SW620 tumour 3D spheroid cell response is different in static compared to microfluidic conditions
Experiments in static plate conditions cannot effectively mimic the PK profile on the spheroids, nor can they easily be explored for treatment scheduling.
SN38 and AZD0156 plasma concentrations in mouse were taken and measured hourly for 96 h, using LC MS. In vitro experimental setup used 8 concentration points translated from the 24 hours plasma concentrations in mouse. Each concentration used for a specific amount of time for in vitro microfluidic setup represented an average of the corresponding time interval from observed and modelled in vivo PK data. Closely related PK profiles between mouse plasma and in vitro concentrations are illustrated in Table 1 .
Table 1
In vivo free drug plasma concentrations (determined by LC -Mass Spectrometry and population-based PK model) and mimicked concentrations in the microfluidic setup.
In vivo drug free concentrations (nM)
|
|
In vitro microfluidic drug concentrations
|
Time (h)
|
AZD0156 (10 mg/kg)
|
SN38 (Irinotecan 50 mg/kg)
|
|
AZD0156
(nM)
|
SN38
(nM)
|
Time (h)
|
Reservoir
|
1
|
187.8
|
5.5
|
|
192
|
5.5
|
1
|
2
|
2
|
192.1
|
4.7
|
|
150
|
4.3
|
3
|
3
|
3
|
173.8
|
4.4
|
|
4
|
154.5
|
4.1
|
|
5
|
137.1
|
3.6
|
|
110
|
3.4
|
2
|
4
|
6
|
121.5
|
3.1
|
|
7
|
107.8
|
2.8
|
|
75
|
2.3
|
3
|
5
|
8
|
95.6
|
2.4
|
|
9
|
84.7
|
2.1
|
|
10
|
75.2
|
1.9
|
|
60
|
1.7
|
3
|
6
|
11
|
66.7
|
1.7
|
|
12
|
59.1
|
1.6
|
|
13
|
52.7
|
1.5
|
|
45
|
1.3
|
4
|
7
|
14
|
46.6
|
1.4
|
|
15
|
41.5
|
1.2
|
|
16
|
36.7
|
1.1
|
|
17
|
32.6
|
1.1
|
|
20
|
0.6
|
8
|
8
|
18
|
28.8
|
0.9
|
|
19
|
25.7
|
0.9
|
|
20
|
22.7
|
0.8
|
|
21
|
20.2
|
0.7
|
|
22
|
17.8
|
0.7
|
|
23
|
15.9
|
0.6
|
|
24
|
14.1
|
0.6
|
|
The concentration of a drug applied to a static set up over time is very different from the in vivo situation and the mouse PK profiles for both SN38 & AZD0156 (Fig. 2b). In addition, under static conditions in the plate format, 3D nude spheroids do not allow washing and media replacement steps due to the high risk of damage to the spheroids. In contrast, the Matrigel encapsulated spheroids in the Ibidi chip can resist the medium flow for longer term experiments (tested up to 14 days) and the Matrigel droplet is permeable to nutrients, drugs, fluorophores and antibodies.
As a first step to compare the response of SW620 cells treated in the 3D chip microfluidic set up versus 3D plate, we explored differences in tumour spheroid size and viability following 6 days of treatment with SN38 and combination with AZD0156 using exposures as shown in Fig. 2a & b. AZD1056 monotherapy was not further explored as there was no spheroid size, morphology or viability effect on SW620 spheroids in static 2D or 3D (Supplementary S1). Tumour spheroid size at day 6 for both static 3D and the microfluidic setup was expressed as fold increase in size over day 1 values and showed that a SW620 spheroid size reduction was indicative of potentiation activity of ATM inhibitor (AZD0156) over SN38 alone, only in the microfluidic Ibidi chip but not in static experimental conditions (Fig. 2c). The untreated spheroid size at day 6 was slightly higher in the static setup possibly as these nude spheroids had no mechanical constraints (as in Matrigel-embedded spheroids, used in the microfluidic setup).
From the same experimental setup, we explored SW620 spheroid viability at 6 days for static and microfluidic setup (Fig. 2d). We observed that continuous exposure to SN38 led to greater reduction in spheroid viability (~ 60%) compared to the effect caused by SN38 dosed to mimic in vivo PK profile (~ 40% reduction in viability). Furthermore, the SN38 + AZD0156 combination in a microfluidic setup reduced viability by 1.4 fold, compared to 2.2 fold in static setup (Fig. 2d), suggesting that the static exposure to a higher concentration of drugs may exaggerate the treatment effect. This experiment demonstrated that exposure of the 3D tumour cells to the drugs in more physiological-like setting leads to a different response than at static exposure conditions, leading us to further explore whether this novel in vitro system could indeed better predict in vivo experiments.
