Perineural invasion (PNI) is a common feature of head and neck squamous cell carcinoma (HNSCC), conferring lower survival rates. Its mechanisms are poorly understood. Utilizing neurites generated from murine dorsal root ganglia confined to a semisolid matrix, the pathways involved in the PNI of HNSCC cell lines can be investigated.
Perineural invasion (PNI) is found in approximately 40% of head and neck squamous cell carcinomas (HNSCC). Despite multimodal treatment with surgery, radiation, and chemotherapy, locoregional recurrences and distant metastases occur at higher rates, and overall survival is decreased by 40% compared to HNSCC without PNI. In vitro studies of the pathways involved in HNSCC PNI have historically been challenging given the lack of a consistent, reproducible assay. Described here is the adaptation of the dorsal root ganglion (DRG) assay for the examination of PNI in HNSCC. In this model, DRG are harvested from the spinal column of a sacrificed nude mouse and placed within a semisolid matrix. Over the subsequent days, neurites are generated and grow in a radial pattern from the cell bodies of the DRG. HNSCC cell lines are then placed peripherally around the matrix and invade preferentially along the neurites toward the DRG. This method allows for rapid evaluation of multiple treatment conditions, with very high assay success rates and reproducibility.
Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer in the US, with 10,000 deaths per year nationally and 300,000 deaths per year worldwide1. The overall prognosis for HNSCC has remained unchanged at 50% for the past several decades. Perineural invasion (PNI) is one of the most prominent pathological features that portend a poor prognosis in patients with HNSCC. Unfortunately, PNI is a frequent occurrence in HNSCC and can be found in up to 40% of HNSCC patients2,3.
PNI is the process by which malignant cells track along nerves to adjacent tissues, allowing for higher rates of local and distant spread. Accordingly, PNI-positive HNSCC tumors have higher rates of locoregional recurrences and distant metastases, resulting in lower overall survival compared to HNSCC patients without PNI4-8.
Although the treatment of patients with PNI is typically maximized by employing surgery, radiation, and chemotherapy, the overall survival rates of these patients are still decreased by up to 40% compared to patients without PNI9-11. Thus, it is clear that the current treatment modalities for HNSCC are ineffective in improving the adverse prognosis associated with PNI. The approach of developing targeted therapy against PNI in HNSCC has been hindered by the poor understanding of the factors that regulate this process. This is, in part, a consequence of the lack of a consistent in vitro model for the study of PNI in HNSCC.
In recent years, several groups have been utilizing an in vitro model for studying PNI in predominantly pancreatic and prostate cancers12-19. This model uses the neurites generated from dorsal root ganglia isolated from mice or rats as a surrogate for large-nerve invasion. The dorsal root ganglia are fixed in a factor-depleted semisolid matrix, which is a solubilized basement membrane protein mixture secreted by Engelbreth-Holm-Swarm mouse sarcoma cells. This matrix allows for the outgrowth of the neurites and the tracking of single cancer cells along these neurites. Described here is the adaption of this model for the examination of PNI in HNSCC.
1. Preparation of Culture Medium and Dishes (10 min)
2. Dissection of Murine DRG (45 min)
3. Preparation of Semisolid Matrix Droplets (< 1 min per plate)
4. Insertion of DRG into Semisolid Matrix Droplets (< 2 min per plate)
5. Preparation of the Head and Neck Cancer Cells
NOTE: Cell lines other than head and neck squamous cell carcinoma cells can be used in this experimental design.
6. Plating Head and Neck Cancer Cells
After the dissection of the DRG and the placement within the matrix droplet, the appearance of the assay should resemble Figure 1. Note that the DRG is not perfectly round, but it is centered within the matrix droplet. This allows for the outgrowth of neurites in 360 degrees, shown partially in Figure 2. Be aware that certain parts of the DRG send out neurites faster and in greater numbers than others, typically corresponding to where the efferent and afferent nerve branches entered and exited the DRG, respectively. We account for this and for size differences between DRGs by randomly plating the DRGs in groups of 4 and then randomly assigning a given plate to each cell condition.
