643 Da for Alexa Fluor 488), and size should be considered in experiments involving steric effects.

Another important consideration for in vivo imaging is to ensure that the desired injection dose is within fluorescence detection limits in the tissue of interest, particularly with smaller proteins such as IGF-I with a dye-to-protein ratio of 1:1 (larger proteins typically have a higher degree of labeling and thus increased fluorescence intensity). This 1:1 degree of labeling is optimal for the small molecular weight of IGF-I because it renders an adequate signal without quenching effects caused by excess dye. We used a relatively low dose of fluorescently labeled IGF-488 (~6 µg/g body wt) because of its relevance to another ongoing study in our laboratory. This dose is within a range known to stimulate bone growth in mice (70), could be visualized in the growth plate, and did not have adverse effects after a single bolus injection. However, smaller doses might be more difficult to detect.

Finally, biological activity should be assessed within the context of the desired application since many circulating molecules interact with binding proteins and cofactors. For example, we used phosphorylation of the protein kinase Akt as a biomarker of IGF-I receptor activation because, of the multiple substrates activated when IGF-I binds to its receptor (39), the Akt pathway is known to be important for skeletal development and bone acquisition (58). We did not test whether the fluorescent tag affected the interaction of IGF-I with binding proteins that carry it in serum (IGFBPs; 34), and so we cannot eliminate the possibility that the labeling process might interfere with IGF-IGFBP interactions. However, there is structural evidence that IGFBPs interact with the same residues of IGF-I that are critical for receptor binding (26, 71), suggesting that IGFBP activity would also be preserved. Furthermore, our timed-entry results in the growth plate are consistent with the known half-life of circulating IGF-I when given at a similar dose (35). Since our results demonstrate that the labeled IGF-488 conjugate retained its ability to both activate the IGF-I signaling pathway (Fig. 3) and enter the growth plate in an expected time frame (Figs. 4–6), we concluded that the bioactivity of IGF-I was suitable for the purposes of our experiments.

Conclusions

Multiphoton imaging offers a unique opportunity to evaluate real-time drug and growth factor delivery to the skeleton at different concentrations and timing to optimize a dosing schedule. This paper describes and validates methods for imaging the transport of biologically active, fluorescently labeled IGF-I into growth plates of live mice and demonstrates, for the first time, IGF-I localization in perichondrium and growth plate chondrocytes in vivo. These results form important groundwork for future studies aimed at targeting and retaining therapeutics in growth plates with local specificity to potentially treat a wide range of skeletal pathologies.

GRANTS

Generous funding was provided by the Marshall University Appalachian Clinical and Translational Science Institute and National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant 1R15-AR-067451-01.

DISCLAIMERS

The content is solely the responsibility of the authors.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

M.A.S. conceived and designed research; M.A.S. and G.I. performed experiments; M.A.S. and G.I. analyzed data; M.A.S. interpreted results of experiments; M.A.S. prepared figures; M.A.S. and G.I. drafted manuscript; M.A.S. and G.I. edited and revised manuscript; M.A.S. and G.I. approved final version of manuscript.