We analyzed cellular and molecular effects triggered by irradiation with photons or 12C-ions. We focused on DNA damage as well as selective and comprehensive gene regulation. For that, we analyzed the two cell lines LNCaP and DU145 that differ genetically with respect to TP53 and androgen signaling. LNCaP have functional TP53 and androgen receptor (AR) signaling whereas DU145 are dysfunctional for TP53 and AR signaling. AR signaling was induced in LNCaP cells by addition of dihydrotestosterone (DHT). A schematic diagram summarizes the experimental workflow (Fig. 1).

Fig. 1

Schematic diagram of the experimental workflow

DNA damage evaluated by immuno-cytofluorescence of γ-H2AX and 53BP1

We measured DNA-damage caused by 12C-ion or photon irradiation by immunofluorescence for γ-H2AX and 53BP1 indicating the DNA-double-strand breaks (DSB) (Fig. 2A). The applied dose ranges for 12C-ions (0, 1, 2, 4 Gy) and photons (0, 3, 6, 12 Gy) were adjusted to take account of the relative biological effectiveness (RBE), which is higher for 12C-ions (approximately 3-times) than for photons [23, 27]. Using the number of DSB-foci as an indicator for DNA damage, the respective 12C-ion and photon dose curves strikingly overlapped for LNCaP and DU145 cells (Fig. 2B). Typically, the number of DSB-foci at 2 h after irradiation correlated with the irradiation dose. The numbers of DSB-foci per nucleus were higher in LNCaP than in DU145 cells. The decrease in DSB-foci over time reflected the process of DNA repair. Alongside, the formation of DSB-foci was paralleled by an increase of protein abundance of γ-H2AX and H2AX in cell lysates of LNCaP and DU145 cells (Fig. 2C) upon photons and 12C-ions.

Fig. 2

Analysis of DNA-damage in LNCaP and DU145 cells following photon and 12C-ion irradiation. A Immunofluorescence of γ-H2AX (red), 53BP1 staining (green) and DAPI-merge (weakly blue) for counterstaining of nuclei. The co-localized red and green foci indicate the DNA-double-strand-breaks (DSB) per nucleus. The white bar scale corresponds to 10 µm. The respective images of dose and time served for counting of DSB-foci. B Number of DSB-foci for photon and 12C-ion irradiation of LNCaP (left) and DU145 (right) time dependently for each dose of photon or 12C-ion irradiation (mean ± SD, n ≥ 95). C Western-blot analysis of γ-H2AX and H2AX, after (2 h) photon or 12C-ion irradiation, dose-dependent in LNCaP cells (left) and DU145 cells (right)

Regulation of mRNAs encoding genes for DNA damage, DNA repair, DNA replication and cell cycle after photon and 12C-ion irradiation in LNCaP and DU145 cells

Next, we analyzed selected mRNAs of DNA-damage, DNA-repair, DNA replication and cell cycle in LNCaP and DU145 cells. An overview of the changes in mRNA levels upon photon and 12C-ion irradiation with all tested doses and time points is displayed (Fig. 3A). It reveals that the mRNA changes in LNCaP cells are strikingly stronger than in DU145 cells.

Fig. 3

RT-qPCR analysis of selected mRNAs related to DNA-damage (green), DNA-repair (cyan), DNA replication (blue) and cell cycle (red) in LNCaP (left) and DU145 cells (right). A Overview of the changes of mRNA levels (ΔΔCt, mean, n = 3) upon photon and 12C-ion irradiations with all tested doses and time points. B Subsets of changes of mRNA levels (ΔΔCt) (mean ± SD, n = 3) in adjusted scales displayed for the irradiation modes with the maximum applied dose for photons (12 Gy) and for 12C-ions (4 Gy). Significant differences (2-way-ANOVA with multiple comparisons) are indicated for each time point (separated by slash) in the insets. C Changes of mRNA levels (ΔΔCt) after DHT treatment (20 nM; 24 h) of LNCaP cells for analysis of mRNAs related to DNA-damage (green), DNA-repair (cyan), DNA replication (blue), cell cycle (red) in addition to androgen receptor (AR) signaling (black). The changes of mRNA levels (ΔΔCt) are displayed (mean ± SD, n = 3). Significant differences between DHT and control were determined by unpaired Welch corrected t-test. Symbols of significances (*p < 0.05, **p < 0.01; ***p < 0.001; ****p < 0.0001)

