2.1.

QDs

For this work, nontargeted methoxy-PEGylated $\left(5000\text{-}\mathrm{MW}\right)$ $655\text{-}\mathrm{nm}$ QDs (maximum fluorescence emission $655\mathrm{nm}$ ) were supplied by Quantum Dot Corporation (Hayward, California). Immediately prior to injection, QDs were sonicated for $10\min$ in an ice bath to break up any aggregates that might have formed and mixed $30∕70$ (vol/vol) with saline to a working concentration of $0.6\mu \mathrm{M}$ . In preliminary studies comparing the fluorescent signal from equimolar solutions of 565- and $655\text{-}\mathrm{nm}$ mPEG-5000 QDs, we found the larger CdSe-ZnS core-shell nanoparticles to be the “brighter” of the two when excited at $900\mathrm{nm}$ . The larger QDs with emission maximum at $655\mathrm{nm}$ also had the advantage of being further red-shifted from the skeletal muscle tissue autofluorescence background signal.

2.2.

Animal and Extensor Digitorum Longus Muscle Preparation

Male CD1 mice, weighing $30\text{to}35\text{grams}$ , were anesthetized by an intraperitoneal injection of ketamine $(112.5\mathrm{mg}∕\mathrm{kg})$ and xylazine $(15\mathrm{mg}∕\mathrm{kg})$ and placed on a $37°\mathrm{C}$ heating pad. The left hind limb was shaved and surface hair removed as it interfered with fluorescent imaging. The extensor digitorum longus (EDL) skeletal muscle was blunt dissected and severed at the tendon. The animal was placed in the recovery position with its leg extended into a WillCo-glass-bottom dish (Amsterdam, The Netherlands), for imaging using an inverted microscope, and the muscle held under tension by a suture tied to the EDL tendon. The muscle was bathed in warm saline, maneuvered into the optical beam path, and covered by a piece of Saran Wrap, which acted as an oxygen permeability barrier. Then 60 pmols of QDs were injected into the circulation via the tail vein. Approximating the mouse blood volume at $2\mathrm{mL}$ , we estimated the initial QD circulating concentration to be $20\mathrm{nM}$ . The animal was then covered with a heating pad to maintain internal core temperature. To assess the effect of an endotoxemic response on QD clearance, a mouse model of bacterial infection was used where gram negative bacterial cell wall lipopolysaccharide (LPS) was mixed with saline ( $20\mathrm{mg}∕\mathrm{kg}$ , Escherichia coli 0111:B4 LPS, Sigma, Saint Louis, Missouri) and injected intraperitoneally. Since the microvascular response to LPS takes several hours to develop, QDs were injected $3.5\mathrm{h}$ after LPS injection. All procedures were approved by the Animal Care Committee at the University of British Columbia.

2.3.

Multiphoton Microscopy

To facilitate in vivo microvascular imaging in the highly light absorbing and scattering skeletal muscle tissue, we used a titanium sapphire mode-locked $100\text{-}\mathrm{fs}$ pulse laser (Spectra Physics, Mountain View, California), with a $80\text{-}\mathrm{MHz}$ repetition rate, tuned to $900\mathrm{nm}$ and focused through a $20\times ∕0.7$ numerical aperture (NA) water-immersion objective (AOBS SP2 TCS laser scanning confocal microscope, Leica Microsystems, Heidelberg GmbH, Mannheim). The power out of the objective was $30\mathrm{mW}$ ; however, the power at the focal plane within the muscle was unknown. Fluorescent images were acquired at $400\mathrm{Hz}$ using either single-frame acquisition or three frame averaging. Optical sectioning was performed using Nyquist sampling. In time course experiments to determine QD circulating half-life, imaging parameters were adopted on the fly to optimize the dynamic range of the initial fluorescence reading. Larger microvessels near the surface of the skeletal muscle were chosen to follow the in vivo QD elimination as they were the easiest vessels to find quickly following injection and provided good imaging targets of not only intravascular fluorescence, but also of local QD accumulation near microvessel bifurcations and for tracking of extravasating QD aggregates. To determine what multiphoton excitation wavelength would best minimize endogenous background autofluorescence in skeletal muscle, tissue was excited over a range of wavelengths from $730\text{to}900\mathrm{nm}$ and spectral emission scans with $10\text{-}\mathrm{nm}$ bandwidths were collected from $450\text{to}625\mathrm{nm}$ .

2.4.