As described above, we plate the HSNCC cell lines once the neurites have extended at least ¾ of the way to the edge of the matrix, which is typically on day 3. The added cells form a circumferential ring around the matrix (Figure 3a). We subsequently photograph the assays on day 4 (Figure 3b) and day 5 (Figure 3c). The cell line shown here (FaDu) demonstrates an above-average ability to track along neurites. The SQCCY1 cell line shown in Figure 4, however, shows little to no proclivity to invade the assays.
When using a new cell line, it is import to utilize several negative controls to examine how that particular cell line behaves around the assay. First, plate cells around a "blank" assay consisting of matrix alone (Figure 5). All cell lines that our lab has examined actively divide around the matrix but do not enter or extend over the top of the matrix. When excessive cells are left on top of the matrix, there can be significant growth over the matrix. This highlights the need to ensure that as many cells fall to the periphery of the matrix as possible, rather than being left to rest and divide on top of the matrix. This can be accomplished by gently tapping on the plates before the cells adhere or by pipetting small amounts of media directly on top of the matrix droplet once the cells have begun to adhere to the glass plate.
A second negative control, wherein the cells are plated on the same day that the DRG is placed within the matrix, can be run. The goal of this approach is to demonstrate that it is not a neurotropic attraction that drives the cells to the DRG, but rather that the presence of the neurites is mandatory. Furthermore, when tumor cells have resided along the periphery of the matrix for greater than 2 days, the matrix edge becomes indistinct. The matrix then begins to lift of off the glass bottom and can be found free-floating in the media shortly thereafter. For this reason, this protocol describes plating the HNSCC cells once the neurites have extended 75% of the distance to the edge of the matrix.
There are innumerable options for quantifying the results of what are visually very apparent differences between assays. In one such method, the images of the assays are divided into four quadrants with a vertical and a horizontal line (Figure 6). One point is assigned for every quadrant that has at least one string of PNI. If the PNI extends beyond 50% of the way from the edge of the matrix to the DRG, then 2 points are assigned instead. In this way, a score of 0 – 8 points can be assigned. The major advantage to this system is that assays that have too many PNI units to count can be rapidly assessed. A disadvantage is that is does not adequately express the degree of PNI (i.e., a quadrant with 1 neurite with invasion 100% of the distance to the DRG receives the same score (2) as a quadrant with 10 similarly-invaded neurites). Comparison of the brightfield and fluorescent images makes this system very easy, even with significant fibroblast efflux.
Sample scoring is shown in Figure 6. Using this four-quadrant scoring system for Figure 3, 0 points, 2 points, and 7 points were assigned for panels Figure 3a, Figure 3a, and Figure 3c, respectively. Figure 4 received a score of 2 points and 2 points in panels Figure 4a and Figure 4b, respectively. The data in Table 1 depicts the results of several experiments using the cell lines FaDu and SQCCY1. Note that even with a limited grading scale such as this, statistically-significant differences in the mean four-quadrant scores are readily obtained using the independent samples t-test.
Figure 1. DRG-matrix Assay. Correct placement of the dorsal root ganglion within the matrix droplet, shown with brightfield microscopy at 4X. Scale bar represents 1 mm. Please click here to view a larger version of this figure.
Figure 2. Outgrowth of Neurites. Neurites are absent on day 0 immediately after placing the DRG in the matrix droplet (A). By day 1 (B), neurite outgrowth is apparent at 10X on brightfield microscopy. Neurite growth continues on day 2 (C) and day 3 (D); however, note the efflux of the fibroblasts. Scale bar represents 0.1 mm. Please click here to view a larger version of this figure.
Figure 3. DRG-matrix Assay with FaDu. Head and neck squamous cell carcinoma line FaDu added peripherally around an assay on day 3 (A), shown in green fluorescent and brightfield microscopy at 4X Progressive neural invasion is seen on day 4 (B) and day 5 (C). Scale bar represents 1 mm. Please click here to view a larger version of this figure.
