SEARCH

SEARCH BY CITATION

Keywords:

  • colorectal neoplasms;
  • unfolded protein response;
  • biological markers;
  • GRP78;
  • tumour suppressor protein p53;
  • retrospective studies

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

Adjuvant fluoropyrimidine-based (5-FU) chemotherapy is a mainstay of treatment for colorectal cancer (CRC), but only provides benefit for a subset of patients. To improve stratification we examined (for the first time in CRC), whether analysis of GRP78 expression provides a predictive biomarker and performed functional studies to examine the role of GRP78 in sensitivity to 5-FU. 396 CRC patient samples were collected in a prospective uniform manner and GRP78 expression was determined by immunohistochemistry on tissue microarrays using a well-validated antibody. Expression was correlated with clinicopathological parameters and survival. The role of GRP78 in 5-FU sensitivity was examined in CRC cells using siRNA, drug inhibition and flow cytometry. GRP78 expression was significantly elevated in cancer tissue (p < 0.0001), and correlated with depth of invasion (p = 0.029) and stage (p = 0.032). Increased overall 5-year survival was associated with high GRP78 expression (p = 0.036). Patients with stage II cancers treated by surgery alone, with high GRP78 also had improved survival (71% v 50%; p = 0.032). Stage III patients with high GRP78 showed significant benefit from adjuvant chemotherapy (52% vs. 28%; p = 0.026), whereas patients with low GRP78 failed to benefit (28% vs. 32%; p = 0.805). Low GRP78 was an independent prognostic indicator of reduced overall 5-year survival (p = 0.004; HR = 1.551; 95%CI 1.155–2.082). In vitro, inhibition of GRP78 reduces apoptosis in response to 5-FU in p53 wild-type cells. GRP78 expression may provide a simple additional risk stratification to inform the adjuvant treatment of CRC and future studies should combine analysis with determination of p53 status.

CRC is a leading cause of cancer-related death worldwide with approximately 1.2 million cases annually and over 600,000 deaths.[1] For stage III (node-positive) disease, there is overwhelming evidence to recommend the use of fluoropyrimidine(5-FU)-based adjuvant chemotherapy with improvements in overall survival of around 10%.[2-5] Unfortunately, a large proportion of patients do not benefit from adjuvant chemotherapy,[6, 7] and there is no reliable biomarker that can be used to determine the likelihood of responsiveness to chemotherapy for individual patients.[8] Meanwhile, for patients with stage II disease (node-negative), the benefit of adjuvant chemotherapy remains unclear, with benefit observed in only a relatively small subset of patients. For example, the QUASAR trial demonstrated only a modest benefit (3.6% improved overall survival) from adjuvant fluoropyrimidine/folinic acid (5-FU/FA) for stage II disease,[9] and large pooled analyses have failed to show a significant benefit,[10] even in patients with high risk clinicopathological factors.[11] This suggests that the current practice of stratification of patients with stage II disease, based upon clinicopathological parameters,[12] requires further refinements to improve the stratification of these patients for treatment. Therefore, for patients with stage II and stage III disease, there exists a need to identify predictive biomarkers to determine which individuals might benefit from the use of adjuvant chemotherapy.

A number of processes regulate cell fate in response to chemotherapy ultimately by regulating cell survival. This can either be through direct regulation of apoptosis or by promoting other pathways that permit such chemically stressed cells to survive. One such pathway, the unfolded protein response (UPR), is coordinated by Glucose-regulated protein 78 kDa (GRP78). GRP78 is an endoplasmic reticulum (ER)-resident molecular chaperone that is essential for correct protein folding and assembly in the ER lumen.[13] Microenvironmental stress, elevated glucose metabolism and a requirement for increased protein synthesis, typical of solid tumors such as colorectal cancer (CRC), lead to a disruption of ER homeostasis and accumulation of misfolded proteins in the ER.[14] Under such conditions, GRP78 dissociates from several important ER-resident transmembrane proteins leading to a cascade of signal transduction pathways, known as the unfolded protein response (UPR), that modulate cell survival.[15]