We further divided the respective subsets of mRNAs and displayed their changes in adjusted scales for LNCaP and DU145 cells (Fig. 3B). Interestingly, in LNCaP cells, the initial increase in H2AX protein that was observed after 2 h (Fig. 3C) was counterbalanced by decreased H2AX-mRNA levels at later time points (24 h, 48 h) (Fig. 3B, 1st lane, left) whereas the 53BP1-mRNA levels were slightly upregulated (< twofold). In DU145 cells, both H2AX-mRNA and 53BP1-mRNA levels were unaffected by irradiation (Fig. 3B, 1st lane, right).

In addition to H2AX and 53BP1, we investigated additional mRNAs by RT-qPCR that are involved in DNA repair (XRCC2, FANCI) (Fig. 3B; 2nd lane), DNA replication (PCNA, RFC3, POLE2, TOP2A) (Fig. 3B; 3rd lane) and cell cycle control (CDKN1A, CDKN1B, TP53NIP1, MAD2L1) (Fig. 3B, 4th lane).

Fig. 4

Analysis of RNA seq data by Principal Component Analysis (PCA) and gene set enrichment analysis (GSEA) from samples of control, photon (6 Gy; 24 h) and 12C-ion (2 Gy; 24 h) irradiation or DHT treatment (20 nM; 24 h). A PCA Biplot of the mRNAs including all samples. The principal component 1 (PC1) covers 58.9% (x-axes) and PC2 covers 10% (y-axes) of the multidimensional data sets of the detected mRNA targets (n > 104). The samples of LNCaP or DU145 cells are grouped as control, photons, 12C-ions or DHT, as indicated by different colors and symbols (see legend). The n-number of controls is n = 9 for LNCaP and n = 6 DU145 since the respective matched control values (n = 3) were merged. The samples are encircled by a line representing the 95% confidence interval (95% CI) for each group. The loading vectors (light blue) represent several selected altered genes and indicate their contribution to PC1 or PC2. (B-E) Gene set enrichment analyses (GSEA) of enrichment score (ES) and false discovery rate (FDR) are displayed as volcano plots (ES vs. − log FDR). In GSEA, ES-values are considered as significant if FDR < 0.25 and p < 0.05. The dotted line at (− log FDR = 0.602) corresponds to FDR = 0.25, meaning that the ES values above this line have an FDR < 0.25. Here, these gene sets have nominal p-values < 0.01 (see supplementary information) and therefore the ES-values above the dotted line represent significant ES-values. The critical ES-values of the treatments are displayed with respect to different gene sets: B Amundson-DNA-damage-response-TP53 (TP53-DNA-damage) for photons and 12C-ions; C WP-G1-to S-cell cycle-control (G1-S-Cell cycle) photons and 12C-ions; D WP-DNA-repair pathway full network (FN-DNA-Repair) for photons and 12C-ions. In E the Hallmark Androgen Response (HM-AR) in DHT-treated LNCaP cells is compared with photon and 12C-ion irradiated LNCaP cells. In E, the gene sets shown in B–D are analyzed with respect to DHT treatment. In F, the ES values of FN-DNA-repair are correlated with those of other DNA-repair modes particularly, DNA-repair-homologue-recombination (HR-DNA-repair), non-homologue-end-joining (NHEJ-DNA-repair), mismatch repair-DNA-repair (MM-DNA-repair) and AR-targeted DNA-repair (AR-DNA-repair). The symbols refer to the DNA-repair mode (see inset table). Treatments and cells are indicated LNCaP (LN) and DU145 (DU)

In LNCaP cells (Fig. 3B, left side), the mRNAs of DNA-repair- (XRCC2, FANCI) and DNA-regulator-genes (TOP2A, RFC3, POLE2; MAD2L1) except for PCNA were downregulated by both 12C-ion and photon irradiation. In particular, TOP2A mRNA was downregulated by almost 100-fold (ΔΔCt > 6). As a critical cell cycle inhibitor, CDKN1A-mRNA was upregulated by approximately 20-fold (ΔΔCt > 4) upon photon and ion irradiation.