Analysis of In Vivo QD Elimination

To quantify QD elimination, fluorescence intensity was measured in defined regions of interest within microvessels and the surrounding tissue and normalized to baseline. Intensity values were corrected for background, which was determined in the extravascular space surrounding the blood vessels. Visual inspection of the fluorescent time-course profiles indicated that QDs were eliminated from LPS-treated mice much more rapidly than controls. To quantify in vivo circulating kinetics we determined elimination rate constants (Kel) and circulating QD half-lives $\left(t_{1∕2}\right)$ in three control and three LPS treated mice. QD clearance was modeled as a first order elimination reaction where

While we found no increase in background fluorescent signal in general, we did observe some evidence of QD aggregates accumulating in and extravasating from the skeletal muscle microcirculation of LPS-treated mice. Since LPS is known to activate components of the reticuloendothelial and immune system including macrophages (liver Kupffer cells and circulating monocytes) and leukocytes that roll and adhere to the microvascular endothelium, we suspect that these cells were accumulating QDs in vivo. To assess whether activated leukocytes could nonspecifically accumulate QDs, we isolated mouse leukocytes from whole blood using 7% dextran ( $\mathrm{100,000}\text{to}\mathrm{200,000}\mathrm{kDa}$ ; Sigma, Ontario) ( $1∕1$ vol/vol). Washed leukocytes were resuspended in RPMI-1640 medium (Invitrogen, Ontario) and incubated at $37°\mathrm{C}∕5\%$ $\mathrm{C}{\mathrm{O}}_2$ for $3\mathrm{h}$ with TNFalpha $(200\mathrm{ng}∕\mathrm{ml})$ in the presence of QDs at the estimated $20\mathrm{nM}$ in vivo concentration used for microvascular imaging. TNFalpha is a proinflammatory mediator that activates leukocytes and macrophages. Cells were fixed using 4% paraformaldehyde and analyzed using flow cytometry (Epics XL-MCL, Beckman Coulter, Fullerton, California).

3.1.

Near IR Excitation Reduces Background Skeletal Muscle Autofluorescence

Tissue autofluorescence arises when a variety of intrinsic molecules including NAD(P)H, flavins, retinol, tryptophan, and its indolamine derivatives are excited by UV or near-IR light.¹⁶ To minimize tissue autofluorescence and maximize microvessel fluorescence contrast, skeletal muscle was excited in the IR at $900\mathrm{nm}$ [Fig. 1a versus Fig. 1b]. We found that exciting the tissue at increasingly longer wavelengths, from $730\text{to}900\mathrm{nm}$ , resulted in a reduction in peak autofluroescent signal at $500\mathrm{nm}$ [Fig. 1c]. Moving to longer excitation wavelengths was also advantageous for imaging skeletal muscle because it enabled deeper tissue penetration with less photon scatter. Since the autofluorescence signal ranged from below our cutoff emission wavelength of $450\mathrm{nm}$ to approximately $600\mathrm{nm}$ , which overlapped with the emission spectrum of smaller and less “bright” nontargeted $565\text{-}\mathrm{nm}$ QDs, we opted to use the brighter $655\text{-}\mathrm{nm}$ QDs, which have an emission maximum at $655\mathrm{nm}$ and a relatively narrow emission bandwidth from $600\text{to}700\mathrm{nm}$ .

Fig. 1

Multiphoton excitation wavelength dependence of mouse hind limb skeletal muscle autofluorescence. (a) Tissue autofluorescence when excited at $900\mathrm{nm}$ compared to (b) the same tissue excited at $730\mathrm{nm}$ , $20\times ∕0.7$ NA water-immersion objective with emission bandwidth $450\text{to}600\mathrm{nm}$ ; power was $30\mathrm{mW}$ out of the objective. (c) Spectral profiles obtained from an imaging series ranging from tissue excitation at $900\text{to}730\mathrm{nm}$ with $10\mathrm{nm}$ emission bandwidths. To test for possible photobleaching, images were reacquired at 750, 800, and $900\mathrm{nm}$ . The scale bar is $20\mu \mathrm{m}$ .

3.2.

In Vivo QD Fluorescent Contrast Imaging

Our first question regarding in vivo QD behavior was how they were distributed within the blood. We found QDs to occupy the plasma space between red blood cells without being incorporated into red blood cells themselves. Figure 2 shows rapid $\left(4000\text{-}\mathrm{Hz}\right)$ single-frame high-resolution images of red blood cells flowing through a $4\text{to}5\text{-}\mu \mathrm{m}$ -diam capillary and a relatively larger $15\text{-}\mu \mathrm{m}$ -diam feeding arteriole. Dark nonfluorescent red blood cells can clearly be distinguished from fluorescent plasma gaps containing QDs, which provide good imaging contrast between plasma, erythrocytes, and background tissue. We found no evidence that QDs were taken up by erythrocytes over the time course of this study as erythrocytes remained nonfluorescent; however, the slight speckle on some erythrocytes seen in high-resolution images suggested that there may have been some minor adsorption to the outer membrane. A thin fluorescent film was detected between the boundary of the erythrocyte membrane and the inner microvessel wall (Fig. 2a)].