A paradox of the UPR is that the same pathways which lead to adaptation and cell survival may also trigger apoptosis, depending upon the nature, severity and extent of the stress. During mild or temporary environmental perturbations, the UPR promotes cell survival by reducing initiation of translation, thus reducing the burden of newly synthesized proteins on the ER.[15, 16] In addition to a reduction in protein load, the UPR promotes adaptation by upregulating genes that can augment the protein folding capacity of the ER, including GRP78 and promoting angiogenesis.[16] Under conditions of severe or prolonged stress, the UPR may activate a number of proapoptotic pathways that lead to cell death which has been linked with disease.[17, 18] The point at which prosurvival mechanisms switch to those promoting apoptosis is not fully understood, but it is clear that GRP78 appears to have a role in promoting apoptosis under specific conditions.[16] GRP78 has been found to be overexpressed in a variety of cancers and is also associated with poor outcome,[19-22] albeit that some studies have used antibodies that cross react with related proteins, as we will demonstrate. In vitro data indicate that GRP78 expression is often associated with an aggressive phenotype and drug resistance.[14] Thus, GRP78 has potential as a biomarker for tumor behavior and treatment response.

Small studies in CRC support a role for GRP78 in tumorigenesis.[23, 24] However, many of these are unreliable being based upon unvalidated antibodies and at least one of the most commonly used antibodies actually detects HSP72 in addition to GRP78 (see Supporting Information figures and discussion). Moreover, no study of CRC has yet examined the influence of GRP78 in the adjuvant setting. A recent report describing a cohort of 262 postsurgical patients treated for colorectal adenocarcinoma demonstrated that patients whose tumors expressed elevated GRP78 had nearly a 10% 5-year survival advantage over those with low GRP78, though this did not achieve statistical significance.[25] These findings, together with the body of evidence proposing a role for GRP78 in modulating drug sensitivity,[14] led us to examine the potential of GRP78 as both a prognostic and/or predictive biomarker. We performed studies of patient samples and in vitro analyses of GRP78 and found that not only is GRP78 expression an independent prognostic indicator with predictive value in the context of fluoropyrimidine-based chemotherapy, but also GRP78 expression levels determine the sensitivity of CRC cells to 5-FU thus providing a deterministic basis for the observed patient association.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

Clinical studies

Tissues and survival data were collected from 396 consented patients (194 colon, 202 rectal) who underwent surgery for CRC at the Royal Liverpool University Hospital, UK between 1993 and 2003. 114 patients received fluoropyrimidine-based adjuvant chemotherapy delivered at Clatterbridge Centre for Oncology NHS Foundation Trust, Wirral, UK. Details of oncological treatment can be found in Supplemental Materials and Methods.

Immunohistochemistry was performed on tissue microarrays using a validated antibody for GRP78 (as described in Supporting Information Materials and Methods). Staining was categorized semiquantitatively using a scale of 0–3: 0 = negative, 1 = weakly positive, 2 = moderately positive and 3 = strongly positive. For analysis, expression of GRP78 was dichotomized around the mean score into either low expression (intensity 0 or 1), or high expression (intensity 2 or 3).

Statistical analysis

The χ2 test, Kaplan–Meier curves and the logrank test were used to compare groups. A Cox proportional hazard regression was performed to assess independent predictors of survival. A multivariate analysis was performed using a stepwise forward selection process informed by the Akaike information criterion,[26] and likelihood ratio χ2 statistic. Statview version 5.0 (SAS Institute, Cary, NC) was used for statistical analysis.