In DU145 cells (Fig. 3B, right side), the alterations of mRNAs after irradiation with 12C-ions and photons were of considerably lower amplitude (< twofold; ΔΔCt < 1) than in LNCaP cells and are adequately displayed in a narrower scale.

Regulation of selected mRNAs by DHT

For evaluating AR-signaling, LNCaP cells were treated with DHT (Fig. 3C). The AR markers KLK3 (prostate specific antigen, PSA) and KLK2 were measured as known AR targets and displayed the typical DHT-dependent induction. In the DNA repair group, mRNAs for XRCC2, FANC1 and H2AX were significantly downregulated (twofold), while 53BP1 was slightly upregulated (1.5-fold) by DHT. Similarly, the DNA regulators PCNA, RFC3, POLE2 and TOP2A were downregulated (twofold). In the cell cycle group, CDKN1A was unaffected, whereas T53INP and MAD2L1 were downregulated by DHT.

In essence, certain genes from functional groups gave typical congruent responses in LNCaP cells to 12C-ion and photon irradiation. In particular, some DHT responses were common to the irradiation responses, most distinctly for DNA-damage and DNA repair genes and partially for DNA-regulators and cell cycle genes (e.g. TOP2A, MAD2L), with the exception of CDKN1A. In contrast to LNCaP cells, DU145 cells showed considerably lower changes in mRNA levels following irradiation.

Transcriptomic analysis by RNA-seq

Subsequently, we analyzed the transcriptomic mRNA response of LNCaP and DU145 cells by RNA-seq. Based on the previous results (Fig. 3A, B), the analysis was performed 24 h after irradiation with a dose of 6 Gy (photons) and 2 Gy (12C-ions). These intermediate irradiation doses as well as the intermediate time period exhibited significant gene regulatory effects and appeared suitable for comprehensive RNA-seq analysis. In particular, the higher selected dose range of photons versus 12C-ions is supported by literature [23, 27]. For analysis of AR response, LNCaP cells were incubated with 20 nM DHT. Each treatment included directly matched naïve control cells for analysis.

The obtained RNA seq data were evaluated using Principal Component Analysis (PCA) (Fig. 4A). PCA projected the huge multidimensional mRNA data (n > 104) of the treated and control samples to a 2-dimensional plane geometry with maximum possible differentiation for the first dimension 1 (X-axis: PC1 = 58.9%) and for the second dimension (Y-axis: PC2 = 10%) (Fig. 4A). The samples (dots) were labeled according to cell type and treatment group (control, photons, 12C-ions, or DHT). The groups of samples are encircled by respective lines indicating the 95% confidence interval (CI) of the treatment. In result, LNCaP cells (allocated on the left side) and DU145 cells (allocated on the right side) were separated. The LNCaP cells (filled shapes) could be separated into (i) control group (green circles) (ii) DHT group (red diamonds) and (iii) the overlapping irradiation modes 12C-ion (blue triangles) and photon (purple squares). Within DU145 cells (empty shapes), no separation was possible since the encircled 95% CI of control (green circles), photon (yellow squares) and 12C-ion (brown triangles) all overlapped. The loading vectors (light blue) refer to the most differentially expressed genes. Some of these have been already mentioned in Fig. 3 (CDKN1A, KLK3, TOP2A).

Gene set enrichment analysis of RNA-seq data

Furthermore, the RNA-seq data were analyzed using Gene set enrichment analysis (GSEA). GSEA statistically analyses gene sets that are preferentially altered as a whole in response to a trigger. Here, we focused on gene sets covering signaling pathways that are related to irradiation and AR signaling (listed in supplementary information). We tested gene sets related to Amundson-DNA-damage-response-TP53 (TP53-DNA-damage), WP-DNA-repair-pathway-full-network (FN-DNA-repair), WP-G1-to S-cell cycle-control (G1-S-cell cycle), and hallmark-androgen-response (HM-AR). Individual members of these pathways have been addressed previously (Fig. 3).

TP53-DNA-damage gene set

LNCaP cells displayed a significant positive enrichment score for the gene set TP53-DNA-damage after photon and 12C-ion-irradiation (Fig. 4B). In DU145 cells, which are dysfunctional for TP53, no significant ES for TP53-DNA-damage by photons or 12C-ions was determined (Fig. 4B).

