A549, HeLa, MCF7, MDA-MB231cells were form CEINGE Cell Culture Facility. A549 cells were grown in Roswell Park Memorial Institute Medium - 2% L-glutamine (RPMI-1640; Cat# R8758; Sigma-Aldrich, St. Louis, MO, USA), HeLa cells were grown in Dulbecco’s Modified Eagle Medium - high glucose (DMEM; Cat# D6429; Sigma-Aldrich), MCF7 cells were grown in Minimum Essential Medium - 2% L-glutamine (MEM; Cat# M4655; Sigma-Aldrich), MDA-MB231 cells were grown in Dulbecco’s Modified Eagle Medium – low glucose (DMEM; Cat# D6046; Sigma-Aldrich), all supplemented with 10% fetal bovine serum (FBS; Cat# CHA30160L; GE Healthcare Life Sciences, Pittsburgh, PA, USA), 1% penicillin/streptomycin (Cat# ECB3001D; Euroclone, Pero, MI, Italy), and incubated in a humidified incubator at 37° C with 5% CO₂.

For biochemical and immunofluorescence studies, cells were seeded at a cell density of 7000/cm² either into 10 cm dishes or onto glass coverslips and treated with vehicle, dimethylsulfoxide (DMSO), as control or Paclitaxel (Plx; Cat# T1912; Calbiochem, Merck Millipore, Billerica, MA, USA) at the indicated doses, after 24 hours from seeding. Cell samples were taken at the indicated time points, washed once in PBS (Cat# ECB4004LX; 10Euroclone) and lysed with 5 volumes of lysis buffer (LB; 0.2% Igepal; 80 mM b-glycerophosphate, 10 mM MgCl₂, 20 mM EGTA, 250 mM NaCl; Sigma-Aldrich) or fixed as described in Immunofluorescence fixation and staining section. Lysates were incubated 30 min on ice and then spun for 20 min at 13.200 rpm in a refrigerated microcentrifuge (4°C; Eppendorf centrifuge 5424R). For cell counting and viability assays, cells were seeded at 7000 cells/cm2 density. After 24 hours of incubation, one cell sample per cell type was trypsinized, collected and resuspended in PBS and loaded into a Bürker counting chamber, cells were counted manually under a microscope for the Time 0 cell count. Other cell samples were treated with either vehicle (DMSO) as control or with 2, 4 and 8 nM Plx and incubated for 48 hours, then cells were trypsinized and resuspended in PBS, mixed to an equal volume of Trypan Blue and counted (Trypan Blue solution 0.4%; Cat# 15250061; Gibco - Thermo Fisher Scientific, Waltham, MA, USA). Cell counting was conducted manually under a microscope and cell viability was determined by Trypan Blue exclusion.

RNA interference via siRNAs was performed using DharmaFECT 1 siRNA Transfection Reagent (Cat# T200103; Dharmacon). For efficient knock-down cells were plated 24 hours prior to treatment and transfected with 25 nM of non-targeting or targeting siRNAs duplex using DharmaFECT 1, according to the manufacturer’s protocol. DharmaFECT 1 and siRNAs were mixed in RPMI-1640 medium and incubated at room temperature (rt) for 20 min, then, the mixture was added to the cells and incubated for 24 hours before Paclitaxel addition. Non-targeting or human cGAS-targeting siRNAs (Non-Targeting SMARTpool Cat# L-009326-00-0020; MB21D1-Targeting SMARTpool Cat# L-015607-02-0020) were purchased from Dharmacon.

Immunoblotting was performed as described (16). Briefly, samples were prepared by adding SDS loading buffer (Laemmli sample buffer; Cat# 1610747; BioRad, MI, Italy) to lysates. Samples were boiled for 10 min at 99°C before being separated on SDS-PAGE (poly-acrylamide percentage spanning from 10 to 12%). Proteins were blotted onto nitrocellulose membrane (Cat# GEH10600002; GE Healthcare) using a wet-transfer system (Cat# EI0001; ThermoFisher). Membranes were incubated with 5% not fat dry milk (NFDM; Cat# A0830; AppliChem GmbH, DA, Germany) or 3% bovine serum albumin (BSA; Cat# A7030; Sigma-Aldrich) in PBS or TBS (tris buffered saline; Cat# T5912; Sigma-Aldrich) supplemented with 0.01% Tween20 (Cat# P9416; Sigma-Aldrich; TPBS or TTBS, respectively) for 1 hour at rt. Then, membranes were incubated with primary antibodies, diluted in TPBS or TTBS, at 4°C overnight. After washing twice with TPBS or TTBS, filters were incubated with secondary peroxidase-conjugated antibodies, diluted in TPBS or TTBS, for 1 hour at rt. Detection was performed using an Enhanced ChemiLuminescence (ECL) kit (Cat# GEHRPN2106; GE Healthcare). Blots were acquired using Canon CanoScan LiDE 300 scanner (Canon) and scanned at 300 dpi. Primary and secondary antibodies used for immunoblotting are listed in Table 1.

