Tumor Microenvironment Alters Chemoresistance of Hepatocellular Carcinoma Through CYP3A4 Metabolic Activity

Variations in tumor biology from patient to patient combined with the low overall survival rate of hepatocellular carcinoma (HCC) present significant clinical challenges. During the progression of chronic liver diseases from inflammation to the development of HCC, microenvironmental properties, including tissue stiffness and oxygen concentration, change over time. This can potentially impact drug metabolism and subsequent therapy response to commonly utilized therapeutics, such as doxorubicin, multi-kinase inhibitors (e.g., sorafenib), and other drugs, including immunotherapies. In this study, we utilized four common HCC cell lines embedded in 3D collagen type-I gels of varying stiffnesses to mimic normal and cirrhotic livers with environmental oxygen regulation to quantify the impact of these microenvironmental factors on HCC chemoresistance. In general, we found that HCC cells with higher baseline levels of cytochrome p450-3A4 (CYP3A4) enzyme expression, HepG2 and C3Asub28, exhibited a cirrhosis-dependent increase in doxorubicin chemoresistance. Under the same conditions, HCC cell lines with lower CYP3A4 expression, HuH-7 and Hep3B2, showed a decrease in doxorubicin chemoresistance in response to an increase in microenvironmental stiffness. This differential therapeutic response was correlated with the regulation of CYP3A4 expression levels under the influence of stiffness and oxygen variation. In all tested HCC cell lines, the addition of sorafenib lowered the required doxorubicin dose to induce significant levels of cell death, demonstrating its potential to help reduce systemic doxorubicin toxicity when used in combination. These results suggest that patient-specific tumor microenvironmental factors, including tissue stiffness, hypoxia, and CYP3A4 activity levels, may need to be considered for more effective use of chemotherapeutics in HCC patients.

Variations in tumor biology from patient to patient combined with the low overall survival rate of 27 hepatocellular carcinoma (HCC) present significant clinical challenges. During the progression of 28 chronic liver diseases from inflammation to the development of HCC, microenvironmental properties, 29 including tissue stiffness and oxygen concentration, change over time. This can potentially impact drug 30 metabolism and subsequent therapy response to commonly utilized therapeutics, such as doxorubicin, 31 multi-kinase inhibitors (e.g., sorafenib), and other drugs, including immunotherapies. In this study, we 32 utilized four common HCC cell lines embedded in 3D collagen type-I gels of varying stiffnesses to 33 mimic normal and cirrhotic livers with environmental oxygen regulation to quantify the impact of these 34 microenvironmental factors on HCC chemoresistance. In general, we found that HCC cells with higher 35 baseline levels of cytochrome p450-3A4 (CYP3A4) enzyme expression, HepG2 and C3Asub28, 36 exhibited a cirrhosis-dependent increase in doxorubicin chemoresistance. Under the same conditions, 37 HCC cell lines with lower CYP3A4 expression, HuH-7 and Hep3B2, showed a decrease in doxorubicin 38 chemoresistance in response to an increase in microenvironmental stiffness. This differential 39 therapeutic response was correlated with the regulation of CYP3A4 expression levels under the 40 influence of stiffness and oxygen variation. In all tested HCC cell lines, the addition of sorafenib 41 lowered the required doxorubicin dose to induce significant levels of cell death, demonstrating its 42 1 Introduction 47 Cancer is the second-highest cause of mortality in the United States, lagging just slightly behind 48 cardiovascular disease in 2019 1 . Among all cancer types, hepatocellular carcinoma (HCC) has the 49 second-lowest 5-year survival rate (17.7%) and has shown the highest increase in mortality among all 50 cancers over the past seven years 2 , 3 . Complicating treatment, HCC is commonly diagnosed at an 51 intermediate or an advanced stage and often occurs secondary to underlying chronic liver disease and 52 cirrhosis 4 . The prognosis is poor as treatment options are limited by compromised liver function due 53 to underlying disease. Despite screening efforts for at-risk patients, most are not surgical candidates 54 for partial resection, and the availability of full liver transplantation is very low relative to the need 5 . 