G1-S-Cell-cycle gene set

LNCaP cells displayed significant negative ES for the gene set G1-S-cell cycle by photons and 12C-ions (Fig. 4C). In DU145 cells, this gene set also displayed significant negative ES, but to a lesser extent, when treated with photons and 12C-ions (Fig. 4C).

DNA-repair gene set

Next, we analyzed the gene set FN-DNA-repair. LNCaP cells showed significant negative ES after photon and 12C-ion-irradiation (Fig. 4D). In DU145 cells, the gene set FN-DNA-repair also displayed significant negative ES for both photons and 12C-ions (Fig. 4D).

DHT, photons and 12C-ions in relation to several gene set pathways

The androgen responsive LNCaP cells that had been treated with DHT were analyzed with respect to androgen receptor signaling using the gene set HM-AR (Fig. 4E). As expected, the gene set HM-AR exhibited a significant positive ES in DHT treated LNCaP. In comparison, the gene set HM-AR showed rather negative ES for photons and 12C-ions without significance. Alongside, TP53-DNA-damage, FN-DNA-repair and G1-S-cell cycle were analyzed for DHT treated LNCaP cells (Fig. 4E). TP53-DNA-damage was not significantly altered by DHT, whereas the ES values for FN-DNA-repair and G1-S-cell cycle control revealed significant negative enrichment by DHT. Complementarily, photons and 12C-ions both had negative ES-values for HM-AR but without significance (Fig. 4E). Furthermore, we compared FN-DNA-repair with subsets of DNA-repair modes including gene sets for kegg-homologues-recombination (HR-DNA-repair), kegg-non-homologues-end-joining (NHEJ-DNA-repair), kegg-mismatch-repair (MM-DNA-repair) and the subset of AR-targeted DNA-repair genes (AR-DNA-repair) [9]. The ES-values of FN-DNA-Repair significantly correlated with the listed DNA-repair modes except for NHEJ-DNA-repair (Fig. 4F).

Differentially expressed genes (DEG) after irradiation by photons or 12C-ions and DHT treatment

Next, DEG were determined from the RNA-seq data. DEG are defined as mRNAs that are altered with a log2FC ≥ 1 and corrected p-values < 0.05. DEG-photon and DEG-12C-ion were determined for LNCaP (Fig. 5A, B) and for DU145 cells (Fig. 5J, K). Furthermore, DEG-DHT were determined for LNCaP cells (Fig. 5C). The DEG are displayed in volcano plots.

Fig. 5

Analysis of differentially expressed genes (DEG) determined by RNA-seq. DEG of LNCaP cells for DEG-photon A, DEG-12C-ion B and DEG-DHT C and in DU145 cells for DEG-photon J and DEG-12C-ion K. DEG are defined as log2FC ≥ 1; corrected p < 0.05. The log2-FC in the heatmaps with the corresponding volcano plots indicate treatment versus control (n = 3). Upregulated DEG are red, downregulated DEG are blue and non-significantly altered mRNAs are grey. D Correlation graphs of log2-FC of DEG-photon and DEG-12C-ion in LNCaP cells. DEG-photon that intersected with DEG-12C-ion are black (n = 577). Unique DEG-photon are orange (n = 232) and unique DEG-12C-ion are green (n = 220). E, F Correlation graphs of DEG-DHT that intersected with DEG-photon or with DEG-12C-ion. The same samples are plotted once with (log2-FC-photon) on the y-axis E and once with (log2-FC-(12C-ion) on the y-axis F. The majority of DEG-DHT were unique (n = 904, olive). A prominent fraction of DEG-DHT intersected with DEG-12C-photon (orange; n = 55), with DEG 12C-ion (cyan; n = 53) or with both (red; n = 292). G–I Correlation graphs of merged DEG-photon, DEG-12C-ion and DEG-DHT that are restricted to those with high expression levels (top 50th percentile of all DEG) and strong fold changes (0.3 > FC > 3; adequately to (log2FC > 1.6) for at least one treatment. The samples are plotted with different XY-axes. G log2-FC of photon versus 12C-ion with Pearson correlation analysis (r = 0.9725, p < 0.0001). H log2-FC of DHT versus photon and I log2-FC of DHT versus 12C-ion. Respective unique and intersected DEG are assigned to different sample colors with the n-number in the legends. The names of strikingly altered DEG are indicated. L Correlation graphs of log2-FC of DEG-photon and DEG-12C-ion in DU145 cells. DEG-12C-ion (orange; n = 26) and DEG-12C-ion (green, n = 26) were all unique without intersection. Some DEG-photon in DU145 cells (n = 5) intersected with DEG-photon in LNCaP cells and are marked with an asterisk