Figure 4. DRG-matrix Assay with SQCCY1. Head and neck squamous cell line SQCCY1 added on day 4 (A), shown in brightfield and green fluorescent microscopy at 4X. Note that this cell line is not as proficient at perineural invasion as FaDu, even by day 5 (B). Scale bar represents 1 mm. Please click here to view a larger version of this figure.
Figure 5. Matrix Assay with FaDu. Head and neck squamous cell carcinoma cell line FaDu added around a matrix droplet without a DRG on day 3. brightfield and green fluorescent microscopy (4X) shown on day 4 (A). Note that the tumor cells do not enter the matrix or grow over the top of it, even by day 5 (B). Scale bar represents 1 mm. Please click here to view a larger version of this figure.
Figure 6. Four-quadrant Method for Assay Quantification. One point is assigned for every quadrant with a PNI less than 50% of the distance from the edge of the matrix to the DRG. 2 points are assigned when the PNI extends beyond 50%. The first example (A) would receive a score of 5 (1 point for every quadrant for a PNI less than 50%, except the bottom right, which has a PNI extending beyond 50%, where 2 points are scored). The second example (B) is scored as an 8 (a PNI beyond 50% in all four quadrants). Scale bar represents 1 mm. Please click here to view a larger version of this figure.
Cell line | n | Day 4 | SD | P-value | Day 5 | SD | P-value |
FaDu | 24 | 2.88 | 1.42 | Ref | 6.04 | 0.35 | Ref |
SQCCY1 | 20 | 0.60 | 0.60 | <0.001 | 1.05 | 0.17 | <0.001 |
Table 1. Comparison of the PNI Between FaDu and SQCCY1. Mean four-quadrant scores as shown in Figure 6 between the cell lines FaDu and SQCCY1 at 24 h (day 4) and 48 h (day 5) after the cells were plated around the DRG-matrix assay. Comparisons are made using the independent sample t-test and presented with the standard deviation (SD) and P-values.
Critical Steps within the Protocol
The most important steps within this protocol are the precise dissection and extraction of the dorsal root ganglia. Proper transection of the spinal column and a midline-longitudinal division into two hemi-spines are critical to obtaining large numbers of DRG. During the dissection of individual DRGs, the ganglion should never be handled directly, but rather the surrounding fascia should be grasped with the microscopic forceps. Failure to do this will result in a crush injury of the DRG, which is likely the major cause of failure for neurite outgrowth. It is far better to under-trim the surrounding neural tissues during dissection than to risk over-handling the DRG and obtaining no neurite growth in the assay.
Modifications and Troubleshooting
The experimental protocol as described above reflects the optimal methodology established from a large number of procedural adjustments made over several years. Listed directly under the corresponding step in the protocol section is a number of notes and pitfalls that resulted from the experiences of the authors when using this methodology. Given the number of steps involved, there are indeed many modifications that can be made to this protocol by future investigators. As experience with these modifications grows, it is anticipated that additional means of troubleshooting will be developed according to the preferences of the investigating team.
Limitations of the Technique
There are three major limitations to this technique. The first is that very fine neurites are used as a surrogate for large neural invasion, which is what pathologists observe in tumor specimens. It is not known what the role of neurites, as opposed to neurons, are in vivo because identifying perineurite invasion is beyond the capabilities of a routine histopathological exam. Secondly, perineural invasion is a form of soft-tissue invasion that may not simply involve tumor cells and an adjacent neuron, as is simulated in this assay. As such, exogenous factors native to the in vitro tumor milieu are lacking in this assay. Finally, the time period during which PNI can be studied is limited to approximately 48 h due to the combination of fibroblast efflux and loss of matrix integrity. In this way, cell lines that are effective at neural invasion but are slow dividing and/or invading will be missed and presumed to not be proficient at PNI.