55 These issues mean that systemic and localized drug-based therapies play a significant role in current 56 standard therapy for HCC. Despite these therapeutic interventions, the survival rate for HCC remains 57 low, partially attributed to the variable efficacy of current treatment methods based on underlying 58 factors 6-8 . As such, stratifying patients for the most effective treatment is critical because of three 59 factors; the degree of tumor burden, the degree of liver dysfunction, and highly variable treatment 60 efficacy between patients 9 . For example, the tyrosine kinase inhibitor sorafenib has shown modest 61 success in selected patients as a systemic treatment 10 , but effectiveness is tempered by poor tolerance 62 of the drug in many instances 9 . Localized delivery of drugs through transarterial chemoembolization 63 (TACE) combines delivery of drugs such as doxorubicin with embolization to promote localized 64 ischemia and hypoxia. TACE blocks the arterial blood supply of a tumor through particulate or viscous 65 liquid agents such as degradable starch microspheres, drug-eluting beads, or ethiodized oil. This is a 66 well-established technique that allows a high local dose while simultaneously increasing the residence 67 of chemotherapeutic drugs in the target area, cutting off the supply of nutrients, and also limiting 68 exposure and toxicity for the rest of the body 11 . This has emerged as the standard of care for 69 intermediate-stage HCC. However, tumor cells in the hypoxic environment may undergo phenotypic 70 adaptations that aid survival. Such changes may account for the high rate of persistent, viable tumor 71 cells observed after TACE in previous studies 12 . Therefore, while a substantial survival benefit can be 72 realized, there is still much room for improvement and understanding of the changes that occur in the 73 tumor cells during embolization 6-8 . 74 It is well established that many of the difficulties in treating HCC may stem from the numerous tumor 75 microenvironment (TME) changes that occur in underlying chronic liver disease and the rapid 76 progression of HCC. The modulation of the TME has been shown to impact drug metabolism 77 significantly and is thought to be a major contributor to the known differential response of patients to 78 chemotherapy 13,14 . Furthermore, induction of hypoxia in the TME due to stiffening of the extracellular 79 matrix (ECM) and embolization during treatment can alter the chemoresistance of the tumor cells, 80 further impacting the treatment efficiency in intermediate and advanced stage HCC 15-17 . The 81 modulation of response and the individual impact of these TME features have yet to be fully 82 characterized in a three-dimensional (3D) HCC-TME model. 83 The increase in microenvironmental stiffness resulting from fibrosis, usually culminating in cirrhosis, 84 is a hallmark of most chronic liver diseases and is observed in 80-90% of HCC patients 4 . The most 85 notable hallmark of liver cirrhosis that impacts cellular and tissue function is increased collagen 86

Confined Compression Test 189
Cirrhotic stiffening in the TME has been shown to alter the chemoresistance of many cancer cell types Cell Titer Blue (G8081, Promega, Fitchburg, WI) cell viability assay to quantify the response of HCC 215 cells to chemotherapeutic treatment under varying TME conditions (such as stiffness and hypoxia). 216 Briefly, cell media was mixed at a ratio of 5:1 with assay solution and incubated at 37˚C for 1 hr. A 217 Cytation 3 plate reader was used to read fluorescence (Ex: 530 nm/Em: 620 nm) of the assay. Findings 218 were normalized to control (cells treated with drug-free media) to obtain percent cell viability. The 219 data gathered from cell viability assays after doxorubicin treatment was fitted using the cftool function 220 of Matlab ® (MathWorks, Natick, MA) and half-maximal inhibitory concentration (IC50) value was 221 This is a provisional file, not the final typeset article calculated and reported as chemoresistance indicator. As sorafenib has been used as a prodrug used for 222 angiogenesis treatment, mainly by inhibiting vascular endothelial growth factor (VEGFR), platelet-223 derived growth factor receptor (PDGFR), and rapidly Accelerated Fibrosarcoma (RAF), but it also has 224 a minor direct cytotoxic side-effect in some instances 15 . For that reason, treatment efficacy of 225 standalone sorafenib and combined treatments were reported as fold viability change compared to 226 untreated control for the two clinically relevant doses used, standard (11 μM), and high (22 μM

Statistical Analysis 247
Two-tailed student's t-test assuming unequal variance was performed in Matlab ® (MathWorks, Natick, 248 MA) to compare samples, and a p-value less than 0.05 was considered significant for variation. Data 249 are reported as mean ± standard deviation unless otherwise indicated. All experiments were replicated 250 a minimum of four times. 251

3
