The present study involved for the first time the effect of crocin supplementation on PAI-1 activity in liver, kidney and plasma, as well as on the expression of SOD1, SOD2, catalase and PAI-1 genes. Although the effect of crocin administration on the antioxidant status of streptozotocin-induced diabetic rats is studied in several publications, only a small part of this literature studies the effect of the per os crocin administration [14,15,16].
Regarding the effect of crocin administration on glucose levels in the blood of healthy rats, Tamaddonfard et al.  reported no hypoglycemic effect. On the contrary, Arasteh et al.  reported that the intraperitoneal administration of a hydromethanolic saffron extract does have such an effect. In our study, glucose levels were decreased by crocin in both Cr20 and Cr50 groups, compared to controls. One might point out that there is a discrepancy between our results and the results by Tamaddonfard et al. . Nevertheless, it is important to comment that when crocin is administered per os, as in this study, it is converted to crocetin and absorbed as such . Therefore, our protocol is closer to that of Arasteh et al. , since they administered a saffron extract, which obviously contained both crocin and crocetin. As for a possible mechanism explaining the effect on blood glucose levels, given that crocin inhibits pancreatic lipase , it is possible that it also reduces lipid absorbance and inhibits gluconeogenesis. Regarding the effect of crocin on glucose levels in diabetic rats, we report no statistically significant outcome. Although Altinoz et al.  showed a decrease of glucose levels by 13%, they started treatment just 3 days after the injection of STZ, while our animals received crocin 2 weeks later. Such a time interval allows for a more stable picture of the damage inflicted to the pancreas, as reflected in glucose levels.
The available literature does not allow for a definite conclusion regarding the effect of crocin supplementation on body weight. While Hazman et al.  reported a weight decrease of overweight rats, in another study , the addition of crocin to a high-fat diet for rats induced a decrease in total as well as epididymal fat, but not in animal weight. In our experiment, no statistical differences were detected, yet p values were very low (0.059 for the Cr20 group and 0.1 for the Cr50 group). On the other hand, although the induction of diabetes expectantly caused a significant decrease in animal weight, crocin administration did not compensate for this effect, which is in accordance with Altinoz et al.  and Hazman et al. . Most probably, the simple administration of an antioxidant such as crocin cannot compensate for the serious weight loss caused by diabetes, at least when insulin levels remain compromised.
The H2O2 decomposing activity in the liver did not change by the administration of crocin to healthy animals. This result agrees with Magesh et al.  and Rahbani et al. . Furthermore, in our experimentation, the induction of diabetes did not induce any significant change of hydrogen peroxide decomposing activity in the liver. Although the H2O2 decomposition rate does not exclusively reflect CAT, this enzyme comprises a major element of the H2O2 decomposing activity. Therefore, the stable H2O2 decomposition rate agrees with the unaffected expression levels of the catalase gene, also noticed in our experiment. These results are supported by Maritim et al. , who suggested that hepatic catalase activity is not consistently altered by a diabetogen (e.g. STZ), nor by the administration of some antioxidants such as melatonin, quercetin or various vitamins.
To our knowledge, our manuscript presents for the first time the effect of long-term crocin administration on the H2O2 decomposing activity in the kidney. Boussabbeh et al.  injected i.p. a single dose of crocin (250 mg kg−1) and reported no alterations in catalase activity. In our experiment, the Cr50 group exhibited a significantly lower H2O2 decomposing activity compared to the controls, while the Cr20 group did not. Carotenoids can be beneficial at low concentrations but harmful at high concentrations (prooxidant action), especially when not administered along with other substances . In the previously mentioned review, Maritim et al.  state that renal catalase activity exhibits a clearer decrease in diabetes, an element reported by Kakkar et al.  and identified in our experiment as well. On the other hand, the administration of antioxidants may alleviate this effect . Such an alleviation did occur in our experiment, but only with the 20 mg kg−1 crocin dose. While it is not always clear why antioxidants do not necessarily work in a dose-dependent manner, it is true that in an oxidative stress state (e.g. diabetes), antioxidant enzymes may be depleted. In this state, antioxidant substances such as crocin can reduce oxidative stress by scavenging ROS and contribute to some extent to the restoration of enzyme levels.
In the only study that relates to our work, Magesh et al.  administered i.p. crocetin with no effect on SOD activity in the rat liver. In our experiment, the administration of crocin to healthy animals increased SOD activity, but only in the DCr50 group. According to Wang et al. , such an increase is possibly due to an increase in Νrf2 (nuclear factor-erythroid 2-related factor 2) activity, which has an important role in the cellular defense against oxidative stress and a proven relationship with the increase of antioxidant enzyme activity after the administration of antioxidants such as vitamin A. On the other hand, various research teams report contradicting results regarding the effect of diabetes induction on the activity of SOD in the liver [e.g. 25], although they always reverse these results by administering some antioxidant other than crocin. In 2011, Rahbani et al.  reported a decrease of SOD activity by the induction of diabetes, a finding that relates to our work since they counterbalanced this decrease by administering a saffron ethanolic extract. Our experimentation only adds extra information to the debate: diabetes tended to increase SOD activity (p = 0.055), and this was reversed by the higher crocin dose (once again, only as a trend, p = 0.078).