In vitro studies

Procedures for cell culture, siRNA transfection, SDS-PAGE and immunoblotting and flow cytometric measurement of apoptosis and cell cycle were as performed previously described,[27, 28] except where modified in Supporting Information Materials and Methods. 5-FU from medac GmbH, Wedel, Germany. EGF-SubA, from SibTech, SibTech Inc., Brookfield, CT, USA, a recombinant protein containing a SubA moiety genetically fused to EGF. SubA is the active A subunit of a subtilase cytotoxin (SubAB), demonstrated to cause highly selective cleavage of GRP78 between leucine residues 416 and 417.[29]

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

GRP78 expression in CRC

We set out to examine whether GRP78, a critical regulator of the UPR with roles in cell survival in response to a range of stresses, might serve as a predictive biomarker of response/benefit to adjuvant 5-FU-based chemotherapy in a cohort of CRC patients. Many studies of GRP78 in clinical samples have used an antibody that we found cross reacts to the apparently more abundant related protein HSP72 (Supporting Information Fig. S1). We therefore tested a number of antibodies in vitro and validated one (sc13968, Santa Cruz, CA; see Supporting Information Fig. S1 and also siRNA knockdown of GRP78 in Supporting Information Figs. S2–S4).

GRP78 expression was examined by immunohistochemistry and was detected ubiquitously in the cytoplasm of colonic carcinoma cells with all tumor cells in an individual tumor staining comparably. Minimal staining was evident in stromal tissue (Fig. 1). Of 41 cases where matched normal colonic epithelium was available to score, GRP78 expression was significantly elevated (intensity 2 or 3) in 16 of 41 cancers versus 1 of 41 matched normal colonic epithelium (p < 0.0001, Supporting Information Table S1).

image

Figure 1. Immunohistochemical analysis of GRP78 expression in representative samples. A tissue microarray (TMA) was stained using a rabbit polyclonal antibody specific for GRP78 (sc-13968, Santa Cruz, CA, see validation in Supporting Information Figures S1–S4). This figure illustrates magnified views of representative cores from the TMA. Control tissues (a) normal colon, (b) testis, also kidney and liver (not shown), provided a range of staining intensity for each section of the TMA. Also displayed are representative cores from CRC samples demonstrating (c) no staining = 0, (d) weak staining = 1, (e) moderate staining = 2 and (f) strong staining = 3. Original magnification ×40. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Correlation of GRP78 with clinicopathological parameters

In keeping with a previous study,[23] no correlation could be found between GRP78 expression and gender, age, nodal status (pN stage), differentiation, tumor size, resection margin status or administration of preoperative radiotherapy (Table 1). However, we did observe an association between GRP78 expression and depth of invasion (pT, p = 0.029) and stage (p = 0.032).

Table 1. Association between GRP78 expression and clinicopathological parameters in 396 patients with colorectal cancer treated by surgery
  All
   GRP78 HighGRP78 Low 
Characteristicn = 396%n = 162%n = 234%P-Value
  1. χ2 test was used for comparison of variables. Stage I, pT1–2/pN0; Stage II, pT3–4/pN0, Stage III, pT1–4/pN+.

Gender       
Male24862.610263.014662.4 
Female14837.46037.08837.60.964
Age       
<68 years178456942.610946.6 
≥68 years218559357.412553.40.433
pT       
T1194.874.3125.1 
T25914.9169.94318.4 
T326566.911067.915566.2 
T45313.42917.9241.30.029
pN       
N022256.1825.614059.8 
N19925.04628.45322.6 
N27518.93421.04117.50.190
Stage       
I6015.2169.94418.8 
II1624.9664.79641.0 
III17443.98049.4944.20.032
Differentiation       
Well102.531.973 
Moderate35489.41479.720788.5 
Poor287.1127.4166.8 
Unrecorded41.00041.70.755
Tumor size       
<60 mm27569.411571.016068.4 
>60 mm1213.64729.07431.60.579
Resection margin      
Clear35188.613885.221391.0 
Involved4411.12314.2219.0 
Unrecorded10.310.6000.099
Adjuvant chemotherapy      
No28271.211067.917274 
Yes11428.25232.162270.226

GRP78 expression, survival and benefit from adjuvant chemotherapy

Overall increased 5-year survival was associated with high GRP78 expression in the whole cohort (53% vs. 42%; p = 0.036; Fig. 2a). Patients with stage II cancers treated by surgery alone with high GRP78 expression had improved survival (71% vs. 50%; p = 0.032; Fig. 2b). Insufficient numbers of stage II patients received adjuvant chemotherapy (n = 32) to perform analysis of the influence of GRP78 expression on this subset of patients.

