Modulating Properties of Piroxicam, Meloxicam and Oxicam Analogues against Macrophage-Associated Chemokines in Colorectal Cancer

The mechanisms underlying the antineoplastic effects of oxicams have not been fully elucidated. We aimed to assess the effect of classic and novel oxicams on the expression/secretion of macrophage-associated chemokines (RTqPCR/Luminex xMAP) in colorectal adenocarcinoma cells, and on the expression of upstream the non-steroidal anti-inflammatory drug (NSAID)-activated genes NAG1, NFKBIA, MYD88, and RELA, as well as at the chemokine profiling in colorectal tumors. Meloxicam downregulated CCL4 9.9-fold, but otherwise the classic oxicams had a negligible/non-significant effect. Novel analogues with a thiazine ring substituted with arylpiperazine and benzoyl moieties significantly modulated chemokine expression to varying degree, upregulated NAG1 and NFKBIA, and downregulated MYD88. They inhibited CCL3 and CCL4, and their effect on CCL2 and CXCL2 depended on the dose and exposure. The propylene linker between thiazine and piperazine nitrogens and one arylpiperazine fluorine substituent characterized the most effective analogue. Only CCL19 and CXCL2 were not upregulated in tumors, nor was CXCL2 in tumor-adjacent tissue compared to normal mucosa. Compared to adjacent tissue, CCL4 and CXCL2 were upregulated, while CCL2, CCL8, and CCL19 were downregulated in tumors. Tumor CCL2 and CCL7 increased along with advancing T and CCL3, and CCL4 along with the N stage. The introduction of arylpiperazine and benzoyl moieties into the oxicam scaffold yields effective modulators of chemokine expression, which act by upregulating NAG1 and interfering with NF-κB signaling.


Introduction
Colorectal cancer (CRC) is one of the most common malignancies worldwide. Its incidence is higher among individuals with a history of polyps and patients with inflammatory bowel disease [1]. Chemokines, which are involved in immune cell trafficking, constitute a link between inflammation and cancer. Moreover, they mediate interactions between tumor Figure 1. Mechanisms of NSAIDs' action: COX-dependent and exemplary COX-independent targets relevant for cancer development. The COX-independent targets evaluated in the present study are bolded, and are indicated by a green frame. NSAIDs, non-steroidal anti-inflammatory drugs; COX, cyclooxygenase; mPGES, microsomal prostaglandin E2 synthase; PGFS, prostaglandin F synthase; PGDS, prostaglandin D synthase; PGIS, prostacyclin synthase; TXS, thromboxane synthase; NAG-1, NSAIDs-activated gene; AP-1, activator protein 1; HSPs, heat shock proteins; PPARs, peroxisome proliferator-activated receptor; CA, carbonic anhydrase; SERCA, sarco (endo)-plasmic calcium ATPase; ETC, electron transport chain.
It is well recognized now that the antineoplastic properties of NSAIDs go beyond COX inhibition [27][28][29], and a better understanding of the underlying molecular mechanisms is crucial to enable the design of improved drugs for molecular-targeted treatment strategies in CRC [30]. The aim of the present study was to determine whether classic oxicams (piroxicam and meloxicam) and their novel analogues affect the expression and secretion of MCPs (CCL2/MCP-1, CCL7/MCP-3, and CCL8/MCP-2) and MIPs (CCL3/MIP-1α, CCL4/MIP-1β, CCL19/MIP-3β, and CXCL2/MIP-2) in colorectal adenocarcinoma cell lines. In order to shed some light on a mechanism of oxicam action, its impact on NAG1 (GDF15)-a growth factor belonging to the tumor growth factor (TGF)-β superfamily and an NSAID-activated gene-as well as on the expression of NFκB-associated RELA (encoding NFκB p65 subunit), MYD88 (myeloid differentiation primary response 88, an adapter protein), and NFKBIA (encoding IκBα, an NFκB inhibitor) was additionally evaluated. In the light of the paucity of clinical data on MCP and MIP in CRC other than prototypical MCP-1 and MIP-1, their transcriptional patterns in transformed and non-transformed colorectal mucosa from CRC patients in reference to normal colonic tissue, cancer pathology, and the concentration of circulating chemokines were investigated. Mechanisms of NSAIDs' action: COX-dependent and exemplary COX-independent targets relevant for cancer development. The COX-independent targets evaluated in the present study are bolded, and are indicated by a green frame. NSAIDs, non-steroidal anti-inflammatory drugs; COX, cyclooxygenase; mPGES, microsomal prostaglandin E2 synthase; PGFS, prostaglandin F synthase; PGDS, prostaglandin D synthase; PGIS, prostacyclin synthase; TXS, thromboxane synthase; NAG-1, NSAIDs-activated gene; AP-1, activator protein 1; hSPs, heat shock proteins; PPARs, peroxisome proliferatoractivated receptor; CA, carbonic anhydrase; SERCA, sarco (endo)-plasmic calcium ATPase; ETC, electron transport chain.

MCP and MIP Expression in Colonic Adenocarcinoma Cell Lines
Two cell lines, namely Caco-2 and hCT 116, were tested by RTqPCR for the expression level of MCP and MIP chemokines. As depicted in Figure 2, there was great variability in the expression level of particular genes. In the Caco-2 cell line, CCL2 and CXCL2 were highly expressed. The expression level of CCL3 and CCL4, although markedly lower, still allowed for their reliable quantification. In the hCT 116 cell line, CXCL2 expression was the highest, although lower than in Caco-2, and that of CCL3 could still be credibly measured. The expression of the remaining genes was too low to be reliably quantified, and therefore they were not analyzed further. still allowed for their reliable quantification. In the HCT 116 cell line, CXCL2 expression was the highest, although lower than in Caco-2, and that of CCL3 could still be credibly measured. The expression of the remaining genes was too low to be reliably quantified, and therefore they were not analyzed further.

Effect of Oxicams on MCP and MIP Expression in Colonic Adenocarcinoma Cell Lines
The Caco-2 and HCT 116 cells were treated with oxicams: the classic drugs piroxicam (compound #6) and meloxicam (compound #7), and their five new analogues, denoted as compounds #1-5. In order to analyze the effect of time, the cells were incubated with 200 μM drugs for 6 and 24 h. In order to analyze the dose-response, the cells were incubated for 24 h with 5, 50 and 200 μM concentrations of oxicams.

Time-Dependent Response to Oxicams
In Caco-2 cells, the classic oxicams tended to downregulate CCL2, CCL3, CCL4 and CXCL2 expression after 24-h treatment (1.3 to 3.0-fold), but the effect was significant only in the case of meloxicam and CCL4 expression (9.9-fold downregulation). A shorter incubation resulted in the slight upregulation of CCL3 and CXCL2 (by 1.3 to 1.6-fold), which was significant in the case of meloxicam and CCL3 (1.7-fold upregulation) ( Figure 3).