Results 252 HCC cells were cultured for three days to reach native morphology in a monolayer in a tissue culture 253 plate (2D) or in rat tail-derived collagen type I hydrogels (3D). Cells were treated with doxorubicin 254 with or without sorafenib for 24 or 48 hours. Viability was assessed 72 hours after the end of drug 255 treatment, as described in Figure 1a. Experimental procedure outline. Cells were allowed to adhere and reach their native morphology 259 before the treatment for 24 or 48 hours. b) Compression modulus of the HCC hydrogels with 4 and 7 260 mg/ml collagen concentrations, which replicate normal and cirrhotic tissues, respectively. The 261 compression modulus did not vary when different HCC cells were cultured in collagen hydrogels. 262 Compression modulus values were significantly different when different collagen concentrations were 263 used to replicate normal and cirrhotic conditions. Dashed lines represent patient compression modulus 264 of healthy (red) and cirrhotic (green) tissues. 43,44 . Selected collagen concentrations replicated native 265 tissues successfully at different preload strains. * p<0.05, ** p<0.01. 266

Native 3D Microenvironment Compression Modulus 267
Hepatocytes uniformly embedded inside rat tail-derived collagen type I hydrogels were allowed to 268 reach their native morphology for three days. Afterwards, stiffness of the collagen hydrogels was 269 quantified using a uniaxial compression test with 5% and 10% preload strains as presented in Figure  270 1b to determine any potential impact of different HCC cell lines on collagen stiffness. Our results 271 demonstrate that increasing collagen concentration significantly elevated the compression modulus of 272 the collagen gels (p<0.05). However, we found no significant difference between compression modulus 273 between collagen hydrogels with the different HCC cell lines over the timeframe considered. Collagen 274 hydrogels at 4 mg/ml concentration produce a compression modulus comparable to tissue in a normal 275 hepatic microenvironment, which has been reported to be at 0.64 ± 0.08 kPa and 1.08 ± 0.16 kPa for 276 5% and 10% preload strains, respectively 43 . At 4 mg/ml collagen concentration, the average 277 This is a provisional file, not the final typeset article compression modulus was found to be 0.66 ± 0.07 kPa and 0.11 ± 0.01 kPa for 5% and 10% preload 278 strains, respectively. Likewise, at 7 mg/ml collagen concentration, collagen hydrogels achieved a 279 compression modulus comparable to a human hepatic tumor microenvironment, which has been 280 reported to be 3 kPa under 5% preload strain 44 . The significant difference (p<0.05) of compression 281 moduli between 7 mg/ml and 4 mg/ml collagen hydrogels showed we could replicate the 282 microenvironment stiffness using these collagen properties. In our findings, using a 7 mg/ml collagen 283 concentration, the average compression modulus was found to be 2.70 ± 0.22 kPa for 5% preload strain. 284 The reported collagen compression moduli for this study are also consistent with our previously 285 reported collagen compression modulus values 26 . In the same study, we also showed that collagen 286 concentration does not alter the diffusivity of solutes, thereby demonstrating that differential response 287 to chemotherapy is not likely due to physical diffusion differences of the collagen concentrations 26 . 288

HepG2 C3Asub28
Hep3B2.1.7 HuH-7 This is a provisional file, not the final typeset article We observed that IC50 of HepG2 cells after a 24-hour doxorubicin treatment was 3.09-fold higher in 320 the 3D normal-normoxic environment (IC50 = 1.13 μM) compared to the 2D-normoxic (IC50 = 3.48 321 μM). The IC50 of doxorubicin against C3Asub28 cells was consistently the highest, but it was the only 322 cell line that did not lead to a significant change in IC50 of doxorubicin in response to the 3D normal-323 normoxic compared to 2D-normoxic. The efficacy of doxorubicin on HuH-7 cells cultured in 3D 324 normal-normoxic microenvironment was higher than the HepG2 and C3Asub28 phenotypes. Overall, 325 we observed that the IC50 of doxorubicin against HuH-7 cells was lower (p<0.05) in 3D normal-326 normoxic compared to 2D-normoxic by 62.61-fold. Conversely, IC50 values of doxorubicin against 327 Hep3B2 cells were the lowest overall and significantly decreased 4.02-fold 3D normal-normoxic 328 compared to 2D-normoxic. 329

The Influence of Microenvironmental Stiffness on Doxorubicin Chemoresistance under 330