image

Figure 2. Kaplan–Meier plots showing overall survival according to GRP78 expression. Kaplan–Meier curves for the whole cohort (a); stage II cancers (excluding patients who received adjuvant chemotherapy) (b) and in stage III cancers with high GRP78 (c) or low GRP78 (d) and in (e) analysis of all patients receiving adjuvant chemotherapy dichotomized by GRP78 expression. Adj.CT = adjuvant chemotherapy. P values according to the logrank test are illustrated.

Download figure to PowerPoint

Amongst patients with stage III cancer, 80 (46 rectal and 34 colon cancers) received 5-FU-based adjuvant chemotherapy with a survival benefit of 10% at 5-years, though this was not statistically significant (40% vs. 30%; p = 0.150; Supporting Information Fig. S5). We noted however, that patients with high GRP78 showed significant benefit from adjuvant chemotherapy (52% vs. 28%; p = 0.026; Fig. 2c), whilst patients with low GRP78 failed to benefit (28% vs. 32%; p = 0.805; Fig. 2d). Indeed patients with low GRP78 performed comparably to those not receiving adjuvant treatment. Overall 5-year survival of stage III patients treated by surgery alone was not significantly different between those with high and low GRP78 expression (28% vs. 30%, p = 0.780), however median survival was 7 months longer in the high GRP78 group compared with the low GRP78 group (2.668 vs. 2.094 years).

In addition, regardless of stage, considering all patients who received adjuvant chemotherapy, those with high GRP78 had increased overall survival than patients with low GRP78 (56% vs. 32%; p = 0.008; Fig. 2e).

Multivariate analysis for the cohort (Table 2) demonstrated elevated GRP78 to be an independent prognostic indicator of overall 5-year survival (p = 0.004; HR = 1.551; 95%CI 1.155–2.082). Positive nodal status (p < 0.0001; HR = 1.871; 95%CI 1.399–2.502) and complete excision (p = 0.010; HR = 0.585; 95%CI 0.390–0.878) were also independent prognostic indicators. Since a key finding of the survival analysis was the observation that patients who benefitted the most from adjuvant chemotherapy were those with elevated GRP78, the univariate and multivariate analysis were repeated on the subgroup of 114 patients who received chemotherapy following surgery. Weak GRP78 expression remained a significant predictor of poor prognosis in the subgroup of patients who received adjuvant chemotherapy (p = 0.016; HR = 1.896; 95%CI 1.128–3.189; Table 3).

Table 2. Association between GRP78 expression and overall 5 year survivala
  UnivariateMultivariate
VariablenHR95%CIχ2p valueHR95%CIχ2p value
  1. a

    Excluding patients who died <30 days after surgery, HR, Hazard Ratio; CI, confidence interval.

Age3891.0121.00–1.0253.1570.0761.0171.004–1.0306.860.009
Tumor size         
<60 mm2710.8930.662–1.2060.5420.462
>60 mm1181
pT         
T1190.4520.187–1.0913.1220.077
T2580.9340.564–1.5450.0720.789
T32620.8360.557–1.2570.7390.390
T4501
pN         
N022011
N1–N21691.931.462–2.54721.55<.00011.8711.399–2.50217.83<.0001
Resection margin        
Clear3450.5380.365–0.7949.7590.0020.5850.390–0.8786.720.010
Involved4311
Differentiation        
Well/Mod35811
Poor271.7171.070–2.7565.0230.0251.7031.039–2.7914.4580.035
Grp78 expression        
Weak2291.3591.019–1.8134.3560.0371.5511.155–2.0828.5250.004
Strong16011
Table 3. Association between GRP78 expression and overall 5 year survival for patients who received adjuvant chemotherapy
  UnivariateMultivariate
VariablenHR95%CIχ2p valueHR95%CIχ2P value
  1. HR, Hazard Ratio; CI, confidence interval.