Effect of Oxicams on MCP and MIP Expression in Colonic Adenocarcinoma Cell Lines
The Caco-2 and hCT 116 cells were treated with oxicams: the classic drugs piroxicam (compound #6) and meloxicam (compound #7), and their five new analogues, denoted as compounds #1-5. In order to analyze the effect of time, the cells were incubated with 200 µM drugs for 6 and 24 h. In order to analyze the dose-response, the cells were incubated for 24 h with 5, 50 and 200 µM concentrations of oxicams.

Time-Dependent Response to Oxicams
In Caco-2 cells, the classic oxicams tended to downregulate CCL2, CCL3, CCL4 and CXCL2 expression after 24-h treatment (1.3 to 3.0-fold), but the effect was significant only in the case of meloxicam and CCL4 expression (9.9-fold downregulation). A shorter incubation resulted in the slight upregulation of CCL3 and CXCL2 (by 1.3 to 1.6-fold), which was significant in the case of meloxicam and CCL3 (1.7-fold upregulation) ( Figure 3).
The expression of CCL2 and CXCL2 was affected primarily by compounds #1 to 3, and the effect was either inhibitory or stimulatory depending on the length of exposure. A shorter incubation with #1 to 3 downregulated CCL2 (3.4-, 2.3-and 1.6-fold, respectively) and with #1 and #2 downregulated CXCL2 (2.5-and 2.1-fold, respectively). In turn, longer incubation with #2 and #3 upregulated CCL2 (5.8-and 4.1-fold, respectively) and CXCL2 (2.4-and 3.7-fold, respectively). Compound #5 had a slightly inhibitory effect on CXCL2 expression as well, although without a difference between the 6-and 24-h treatment (1.6-and 1.5-fold, respectively) ( Figure 3). The bars and whiskers represent the mean ± SD expression ratio of the normalized relative quantities [NRQ] obtained for cells treated with 200 μM oxicams for 6 and 24 h, and the corresponding untreated controls [T/U]. The significant differences between the treated and untreated cells, analyzed using a t-test for paired samples, are indicated by asterisks (*). The expression ratio of 1 (no change) is indicated by a red solid reference line, while the expression ratio of 5 (5-fold upregulation) and the expression ratio of 0.2 (5-fold downregulation) are indicated by an orange dashed reference line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.
The expression of CCL2 and CXCL2 was affected primarily by compounds #1 to 3, and the effect was either inhibitory or stimulatory depending on the length of exposure. A shorter incubation with #1 to 3 downregulated CCL2 (3.4-, 2.3-and 1.6-fold, respectively) and with #1 and #2 downregulated CXCL2 (2.5-and 2.1-fold, respectively). In turn, longer incubation with #2 and #3 upregulated CCL2 (5.8-and 4.1-fold, respectively) and CXCL2 (2.4-and 3.7-fold, respectively). Compound #5 had a slightly inhibitory effect on CXCL2 expression as well, although without a difference between the 6-and 24-h treatment (1.6-and 1.5-fold, respectively) ( Figure 3). (d) CXCL2. The bars and whiskers represent the mean ± SD expression ratio of the normalized relative quantities [NRQ] obtained for cells treated with 200 µM oxicams for 6 and 24 h, and the corresponding untreated controls [T/U]. The significant differences between the treated and untreated cells, analyzed using a t-test for paired samples, are indicated by asterisks (*). The expression ratio of 1 (no change) is indicated by a red solid reference line, while the expression ratio of 5 (5-fold upregulation) and the expression ratio of 0.2 (5-fold downregulation) are indicated by an orange dashed reference line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.
In the hCT 116 cell line, CCL3 was downregulated by compound #2 (3-and 7.4-fold after 6 and 24-h treatment) and compound #3 (2.6-and 3.7-fold, respectively) in a timedependent manner. Compound #5 downregulated CCL3 to the same degree (3.2-fold) independently from the length of treatment, and the inhibitory effect of compound #4 (3.6-fold) was significant only in the 24-h cultures ( Figure 4).

Dose-Dependent Response to Oxicams
In Caco-2 cells, compounds #2, #4 and #5 dose-dependently downregulated CCL3 and CCL4 expression. The inhibitory effect of the 50 μM concentration was significant for all of these compounds in the case of CCL3 expression (6.2-, 2.3-, and 2.1-fold, respectively), and only for #2 (4-fold) in the case of CCL4 expression ( Figure 5). The bars and whiskers represent the mean ± SD expression ratio of the normalized relative quantities [NRQ] obtained for cells treated with 200 µM oxicams for 6 and 24 h, and the corresponding untreated controls [T/U]. The significant differences between the treated and untreated cells, analyzed using a t-test for paired samples, are indicated by asterisks (*). An expression ratio of 1 (no change) is indicated by a red solid reference line, while an expression ratio of 5 (5-fold upregulation) and an expression ratio of 0.2 (5-fold downregulation) are indicated by an orange dashed reference line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.

Effect of Oxicams on MCP and MIP Secretion by Caco-2 Cells
The oxicam effect on MCP and MIP chemokines at the protein level was tested on Caco-2 cells. The concentrations of MCP-1/CCL2, MIP-1α/CCL3, MIP-1β/CCL4 and MCP-3/CCL7 released into the medium were simultaneously determined using Luminex xMAP technology.
Compared to untreated cells, compounds #1 to 5 downregulated MIP-1β/CCL4 secretion. Likewise, compounds #1, #2, #4 and #5 downregulated the secretion of MIP-1α/CCL3. Except for MIP-1α/CCL3 and compound #1, the effect was more marked in the 24-h cultures. Compound #2 and meloxicam (#7) also had a slight inhibitory effect in 24-h cultures on MCP-3/CCL7 secretion (Figure 7). The bars and whiskers represent the mean ± SD expression ratio of the normalized relative quantities [NRQ] obtained for cells treated with 5, 50, and 200 μM oxicams for 24 h, and the corresponding untreated controls [T/U]. The significant differences between the treated and untreated cells, analyzed using a t-test for paired samples, are indicated by asterisks (*). An expression ratio of 1 (no change) is indicated by a red solid reference line, while an expression ratio of 5 (5-fold upregulation) and an expression ratio of 0.2 (5-fold downregulation) are indicated by an orange dashed reference line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.
In a dose-dependence experiment, an increasing concentration of compounds #2, #4, and #5 decreased the MIP-1α/CCL3 secretion in a stepwise manner. Likewise, the MIP-1β/CCL4 secretion decreased along with the increasing concentration of compounds #1 to 5 ( Figure 8). Compounds #1 to 3 had a stimulatory effect on CXCL2 expression. The 5 μM dose was slightly inhibitory in the case of compound #1 (1.5-fold), and had no impact in the case of compounds #2 and #3. In turn, the 50 μM dose was significantly stimulatory for all of these compounds (2.7-, 3-, and 4.7-fold, respectively) ( Figure 6).