Tumour-on-chip microfluidic platform can be used to explore efficacy and biomarker response to guide optimal design of treatment schedules
Next, we assessed the response of Matrigel encapsulated SW620 spheroids in our microfluidic set up to SN38 monotherapy and in combination with AZD0156 using various treatment schedules (Table 2). The tested schedules were designed to recapitulate mouse plasma PK profiles from similar schedules used in in vivo efficacy studies (Table 1) with a desire to reverse translate in vivo data and to evaluate the predictive ability of our microfluidic chip setup.
Table 2. Treatment schedules for SN38 and AZD0156 assessed in the microfluidic setup, on Matrigel-encapsulated SW620 spheroids.
|
1 Cycle = 7 Days
|
Treatment type details
|
Drug
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
|
SN38
|
|
|
|
|
|
|
|
|
Mono
|
SN38
|
AZD0156
1/7
|
|
|
|
|
|
|
|
Combo (1+1/7, intermittent)
|
SN38
|
AZD0156
7/7
|
|
|
|
|
|
|
|
Combo (1+7/7, continuous)
|
SN38
|
AZD0156
3/7 (24h gap)
|
|
|
|
|
|
|
|
Combo (1+3/7, intermittent) early gap
|
SN38
|
AZD0156
3/7 (72h gap)
|
|
|
|
|
|
|
|
Combo (1+3/7, intermittent) late gap
|
SW620 spheroids response to treatment was assessed after 7 days. We investigated changes in the spheroid volume, as well as viability using the CellTiterGlo assay. To evaluate whether our system could also be used to monitor treatment impact on pharmacodynamic biomarkers, we collected individual spheroids at the end of the treatment period and performed immunofluorescence analysis for γH2AX (DNA double strand break marker), cleaved caspase 3 (apoptosis) and Ki67 (proliferation marker), as detailed in Methods and corresponding Supplementary sections (Supplementary S2). Each treatment condition was repeated at least 3 times and measurements for each condition were pooled from at least 6 spheroid replicates.
Spheroids were imaged on day 1 and 7 (Fig. 3a) and their volume measured on day 7 following indicated treatment conditions (Table 1). A treatment effect at day 7 was assessed by calculating the volume of the spheroids following treatment as a percentage of the average volume (n = 7) of untreated spheroids. (Fig. 3b).
SN38 monotherapy led to a reduction in average volume size to 55% that of control. Concurrent combination treatment with SN38 (dosed on day 1) and AZD0156 (dosed on days 1–7) was more potent and caused spheroid volume reduction to 17% of control. We also explored whether the duration of AZD0156 dosing or a gap between SN38 and AZD0156 affected the combination efficacy for these particular agents. Treatment with SN38 (dosed on day 1) plus AZD0156 (dosed on day 1 only) led to a small drop in efficacy compared to a continuous 7-day treatment with AZD0156 as shown by average relative spheroid volume of 22% (1-day schedule) versus 17% (7-day schedule). Introducing a dosing gap between the SN38 and initiation of AZD0156 treatment proved to have a more significant consequence on the efficacy benefit. While a 24 h delay of AZD0156 dosing was still able to cause significant spheroid growth inhibition ( 25% of control volume), with a 72 h gap we measured spheroids volume was only 40% of control (Fig. 3b). Moreover, a similar outcome was observed using spheroid viability as a readout (Fig. 3c), providing additional evidence on the different impact of various treatment schedules.
SN38 is a TOP1 inhibitor that that prevents relegation of the DNA strand by binding to TOP1–DNA complex and causes DNA double strand breaks (DSBs). Those DNA DSB lesions trigger activation of the ATM kinase for effective repair. Inhibition of ATM and consequent accumulation of unrepaired breaks leading to increased cell death provides the rationale for developing this combination for cancer treatment. AZD0156 is a potent and selective inhibitor of ATM kinase, shown to provide robust efficacy against tumour models when combined with DNA DSB inducing agents14. Here, we tested whether different schedules lead to different activation of pharmacodynamic markers that are associated with DNA DSB damage, apoptosis and cell proliferation. The γ-H2AX associated foci accumulate in cells exposed to DBS inducing agents15, 16. These foci can be detected by immunostaining to determine their, size and morphology16. However, spheroid imaging using a 20x objective on a CV7000 confocal microscope cannot accurately explore foci morphology, so we introduced customized ImageJ macros (described in Methods and Supplementary S2, S10) to quantitate positive cells within a spheroid. The same method was applied to quantitate CC3-positive (apoptotic) cells, while normalization by spheroid area was not required for Ki67, as this biomarker is present in all proliferating cells.