Age1141.0270.996–1.0582.9310.0871.030.998–1.0633.3290.068
Tumor size         
<60 mm781.2380.719–2.1330.5930.441
>60 mm361
pT         
T120.9240.121–7.0660.0060.939
T2101.4240.567–3.5760.5650.452
T3810.8980.484–1.6680.1160.734
T4211
pN         
N0341
N1–N2801.3720.789–2.3861.2550.263
Resection margin        
Clear931.3380.681–2.6270.7140.398
Involved201
Differentiation        
Well/Mod10711
Poor53.2401.294–8.1126.3000.0122.9271.145–7.4815.0320.025
Grp78 expression        
Weak621.9641.179–3.2716.7160.0101.8961.128–3.1895.8250.016
Strong5211

Inhibiting GRP78 by siRNA knockdown or EGF-SubA renders cells resistant to 5-FU

To investigate whether the association with GRP78 expression and benefit from 5-FU-based treatment we had observed might be causal, we examined the effect of inhibiting endogenous GRP78 using siRNA or EGF-SubA in HCT116 cells treated ±100 µM 5-FU (concentration required to inhibit proliferation by 50% by 24 hr). Cells transfected with siRNA for GRP78 and exposed to 5-FU displayed substantially reduced apoptosis compared to cells transfected with a scrambled control siRNA or mock-treated (no siRNA) cells (Fig. 3a). Note that the level of apoptosis observed in GRP78 siRNA transfected cells exposed to 5-FU was comparable to the level observed in cells not exposed to 5-FU (Fig. 3a), which suggests that GRP78 plays a substantial role mediating 5-FU-induced apoptosis in these CRC cells. As expected from previous studies, treatment with 5-FU resulted in an increase in the number of cells in S-phase, together with a reduction in the number of cells in G1 and G2/M phases and this was unaffected by GRP78 downregulation with siRNA (Fig. 3b).[30] We also noted an accumulation of sub-G1 events following exposure to 5-FU under all conditions examined (Fig. 3b). In accordance with expectations due to the ability of Annexin V staining to detect earlier apoptotic events, the number of cells with a sub-G1 DNA content was lower than the number of Annexin V positive cells detected in Figure 3a. Finally, as Supporting Information Figure S2 shows, PARP cleavage, a further indicator of apoptosis, was detected in 5-FU treated cells.

image

Figure 3. GRP78 expression determines sensitivity to 5-FU-induced apoptosis in CRC cells. For all panels A–D the upper panel displays the data collected and the lower panel displays a histogram summarising the data. (a) HCT116 cells were transfected with the indicated siRNA (40 nM) or mock treated (Mock Tx) by omitting oligonucleotides. 48 hr later, cells were treated as indicated with 100 µM 5-FU or vehicle control (PBS) for 24 hr and then harvested for flow cytometry. Bivariate distributions of Annexin V (AV)-positivity vs. propidium iodide (PI)-positivity were generated for each population and the data used to determine the percentage of viable (PI/AV –ve), apoptotic (AV +ve) or necrotic (PI +ve/AV−ve) cells present. (b) DNA content was measured in fixed asynchronously growing cells using propidium iodide (PI) staining 72 hr after transfection with siRNA GRP78 or mock/scrambled controls. Cells were exposed to 100 µM 5-FU or vehicle control for 24 hr prior to harvesting. The percentage of cells corresponding to each phase of the cell cycle based on DNA content is displayed. (c) HCT116 cells were incubated as indicated with EGF-SubA (2 nM) and/or 5-FU (380 µM) or vehicle control for 24 hr and then harvested and analyzed by flow cytometry for Annexin V and PI staining as described for panel A. (d) Cells treated as in (c) were stained and analyzed by flow cytometry for DNA content as described in (b).