Fig. 2

Optical sections of skeletal muscle microvessels. Multiphoton rapid scans $\left(4000\mathrm{Hz}\right)$ of (a) a single capillary and (b) microvascular arterioles using an upright laser scanning confocal microscope configuration show that QDs are confined to the plasma space between red blood cells ( $900\text{-}\mathrm{nm}$ excitation, fluorescent emission $600\text{to}700\mathrm{nm}$ , $40\times ∕0.8$ NA dipping objective). In (a) we see a train of red blood cells flowing through a capillary; red blood cells (arrows A, B, and C) are nonfluorescent and detected by contrast with the fluorescent QDs filled plasma gaps (arrow D). The characteristic red blood cell “tail” is indicative of red blood cell deformation as it passes through a capillary diameter smaller than its own diameter. A thin fluorescent layer can be seen between the red blood cell and the capillary cell wall (arrow E). In larger diameter arterioles (b) red blood cells retain their discoid shape (arrow G). Arrow F indicates fluorescent plasma. The scale bars are 8 and $16\mu \mathrm{m}$ , respectively.

3.3.

In Vivo Microvascular Imaging Using QDs and Multiphoton Microscopy

Since QDs provided good microvascular imaging contrast, we next evaluated their ability to facilitate deep tissue mapping of the functional microcirculation. To do this, we used multiphoton microscopy $\left(900\text{-}\mathrm{nm}\right)$ optical sectioning and three-frame averaging to obtain images of the mouse skeletal muscle microcirculation to depths of $150\text{to}200\mu \mathrm{m}$ . Figures 3a, 3b, 3c, 3d depict a reconstructed pseudocolored (for depth) composite image of a complete skeletal muscle microvascular unit, comprising feeding arteriole, capillary network, and collecting venule. Single-frame-reconstructed 3-D black and white images [rotated to show top and side views of Fig. 3d] show larger dark (fluorescent) feeding arterioles branching off to individual capillaries, which run perpendicular to the feeding arterioles and parallel to muscle fibers (not shown in the image). In individual capillaries, fluorescent plasma gaps appear as dark objects, while white spaces indicate the presence of nonfluorescent red blood cells. Accordingly, the single-frame image provides static information on capillary red cell lineal density (LD, RBC/mm capillary length) and maps the 3-D microvascular architecture; however, it does not provide any dynamic information. To obtain dynamic information from a single capillary, individual line scans can be performed to generate space-time images^{5, 17} (STIs), which provide hemodynamic information on red cell velocity and lineal density. An example is shown in Fig. 3e.

Fig. 3

In vivo three-frame average composite multiphoton image of the skeletal muscle microcirculation using fluorescent $655\text{-}\mathrm{nm}$ QDs with emission range of $600\text{to}700\mathrm{nm}$ . Excitation at $900\mathrm{nm}$ was delivered through a $20\times ∕0.7$ NA water-immersion objective lens with $30\mathrm{mW}$ out of the objective. The capillary network is orientated within the dashed lines from 7 to 2 o’clock. In the centre of the image, a deep collecting venule (V) is running beneath the capillary network (CN), which is supplied by feeding arterioles (FA) on the left and drained by a collecting venule on the right. Color bar indicates depth. Scale bar is $20\mu \mathrm{m}$ . The enlarged panel (d) distinguishes between flowing capillaries with continuous fluorescent signal (arrow B) and stopped-flow capillaries with a punctate fluorescent signal (arrow A). Different perspectives of a single frame image from panel (d) show both fluorescent plasma gaps (dark objects, liquid phase) and nonfluorescent red blood cells (bright spaces, solid phase), indicative of the two phase nature of capillary blood flow. (e) Space time image (STI) generated from a line scan (milliseconds per line) across a single capillary, where $x$ and $t$ are the spatial and time dimensions, respectively. Red cell velocity can be calculated from the slope of the dark bands, while red cell lineal density can be determined from the number of red cells (dark bands) at a given time.

