Generation of MRMT1 cells expressing reporter proteins was achieved by genomic integration of the Luc2-copGFP reporter gene construct using lentiviral transduction. The lentiviral vector contains the EF1α promoter sequence driving the equimolar expression of a codon-optimized firefly luciferase (Luc2) gene and a copepod green fluorescent protein (copGFP) thanks to the presence of a T2A sequence. Vector production and cell transduction were performed under appropriate biosafety level conditions (ML-II) in accordance with the National Biosafety Guidelines and Regulations for Research on Genetically Modified Organisms. Procedures and protocols were reviewed and approved by the LUMC Biosafety Committee (GMO permit 08-129). Vector production and transduction procedures have been described in detail elsewhere ¹⁰ . In brief, cells were seeded in 24-well plate at a cell density of 7.5×10 ⁴ cells/well and maintained in a humidified incubator at 37°C and 5% CO ₂. After attachment was accomplished, the cells were transduced using a MOI (multiplicity of infection) of 1. Subsequently cells were FACS sorted for high expression of copGFP and cell culture was expanded for experiments.

Syngeneic MRMT1-Luc2 rat mammary gland carcinoma cells (Leiden University medical Center, The Nederlands) were cultured in RPMI 1640 medium without phenol red (Invitrogen, Paisley, UK/Nærum, Denmark) supplemented with 10% heat-inactivated foetal bovine serum, 1% L-glutamine and 2% penicillin/streptomycin (Invitrogen, Paisley, UK/Nærum, Denmark). On the day of surgery, MRMT1-Luc2 carcinoma cells were released by brief exposure to 0.1% w/v trypsin-EDTA and centrifuged in medium for 3 min at 1200rpm. The pellet was washed twice with Hanks’ balanced salt solution (HBSS) without calcium, magnesium or phenol red (Invitrogen, Paisley, UK/Nærum, Denmark) and centrifuged for 3 min at 1200rpm. Cells were re-suspended in HBSS and kept on ice until use.

The inoculation of carcinoma cells was performed modified from 11 as previously described ¹² . Hence, following induction of isoflurane anaesthesia (induction 4%, maintenance 1.5–2%) the animal was placed with the abdominal side up. A small incision was made in a shaved and disinfected area on the anterior-medial surface of the tibia and the tibia carefully exposed. A hole was made in the tibia with a 0,7mm dental drill through which a thin polyethylene tube was fed 1cm into the proximal intramedullary cavity. Animals were injected with 5×10 ³ MRMT1-Luc2 carcinoma cells in 10µl HBSS. Following removal of the tube, the hole was plugged with bone restorative material (IRM, Dentsply, Surrey, UK/Vallensbæk, Denmark). The wound was irrigated with saline and closed with two metal clips. The animals were placed under a heat source until fully awake. Postoperative analgesia was administrated by injection of Rimadyl (s.c. 5mg/kg, Pfizer, Denmark) and application of 2% Lidocaine gel (AstraZeneca, Denmark) to the wound.

Behavioral responses were assessed prior to surgery and on day 7, 10, 14, 17, 21 and 23. All animals were introduced to all behavioral tests 2–3 times prior to the start of the experiment. Animals reaching humane endpoint before day 23 (n=7) (predefined as limb use score 0) were euthanized by decapitation following brief exposure to isoflurane, and the behavioral scores from the last day carried forward for data analysis.

Von Frey test. Mechanical hypersensitivity, detected by von Frey filaments (North Coast Medicinal, Inc., Camino Arroyo, Gilroy, CA, USA), was used as an indicator of early pain behavior. The threshold was determined by the up and down method, as described previously ¹³ . Briefly, rats were placed in acrylic enclosures on a wire mesh floor and allowed to acclimatize for minimum 30 min. Starting at 6 g, filaments ranging from 0.4–26 g were applied to the plantar surface of the hind paw with a minimum of 3 min intervals between two stimulations. A stimulus was recorded as positive if paw withdrawal was observed within 3 s of stimulation with a given filament; if no response was observed it was recorded as negative. Following a positive response, the next stimulation was performed with a filament with a decreased bending strength, whereas a negative response was followed with stimulation with a filament with increased bending strength. According to the original protocol, optimal threshold calculation by this method requires 6 responses in the immediate vicinity of the 50% threshold, therefore recording of the 6 data points did not begin until the response threshold was first crossed (positive response changes to negative response or inverse). The 2 responses detecting the threshold were designated as the first 2 responses of the series of 6. In cases where continuous negative responses were observed to the maximum of the stimulus set, a value of 26 g was set as the cut-off and used as withdrawal threshold for data analysis. In all other cases the resulting pattern of positive and negative responses was calculated into a 50% response withdrawal threshold using the formula: 50% g threshold = (10Xf+kδ)/(10,000), where Xf = value (log unit) of the final von Frey hair used; k = tabular value for the pattern of positive/negative responses; and δ = mean difference (in log units) between filaments.