According to Maritim et al. , the effect of diabetes on SOD activity in the kidney is not consistent either. In our experiment, the enzyme activity did not change after the induction of diabetes. This result goes along with Maritim et al. , as well as with Bandegi et al. . On the other hand, crocin administration decreased renal SOD activity in diabetic animals. Kataya and Hamza  presented a similar effect by administering an antioxidant extract of red cabbage.
The concentration of protein carbonyls (liver and kidney) did not change in any group, diabetic or not. In a similar way, in the only previous study on the effect of crocin on healthy animals, El-Beshbishy et al.  reported no alteration of the protein carbonyls concentration after the i.p. administration of the substance. Regarding the kidney, Palsamy and Subramanian  administered per os resveratrol to healthy rats with no effect on protein carbonyls concentration. Diabetes tends to increase the concentration of protein carbonyls in the rat liver [35, 36] due to the formation of active carbonyl forms. Our findings do not confirm these reports, but this could be just due to the high standard deviation of our results.
The concentration of GSH in the liver of healthy animals was increased by the administration of 50 mg kg−1 of crocin. This is in accordance with Gedik et al. , and verifies the results reported in 2004 by Ochiai et al. , where crocin administration increased GSH concentration in PC-12 cells. Skrzydlewska et al.  specifically reported for the rat liver an induction—at gene level—of enzymes that metabolize drugs after the administration of tea polyphenols. Alternatively, such an increase in enzyme activity could be attributed to (a) an increase in the activity of enzymes responsible for GSH synthesis (e.g. g-glutamylcystein ligase or GSH synthetase), or (b) an increase in glutathione reductase activity, which resynthesizes GSH through its oxidized form . Furthermore, in our experiment, GSH did not change in any organ of the diabetic controls. It seems that at their sacrifice time-point, the animals had neither a severe oxidative stress level to the point of GSH significant reduction, nor a stimulus for a compensatory rise of GSH levels.
As far as hepatic TAC in the liver of healthy rats is concerned, Yuan et al.  administered a mixed fruit and vegetable juice for 5 weeks and reported an increase in liver TAC. Our result was similar regarding the Cr50 group. Such an outcome can be attributed to the increased activity of both GSH and SOD. Indeed, Bartosz  mentioned that in tissue homogenates, hepatic included, TAC is high due to glutathione. Furthermore, a strong correlation was determined between TAC and SOD (Fig. 3c). Finally, regarding the Cr20 and Cr50 groups, a strong correlation was determined between TAC and GSH as well (rs = 0.736, p = 0.001). Nevertheless, such a correlation between TAC and GSH was not confirmed in the diabetic groups. This suggests a negative impact of diabetes induction on the oxidative status. According to Hosseini et al. , hepatic TAC is not necessarily altered by the induction of diabetes, nor by the treatment of diabetics with saffron. With reference to the diabetic controls and the DCr20 group, our results agree to the aforementioned authors. However, in our protocol, the DCr50 group exhibited a statistically significant decrease of TAC levels in the liver compared to diabetic controls, an effect which should not be considered as prooxidant since protein carbonyls remained stable. A question that may arise is why TAC in the DCr50 group is low, even though H2O2 decomposing activity, SOD, GSH and protein carbonyls are not different from the controls. A possible explanation is that TAC comprises not only the parameters we determined, but other molecules as well (e.g. vitamin C, E, etc.). As mentioned previously, the increase of some antioxidants may induce the decrease of others. Indeed, when comparing the diabetic group with the DCr50 group, TAC alterations are paralleled by the changes in SOD activity: although superscript letters in Table 2 do not indicate any statistical difference because p values were slightly higher than 0.05, both TAC (p = 0.076) and SOD (p = 0.068) are increased in the D group, and both return back to normal levels (TAC; p = 0.018, SOD; p = 0.1) in the DCr50 group. Hence, we believe that SOD does contribute to the low TAC levels in the DCr50 group.
In the kidney of healthy animals, TAC was not affected by the administration of crocin. This result agrees with Nasiri et al. , who reported that the administration of antioxidant compounds did not affect renal TAC in healthy animals. The induction of diabetes did not have any significant impact either. In a study which is the closest available to ours, Karamouzis et al.  found that TAC levels in the plasma of patients with chronic kidney disease remained stable in any stage apart from stage five. In our protocol, TAC in kidney was indeed unaffected by the diabetic state. According to Kusano and Ferrari , it is possible that TAC is not influenced as long as vitamins A and E remain unaffected.