DEG in LNCaP cells

Essentially, LNCaP cells (with functional TP53 and AR signaling) displayed considerable alterations of certain mRNA levels upon photon and 12C-ion irradiation. A higher proportion of DEG-photon (n = 809) and of DEG-12C-ion (n = 797) were downregulated (blue) than upregulated (red) (Fig. 5A, B). Concerning DEG-DHT (n = 1304), a higher proportion was upregulated (red) than downregulated (blue) (Fig. 5C). The individual names of DEG are listed (supplementary information).

Furthermore, correlation plots depicting the fold-change (FC) of DEG-photon, DEG-12C-ion and DEG-DHT were constructed. DEG-photon considerably overlapped with DEG-12C-ion (black; n = 577) (Fig. 5D). DEG-photon that were unique are labeled in orange (n = 232) and those that were unique for DEG-12C-ion are labeled in green (n = 220) (Fig. 5D). Moreover, we compared DEG-DHT with DEG-photon or DEG-12C-ion; once with log2-FC-photon on the Y-axis (Fig. 5E) and once with Log2-FC-12C-ion on the Y-axis (Fig. 5F). We found that the majority of DEG-DHT were unique (n = 904, olive). A prominent fraction of DEG-DHT intersected with DEG-12C-photon (orange; n = 55), with DEG 12C-ion (cyan; n = 53) or with both (purple; n = 292).

Next, we focused on those DEG that have high expression levels and are strongly changed upon treatments. To that end, we arbitrarily restricted the comparison to the DEG of the top 50th percentile with a minimum threefold change (log2FC > 1.6) under photon, 12C-ion or DHT treatment. These DEGs were then merged and displayed in three XY-graphs (Fig. 5G, H, I). Strikingly, a strong correlation between DEG-photon versus DEG-12C-ion became obvious in LNCaP cells (Pearson r = 0.9725; p < 0.0001) reflecting the particularly high degree of congruent regulation by 12C-ion and photon in the top level DEG. The extremely altered mRNAs are named (Fig. 5G). Alongside, we displayed the values of the same samples for log2-FC DHT (X-axes) versus log2-FC photon (Y-axes) (Fig. 5H) and for log2-FC DHT (X-axes) versus log2-FC 12C-ion (Y-axes) (Fig. 5I). It is noteworthy that the genes that are predominantly induced by DHT remain unaffected by photon- or 12C-ion-irradiation (e.g.: SGK1, NPPC, HPGD, NDRG1). On the other hand, those genes that are majorly induced by photons and 12C-ions remain unaffected by DHT (e.g. CDKN1A, ZMAT3, MDM2) (Fig. 5H, I).

DEG in DU145 cells

In DU145 cells (dysfunctional TP53), the number of DEG-photon (n = 26) and DEG-12C-ions (n = 26) (Fig. 5J, K) were significantly lower (1–2 magnitudes) than in LNCaP cells (Fig. 5A, B). In addition, the amplitude of the fold changes was considerably lower and the triplicate values of fold changes appear less conform. Furthermore, all DEG-photon and DEG-12C-ion were unique in DU145 cells without intersection. The DEG showing the most extreme fold changes are identified with labels. Some mRNAs (n = 5) of DEG-photon in DU145 cells intersected with DEG-photon in LNCaP cells and are indicated by an asterisk in the correlation graph (Fig. 5L). For DEG-12C-ion, no intersection was observed between DU145 and LNCaP cells.

Identification of putative 12C-ion and photon dependent pathways in LNCaP and DU145 cells

As a direct approach to identify putative 12C-ion and photon dependent pathways, GSEA was applied to 12C-ion irradiated versus photon irradiated samples. This allowed us to identify Notch signaling as significantly enriched in 12C-ion versus photon irradiated LNCaP cells. In DU145 cells, we identified the unfolded protein response, the reactive oxygen species pathway and oxidative phosphorylation as significantly enriched in 12C-ion versus photon irradiated samples (supplementary information).