Normoxic Conditions 331 To isolate the impact that microenvironmental stiffness plays in the regulation of chemoresistance of 332 different HCC cell lines, we analyzed the impact that the shift from normal (4 mg/mL) to cirrhotic (7 333 mg/mL) collagen concentration has on doxorubicin IC50 values under normoxic conditions for a 24-334 hour treatment duration. Overall, we found that the increase in microenvironmental stiffness, modeled 335 by the higher collagen concentration, increased the IC50 values of doxorubicin against the HCC cell 336 lines, HepG2 and C3Asub28, that had a higher basal chemoresistance to doxorubicin. Variations in 337 fold changes for 24-hour treatment were reported in Figure  cirrhotic-normoxic conditions increased compared to cells cultured in 3D normal-normoxic conditions. 341 However, for both HuH-7 and Hep3B2.17, we did not see any significant difference in IC50 values of 342 doxorubicin against these cells (p>0.05), between 3D normal-normoxic and 3D cirrhotic-normoxic 343 conditions. 344

Influence of Hypoxia on Doxorubicin Chemoresistance 345
Oxygen concentration in HCC liver tumors can change due to reduced blood flow, increased 346 cell density, and environmental stiffening 17,23 . This has been shown to alter both the proliferation rate 347 and chemoresistance of tumor cells. We isolated and quantified the influence of hypoxia on 348 chemoresistance, as measured by IC50 values, when cells were cultured in 3D with normal liver 349 stiffness (4 mg/ml) and in 2D monolayers both under hypoxic conditions. Overall, the introduction of 350 hypoxic conditions to 2D monolayers or 3D collagen hydrogels with normal stiffness (4 mg/ml) 351 showed a change in response between the HCC cell lines for 24-hour treatment durations and saw 352 minimal changes for 48-hour treatment durations as shown in Figure S-

Sorafenib Improves Minimum Required Dose of Doxorubicin for Acute Toxicity 390
Combined administration of doxorubicin and sorafenib in HCC patients has been commonly used due 391 to their potential synergistic effect 41,46 . Doxorubicin is used to inhibit tumor proliferation 47 , while 392 sorafenib is used to inhibit angiogenesis and tumor cell proliferation in the TME 15 . Cells cultured in 393 3D normal and cirrhotic hydrogels were investigated for the impact of doxorubicin-sorafenib 394 combination therapy to determine how the introduction of sorafenib influenced the resulting cell 395 viability. Cell viability in normal and cirrhotic tissue with a 24-hour treatment duration of both drugs 396 and the influence of hypoxia are presented in Figure 3 and discussed in this section. Treatment efficacy 397 is reported as fold viability change compared to untreated controls reported in Figure S-IV. In addition, 398 we explore how the addition of sorafenib can lower the minimum required doxorubicin dose to produce 399 a significant change in cell viability from untreated controls. For this analysis, statistical comparison 400 (p<0.05) of viability under tested doses and untreated samples were compared, and minimum dose to 401 induce cytotoxicity was reported. We found minimal differences between 24-and 48-hour treatments 402 with these drugs and present 48-hour treatment results in Figure S-  This is a provisional file, not the final typeset article

Cirrhosis and Hypoxia Regulated HCC Metabolic Activity 457
The CYP3A4 enzyme is one of the major mechanisms of drug metabolism for cancer therapeutics, 458 including sorafenib and doxorubicin, in the liver. These enzymes can metabolize drugs before they 459 have the chance to cause their intended direct cytotoxic effects on the cells. 14 . CYP3A4 metabolic 460 activity of the HCC cell lines was measured in 2D monolayers and in 3D collagen I hydrogels in 461 response to different stiffness and hypoxic conditions. Regulation of CYP3A4 by cirrhosis and hypoxia 462 is presented in Figure 4. In agreement with the previously published literature, the C3Asub28 cell line 463 expressed much higher CYP3A4 expression than other HCC cell lines 38 . In our study, C3Asub28 464 CYP3A4 expression is 7.17 ± 2.73 fold higher than the HepG2 cell line, which is within the range of 465 previously published work (6.1 ± 0.2 fold) 38 . The introduction of hypoxia significantly downregulated 466 CYP3A4 expression compared to normoxia in all microenvironments (p=0.03 In this study, we establish that chemoresistance can be regulated by hypoxic and cirrhotic conditions 491 in the TME through direct modulation of CYP3A4 expression. This regulation can differentially alter 492 the efficacy of chemotherapeutic drugs in HCC cell lines, which potentially has clinical translation to 493 patient-specific HCC treatments. We examined the direct impact of TME stiffness and oxygen 494 concentration, variables commonly associated with HCC tumors on cellular response to 495 chemotherapeutics. This impact was measured by determining the