In contrast to the single-frame image, acquiring microvascular images using three-frame averaging (images were acquired at $400\mathrm{Hz}$ using bidirectional scan mode with off-line phase correction) identifies putative functional or flowing capillaries as the temporal and spatial variation in fluorescent plasma gaps results in a continuous fluorescent signal along the length of the capillary. Note that stopped-flow vessels, which increase in various disease states, generate a discontinuous or punctate fluorescent signal when acquired as a three-frame average because stationary nonfluorescent red blood cells produce nonfluorescent gaps in the fluorescent signal. By comparison, vessels with only plasma flow produce a continuous fluorescent signal; however, these vessels can not be deemed “functional” since they do not contain erythrocytes. In the panel in Fig. 3d, enlarged, we see evidence of both continuous-flow (continuous fluorescent signal along the capillary) and stopped-flow capillaries (discontinuous fluorescent signal along the capillary). To determine microvascular functionality, one must assess the three-frame average image relative to the single-frame image. Comparing the nature of the fluorescent signals in three-frame and single-frame images [Fig. 3d], we determined that most microvessels were in fact functional, that is they showed both a continuous fluorescent signal and the presence of red blood cells.

3.4.

In Vivo QD Stability for Microvascular Imaging

During our initial imaging studies of the inflamed microcirculation, we were surprised to find an absence of QD fluorescent signal $5\mathrm{h}$ after LPS injection. To investigate the disappearance of QDs from the circulation, we performed in vivo kinetic experiments by sampling the fluorescent signal in single microvessels ( $20\text{to}30\text{-}\mu \mathrm{m}$ diameter) over time. Figure 4 shows the time course of mPEGylated $\left(5000\text{-}\mathrm{MW}\right)$ QD $655\text{-}\mathrm{nm}$ clearance from the circulation. We found the elimination rate constant (Kel) to be almost 7 times greater in the LPS treated mice compared to control ( $1.6\pm 0.06{\mathrm{h}}^{-1}$ in LPS versus $0.23\pm 0.05{\mathrm{h}}^{-1}$ in control, $P<0.05$ ). The increased Kel in LPS treated mice was reflected by a significant decrease in circulating QD half-life compared to control ( $27\pm 2\min$ in LPS versus $185\pm 36\min$ in control, $P<0.05$ ). Since the normalized QD signal in inflamed mice decreased to less than 10% by $90\min$ , the rapid clearance of QDs offers an explanation for why no fluorescent signal was detected by $5\mathrm{h}$ .

Fig. 4

QD clearance from mouse circulation: (a) example of surface skeletal muscle microvessels used to follow the in vivo time course of QD clearance form the circulation. Microvessels were imaged using multiphoton microscopy ( $900\text{-}\mathrm{nm}$ excitation, $63\times ∕1.2$ NA water-immersion objective, $600\text{to}700\text{-}\mathrm{nm}$ emission fluorescence). Circles indicate regions of interest for intravascular fluorescence and background measurements. (b) The time course of QD elimination for control and LPS treated mice. The scale bar is $20\mu \mathrm{m}$ .

While we confirmed ex vivo that QDs accumulate in a number of tissues including the liver, spleen, kidney, and lung (Fig. 5 ), we also observed the in vivo accumulation of QD aggregates at microvascular bifurcations in skeletal muscle (Fig. 6 ). Additionally, we observed the extravasation of QD aggregates from the microcirculation (Fig. 6). To assess whether activated leukocytes might enhance nonspecific QD uptake and accumulation, we isolated white blood cells and treated them with TNFalpha in the presence of QDs. Flow cytometry analysis revealed that activated monocytes (circulating phagocytes) were more likely to take up QDs than nonactivated monocytes (Table 1 ).

Fig. 5

Accumulation of QDs in mouse liver, lung, spleen, and kidney $5\mathrm{h}$ after treatment with LPS. Organs were removed and imaged in cross section using high-resolution multiphoton imaging ( $900\text{-}\mathrm{nm}$ excitation, $63\times ∕1.2$ NA water-immersion objective, $600\text{to}700\text{-}\mathrm{nm}$ emission fluorescence). While QD fluorescence was virtually eliminated from the circulation, we found evidence of QD accumulation in all organs examined. The color bar indicates fluorescent intensity and the Scale bar is $8\mu \mathrm{m}$ .

Fig. 6

QD accumulation at a microvascular bifurcation and extravasation from the skeletal muscle microcirculation in LPS-treated mice. QDs were imaged using multiphoton microscopy ( $900\text{-}\mathrm{nm}$ excitation, $63\times ∕1.2$ NA water-immersion objective, $600\text{to}700\text{-}\mathrm{nm}$ emission fluorescence). QD aggregates can be seen to accumulate at multiple locations near the microvascular bifurcation (arrow A) and extravasate from the microcirculation (arrow B). The color bar indicates fluorescence intensity and the scale bar is $20\mu \mathrm{m}$ .

Table 1

Percent of isolated leukocytes that accumulate QDs.