We collected and imaged individual spheroids from the chip at the end of the study (day 7), following an on-chip fixing step in formaldehyde and immunofluorescence staining for γH2AX in all treatment groups (Fig. 4a & b) (Supplementary S2, S9). We did not observe a significant change in γH2AX in spheroids treated with SN38 monotherapy compared to the control group, suggesting that the effect of single dose of SN38 on DNA damage is likely resolved by day 7. However, even a single dose of AZD0156 at concentration mimicking the mouse plasma exposure profile from a 10 mg/kg dose when combined with SN38 resulted in 68% increase in γH2AX which was still detectable at day 7 compare to the untreated control. This was further accentuated by increasing the number of AZD0156 doses with maximum effect over 400% increase observed with 7 days of continuous exposure to ATM inhibitor after SN38 dose (Fig. 4b). Greater increase in DNA damage also resulted in potentiation of apoptosis in these cells as demonstrated by 500% increase in cleaved caspase 3 (Fig. 4c) (Supplementary S3) and it is consistent with the observed effect on spheroids size/ viability (Fig. 3b, c). Proliferation of SN38 treated spheroids was about 20% decreased compared to the control cells with significantly more (50–60%) decrease in all the combination treated groups. The most obvious effect on reduction (64%) in cell proliferation was observed in the 7-day continuous treatment schedule (Fig. 4d). Our data therefore suggest that our microfluidic system not only detects the potentiation of TOP1 inhibitor induced DNA damage and apoptosis by the ATM inhibitor AZD0156 but also the impact of the duration of AZD0156 dosing after SN38 treatment, as well as the gap between the two agents.
Microfluidic tumour-on-chip platform can predict efficacy response in vivo
To assess the translational potential of the microfluidic platform, we designed an in vivo experiment using SW620 xenograft to assess the efficacy response to irinotecan (50 mg/kg) and its combination with AZD0156 (10 mg/kg). We tested different weekly in vivo dosing schedules, previously evaluated in SW620 spheroids on the chip (Fig. 5a) using matched exposure profiles to those previously observed in mouse plasma. (Fig. 2b, Table 1).
Tumour xenograft volumes (in vivo) and tumour spheroids volumes (in vitro microfluidic setup) for irinotecan, as well as irinotecan plus AZD0156 dosed with a 24 h or 72 h gap, were compared at day 7. The concurrent combination schedule was not tolerated in mice, therefore in vivo efficacy data could not be generated for this. For tumour xenografts, tumour volumes were measured twice weekly for 38 days for all treatments groups, expect vehicle control which had to be terminated on day 15 due to the tumour volume limit being reached. .
There were no significant differences observed in tumour size amongst the treatment groups at day 7, all reaching approximately 80% of the control group size. However, at day 15 the combination with the 24 h gap showed greater efficacy (39% volume of control) than irinotecan alone (54% volume of control) or a combination schedule using a 72 h gap between irinotecan and AZD0156 (59% volume of control). Even more significant separation of the tumour growth curves became apparent at day 35, with combination using the 24 h gapped schedule providing the best response, followed by a combination treatment using a 72 h gap, which was still more potent than irinotecan monotherapy (Fig. 5b).
Tumour xenograft versus tumour spheroid percent growth inhibition (% GI) for the treated samples compared to untreated controls followed the same trend at 7 days for in vitro microfluidic chip setup (52% GI) and at 15 days for the in vivo experiment (53% GI) (Supplementary S5). In combination with 24 h gap treatment schedule, we observed % GI of 69% (at 15 days, in vivo) and 88% (at 7 days, in vitro microfluidic). Combination with 72 h gap showed 46% GI (at 15 days, in vivo) and 70% GI (at 7 days, in vitro microfluidic chip). The differences in % GI between 24 h and 72 h combination schedules were 23% (in vivo) and 18% (in vitro, microfluidic chip).