2.1 Participants

The study was approved by the Institutional Review Board at the University of Michigan Medical School. After providing written informed consent, healthy lightly pigmented volunteers (60–88 years-old, 13 females, 21 males) were administered 4 injections of CL-HA (Restylane, Galderma, Fort Worth, TX) into the dorsal skin of one randomly assigned forearm and 4 injections of vehicle (0.9% NaCl in sterile water) into the other. Injections (each 0.5 mL) were intradermal, as typically performed when injecting into the nasolabial folds, and spaced 2–4 cm apart. Using a clear plastic template to record their locations, injection sites were located reliably and punch biopsies (4 mm) were obtained under local anaesthesia (lidocaine) at 1, 2, 3 and 4 weeks post-injection for an early group of subjects (15/16 completed the study), and 3, 6, 9 and 12 months for a late group (16/18 completed the study). Specimens were stored in optimum cutting temperature (OCT) medium at −80°C for later analysis. Only participants who completed the study were included in the analysis. There were no major adverse effects.

2.2 Immunostaining

As previously described,^{21, 29-31} OCT-embedded sections were stained for heat shock protein 47 (HSP47, Stressgen Biotechnologies, Victoria, Canada) and HA (Calbiochem, Burlington, MA). Staining for the latter was performed using a biotinylated HA binding protein, which according to the manufacturer, is derived from bovine nasal cartilage, binds specifically and strongly to HA, and is composed of two binding polypeptides derived from N-terminal regions of the HA binding proteoglycan and linkage proteins.²² Quantification of staining and dermal area occupied by injected CL-HA was analysed using Image-Pro Plus v4.1 (Media Cybernetics, Silver Spring, MD).

2.3 Multiphoton microscopy

As described previously,³¹ multiphoton microscopy was used to perform second-harmonic generation (SHG) and immunofluorescence imaging during separate scans. Briefly, skin sections (up to 100 μm) were incubated with HSP47 monoclonal antibody (Enzo, Farmingdale, NY) or HA binding protein (Calbiochem, Burlington, MA), followed by anti-mouse or streptavidin secondary antibodies conjugated with Alexa Fluor 488 (2 μg/mL for 45 min; Jackson ImmunoResearch, West Grove, PA). Fluorescent emissions were detected by a HyD detector between 440–470 nm (autofluorescence), 490–535 nm (Alexa Fluor 488) and 565–605 nm (SYTOX Orange for staining of cell nuclei, Molecular Probes, Eugene, OR). SHG emissions were detected through an externally mounted 410/10 or 483/32 bandwidth filter. Forward emissions were detected by an externally mounted forward-directed photomultiplier tube. Images were analysed using ImageJ v.1.49t (http://imagej.nih.gov/ij/), 3-D Gaussian blurred, autofluorescence subtracted, and projected using maximum projection or + 1 SD projections. Three dimensional renders were generated with FluoRender (v.2.25, University of Utah, Salt Lake City, UT).

2.4 Gene expression

As described previously,^31-34 quantitative polymerase chain reaction (PCR) was performed.

2.5 Statistical analysis

Data are displayed as means + SEM. When appropriate logarithmic transformation of data was performed to achieve normality. Comparisons were assessed at each timepoint using paired, two-sided t-tests. An overall α-level of 0.05 was used to determine statistical significance.

3.1 Increased fibroblast synthetic activity as early as 1 week post-filler injection

To investigate when fibroblasts become synthetically active following CL-HA injection, we measured protein expression (immunostaining) of a specific marker of dermal collagen synthesis, heat shock protein 47 (HSP47),³⁵ which is expressed in fibroblasts actively producing type I procollagen. In vehicle-injected skin, dermal staining was minimal at 1, 2, 3 and 4 weeks post-injection (all n = 15, Figure 1A). In CL-HA-injected skin, we noticed droplets/pools of injected filler localized to discrete pockets within the mid-to-lower dermis (Figure 1A). Around these pockets, staining occurred within fibroblasts, beginning at 1 week post-injection and appearing more intense and abundant at subsequent weeks. These fibroblasts were mostly embedded within the dermal ECM surrounding pools of injected CL-HA, but not directly contacting the filler material. Based on image analysis, staining in filler-injected skin was significantly increased 11.3 ± 4.8-fold at 1 week, and remained increased at 2, 3 and 4 weeks (28.4 ± 10.3−, 19.9 ± 5.1−, and 28.7 ± 10.6-fold, respectively), compared with vehicle-injected skin (all p < 0.05, n = 15, Figure 1B).