Weight-bearing test. Rats were placed in the incapacitance tester (MJS Technology Ltd., Buntingford, Herfordshire, UK) and allowed to acclimatize until calm. Measurements were performed over a period of 4 s and in triplicates. An average weight-bearing ratio was subsequently calculated as the amount of weight placed on the cancer-bearing leg divided by the total amount of weight placed on both legs and used for data analysis.

Limb use test. Rats were allowed to move freely in a transparent plastic cage without bedding (500 mm × 300 mm × 500 mm). Following 10 min of acclimation, each animal was observed for 3 min and assigned a limb use score from 3 to 0 as follows: 3: Normal use of leg, 2: mild or insignificant limping, 1: significant limping, 0: significant limping and part lack of use.

Animals were anesthetized in an induction chamber with 2.5% isoflurane (Isobar Vet; 100%, Nomeco, Copenhagen, DK). D-Luciferin (PerkinElmer, Skovlunde, Denmark) dissolved in PBS was administered by intraperitoneal injection (40mg/kg). 10 min after injection, animals were placed on their back in a nose cone in a Lumina XR instrument (Caliper Life Sciences, Teralfene, Belgium) and anesthesia was maintained with a 2.5% isoflurane/oxygen mix. Image capture was performed with binning: M(4), F/stop: 1 and exposure time from 1 s to 60 s according to the power signal. The signal was adjusted according to the exposure time prior to data analysis. For each animal, an average of three images was used for analysis. Between each capture, the animal was repositioned to minimize bias caused by placement of the animals in the machine. Bioluminescence images were analyzed using IVIS Imaging Software (Living Image©, version 4.0.0.9801; Caliper Life Sciences, Teralfene, Belgium). For each image, the threshold was adjusted to 35% of the signal, and the readout was measured in total flux, photos/s.

X-Ray images were captured subsequent to the bioluminescence images. The severity of bone degradation was analyzed using ImageJ (ImageJ 1.47v). Each X-Ray image was calibrated to a standard aluminum wedge. The mean grayscale value of a standard region of interest within the trabecular bone of the proximal tibia was measured and the average of two corresponding background regions in the soft tissue proximate to tibia was subtracted. The grayscale value was translated into millimeter aluminum equivalents (mmAl) according to the standard wedge and used as estimate of the relative bone density of the distal femur. Data analysis was blinded for sex of the animals.

Animals with no bioluminescence signal or lack of osteolysis, hence no active cancer growth, were excluded from the active state group. All animals included in this study were assigned to an active or restored state based on presence of bioluminescence signal and osteolysis (active) or lack of bioluminescence signal and osteolysis (restored). Four previous in-house experiments were reanalyzed with regard to active and restored state based on presence bioluminescence signal and osteolysis or lack of bioluminescence signal and osteolysis, or based on osteolysis or lack of osteolysis alone. One study was similarly using bioluminescence as exclusion criteria, whereas the data from the three previous studies was analyzed with respect to lack of osteolysis. Lack of osteolysis was based on the last measure day, usually day 21–23, whereas one experiment was terminated at day 14, however on this day 10 out of 11 animals showed clear osteolysis.

No adverse events were observed during the experiment.

The bioluminescence signal was evaluated to validate successful transfection of Luc2 into the MRMT-1 carcinoma cells. In vitro, the signal increased linearly with the number of cells, p <0.0001 ( Figure 1A). In vivo, i.p. administration of D-Luciferin induced a robust plateau of the bioluminescence signal 5 min after injection and lasting approximately 20min ( Figure 1B). Luc2 was hence successfully transfected into the carcinoma cells, linearly reflecting the number of living cells in vitro, and produced a robust signal in vivo.