Cells were plated onto glass coverslips 24 hours prior to Paclitaxel addition. After 48 hours Plx treatment, coverslips were briefly washed in PBS and cells fixed with 4% paraformaldehyde (Cat# P6148; Sigma-Aldrich) in PBS (Euroclone) for 10 min at rt. Cells were washed twice with PBS and permeabilized with 0.25% Triton X- 100 (Cat# T9284; Sigma-Aldrich) in PBS for 15 min. The permeabilization step was omitted for PD-L1 immunofluorescence. Then, cells were washed once with PBS and incubated with 3% bovine serum albumin (BSA; Sigma-Aldrich) in PBS for 1 hour at rt. Coverslips were transferred into a humidity chamber and incubated with primary antibodies in 1.5% (w/v) BSA-PBS solution for 2 hours at rt, except for PD- L1 immunofluorescence for which incubation with primary antibody was performed overnight at 4°C. After incubation, cells were washed three times with PBS and incubated with fluorescently labelled secondary antibodies, diluted in 1.5% BSA-PBS solution, for 1 hour at rt. DNA was stained with Hoechst 33258 (1 mg/mL; Cat# 94403; Invitrogen, Waltham, MA, USA) by incubation for 10 min. Finally, cells were washed four times with PBS and slides mounted with Mowiol 40-88 (Cat# 81381; Sigma-Aldrich). Primary and secondary antibodies used for immunofluorescence are listed in Table 1.

Fixed cells were photographed using an inverted confocal fluorescence microscope LSM 980 (Zeiss) equipped with a 63X/1.4 oil objective (Zeiss). Representative images were obtained collecting 5 Z-stack series. The acquisitions were deconvoluted and projected into one plane using the ZEN3.1 software.

It has been proposed that chromosome segregation defects in chromosomally unstable cancer cells or upon treatment of cancer cells with Taxanes can cause formation of cGAS-positive micronuclei and activation of proinflammatory pathways (1, 15, 17). To investigate potential immunomodulatory effects of the Taxane Pxl, we started asking whether treatment with low doses of Pxl promoted formation of cGAS-positive micronuclei in various cancer cell lines: lung adenocarcinoma A549 cells, triple negative breast cancer (TNBC) MDA-MB231 cells, hormone-responsive breast cancer MCF7 cells and cervical cancer HeLa cells. We chose a low Pxl concentration range, between 2 and 8 nM, because it is known that Pxl concentrates within cells about 50 to 1000 folds, depending on the cell type, so that after a 20-hour treatment with Pxl concentrations in the low nM range in the cell culture medium, Pxl can reach intracellular concentrations similar to those reached, in vivo, in tumor cells of patients treated with Pxl infusions (5). In addition, since Pxl-induced micronucleation requires passage through mitosis, we chose a treatment time of 48 hours since, under our experimental conditions, the cell doubling time was approximately 22 hours for A549, 24 hours for HeLa, 27 hours for MCF7 and 29 hours for MDA-MB231 in control cells (treated just with dimethylsulfoxide, DMSO, vehicle; Supplementary Figure 1 ), so that by 48 hours all cell types had undergone mitosis at least once (1, 5). Moreover, a 48-hour Pxl treatment had mostly a cytostatic effect at the doses used in all cell lines tested, while cytotoxicity was relatively modest since the highest drop of cell viability (measured by trypan blue exclusion) was around 20% in MDA-MB231 cells treated with 8 nM Pxl ( Supplementary Figure 1 ).

Thus, to determine whether Pxl treatment promoted formation of cGAS-positive micronuclei, the four cell lines were treated with vehicle, as control, or with 4 nM Pxl for 48 hours and the presence of cGAS-positive micronuclei assessed by immunofluorescence (IF; Figure 1 ). In all cell lines tested, including MDA-MB231 that are very genetically unstable and show a higher basal level of cGAS-positive micronuclei, Pxl treatment strongly increased the number of cells with cGAS-positive micronuclei ( Figure 1 ).