Next, we measured HSP47 gene expression, which exhibited similar changes. In filler-injected skin, expression increased significantly at 1 week (1.3 ± 0.1-fold, p < 0.05, n = 14), compared with vehicle-injected skin (Figure 1C). Gene expression of HSP47 continued increasing at 2, 3 and 4 weeks (2.0 ± 0.3−, 2.8 ± 0.5− and 3.8 ± 1.0-fold, respectively; all p < 0.05, n = 14). These data indicate that fibroblast activation begins rapidly after filler injection (as early as 1 week).

3.2 Elongation of fibroblasts, activation of the TGF-β pathway and increased type I procollagen synthesis as early as 1 week post-filler injection

To investigate mechanisms of rapid fibroblast stimulation following CL-HA injection, we performed immunofluorescence staining of HSP47 and examined the morphology of positively stained fibroblasts at high magnification. In vehicle-injected skin at 1, 2, 3 and 4 weeks, dermal staining was minimal and localized to small fibroblasts displaying a contracted/collapsed appearance, indicative of reduced mechanical support in the ECM (Figure 2A). In CL-HA-injected skin, particularly around pockets containing the injected filler, staining occurred as early as 1 week, and was robust at 4 weeks post-injection (Figure 2A). Staining was localized to large fibroblasts displaying an elongated/spread morphology, indicative of enhanced dermal mechanical support.

As transforming growth factor (TGF)-β is a cytokine required for type I procollagen synthesis,^{33, 34, 36, 37} we next measured gene expression of components of the TGF-β pathway (Figure 2B). TGF-β binds to cell-surface receptors on fibroblasts,³⁸ and results in production of connective tissue growth factor/cellular communication network factor 2 (CTGF/CCN2), a growth factor and marker of TGF-β pathway activation that participates in upregulation of type I procollagen production.³²

We found that two isoforms of TGF-β were increased at 1 and 2 weeks post-CL-HA injection, including TGF-β1 (1.6 ± 0.3-fold) and TGF-β3 (2.5 ± 0.7-fold), respectively (both p ≤ 0.05, n = 11). Expression of CTGF/CCN2 increased significantly at 1 week (1.4 ± 0.2-fold), and remained elevated at subsequent weeks (all p < 0.05, n = 14). Expression of the type I and type II receptors for TGF-β was elevated as early as 2 and 3 weeks, respectively (1.6 ± 0.3− and 1.6 ± 0.3-fold; both p ≤ 0.05, n = 14).

Additionally, we examined gene expression of type I procollagen, which showed significant elevation (1.8 ± 0.3-fold) at 1 week, and continued increasing at 2, 3 and 4 weeks (7.9 ± 2.1−, 13.5 ± 4.6− and 14.6 ± 2.9-fold, respectively; all p < 0.05, n = 14, Figure 2B). These data indicate that injection of CL-HA rapidly enhances mechanical support in the dermal ECM and activates the TGF-β pathway, stimulating type I procollagen synthesis.

3.3 Deposition of thick collagen bundles as early as 4 weeks post-CL-HA injection

To investigate whether early fibroblast elongation and synthetic activation following CL-HA-injection lead to deposition of collagen bundles in the dermal ECM, we measured gene expression of enzymes required for proper assembly of type I collagen, including procollagen N-proteinase and C-proteinase.³⁹ Following CL-HA-injection, expression of both enzymes increased significantly at 1 week, and remained elevated at subsequent weeks, compared with vehicle-injected skin (all p < 0.05, n = 14, Figure 3A).

Next, we performed SHG imaging, which provides high-resolution, three-dimensional images of collagen based on its inherent optical properties.⁴⁰ As type I collagen is the most abundant type of collagen in the dermal ECM, SHG imaging predominantly displays the structure and organization of this type of collagen.