Effect of Oxicams on MCP and MIP Secretion by Caco-2 Cells
The oxicam effect on MCP and MIP chemokines at the protein level was tested on Caco-2 cells. The concentrations of MCP-1/CCL2, MIP-1α/CCL3, MIP-1β/CCL4 and MCP-3/CCL7 released into the medium were simultaneously determined using Luminex xMAP technology.

Effect of Oxicams on NAG1 Expression in Caco-2 and hCT 116 Cells
In order to explore the COX-independent mechanism of oxicam action, the potential effect of classic oxicams and their novel analogues on the expression of NSAID-activated gene-1 (NAG1) was determined.
3/CCL7 released into the medium were simultaneously determined using Luminex xMAP technology.
Compared to untreated cells, compounds #1 to 5 downregulated MIP-1β/CCL4 secretion. Likewise, compounds #1, #2, #4 and #5 downregulated the secretion of MIP-1α/CCL3. Except for MIP-1α/CCL3 and compound #1, the effect was more marked in the 24-h cultures. Compound #2 and meloxicam (#7) also had a slight inhibitory effect in 24-h cultures on MCP-3/CCL7 secretion (Figure 7). . The bars represent the mean ± SD (whiskers) ratio of the chemokine concentration in the medium from cells treated with 200 μM oxicams for 6 and 24 h, as compared to untreated controls, expressed as a percentage. The significant differences between the secretion from treated and untreated cells, analyzed using a t-test for one mean, are indicated by asterisks (*). The reference secretion of the control cells is indicated by a red solid line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.
In a dose-dependence experiment, an increasing concentration of compounds #2, #4, and #5 decreased the MIP-1α/CCL3 secretion in a stepwise manner. Likewise, the MIP-1β/CCL4 secretion decreased along with the increasing concentration of compounds #1 to 5 ( Figure 8). . The bars represent the mean ± SD (whiskers) ratio of the chemokine concentration in the medium from cells treated with 200 µM oxicams for 6 and 24 h, as compared to untreated controls, expressed as a percentage. The significant differences between the secretion from treated and untreated cells, analyzed using a t-test for one mean, are indicated by asterisks (*). The reference secretion of the control cells is indicated by a red solid line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.

Effect of Oxicams on NAG1 Expression in Caco-2 and HCT 116 Cells
In order to explore the COX-independent mechanism of oxicam action, the potential effect of classic oxicams and their novel analogues on the expression of NSAID-activated gene-1 (NAG1) was determined.
In Caco-2 cells, piroxicam (#6), and meloxicam (#7), as well as compounds #1 and #2, had no significant impact on NAG1 at 6 or 24 h, while compound #3 upregulated the gene expression 9-fold after 6-h treatment without an effect after 24 h. In the case of compounds #4 and #5, the expression of NAG1 was significantly upregulated, more markedly following the 6-h than 24-h treatment (respectively, 3.4-and 2.3-fold for #4, and 4.1-and 1.4-fold for compound #5) ( Figure 9). . The bars represent the mean ± SD (whiskers) ratio of the chemokine concentration in the medium from cells treated with 5, 50 and 200 µM oxicams for 24 h, as compared to untreated controls, expressed as a percentage. The significant differences between the secretion from treated and untreated cells, analyzed using a t-test for one mean, are indicated by asterisks (*). The reference secretion of the control cells is indicated by a red solid line. Compound #6 denotes piroxicam, and #7 denotes meloxicam.

Effect of Oxicams on NAG1 Protein Expression in Caco-2, hCT 116, and hT-29 Cells
In order to confirm the stimulatory activity of oxicams on NAG1 protein expression and the protein level, Caco-2, hCT 116, and hT-29 were treated with 50 and 200 µM drug concentrations for 48 h, and the cell protein content was analyzed by Western blotting. The hT-29 line was examined additionally, as the Caco-2 cells did not show any reactivity with anti-NAG1 antibodies.
In hCT 116 cells, all of the oxicams at both concentrations upregulated NAG1, but the effect was statistically significant solely for compound #2, 2. gene-1 (NAG1) was determined.

Effect of Oxicams on the Expression of NFκB-Associated Genes in Caco-2 Cells
In addition, the potential effect of the investigated oxicams on genes encoding proteins associated with NF-κB signaling-namely, NFKBIA, MYD88, and RELA-was examined in Caco-2 cells.

Effect of Oxicams on the Expression of NFκB-Associated Genes in Caco-2 Cells
In addition, the potential effect of the investigated oxicams on genes encoding proteins associated with NF-κB signaling-namely, NFKBIA, MYD88, and RELA-was examined in Caco-2 cells.

Effect of Oxicams on the Expression of NFκB-Associated Genes in Caco-2 Cells
In addition, the potential effect of the investigated oxicams on genes encoding proteins associated with NF-κB signaling-namely, NFKBIA, MYD88, and RELA-was examined in Caco-2 cells.

Pairwise Analysis of MCPs and MIPs in Transformed and Non-transformed Tumor-Adjacent Colorectal Mucosa
The MCP and MIP chemokines displayed different expression patterns, as determined with a reversely-transcribed quantitative polymerase chain reaction (RTqPCR). The expression of CCL2 and CCL8 from MCP chemokines and CCL19 from MIP chemokines was significantly downregulated in tumors compared to patient-matched macroscopically normal tissue by 2.1-fold, 5.1-fold, and 2.2-fold, respectively. The expression of CCL4 and CXCL2 from MIP chemokines was significantly, and CCL3 insignificantly, upregulated in tumors compared to tumor-adjacent non-transformed tissue by 1.4-fold, 2.0fold, and 1.3-fold, respectively (Table 1). In order to establish whether the chemokine expression in tumors is indeed downregulated-or rather, if it is upregulated in tumor-surrounding tissue moreso than in The MCP and MIP chemokines displayed different expression patterns, as determined with a reversely-transcribed quantitative polymerase chain reaction (RTqPCR). The expression of CCL2 and CCL8 from MCP chemokines and CCL19 from MIP chemokines was significantly downregulated in tumors compared to patient-matched macroscopically normal tissue by 2.1-fold, 5.1-fold, and 2.2-fold, respectively. The expression of CCL4 and CXCL2 from MIP chemokines was significantly, and CCL3 insignificantly, upregulated in tumors compared to tumor-adjacent non-transformed tissue by 1.4-fold, 2.0-fold, and 1.3-fold, respectively (Table 1). In order to establish whether the chemokine expression in tumors is indeed downregulated-or rather, if it is upregulated in tumor-surrounding tissue moreso than in tumors-the chemokine expression in tumors and the adjacent tissue was compared with normal tissue from patients undergoing polypectomy (n = 20).
The non-paired analysis confirmed significantly lower CCL2 and CCL8 expression in tumors compared to tumor-adjacent tissue. however, compared to the chemokine expression in normal tissue from non-cancer patients, all of the MCPs were significantly upregulated in both tumors and tumor-adjacent tissue. The expression of CCL2 was upregulated 11.1-fold in tumor-adjacent tissue, and 5.3-fold in tumors compared to normal mucosa, in CCL7 2.3-and 1.9-fold, and in CCL8 8.3-and 2.1-fold ( Figure 14). tumors-the chemokine expression in tumors and the adjacent tissue was compared with normal tissue from patients undergoing polypectomy (n = 20). The non-paired analysis confirmed significantly lower CCL2 and CCL8 expression in tumors compared to tumor-adjacent tissue. However, compared to the chemokine expression in normal tissue from non-cancer patients, all of the MCPs were significantly upregulated in both tumors and tumor-adjacent tissue. The expression of CCL2 was upregulated 11.1-fold in tumor-adjacent tissue, and 5.3-fold in tumors compared to normal mucosa, in CCL7 2.3-and 1.9-fold, and in CCL8 8.3-and 2.1-fold ( Figure 14). The non-paired analysis confirmed significantly higher CCL19 and significantly lower CCL4 and CXCL2 expression in tumor-adjacent tissue than in tumors, and a lack of difference in the case of CCL3. Compared to normal tissue from non-cancer patients, the CCL3, CCL4 and CXCL2 expressions were upregulated in both tumor and tumor-adjacent tissue, while CCL19 was upregulated solely in tumor-adjacent tissue (Figure 15). with 95% confidence intervals. The data were analyzed using one-way ANOVA with a Student-Newman-Keuls post hoc test or a Kruskal-Wallis H test with a Conover post hoc test, respectively. The significant between-group differences are indicated by the same type of symbol: *, +, orˆ. NRQ, normalized relative quantities.
The non-paired analysis confirmed significantly higher CCL19 and significantly lower CCL4 and CXCL2 expression in tumor-adjacent tissue than in tumors, and a lack of difference in the case of CCL3. Compared to normal tissue from non-cancer patients, the CCL3, CCL4 and CXCL2 expressions were upregulated in both tumor and tumor-adjacent tissue, while CCL19 was upregulated solely in tumor-adjacent tissue ( Figure 15).