Download figure to PowerPoint

The data in Figure 3 show that GRP78 expression levels determine the sensitivity of these cells to induction of apoptosis by 5-FU. The tumor suppressor p53 is frequently mutated in CRC,[31] and is a key regulator of apoptosis with a well-documented role in apoptotic signalling in response to 5-FU.[32] Moreover, on-going clinical trials in other cancers suggest that p53 can provide a useful biomarker for patient stratification (NCRI trial CLL-206). We therefore investigated whether p53 was required for GRP78-dependent 5-FU-induced apoptosis by taking advantage of the availability of isogenic HCT116 cells that lack p53 (p53-null). As expected, the ability of 5-FU to induce apoptosis was compromized in p53-null cells regardless of the level of GRP78 expressed (Supporting Information Fig. S3a). These data suggest that the apoptosis that ensues following treatment with 5-FU in the presence of GRP78 is substantially p53-dependent, which has implications for future studies of GRP78 in CRC (see “Discussion” Section). In contrast to p53 wild-type cells, p53-null HCT116 cells demonstrated no S-phase arrest after exposure to 5-FU, but did display a reduction G2/M that has been reported previously (Supporting Information Fig. S3b).[33]

5-FU has both cell cycle-dependent and independent effects on cells. Since GRP78 can allow cells to escape cell cycle-dependent drug-induced death by promoting growth arrest,[34] we speculated that one way for GRP78 to protect cells from 5-FU-induced apoptosis could be through altering the cell cycle. In Figure 3b, under conditions in which GRP78 levels were substantially reduced, we observed that GRP78 suppression had little impact on the cell cycle ±5-FU. It is not possible to completely suppress GRP78 using RNAi, therefore, to achieve a more effective inhibition of GRP78 function we used a novel compound which specifically targets GRP78 in EGFR expressing cells: EGF-SubA.[35] Supporting Information Figure S4 shows that EGF-SubA promotes cleavage of GRP78 in HCT116 cells in a dose-dependent manner. We therefore examined the effect of EGF-SubA-mediated inhibition of GRP78 on 5-FU-induced apoptosis and also on the cell cycle. Figure 3c shows that EGF-SubA treatment induces a small increase in apoptosis, but considerably less than that induced by 5-FU. Importantly, as we had observed with siRNA-mediated downregulation of GRP78, inhibition of GRP78 using EGF-SubA results in a substantial reduction in 5-FU-mediated apoptosis. Thus comparable results are obtained using two different approaches to inhibit GRP78 and both demonstrate that GRP78 expression/function contributes substantially to apoptosis induced by 5-FU. Note that analysis of the cell cycle profile of these cells indicates that treatment with EGF-SubA leads to an increase in the G1 phase with concomitant reductions in G2/M (Fig. 3d). As we observed similar effects on 5-FU-mediated apoptosis in Figures 3a and 3c, but did not observe any detectable effect on the cell cycle using siRNA to downregulate GRP78, it appears unlikely that the protective effect of loss/reduced GRP78 function/expression is due to a reduction in cell cycling. It is possible that these cell cycle differences may reflect the different consequences resulting from inhibition of GRP78 which can variously lead to activation of the UPR as a survival measure or as a promoter of apoptosis depending upon the nature/extent/duration of the stimulus[15-17] (discussed above) and this requires further investigation. In summary, these results demonstrate a role for GRP78 expression in determining apoptosis in response to 5-FU. Taking all of our results together, we find that GRP78 predicts benefit from 5-FU-based adjuvant chemotherapy and determines response to 5-FU in CRC cells.

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

This study represents the largest CRC series examined for GRP78 expression and the first to examine survival and response to adjuvant chemotherapy. We find that GRP78 is elevated in CRC, is an independent prognostic indicator, a positive predictor of benefit from fluoropyrimidine-based adjuvant chemotherapy, and determines sensitivity to 5-FU in vitro.

The reported 5-year survival rate for stage II disease is 75% after surgery, but a subgroup of patients have survival similar to stage III patients of 50%.[36] Our data are in keeping with this, with low GRP78 stage II patients showing overall 5-year survival of 50% (compared to 71% with high GRP78). Contemporary clinicopathological high risk features for stage II disease such as elevated carcinoembryonic antigen, obstruction or perforation, T4 stage, inadequate nodal resection (<12 nodes), lymphovascular invasion and high-grade disease may predict worse outcome, but they do not predict response to chemotherapy.[10-12] Importantly, whilst low GRP78 can predict a poor prognostic group, our data from stage III disease further suggests these patients do not benefit from a fluoropyrimidine-based regimen.