In vehicle-injected skin, collagen bundles appeared shortened/fragmented, thin, loosely packed and separated from each other (Figure 3B). As early as 4 weeks in filler-injected skin, collagen bundles appeared thick and densely packed, particularly around dermal pockets containing injected filler (Figure 3B). Compared with vehicle-injected skin, these bundles appeared bright, suggesting that they were comprised of compactly arranged type I collagen fibrils. These data indicate that the enzymatic machinery needed for proper assembly of intact type I collagen is increased as early as 1 week post-filler injection, and thick, densely packed collagen bundles accumulate in the ECM by 4 weeks.

3.4 Prolonged fibroblast elongation and synthetic activity post-filler injection

To investigate the duration of fibroblast elongation and activation following CL-HA injection, we performed HSP47 immunostaining at late timepoints. In vehicle-injected skin, we observed minimal dermal staining at 3, 6, 9 and 12 months post-injection (Figure S1A). In CL-HA-injected skin, we continued to observe dermal pockets containing the injected filler material (Figure S1A). Embedded within the ECM surrounding these pockets, numerous fibroblasts remained intensely stained and markedly elongated, especially at 3 and 6 months post-injection. Image analysis revealed that staining (protein expression) remained significantly elevated at these timepoints (13.6 ± 5.9− and 13.2 ± 4.7-fold, respectively; both p < 0.05, n = 8), attenuating by 9 months (Figure S1B).

To complement these findings, we measured HSP47 gene expression, which exhibited similar changes and remained significantly increased as long as 6 months post-CL-HA injection (2.2 ± 0.5-fold, p < 0.05, n = 9), attenuating by 9 months (Figure S1C). At all late timepoints, gene expression of collagen-fragmenting MMPs was barely detectable (near the limit of detection, data not shown). These data indicate that fibroblast elongation and activation remain prolonged following CL-HA injection.

3.5 Prolonged activation of the TGF-β pathway and type I procollagen synthesis following CL-HA injection

To further investigate mechanisms promoting prolonged fibroblast activation at late timepoints following CL-HA injection, we measured gene expression of TGF-β pathway components. Expression of TGF-β1 and CTGF/CCN2 remained significantly elevated as long as 3 months post-filler injection (1.5 ± 0.2- and 1.6 ± 0.3-fold, respectively), while expression of TGF-β3 remained elevated as long as 6 months (1.9 ± 0.4-fold) (all p ≤ 0.05, n = 9, Figure S1C). Gene expression of type I procollagen was increased at 3 and 6 months (10.0 ± 2.3− and 8.1 ± 2.3-fold, respectively), and remained significantly elevated at 9 months (2.3 ± 0.5-fold) (all p < 0.05, n = 9, Figure S1C). Gene expression of procollagen N-proteinase exhibited similar changes (Figure S1C). These data indicate that prolonged fibroblast activation following CL-HA injection is associated with stimulation of the TGF-β pathway, type I procollagen, and enzymes required for assembly of type I collagen.

3.6 Accumulation of collagen bundles for at least 1 year following CL-HA injection

To investigate whether collagen bundles accumulate over time following CL-HA injection, we performed high-powered microscopy at 12 months post-injection. We used SHG imaging to examine the organization of dermal collagen bundles, coupled with immunofluorescence staining to localize injected CL-HA. In vehicle-injected skin, collagen bundles appeared fragmented, thin and loosely packed (Figure 4A). Immunofluorescence staining revealed the presence of endogenous (natural) HA, which appeared grainy and dot-like throughout the dermal ECM (Figure 4B,C).

In filler-injected skin at 12 months, SHG imaging revealed numerous bright collagen bundles, which appeared long, thick and densely packed in the ECM, particularly around discrete dermal pockets (Figure 4A). Within these pockets, immunofluorescence staining demonstrated substantial amounts of injected filler appearing as droplets (Figure 4B,C). These findings are clearly shown using 3-D video rendering of images (Video S1).

We quantified the amount of space (area) in the dermis occupied by the filler. Compared with 3 months post-injection, the filler occupied just as much space in the dermis at 6, 9 or 12 months (Figure 4D). These observations indicate that collagen bundles accumulate stably for at least 1 year following filler injection, during which time substantial amounts of the filler remain physically present.