Impact of Cancer Pathology on Chemokine Expression
The chemokine expression in tumors was dependent on the depth of tumor invasion. The expression of CCL2 and CCL7 increased significantly along with the advancing T stage, and that of CCL3 and CCL8 increased non-significantly ( Figure 16).
The chemokine expression in tumors was dependent on lymph node involvement. The expression of CCL3 and CCL4 increased along with the advancing N stage ( Figure 17). The expression of CCL8 (2.6-fold) and CCL19 (2.5-fold) in the tumors of patients with lymph node involvement (N1 or N2) was non-significantly higher than that in the tumors of patients without lymph node metastasis ( Figure 16). In the case of CCL8, the fold-change in expression (tumor-to-adjacent) in N0 vs. N1/2 patients differed significantly by 3.7-fold (0.1 vs. 0.37, p = 0.021).
The local expression of MCPs and MIPs was not dependent on the presence of distant metastases (stage M) or the tumor histopathological grade (G).
The expression of CCL7 tended to be higher in rectal than colonic tumors by 1.8-fold (p = 0.086), without changes in the non-affected mucosa with respect to the tumor location. The CCL8, in turn, was significantly (p = 0.002) more markedly expressed in non-affected mucosa from right-sided tumors, 2.7-fold compared to rectal tumors and 1.8-fold compared to left-sided tumors.

Interrelationship between the Chemokine Expression in the Bowel and Circulating MCPs and MIPs
In general, the expression of MCP and MIP chemokines was more tightly interrelated in tumor tissue than normal or non-transformed adjacent tissue. The MCPs were more tightly related to each other than to MIPs. The expression of CCL2 correlated with all of the other chemokines (also in tumor-adjacent tissue), while CCL19 was the least correlated gene. Except for an inverse relation between CXCL2 and CCL19 in non-transformed tumor-adjacent tissue, all of the correlations between the chemokine genes were positive ( Table 2).
cancer patients: (a) CCL2; (b) CCL7; (c) CCL8. The data are presented as means (CCL2) or medians (CCL7 and CCL8) with 95% confidence intervals. The data were analyzed using one-way ANOVA with a Student-Newman-Keuls post hoc test or a Kruskal-Wallis H test with a Conover post hoc test, respectively. The significant between-group differences are indicated by the same type of symbol: *, +, or ^. NRQ, normalized relative quantities.
The non-paired analysis confirmed significantly higher CCL19 and significantly lower CCL4 and CXCL2 expression in tumor-adjacent tissue than in tumors, and a lack of difference in the case of CCL3. Compared to normal tissue from non-cancer patients, the CCL3, CCL4 and CXCL2 expressions were upregulated in both tumor and tumor-adjacent tissue, while CCL19 was upregulated solely in tumor-adjacent tissue (Figure 15). . The data are presented as means (CCL19) or medians (CCL3, CCL4, and CXCL2) with 95% confidence intervals. The data were analyzed using one-way ANOVA with a Student-Newman-Keuls post hoc test or a Kruskal-Wallis H test with a Conover post hoc test, respectively. The significant betweengroup differences are indicated by the same type of symbol: *, +, or ^. NRQ, normalized relative quantities.

Impact of Cancer Pathology on Chemokine Expression
The chemokine expression in tumors was dependent on the depth of tumor invasion. The expression of CCL2 and CCL7 increased significantly along with the advancing T stage, and that of CCL3 and CCL8 increased non-significantly ( Figure 16). . The data are presented as means (CCL19) or medians (CCL3, CCL4, and CXCL2) with 95% confidence intervals. The data were analyzed using one-way ANOVA with a Student-Newman-Keuls post hoc test or a Kruskal-Wallis H test with a Conover post hoc test, respectively. The significant between-group differences are indicated by the same type of symbol: *, +, orˆ. NRQ, normalized relative quantities.
The local expression of CCL2, CCL3 and CCL4 was compared to the respective chemokine concentration in the systemic circulation (determined using Luminex xMAP technology). There was a positive association between the CCL4 expression in tumors and its systemic concentration (ρ = 0.35, p = 0.022, n = 43). man-Keuls post hoc test or a Kruskal-Wallis H test with a Conover post hoc test, respectively. The significant betweengroup differences are indicated by the same type of symbol: *, +, or ^. NRQ, normalized relative quantities.