Chromatin bridges that form during altered mitosis exit induced by Pxl have also been shown to recruit cGAS and activate the cGAS-STING pathway involved in activation of cGAS (17). We also found evidence of Pxl-induced cGAS-positive chromatin bridges under our experimental conditions (an example is shown in Supplementary Figure 2 ).

Next, we analyzed whether Pxl treatment led to activation of a proinflammatory cascade in cancer cells. To this end we analyzed the phosphorylation status of STING at serine 366 (S366), a site known to be phosphorylated by TBK1 and required for further activation of IRF3, and the STING-TBK1-dependent, activating, phosphorylation of IRF3 at serine 386 (S386) upon treatment of A549, MDA-MB231, MCF7 and HeLa cells with increasing doses of Pxl in the low nanomolar range (2, 4 and 8 nM; Figure 2 ) (18, 19). Indeed, the Pxl treatment induced S366-STING and S386-IRF3 phosphorylations in all cell lines ( Figure 2 ). The phosphorylated forms of STING and IRF3 tended to decline at 8 nM relatively to 4 nM Pxl treatment ( Figure 2 ; see quantitative graphs of the phosphorylated forms). We do not have an exact explanation for this phenomenon, but we believe that it may possibly indicate that negative feedbacks, that attenuate excess proinflammatory signals, are initiated at the highest Pxl concentration (20).

Following IRF3 activation, IFN type 1 genes are potently transcribed and IFN proteins produced and secreted (21). IFNs activate the JAK/STAT signaling cascade and phosphorylation of STAT1 at tyrosine 701 (Y701) is a hallmark of IFN pathway activation (22). Moreover, IFN and cGAS have been shown to be linked by positive feedback loops in which cGAS-dependent IFN induction further stimulates cGAS expression (23). In addition to the innate immunity-promoting action of cGAS, also recruitment of cytolytic CD8 T cells in the tumor microenvironment has been shown to be very dependent on cGAS activity (24). Considering that Pxl has also been shown to upregulate MHC class I, these effects of Pxl treatment may, indeed, strongly favor antitumor adaptive immunity (14).

Nevertheless, several lines of evidence also link the cGAS-STING pathway, STAT1 activation and IFN signaling to the upregulation of the expression of the immune checkpoint protein PD-L1 in cancer cells as well in cells of the immune system (25, 26). Thus, if the Pxl-activated cGAS/STING pathway would increase PD-L1 expression in cancer cells, this might promote, on the other hand, cancer cell escape from surveillance by the adaptive immune system (25, 26).

We thus asked whether Pxl treatment led to Y701-STAT1 phosphorylation and upregulated cGAS and PD-L1 protein levels. To this end, A549, MDA-MB231, MCF7 and HeLa cells were treated for 48 hours with low doses of Pxl and the levels of phosphorylated Y701-STAT1 (p-Y701-STAT1) and of cGAS and PD-L1 protein analyzed ( Figure 3A ). Indeed, in all cell lines tested Pxl induced Y701-STAT1 phosphorylation, marker of IFN pathway activation, and significantly upregulated cGAS and PD-L1 expression ( Figure 3A ). In addition, Pxl treatment induced a substantial increase of PD-L1 at the cell surface in all cell types, as shown by PD-L1 immunofluorescence of non-permeabilized cells ( Figure 3B ).

We asked whether Pxl-induced upregulation of STAT1 phosphorylation and of PD-L1 expression were mediated by the cGAS/STING pathway. To this end, MDA-MB231 cells were treated with non-targeting (NT-siRNAs) or cGAS-targeting small interfering RNAs (cGAS-siRNAs) 24 hours before Pxl (4 nM) or vehicle (Pxl 0 nM; as control) addition, then, cells were taken after further 48 hours incubation ( Figures 4A, B ). The siRNAs treatment resulted in more than 90% downregulation of the cGAS protein expression in cGAS-siRNA-treated cells compared with NT-siRNAs-treated cells ( Figure 4A ). Moreover, induction of Y701-STAT1 phosphorylation as well as the increase of PD-L1 protein expression induced by Pxl in control (NT-siRNA-treated) cells were completely blunted in the cGAS-downregulated (cGAS-siRNA-treated) cells, while the levels of total STAT1 protein were not affected by cGAS downregulation ( Figure 4B ). Similar results were obtained by downregulating cGAS expression in A459 cells under similar treatment conditions as described for MDA-MB231 cells ( Figure 4C ). These data indicate that IFN pathway activation and upregulation of PD-L1 protein in cancer cells treated with Pxl are indeed cGAS-dependent.