Significance of the Technique with Respect to Existing/Alternative Methods
To our knowledge, this is the only in vitro model currently available to examine PNI in HNSCC. Given the frequency with which PNI occurs in HNSCC and the limited knowledge of mechanisms that currently exists, this method is advantageous for several reasons. First, it has a very high success rate and excellent reproducibility. The total experiment time is also relatively short compared to similar protocols12,17-19. Finally, with the large number of assays that can be generated from a single mouse, many conditions can be tested with scientifically-appropriate replicates. The combination of these factors allows for the rapid assessment of many different conditions with consistent results.
At the same time, the high quality of the assay permits more detailed examination of the interaction between the tumor cells and the neurites. The use of live-cell imaging allows for the appreciation of the neurite outgrowth process and the HSNCC cell invasion in time-lapse video form. The addition of fluorescently-stained cells creates very clear differentiation between the cancer cells and background material, such as fibroblasts and dense neurite outgrowth (which auto-fluoresce red). Whether following this protocol or those described by others, this general experimental design is in its relative infancy, and there is far from a gold standard for the in vitro study of PNI. For these reasons, the more approaches to this assay that are described in detail, the greater the opportunity for the collaborative development of an ideal methodology.
Future Applications or Directions after Mastering This Technique
Just as there are several different approaches to setting up the assays, there are several approaches to measuring the results. In the above text, a single method for rapidly quantifying the degree of perineural invasion in each assay was presented. This is ideal for screening the impact of numerous different pathways on PNI, especially given that one can isolate 32 – 40 DRGs per mouse. There are a number of advanced imaging techniques to assess PNI, including the amount of fluorescence within a given area, the distance and rate of cell invasion, and the raw number of cells within the assay. Once the dynamic portion of the experiment is finished, the cells within the assay and/or the supernatant can also be collected for further molecular analysis.
This experimental methodology offers the opportunity to elucidate mechanisms involved in the PNI of HNSCC and other cancers. This knowledge can drive the development of therapeutics to target the pathways of PNI, which, at the present time, are lacking in HNSCC. Specific targeting of PNI when identified as an adverse pathological feature can potentially obviate the need for the non-specific adjuvant treatments, such as radiation therapy and chemotherapy, that are currently utilized.
The authors have nothing to disclose.
This work was supported in whole by funding from the NIH through the R21 grant, “Mechanisms of Perineural Invasion in Head and Neck Cancer” and the NCI T32 training grant, “Post-Doctoral Research Training in Head and Neck Oncology (2T32CA060397-21).” Thank you to Richard Steiman, MD, PhD and lab staff.
DMEM/F-12 50/50 Mix with L-glutamine & 15mM HEPES | Corning Cellgro | 10-090-CV | Manassas, VA |
Fetal bovine serum | Atlanta biologicals | S11150 | Flowery Branch, GA |
0.25% Trypsin-EDTA (1x) | Life Technologies Corporation | 25200056 | Grand Island, NY |
Phosphate buffered Saline 1x | Corning | 21-040-CM | Manassas, VA |
Matrigel hESC-Qualif Mouse | Corning Incorporated | 354277 | Bedford, MA |
Gamma Irradiated 35mm glass bottom culture dishes | MatTek Corporation | P35G-1.5-14-C | Ashland, MA |
SteREO Discovery.V8 Operating Microscope | Carl Zeiss Microimaging | 495015-0021-000 | Thornwood, NY |
Schott ACE I light source | Schott | A20500 | Germany |
CellTracker | Life Technologies Corporation | C2925 | Carlsbad, CA |
BD PrecisionGlide Needle 18G and 21G | BD | 305195 | Franklin Lakes, NJ |
Premium Microdissecting Tweezer | Harvard Apparatus | 60-3851 | Holliston, MA |
Premium Fine Operating Standard Scissors | Harvard Apparatus | 52-2789 | Holliston, MA |
Premium Spring Scissors | Harvard Apparatus | 60-3923 | Holliston, MA |
Dressing Forceps | Harvard Apparatus | 72-8949 | Holliston, MA |
Athymic nude mice (002019) | Jackson Laboratory | 002019 | Bar Harbor, ME |