Pain behavior was quantified by detection of mechanical hypersensitivity, as well as limb use and weight-bearing deficits on days 7, 10, 14, 17, 21 and 23. Compared to baseline level, a significant decline in the 50% withdrawal threshold, reflecting mechanical hypersensitivity, was observed in both females and males from day 10 to 21 ( Figure 2A, Dataset 1: rawdata_vonfrey). Limb use scoring was significantly reduced only on day 23 for the females, but on day 17–23 for the males ( Figure 2B, Dataset 2: rawdata_limbuse). Decline in weight-bearing ratio was significantly reduced on day 17 and 23 for the females and on day 17–23 for the males ( Figure 2C, Dataset 3: rawdata_weight-bearing). No significant difference was observed between the sexes on any of the test days, however females tended to present with a less pronounced limb use and weight-bearing deficit in the later phase of disease progression, days 17–23 ( Figure 2B and C).

The relative bone density was significantly reduced compared to baseline measures in the males from day 14 and throughout the study ( Figure 2D, Dataset 4: rawdata_xray). A similar, however insignificant, tendency was observed in the females ( Figure 2D). No difference was found between males and females at any time point. Overall, the data suggest a less severe progression in females compared to males.

All animals had a detectable bioluminescent signal on days 7 and 10, hence demonstrating successful inoculation of living cancer cells ( Figure 3, Dataset 5: rawdata_BLI, Dataset 6: rawdata_bioluminescence). From day 14 an increasing number of animals lost the bioluminescence signal, suggesting recovery in a subset of animals. Compared to females with consistent bioluminescence signal, a subset of females losing bioluminescence signal displayed a significant decrease in the signal from day 17 ( Figure 3). A similar but later occurring tendency was observed in the males ( Figure 3).

Loss of bioluminescence signal was observed in 40% of the females, but only in 20% of the males ( Figure 4A and B). A significant trend for loss of signal was observed in both females and males, p=0.0079 and p=0.0353 respectively. A similar pattern in recovery was observed in previous independent in-house experiments. In these experiments, lack of osteolysis (determined by x-ray analysis) was used as an exclusion criterion. In one study, 5 out of 16 female animals displayed a lack of osteolysis ( Figure 4C) ¹⁴ . Three additional experiments using male animals demonstrated lack of osteolysis at the end of the experiment in 3 out of 21, 1 out of 10 and 1 out of 11 animals, respectively ¹² ( Supplementary figure S1 and Supplementary figure S2). Pooling the data from the five experiments demonstrated a significant difference in the odds ratio between females and males, reflecting a significantly greater proportion of females clearing active cancer growth compared to males, as indicated by loss of bioluminescence signal and/or lack of cancer-induced osteolysis, p=0.029, ( Figure 4D).

Despite the presence of living cancer cells during the initial 10 days, neither pain behavior nor relative bone density was significantly changed in animals losing bioluminescence signal ( Figure 5A–D). A transient but insignificant decrease in the 50% withdrawal threshold was observed in animals later losing bioluminescent signal ( Figure 5A, rawdata_vonfrey). In addition, a slight decrease in limb use, weight-bearing and relative bone density was observed on day 14, however the decrease was insignificant and returned to baseline levels on the next measured day ( Figure 5B–D, rawdata_limbuse, rawdata_weight-bearing, rawdata_xray). Overall this suggests initial normal disease progression, followed by subsequent recovery and hence normal relative bone density and lack of pain behavior. In addition, a strong correlation between bioluminescence signal and relative bone density was observed ( Figure 6A) supporting the division of animals into two groups; animals with a clear bioluminescence signal and osteolysis, indicating active cancer growth ( Figure 6A, B and C) and animals with no bioluminescence signal and no osteolysis, indicating recovery ( Figure 6A, E and D).

Exclusion of animals successfully recovering aligned the onset and severity of pain behavior in both sexes ( Figure 7A– C). In addition, a similar effect was seen on the relative bone density, reflecting similar bone degradation in both females and males with active growing cancer ( Figure 5D).