Impact of Cancer Pathology on Chemokine Expression
The chemokine expression in tumors was dependent on the depth of tumor invasion. The expression of CCL2 and CCL7 increased significantly along with the advancing T stage, and that of CCL3 and CCL8 increased non-significantly ( Figure 16).  (d) CCL8. The data are presented as means with 95% confidence intervals, and as Spearman correlation coefficients (ρ). NRQ, normalized relative quantities. The chemokine expression in tumors was dependent on lymph node involvement. The expression of CCL3 and CCL4 increased along with the advancing N stage (Figure 17). The expression of CCL8 (2.6-fold) and CCL19 (2.5-fold) in the tumors of patients with lymph node involvement (N1 or N2) was non-significantly higher than that in the tumors of patients without lymph node metastasis ( Figure 16). In the case of CCL8, the foldchange in expression (tumor-to-adjacent) in N0 vs. N1/2 patients differed significantly by 3.7-fold (0.1 vs. 0.37, p = 0.021). . The data are presented as means with 95% confidence intervals, and as a Spearman correlation coefficient (ρ) (CCL3 and CCL4). The data regarding CCL8 and CCL19 were analyzed using a t-test for independent samples. NRQ, normalized relative quantities.
The local expression of MCPs and MIPs was not dependent on the presence of distant metastases (stage M) or the tumor histopathological grade (G).
The expression of CCL7 tended to be higher in rectal than colonic tumors by 1.8-fold (p = 0.086), without changes in the non-affected mucosa with respect to the tumor location. The CCL8, in turn, was significantly (p = 0.002) more markedly expressed in non-affected mucosa from right-sided tumors, 2.7-fold compared to rectal tumors and 1.8-fold compared to left-sided tumors.

Interrelationship between the Chemokine Expression in the Bowel and Circulating MCPs and MIPs
In general, the expression of MCP and MIP chemokines was more tightly interrelated in tumor tissue than normal or non-transformed adjacent tissue. The MCPs were more tightly related to each other than to MIPs. The expression of CCL2 correlated with all of the other chemokines (also in tumor-adjacent tissue), while CCL19 was the least correlated gene. Except for an inverse relation between CXCL2 and CCL19 in non-transformed The data are presented as means with 95% confidence intervals, and as a Spearman correlation coefficient (ρ) (CCL3 and CCL4). The data regarding CCL8 and CCL19 were analyzed using a t-test for independent samples. NRQ, normalized relative quantities.