difference in cell viability in 496 response to chemotherapeutic treatment and regulation of the expression of the primary drug-497 metabolizing enzyme CYP3A4. Using a collagen-based hydrogel system, we observed that 3D culture 498 alone significantly modulates resistance to doxorubicin and sorafenib in HepG2, C3Asub28, and HuH-499 7 cells but not for Hep3B2 cells. This stiffness-dependent resistance was not observed in similar 500 Hep3B2 cultures, which warrants further investigation into potential phenotypic and genotypic 501 differences in this specific cell line that might elucidate this response. This differential regulation of 502 chemoresistance in different cell lines of the same cancer type is not unique to HCC. This phenomenon 503 has also been observed in other cancers: The chemoresistance indicator, ethoxyresorufin, was 504 upregulated in 3D culture conditions compared to 2D culture in the C3A cell line. This is in agreement 505 with HepG2 and C3Asub28 doxorubicin treatment findings 48 . Equivalently, the IC50 of C3A cells 506 after treatment with paracetamol, trovafloxacin, and fialuridine were found to be higher in 2D culture 507 conditions compared to 3D culture conditions. This result corresponds with our HuH-7 and Hep3B2 508 doxorubicin chemoresistance findings 24 . Overall, these studies suggest that drug efficacy in 2D vs. 3D 509 conditions depends on cell phenotype and drug type. This coincides with our findings in which HepG2 510 and C3Asub28 cell lines had greater chemoresistance in 3D culture compared to 2D but showed the 511 converse trend for HuH-7 and Hep3B2 cells. 512 Cirrhosis and desmoplastic stiffening in the TME has been shown to be potential factors regulating 513 drug chemoresistance 49 . In our study, HepG2 chemoresistance to sorafenib and doxorubicin increased This is a provisional file, not the final typeset article in response to cirrhotic stiffness relative to their culture in a matrix of normal stiffness. The C3Asub28 515 cell line demonstrated a higher chemoresistance to doxorubicin in response to cirrhotic conditions than 516 other tested HCC cells. Similar to the HepG2 cell line, a rise in stiffness also increased the 517 chemoresistance to doxorubicin of the C3Asub28 cell line. However, sorafenib alone did not alter the 518 cell viability of the C3Asub28 cell line for the considered doses potentially attributed to the cell line's 519 high baseline CYP3A4 metabolic activity. The resistance of HuH-7 and Hep3B2 cells to doxorubicin 520 was not altered in response to cirrhosis, whereas CYP3A4 expression of these cell lines did not change 521 in response to cirrhosis. However, the same cell lines were shown to have higher chemoresistance to 522 sorafenib in response to cirrhosis. Although CYP3A4 carries out the majority of metabolic activity in 523 hepatocytes, other minor cytochromes, such as CYP1A2, 2A6, 1A2, and 2C9, exist and may also alter 524 the drug chemoresistance to an as yet unknown extent. The additional CYP expression present in HCC 525 cells may be responsible for the differential effect of cirrhosis on chemoresistance differences between 526 sorafenib and doxorubicin 50 . Furthermore, studies in literature showed that a rise in stiffness does not 527 always increase the chemoresistance of cancer cells 49 . The increase in TME stiffness may improve 528 IC50 of MDA-MB-231 triple-negative breast cancer cells to doxorubicin 27 . However, the same study 529 showed that MCF-7 HER2+ breast cancer cells did not show a stiffness-dependent resistance to 530 doxorubicin. This study hypothesized the increase of stiffness altered chemoresistance differently 531 because MCF-7 remained in an epithelial phenotype, but MDA-MB-231 had a mesenchymal 532 phenotype. Similarly, stiffness induces chemoresistance of BxPC-3 and Suit2-007 pancreatic cancer 533 cells to paclitaxel, but not to gemcitabine 51 . These studies hypothesized the differential effect of 534 chemoresistance to different drugs between cell lines could be related to phenotypical differentiation 535 from epithelial to mesenchymal phenotype. In addition, HepG2, HuH-7, and Hep3B2 cell lines have 536 different phenotypic profiles and differentiation levels, potentially explaining differences in their 537 chemoresistance and metabolic activity in response to cirrhosis 52 . 538 Hypoxia is known to be one of the regulating factors of chemoresistance 53 . In our study, HepG2 539 chemoresistance to doxorubicin and sorafenib increased in response to hypoxia compared to the 540 normoxic condition as measured both by increased IC50 values and CYP3A4 expression. However, 541 for C3Asub28 and HuH-7, chemoresistance decreased in response to hypoxia compared to normoxia.