The effect of diabetes on catalase gene expression in the liver is an element not thoroughly studied and the few existent works have no consistent results. In humans, catalase gene expression regulates catalase activity at different levels (transcription, post-transcription, post-translation) . Nevertheless, in our experiment, catalase gene expression was not affected by diabetes or by crocin administration. Our result agrees with Ahmed et al. , who showed that the diabetic state does not influence catalase gene expression. In another case, the expression of this gene was downregulated by the diabetic state , an effect not compensated for by the administration of antioxidants (e.g. ascorbic, lipoic acid). The differences of the protocols concerning diabetes induction might explain the discrepancies of these results.
As far as SOD1 gene expression in the liver is concerned, there is no consistent effect of the induction of diabetes and/or of the administration of antioxidants. For instance, Sadi et al.  found that the diabetic state decreased SOD1 gene expression and this effect was not changed by resveratrol injection. This is in accordance with the administration of crocin in our experiment. Furthermore, Sadi et al.  reported that lipoic acid and vitamin C did not alter the expression of this gene. Nevertheless, in our experiment crocin did alter SOD1 gene expression and compensated for the significant increase induced by diabetes.
In our study, the expression of SOD2 gene in the liver did not change in the crocin-treated healthy animals. Oliveras-López et al.  carried out a study with mice, which received extra virgin olive oil rich in polyphenols, and reported that SOD2 gene expression in pancreatic islets was not affected. Moreover, resveratrol administration in healthy rats does not alter significantly SOD2 gene expression in the liver . On the contrary, this expression increased in the D group compared to the controls, an observation that partly agrees with Sadi and Güray  who presented an increase of expression (albeit non-significant). As for the administration of antioxidants, neither resveratrol  nor the injection of lipoic acid  altered SOD2 gene expression, in accordance with our outcomes.
Compared to type-2 diabetes, the effect of type-1 diabetes on fibrinolysis is relatively under-researched and probably underestimated, possibly due to the clinical significance of thromboembolism in type-2 diabetes. Nevertheless, there is a consensus that the two different types of diabetes have an overlapping pathophysiology . Regarding the mechanisms involved in the alteration of PAI-1 activity by diabetes, a role for hyperglycemia per se (which applies to type-1 diabetes as well) is strongly suggested .
The expression of PAI-1 gene in the liver seems to increase in diabetes mellitus, as well as in several protocols of hepatic injury, and is mainly attributed to mechanisms mediated by glucagon and/or reactive oxygen species . A known pathway of activation of genes, among them PAI-1, includes the phosphorylation of Smad proteins . In that study, the expression of PAI-1 gene was induced by TGF-β and α-lipoic acid inhibits this action in hepatic cells by inhibiting TGF-β-mediated molecular mediators such as Smad3. In our experiment crocin, being an antioxidant, may have acted in a similar way.
Neither the induction of diabetes nor crocin supplementation did affect PAI-1 activity in the liver in any group. The same applies with PAI-1 activity in plasma. Although it might be considered as contradictory that the increase of gene expression was not accompanied by a respective increase of PAI-1 activity, this is not necessarily so. Fibrinolysis involves complex mechanisms and numerous activating and inhibiting factors, and fibrinolytic agents can be readily consumed or inhibited upon release. Additionally, the molecule of PAI-1 has a short half-life of approximately 1 h under physiological conditions . Therefore, our findings might be explained by significant differences in tPA and tPA/PAI-1 ratio between diabetic and normal subjects . Besides, PAI-1 inhibits the action of plasminogen activators such as tPA . That said, a more thorough investigation of the effects of type-1 diabetes on tissue fibrinolysis would be of interest. As far as the administration of antioxidants is concerned, Chan et al.  did not notice any significant variation of the plasma PAI-1 activity in diabetic rats after the administration of astaxanthin, an outcome in accordance with our results. The fact that crocin impressively corrected the effect of diabetes on PAI-1 gene expression but had no similar effect on hyperglycemia allows us to suggest that, although the elevated glucose levels might have played a role in the expression of the gene, some other factors that affect gene expression must be involved.
With respect to the kidney, PAI-1 activity is reported to increase in the diabetic state, but this rise accompanies the depletion of GSH . In our experiment, PAI-1 activity in kidney did not increase in the D group, either due to the stable GSH levels in the kidney or because of a compensatory increase in tPA. Indeed, Fisher et al.  reported that tPA levels increased in the presence of high glucose concentration in renal mesangial cells. However, an increased PAI-1 activity in the kidney is related to diabetic nephropathy .