Discussion
Piroxicam and meloxicam are approved as non-cancer drugs with anti-tumor activity, and are listed in the Repurposing Drugs in Oncology (ReDO) database [31], although the exact mechanisms of their antineoplastic activity are unclear. What seems to be the most important, i.e., the efficacy of piroxicam as a chemopreventive agent, has been demonstrated in animal models [32] as well as in human clinical trials [33]. herein, we sought to examine its effect on the expression of MCPs and MIPs. Of all of the studied MCP chemokines, the expression of CCL2 in Caco-2 cells was the highest, while its levels were barely detectable in hCT 116, corroborating the notion of undetectable chemokine expression in another colon adenocarcinoma cell line, LS174T [34]. Unfortunately, the expressions of CCL7 and CCL8 were too low for analysis in both cell lines as well. Contrary to novel oxicam analogues, neither piroxicam nor meloxicam had a significant impact on MCP-1/CCL2 expression or secretion. The CCL2 in Caco-2 cells displayed an intriguing expression pattern in response to treatment with compounds #1-3. It was downregulated by 24-h incubation with a 50 µM dose or 6-h incubation with a 200 µM dose, but was substantially upregulated by 24-h incubation with a 200 µM dose. Such a biphasic response may imply the involvement of two different regulatory mechanisms: one operating at a low dose or shorter exposure, and the other at a high dose and longer exposure. Importantly, the observation was confirmed at the protein level and mimicked by CXCL2, an MIP chemokine, which was also measurable in hCT 116. Like piroxicam and meloxicam, compounds #4 and #5 had no significant impact on MCP-1/CCL2 expression and secretion, except the inhibitory effect of compound #5 on gene expression in Caco-2.
Structurally, studied novel analogues are distinguished from classic oxicams by two substitutions in the thiazine ring of the oxicam scaffold: arylpiperazine pharmacophore at position 2 replacing the methyl substituent at position 2, and a benzoyl moiety replacing 2-peridocarbamoyl at position 3. Arylpiperazine has high electron-withdrawing properties, which translate into enhanced anti-inflammatory properties [35]. While an upgraded anti-inflammatory activity may explain the inhibitory effect of oxicam analogues on MIP-1α/CCL3 and MIP-1β/CCL4 observed for all of the novel compounds, it does not clarify the selective effect of #1 to 3 on MCP-1/CCL2. Compounds #1 to 3 differ from non-effective compounds #4 and 5 with a linker between the thiazine and arylpiperazine rings. Our results imply that a three-carbon propylene linker is crucial for the modulation of MCP-1/CCL2 expression and secretion, while the two-carbon oxyethylene linker present in compounds #4 and 5, shown to enhance analgesia [36], is ineffective. Compounds #2 and #3 were more effective than compound #1, indicating the significance of electronwithdrawing fluoro-substituents at the arylpiperazine pharmacophore (compounds #2 and #3), and the benzoyl moiety (compound #3), which are absent in compound #1. Regarding the upregulation of MCP-1/CCL2 in response to a high oxicam dose and longer exposure, the expression of MCP-1 has been shown to be upregulated several-fold in ovarian cancer by chemotherapy, and the effect has been reproducible at the mRNA and protein levels, both in vitro and in vivo [34]. Likewise, radiation has been demonstrated to induce MCP-1 secretion from breast cancer cells [37]. Interestingly, the MCP-1 response to chemotherapy could be mimicked by the inhibition of the JAK/STAT signaling pathway, implying a suppressive role of this pathway towards MCP-1 [34]. Therefore, the overexpression of JAK/STAT in colorectal tumors, reported elsewhere and positively associated with cancer advancement and poor prognosis [38], may explain the lower CCL2 expression in tumors compared to tumor-adjacent tissue observed in the present study. In turn, the inhibitory effect of NSAIDs on JAK/STAT signaling, potentially alleviating their suppressive activity towards MCP-1, might explain the upregulating effect of compounds #1 to 3 on MCP-1/CCL2. It is also possible that MCP-1/CCL2 overexpression is a part of the stress response, especially given that these particular oxicam analogues are effective inducers of apoptosis and cell cycle arrest (unpublished results). They have also induced such a biphasic response in colonic cell lines regarding arginase-2 expression from the L-arginine/nitric oxide pathway [25].
The expression and secretion of MIP-1α/CCL3 and MIP-1β/CCL4 were only nonsignificantly decreased by the classic oxicams, with the exception of a 9.9-fold downregulation of CCL4 in Caco-2 cells and a weak downregulation of CCL3 in hCT 116 by meloxicam. In turn, MIP-1α/CCL3 and MIP-1β/CCL4 were consistently downregulated by novel oxicam analogues. Compounds #1 to 5 inhibited the MIP-1α/CCL3 and MIP-1β/CCL4 expression in both Caco-2 and hCT 116 cells, and secretion in Caco-2. Intriguingly, compound #3 suppressed CCL3 expression dose-and time-dependently solely in hCT 116, but consistently had no effect in Caco-2 cells, either on chemokine expression or secretion. It has also failed to reduce the CCL4 expression in Caco-2, which, owing to too-low gene expression, could not be confronted with hCT 116. however, the lack of effect at the transcript level did not translate to the protein level, as compound #3 reduced MIP1β/CCL4 secretion by 30%.
The inhibitory effect on MIP-1α/CCL3 and MIP-1β/CCL4 was the strongest and the most consistent in the case of compound #2. This may imply the superiority of three-carbon propylene linker between thiazine and arylpiperazine nitrogen atoms over the oxoethylene linker present in compounds #4 and #5. It may also indicate that one electron-withdrawing substituent is more effective than none (compound #1) or two such substituents (compound #3). The impact of the studied oxicams on CXCL2 expression resembled the effect observed for CCL2. The gene expression in Caco-2 was downregulated during a 6-h incubation with a 200-µM dose of compounds #1 to 3 and a longer exposure to 50 µM compound #2, but was upregulated by a 200-µM dose of #2 and #3 upon 24-h exposure. In hCT 116, compounds #1 to 3 had a positive time-and dose-dependent effect on CXCL2 expression without an inhibitory impact at lower doses or upon shorter exposure. The exception was a slight inhibition by the 5 µM compound #1. As MIP-2/CXCL2 was not included in the cytokine panel, the effect could not be confirmed on the protein level. In addition, compound #5 inhibited CXCL2 expression at 50 and 200 µM. The literature data on MIP-2/CXCL2, also known as GRO-2 (or β), are scanty and confusing concerning its cancer-related up-or downregulation [39,40] or activity [41]. Nonetheless, the cumulative evidence ascribes MIP-2 a pro-tumorigenic role as a proangiogenic factor and a potent chemoattractant for neutrophils and MDSCs, involved in cancer-related immunosuppression. Moreover, only recently, soluble MIP-2 has been proposed as a marker of the treatment response to immune checkpoint inhibitor therapy against programmed death (PD)-1 in advanced or recurrent non-small-cell lung cancer [42]. Although an early study [43] found the CXCL2 expression in normal and neoplastic colonic tissue to be only minimal, subsequent analyses of McLean et al. [44] and Doll et al. [45] demonstrated significant chemokine upregulation in colonic adenocarcinomas.
In addition to being COX inhibitors, NSAIDs, including oxicams, have been shown to operate via mechanisms independent of the COX/prostaglandin E pathway. Among others, they have been shown to induce the expression of the NSAID-activated gene (NAG1), which is also known as a growth differentiation factor (GDF)-15 or macrophage inhibitory cytokine (MIC)-1 [46]. NAG1/GDF-15/MIC-1 belongs to a TGF-β super-family, and controls the expression of several inflammatory mediators, including MCP chemokines [47]. Its activation has been linked with a reduced propensity to develop intestinal polyps and tumors in mice (reviewed in [48]), which, combined with its expression being regulated by NSAIDs, evoke an interest in NAG1/GDF-15/MIC-1 as a chemopreventive agent. Therefore, we examined the oxicam effect on NAG1 and demonstrated that novel oxicams, but not classic drugs, stimulate its expression. The induction of NAG1 was observed in both COX-expressing (Caco-2), and non-expressing (HCT 116) cell lines, although the effect seemed to be more marked in hCT 116 cells. The greater responsiveness of hCT 116 observed in the present study is consistent with the inverse relationship between NAG1 and COX expression reported elsewhere [49]. All of the studied compounds, to a varying degree, upregulated NAG1, although compounds #2 and #3 seemed to be slightly more effective, indicating the significance of both the propylene linker between thiazine and arylpiperazine nitrogen atoms and fluoride substituents at arylpiperazine and benzoyl moieties. Compounds #2 and #3 were also the ones to significantly upregulate the NAG1 protein in hCT 116 and hT-29 cells. The ability of novel oxicam analogues to act by inducing NAG1 expression is of particular clinical relevance in regard to IBD patients, both in terms of chemoprevention and mucosal healing. Recently, a muco-protective role in the gastrointestinal tract was attributed to NAG1/GDF-15/MIC-1, as it has been shown to promote early epithelial restitution following intestinal injury [50]. As such, the upregulation of NAG1 expression observed in this study might be one of the molecular mechanisms behind the declared reduced gastrointestinal toxicity of novel oxicam analogues [24].
As Caco-2 cells displayed no NAG1 immunoreactivity, we examined oxicam's effect on their expression of NFKBIA, MYD88 and RELA, i.e., genes associated with NF-κB signaling, known to control MCP and MIP expression as well as being regulated by NSAIDs. The analogues characterized by the presence of the three-carbon propylene linker (compounds #1-3), but not classic oxicams, substantially upregulated the NF-κB inhibitor (NFKBIA) while downregulating an adaptor protein (MYD88). The effect was strengthened by the presence of fluoro-substituents at the arylpiperazine and benzoyl moieties.
Leukocyte-attracting chemokines are key mediators of inflammation, displaying a number of pro-neoplastic activities [2][3][4][5]. however, there is a paucity of clinical reports regarding MCP and MIP chemokines, especially those conducted on representative populations/sample sets, although data derived from in vitro and animal studies are rapidly accumulating. In this study, we showed that none of the evaluated MCP chemokines were upregulated in tumors compared to non-transformed tumor-adjacent tissue. Similar results, obtained in an RNA-Seq analysis of five patient-matched samples of normal mucosa, adenoma, and carcinoma have recently been published [51]. In fact, the expression of CCL2 and CCL8 was downregulated in the present study two-and five-fold, respectively. Although surprising at first, it transpired not to be synonymous with chemokine downregulation in cancer. As revealed by comparison with colonic mucosa from non-cancer individuals, both transformed and non-transformed tissue from CRC patients had upregulated CCL2 (5.3-and 11.1-fold, respectively), CCL7 (1.9-and 2.3-fold), and CCL8 (2.1-and 8.3-fold), although non-significantly in the case of tumor CCL8. Moreover, the expression of all of the MCPs increased along with the advancing T stage and a fold-change in CCL8 expression (tumor-to-adjacent) was higher by 3.7-fold in patients with lymph node involvement. Corroborating our observation of CCL2 s association with tumor progression, the limited reports on MCPs in CRC have indicated, albeit not unanimously [52], an increase in CCL2 expression along with an advancing clinical stage, which has been accompanied by the accumulation of TAMs [53,54]. The particularly close association of MCP with metastatic CRC [55,56] has been indicated in functional studies, in which tumor-derived MCP-1 enabled efficient cancer cell extravasation [55] and mediated the stimulatory effect of alcohol on invasion and metastasis [57].
Non-transformed tumor-surrounding tissue is not passive, and has repeatedly been shown to have an altered molecular landscape [25,[58][59][60] despite the lack of morphological and histological changes. Pre-existing cancer-initiating changes in apparently normal mucosa are believed to contribute to the failure of surgical treatment and local tumor recurrence, as well as to the occurrence of synchronous tumors [61]. The upregulation in tumors and in the adjacent tissue may serve different purposes, as was suggested for IL32, the overexpression of which in gastrointestinal tumors reflected proliferative and metastatic potential, and antiapoptotic and immunosuppressive capacity in nontransformed mucosa [58]. Like IL32, the expression of CCL8 was upregulated in the non-transformed tissue of right-sided cancers, which are known to be more aggressive. This may imply the contribution of CCL8 to the molecular mechanism behind the higher recurrence rates of right-compared to left-sided cancers [62]. Moreover, the existence of the CCL8 gradient, that is, an increasing concentration from neoplastic epithelium via stroma to the periphery, has been reported and found to be a driving force of breast cancer metastasis [63]. The significance of MCP-1 expression in non-transformed mucosa has been demonstrated in an animal model of colon carcinogenesis employing MCP-1 KO mice, in which gene deletion downregulated the expression of proinflammatory cytokines and upregulated the suppressor of cytokine signaling (SOCS)-1 in both tumors and the adjacent tissue [64].
The role of MCP-3 in cancer seems to be ambiguous. On one hand, transfection with CCL7 has been shown to retard or completely inhibit tumor growth and prevent formation of metastasis, on the other-MCP-3 facilitates the formation of an environment favoring carcinoma progression and promotes phenotypic transformation in monocytes (reviewed in [65]). Clinically, the association of CCL7/MCP-3 with CRC development has been demonstrated by an over 9-fold up-regulation of its expression in liver-metastases compared to patient-matched primary tumors [66].
Unlike MCPs, the MIP chemokines were upregulated in tumors compared to nontransformed adjacent tissue, although CCL3 overexpression was marginal and non-significant. They were also overexpressed in both tumors (CCL3 by 4.9-fold, CCL4 by 18.6-fold, and CXCL2 by 8.6-fold), and adjacent tissue (CCL3 by 4.9-fold, CCL4 by 11.2-fold, and CXCL2 by 2.7-fold) compared to normal mucosa. The exception was CCL19, expressed at a similar level in tumor and normal tissue, and significantly downregulated in tumors compared to normal mucosa. The overexpression of MIP was associated with lymph node metastasis. Unlike with other chemokines, there seems to be a consensus on the anti-tumor role played by MIP-3β/CCL19, at least in CRC. Its downregulation observed in this study corroborates the findings of others regarding chemokine transcripts [67] as well as protein [67,68]. The CCL19 gene has also been listed among 10 hub genes identified by integrated bioinformatics analysis as having high diagnostic potential for the detection of CRC, in which it was consistently down-regulated [69]. Functional studies have shown the anti-tumor activity of MIP-3β, resulting from its ability to increase the number of anti-tumor immune cells [70], to suppress angiogenesis [68], and to increase the expression and secretion of IFN-γ and IL-12 [71].
The MIP-1α/CCL3 and MIP-1β/CCL4 chemokines may display either anti-tumor or tumor-promoting activity, and the available data seem to support both roles in equal measures. An upregulation of MIP-1α and MIP-1β proteins in colorectal tumors [72] and chemokine elevation in the circulation was associated with disease advancement [73] and a worse prognosis [74]. The CCL4 expression is higher in patient-matched liver metastases than primary tumors [56]. Functional studies have attributed MIP-1α an angiogenesispromoting activity [75], thus corroborating the association of CCL3 with lymph node involvement demonstrated in the present study. MIP-1α/CCL3 exacerbates immunemediated colitis [76], and gene deletion reduces colonic tumorigenesis [77]. Nonetheless, there seems to be equally strong evidence of the anti-tumor activity of MIP-1α and β in CRC. In a large prospective study, a low MIP-1α concentration has been associated with an increased risk for CRC [65]; in another one, increased serum concentrations of MIP-1β were found to predict improved disease-free survival [78]. A comparative analysis of the gene expression patterns between CRC microsatellite-instable (MSI) and microsatellite-stable tumors indicated that CCL3 and/or CCL4 were more strongly expressed in MSI. Indirectly, this links both genes with a favorable prognosis, as MIS in CRC is associated with prolonged survival and lower recurrence rates [79]. Functional studies support the anti-tumor activity of MIP-1α and β in CRC as well. MIP-1α/CCL3 has inhibited the progression of established tumors by the preferential recruitment of anti-tumor immune cells and the production of anti-tumor chemokines [80]. Correspondingly, CCL4 gene transfer has inhibited tumor growth and prolonged the survival of tumor-bearing mice [81].