542
Previous studies have also demonstrated the differential effect of hypoxia on drug efficacy, depending 543 on the cell line and phenotype. The presence of hypoxia has been shown to upregulate hypoxia-induced 544 factor (HIF1-α), but this alters the CYP isoforms differently in various medulloblastoma cell lines 54 . 545 The molecular pathway still could not be explained in this study, but it has been hypothesized that 546 nuclear receptors, namely PPARα, PPARγ, or ER-α, as well as the constitutive androstane and 547 pregnane X receptors, have found to be altered differently under hypoxia 54 . In addition to this, 548 chemoresistance does not always increase in response to hypoxia 53 . The same study also showed the 549 regulation of chemoresistance under hypoxia is not universal between ovarian, renal, breast, lung, and 550 lymphoma cancer cell lines and varies for different drugs. This supports our data showing the 551 differential effects of hypoxia on doxorubicin IC50 values of the tested HCC cell lines 53 . Also, hypoxia 552 increases HepG2 chemoresistance to doxorubicin, in confirmation with our study, but not to rapamycin 553 23 . In parallel to this, a significant decrease in apoptotic cells induced by cisplatin was reported under 554 hypoxic conditions for HepG2 and MHCC97L cell lines, which is in agreement with what we observed 555 with increased chemoresistance of HepG2 cells under hypoxic conditions 55 . It has also been showed 556 that hypoxia downregulates drug-metabolizing enzymes and subsequently the chemoresistance of the 557 HepaRG hepatoma cell line, which agrees with our findings of C3Asub28 CYP3A4 modulation under 558 hypoxia 56 . 559 Consequently, the differential role of hypoxia on molecular chemoresistance expressions and drug 560 efficacy has been reported. It has been shown that HIF1-α is upregulated due to a lack of oxygen in TME. This may or may not induce drug transporters such as MDR1 and targets of delivered drugs 562 (topoisomerase II) in each cell line 20 . Additionally, possible nuclear receptors have been proposed to 563 regulate CYP3A4 in response to hypoxia through the expression of HIF1-α and p53 expression 57 . 564 Alternation of molecular drug transport mechanisms could be the reason why we observed variable 565 chemoresistance between different HCC cell lines under hypoxia. 566 Our study showed CYP3A4 expression is regulated by microenvironment stiffness and hypoxia for 567 HepG2, C3Asub28, and HuH-7 cell lines providing a potential mechanism connecting the TME to the 568 chemotherapeutic response. The regulation of CYP3A4 resulted in a significant impact on the efficacy 569 of doxorubicin and sorafenib, whose trends in the regulation mirror the observed changes in cell 570 viability in response to the drugs. Doxorubicin IC50 was higher, and sorafenib terminated less HCC 571 population for HepG2 and C3Asub28 cells in cirrhotic, 7 mg/ml, microenvironments in general 572 compared to healthy, 4 mg/mL, stiffness reflecting the CYP3A4 expression in those 573 microenvironments. However, we did not see any significant IC50 change in response to doxorubicin 574 for the HuH-7 cell line when cultured in normal and cirrhotic 3D microenvironments. In addition, 575 CYP3A4 expression of HuH-7 did not change when cultured in normal and cirrhotic 3D 576 microenvironments, which is parallel with IC50 findings. 577 Moreover, in general, we demonstrate that hypoxia increases doxorubicin IC50 against HepG2 cells 578 but C3Asub28 cells. studies agree with our findings on the regulation of chemoresistance based on CYP3A4 activity. 