The activity of ALT in blood serum decreased significantly by the administration of crocin to healthy animals in a dose dependent manner. This finding agrees with El-Beshbishy et al. , who reported reduced ALT activity after the i.p. administration of crocin at the dose of 200 mg kg−1 for 7 days, as well as with Asdaq and Inamdar , who found that that the per os crocin treatment to healthy rats at the dose of 19.34 mg kg−1 for 5 days decreased ALT activity as well. Regarding AST activity in serum, the higher dose of crocin resulted in a strong trend of decrease (p = 0.065), a result that again agrees with El-Beshbishy et al. . Such a decrease could be due to the antioxidant effect of crocin, a hypothesis supported by Djordjevic et al. , who administered another carotenoid (astaxanthin) to football players and found decreased AST activity along with a decrease in the production of reactive oxygen species. Furthermore, the activities of ALT and AST in blood serum significantly increased in the diabetic animals compared to controls. In diabetes, the necessary energy exploitation of amino acids through protein catabolism is achieved via transamination with the assistance of aminotransferases, which are elevated in this case. Referring to our experiment, crocin administration managed to decrease ALT and AST activities in the DCr20 group, whereas the 50 mg kg−1 dose did not have the same effect. It is common for an antioxidant to affect some parameters in a positive way at a certain dose, yet have a reverse effect when the dose is higher (probably acting in a prooxidant way). Indeed, Altinoz et al.  noticed that crocin at the dose of 20 mg kg−1 induced a decrease in ALT and AST activities, while Kianbakht and Hajiaghaee  did not detect any decrease after the administration of 50 mg kg−1 of crocin.
Regarding BUN, there was a significant increase in the diabetic controls compared to the C group, an element in accordance with Kulina and Rayfield . Diabetes mellitus triggers gluconeogenesis, hence muscle tissue amino acids are mobilized and used as energy source. For their uptake and exploitation by the liver, these amino acids need to be catabolized to alanine. Alanine plays a dual role: as a precursor used for gluconeogenesis, as well as a transporter of nitrogen into the liver, where it is used for the formation of urea . In our experiment, although the decrease induced by either dose of crocin was not statistically significant, the margin was close (p = 0.107 in the DCr20 group and p = 0.128 in the DCr50 group). From a biological point of view, and given that these results refer to different doses of the same substance, the probability that both null assumptions are wrong should be even lower. Indeed, such an extrapolation is supported by the results of Altinoz et al. . The administration of crocin induced a decrease in ALT activity, which allows for the speculation that protein catabolism decreased, resulting to decreased alanine production and BUN levels. Besides, in a study with diabetic mice , the administration of more broadly used antioxidant compounds (i.e. vitamin C, E) did decrease BUN levels. These authors attribute this outcome to the antioxidant and anti-inflammatory effects of these substances.
Oxidative stress in diabetes mellitus is due to increased lipid oxidation and production of reactive oxygen and nitrogen species, and induces direct damage to nephrons giving rise to vasoconstriction, platelet aggregation and cellular toxicity . Possibly, this mechanism of renal damage gradually leads to kidney disease resulting in creatinine accumulation in the blood stream. Creatinine is a simple kidney status biomarker with its concentration being elevated in the blood when there is renal disease . In a series of studies [71,72,73], various authors reported that crocin administration did not influence creatinine in a few species, rat included. We believe that the statistically significant decrease in the Cr20 group is mainly due to the coincidentally low standard deviations, given that the means are very close. Furthermore, creatinine significantly increased in the D group compared to the C group, and neither dose of crocin could compensate for this increase, results that agree with Altinoz et al. .
Regarding cholesterol and triglyceride concentration in blood serum, Asdaq and Inamdar reported that the per os supplementation of crocin diluted in a H2O vehicle containing carboxymethylcellulose for 5 days reduced cholesterol and triglyceride levels in healthy rats . In our study, no group of healthy rats exhibited such an effect, but the differences of the two protocols regarding the duration of the treatment and the composition of the vehicle constitute difficult any comparison. The elevated cholesterol and triglyceride levels observed in the diabetic controls agree with Young et al. . Reduced insulin secretion occurring in diabetes mellitus increases lipolysis. Insulin inhibits hepatic VLDL production and promotes the catabolism of lipoproteins, which are rich in triglycerides. On the other hand, crocin did not reduce cholesterol or triglyceride levels, a result that disagrees with Altinoz et al. . This discrepancy could be due to the protocol differentiation: we started crocin treatment 2 weeks after STZ injection, while Altinoz et al.  supplemented crocin as early as 3 days after STZ. Our 2-week interval between STZ injection and the beginning of crocin treatment must have induced a more severe potentiation of gluconeogenesis, hence higher levels of cholesterol and triglycerides, an effect that crocin could not reverse.