Study Population
Enrolled patients with histologically confirmed adenocarcinomas of the colon or rectum were admitted to the Department of Surgical Oncology, Regional hospital in Wroclaw (2013-2015) or the First Department and Clinic of General, Gastroenterological and Endocrinological Surgery of Wroclaw Medical University (2012-2013) for curative tumor resection. The patients were subjected to the routine preoperative workup, which included colonoscopy, abdominal and pelvic CT, and a pelvic MRI for rectal cancer. The exclusion criteria included: age < 18 years, ASA physical status classification system >3, emergency surgery, locally advanced cancers which were not amenable to curative resection or gross metastatic disease, tumors requiring en bloc multi-visceral resection, the presence of other synchronous cancers, severe cardiovascular or respiratory disease, severe mental disorders, and immunological diseases requiring the systemic administration of corticosteroids. The resected tumors were staged using the Union for International Cancer Control (UICC) tumor-node-metastasis (TNM) system. The patients' characteristics are summarized in Table 3. As the analyzes were conducted on biobanked material, the cDNA template was not always available for all of the quantified genes. Therefore, the patients' characteristics are presented separately for the CCL2/CCL4, CCL3/CCL7 and CCL8/CCL19/CXCL2 cohorts.
The oxicam dose and the length of treatment were chosen based on literature data and the previously published results of a sulforhodamine B (SRB) assay [25].
A phenol-chloroform extraction was applied to isolate RNA from the clinical samples and cell culture lysates. The isolated RNA was subsequently purified using a Pure-Link™ RNA Mini Kit (Invitrogen). Potential contamination with genomic DNA was avoided owing to the on-column treatment with DNase (PureLink™ DNase Set, Invitrogen). The purified RNA was quantified using a NanoDrop 2000 (Thermo-Fisher Scientific) with the concomitant evaluation of RNA purity (absorbance ratios: 260/280 nm and 260/230 nm). The RNA integrity, in turn, was assessed with LabChip microfluidic technology on the Experion platform, using dedicated Experion RNA StdSens analysis kits (BioRad, herkules, CA, USA).
Aliquots of 500 ng or 1000 ng RNA from clinical samples or cell cultures, respectively, were reversely transcribed following the manufacturer's instructions with an iScript™ cDNA Synthesis Kit (BioRad).
The technical replicates were averaged. The geometric mean of all of the Cq values in a given sample set was calculated. Individual sample Cq values were subtracted from this mean, yielding ∆Cq, and subsequently linearized by a 2 ∆Cq conversion and normalized to the geometric mean of PPIA and RPLP0 (in the case of clinical samples), or to GAPDH (in the case of the cell culture experiments). The resulting values are referred to as a normalized relative quantity (NRQ) [85] and subjected to statistical analysis. The reference gene selection was based on our previous research indicating that PPIA and RPLP0 are the most suitable pair of genes for studies on bowel tissues from CRC patients [86], while GAPDH is stably expressed in Caco-2 and colonic adenocarcinoma cell lines under stressinducing conditions [87].