588 However, our work presents the significant finding that the regulation of CYP3A4 expression can be 589 directly tied to the tumor microenvironment. We demonstrate that CYP3A4 activity can be regulated 590 by oxygen concentration and TME stiffness, subsequently altering the metabolism of the 591 chemotherapeutic drugs in HCC cell lines. Possible nuclear receptor pathways regulating CYP3A4 in 592 response to hypoxia through HIF1-α, p53, PPARα, VDR, FXR, and LXR have been proposed and 593 hypothesized that hypoxia could affect CYP3A4 at different degrees 57 . Furthermore, TME stiffness 594 has been hypothesized to alter CYP3A4 differentially through yes-associated protein (YAP) pathway 595 64 . This likely has a direct clinical translation to in vivo HCC regulation enforced by the clinical 596 observations of highly variable patient to patient HCC CYP3A4 expression 65 . The majority of the 597 current HCC treatments result in poor treatment outcomes, as such, the consideration of tumor 598 microenvironment properties (such as stiffness variation due to the changing fibrosis scores of patients, 599 presence of different levels of hypoxia as a result of this desmoplastic stiffening in the TME), and CYP 600 expression levels could potentially bring benefits to outcomes of HCC treatment, and provide a basis 601 for personalized HCC treatment 62,66 . 602 Overall, this work demonstrates TME stiffness and oxygen concentration modulates CYP3A4 603 expression of HCC cells and, consequently, their chemoresistance to doxorubicin and sorafenib 604 treatment. We determined the existence of a stiffness-dependent resistance to doxorubicin and 605 sorafenib, depending on the differential genetics of the HCC, such as phenotypical changes from 606 epithelial to mesenchymal 27 . HepG2 and C3Asub28 cells showed a higher chemoresistance to 607 This is a provisional file, not the final typeset article doxorubicin and sorafenib under cirrhotic conditions. Conversely, we did not observe a change in 608 chemoresistance for HuH-7 and Hep3B2 cell lines to doxorubicin in response to cirrhotic conditions, 609 which may be due to underlying genotypic differences including differentiation levels, which alter 610 metabolism pathways through glucose, glutamine, and glutamate 52 . Hypoxia demonstrated a much 611 different impact on the HCC cells, upregulating the chemoresistance of HepG2 and HuH-7, but 612 downregulating chemoresistance of the C3Asub28 cell line. We did not observe a significant variation 613 of chemoresistance in the Hep3B2 cell line in 3D culture. Drug metabolism, measured by CYP3A4 614 expression, mirrored effective chemoresistance measured by IC50 values in cell vitality assays. We 615 saw an increase in CYP3A4 expression in the 3D culture of C3Asub28 and HepG2 cell lines compared 616 to 2D, but this expression decreased for the HuH-7 cell line for both hypoxic and normoxic conditions. 617 At a minimum, the presence of a 3D culture system significantly modulates the response of HCC to 618 chemotherapy over the standard 2D methods. Previously it has been shown 3D culture alters the 619 integrin ligands (such as AKT and RAF), which are the targeting for doxorubicin and sorafenib, 620 differently among different cell lines compared to 2D culture 49 . 621 Further, the stiffness of the microenvironment, seen in HCC patients with cirrhosis, modulates drug 622 resistance and should be taken into consideration when determining treatment options and doses. While 623 CYP3A4 maintains the majority of the drug metabolism in the liver, other enzymes such as CYP2A6, 624 1A2, and 2C9 has shown to have a minor contribution to drug metabolism 62 , and provide further 625 insight in IC50 of Hep3B2 variation we observed between 2D and 3D microenvironments but not 626 captured a difference in CYP3A4 expression. Further expansion of this work is needed to investigate 627 the varied response between existing HCC lines or patient-specific primary tumor cells that might 628 provide insight into the known differential effectiveness of standard HCC treatments in vivo 66 . 629 Potentially, other drugs used to treat HCC in clinical practice such as lenvatinib could be tested in this 630 system to observed its potential effect on HCC treatment 67 . 631 5 performing the studies and analyzing the experimental data. Writing: A.O. wrote the initial draft of the 651 paper. All authors discussed the results and revised the manuscript. 652