Protein Analysis
The cell suspensions were sonicated on ice in two 30 s cycles with 40% amplitude using the ultrasonic processor UP 200 (Hielscher Ultrasound Technology, Teltow, Germany). Cell debris was removed by centrifugation (12,500× g, 10 min at 4 • C). The total protein concentration was measured colorimetrically at λ = 562 in the supernatants using the bicinchoninic acid assay (Thermo Fisher Scientific, Waltham, MA, USA) with an Infinite ® M200 plate spectrophotometer (Tecan Group Ltd., Männedorf, Switzerland). Protein samples (10 µg) were diluted 1:1 with 2 × Laemmli sample buffer (Bio-Rad, hercules, CA, USA) containing 5% 2-mercaptoethanol, denatured (5 min at 95 • C), and resolved by SDS-PAGE electrophoresis. Gels (0.75 mm, 4% stacking gel and 10% resolving gel) were run using Mini-Protean Tetra Cell (Bio-Rad) at a constant voltage (150 V). Subsequently, the separated proteins were transferred (5 min at the constant voltage of 25 V) with the Trans-Blot Turbo transfer system (Bio-Rad) to the Immobilon P membrane (Merck-Millipore, Burlington, MA, USA). A dedicated transfer buffer (25 mM Tris, 190 mM glycine, 20% methanol) was used. After the transfer, the membranes were stained with Pierce™ Reversible Protein Stain Kit for PVDF Membranes (Thermo Fisher Scientific) for the total protein according to the manufacturer's instructions. Thereafter, the membranes were blocked with 1% casein blocking buffer (Sigma-Aldrich, Saint Louis, MO, USA) at room temperature for 1 h. After rinsing with PBS-T buffer consisting of PBS (VWR International, Radmor, PA, USA) and 0.05% Tween-20 (Thermo Fisher Scientific, Waltham, MA, USA), the membranes were incubated (overnight, at 4 • C) with specific primary antibodies against NAG-1 (sc-3771961; Santa Cruz Biotechnology, Dallas, TX, USA) diluted 1:200 in PBS-T buffer pH 7.3. The unbound antibodies were removed by washing with PBS-T (3 times, 5 min). Subsequently, the membranes were incubated with hRP-conjugated goat anti-mouse secondary antibodies (Dako A/S, Glostrup, Denmark, cat. no. P044701-2) at a dilution of 1:20,000. The NAG-1 protein bands were visualized using an Immobilon ECL Ultra Western hRP Substrate (Merck-Millipore, Burlington, MA, USA) and ChemiDoc MP Imaging System (Bio-Rad). The obtained Western blot membrane images were analyzed with Image Lab software version 6.0.1 build 34 (Bio-Rad). The intensity of the chemiluminescence signal was normalized to the total protein signal intensity recorded for individual lanes.

Chemokine Profiling
For the purpose of the correlation analysis, the previously published [88,89] data on circulating chemokines were obtained from the patients' database. In the original research, concentrations of MCP-1, MIP-1α and MIP-1β were measured in sera in duplicates/triplicates using a flow cytometry-based method with a BioPlex 200 platform (Bio-Rad) and Bio-Plex Pro™ human Cytokine, Chemokine, and Growth Factor Magnetic Bead-Based Assays, as instructed by the manufacturer.
For the purpose of this study, the concentrations of MCP-1, MCP-3, MIP-1α, and MIP-1β were measured in cell culture supernatants using the same methodology and a custommade cytokine panel ProcartaPlex Immunoassay from Invitrogen (Thermo Fisher Scientific).
Standard curves were prepared using 5-PL logistic regression, and the collected data were analyzed using BioPlex Manager 6.0 software.

Statistical Analysis
The statistical analysis was conducted with MedCalc ® Statistical Software version 20.008 (MedCalc Software Ltd., Ostend, Belgium). The Kolmogorov-Smirnov and Levene tests were applied to test the data distribution and homogeneity of variances, respectively. The RTqPCR data were analyzed as logarithms. Matched tissue samples and treateduntreated cells were analyzed using a t-test for paired samples. Multigroup comparisons were conducted with one-way ANOVA followed by a Student-Newman-Keuls post hoc test for normally distributed data with homogeneous variances, or with a Kruskal-Wallis H test followed by a Conover post hoc test for non-normally distributed data or/and with nonhomogeneous variances. Two-group comparisons of unpaired samples were conducted using a t-test for independent observations. The correlation analysis was conducted using a Spearman rank correlation yielding correlation coefficient rho (ρ) or a Pearson correlation yielding correlation coefficient r. The significance of altered chemokine secretion compared to the control cells was tested using a t-test for one mean. A probability (p) equal to or below 0.05 (two-sided) was considered to be indicative of the statistical significance of observed differences or trends.

Conclusions
Newly synthesized analogues of the oxicam class of NSAIDs labeled with reduced gastrotoxicity, differing from piroxicam and meloxicam with an arylpiperazine moiety at the thiazine ring, are able to downregulate MCP and MIP chemokines to varying degrees, while the classic oxicams had a mostly negligible or non-significant effect on their expression and secretion. The presence of a propylene linker between thiazine and piperazine nitrogen atoms and one fluorine substituent at arylpiperazine pharmacophore characterizes their most effective and consistent inhibitor. Propylene linker-possessing analogues elicit a biphasic response in terms of CCL2 and CXCL2 expression, implying that these analogues might operate by two different mechanisms: an anti-inflammatory mechanism at lower doses and shorter exposures, and evoking a stress response at higher concentrations and longer exposures.
The NSAID-activated gene (NAG1), and NFKBIA and MYD88 from the NFkB signaling pathway are molecular targets of the investigated oxicams located upstream of chemokines. As is consistent with findings regarding chemokines, the analogs with a three-carbon propylene linker are more effective in the upregulation of NFKBIA and the downregulation of MYD88 than oxicams with a two-carbon oxyethylene linker.
Colorectal cancer is associated with aberrantly expressed MCP and MIP chemokines. They are all, with the exception of CCL19, upregulated in both transformed and nontransformed mucosa from cancer patients compared to normal tissue from non-cancer patients. Interestingly, while the upregulation in tumors is markedly higher for CCL4 and CXCL2, the CCL2, CCL8 and CCL19 transcripts are relatively downregulated in tumors compared to tumor-adjacent tissue. high chemokine expression in morphologically and histologically normal mucosa is clinically relevant, as it may predispose to neoplastic transformation and contribute to local cancer recurrence following surgery, or may drive an occurrence of synchronous tumors.