Following recent negative reports,[37, 38] the role of targeted therapies for adjuvant treatment is uncertain and thus there continues to be reliance upon fluoropyrimidine-based chemotherapy as a mainstay of drug treatment for CRC. Patients with stage III disease and high GRP78 had a >20% higher 5-year survival compared to patients with low GRP78 who received the same adjuvant treatment (p = 0.026). Indeed, in this cohort, patients with low GRP78 failed to benefit from adjuvant chemotherapy. The possibility of predicting a treatment group with a higher risk of treatment failure, such as those with tumors expressing low levels of GRP78, could allow better selection and counselling of patients for targeted combination therapy, although this concept clearly requires validation in prospective clinical trials.

Another group has recently reported that GRP78 can serve as a positive predictor for response to doxorubicin/taxane-based adjuvant chemotherapy in breast cancer.[39] When considered with our data, this may necessitate a re-evaluation of our understanding of the role of GRP78 in cancer. With a few exceptions,[40] GRP78 expression in human cancers has typically been associated with more aggressive phenotype and poor prognosis.[20, 21, 41] The influence of GRP78 is likely to be tissue and drug specific, however, some studies suffer from small numbers of enrolled patients, heterogeneous tissue preparation, use of different GRP78 antibodies (in some cases not specific for GRP78 as we have demonstrated; see Supporting Information figures), and different scoring systems (NB* regarding antibody specificity: As HSP72 is typically detected at higher [apparent] levels than GRP78 in most of the cells types that we have examined using the cross-reacting antibody (sc-1050), and this antibody has been one of the most commonly used for studies of GRP78 by IHC, a considerable body of data regarding GRP78 may actually reflect HSP72 levels) {Takahashi, 2011 #1167}. Furthermore, in many cases, analysis did not extend to patients who had received adjuvant therapy. Similarly, in vitro studies often use a variety of conditions to induce ER-stress that, in many cases, result in cell cycle arrest,[42] a well recognized cause of drug resistance,[43] and therefore make interpretation of simple assays that only measure proliferation problematic. These conditions almost certainly fail to accurately model the prolonged physiological stress encountered during tumor growth. Methods not reliant upon cell proliferation, such as flow cytometry, may therefore be superior for detecting alteration in chemosensitivity upon manipulation of GRP78 in some settings.

Much remains to be elucidated regarding how UPR activation elicits such different cellular responses as survival and apoptosis. Unlike other approaches which can reduce expression of GRP78, EGF-SubA causes highly specific cleavage of this protein,[35] and therefore provides a useful tool for investigating the UPR. Downregulation of GRP78 with siRNA is a less efficient means of inhibition than promoting cleavage with EGF-SubA. Under such conditions (i.e., siRNA transfection), levels of GRP78 may still be sufficient to maintain ER homeostasis in unstressed cells. We envisage a model in which cells that have successfully adapted to the tumor microenvironment demonstrate persistently elevated expression of GRP78 and are therefore better equipped to process or remove proteins from the ER, thus preventing accumulation of misfolded proteins and avoiding activation of the UPR and cell-cycle arrest. This allows continued proliferation and thus sensitivity to cytotoxic agents (NB: in Supporting Information Fig. S4, full-length GRP78 is not induced at the highest levels of EGF-SubA, perhaps because further induction of constitutively hyper-activated GRP78 expression may not be possible in these tumor cells). In contrast, the ER has a reduced capacity to handle protein in cells with low levels of GRP78, so although cellular function can be maintained in the absence of increased protein burden, these cells are more sensitive to perturbations in the tumor microenvironment that trigger the prosurvival arm of the UPR. Whilst in vitro this can lead to inhibition of translation and cell cycle arrest, the in vivo consequences of this are not yet understood and may well lead to altered tumorigenicity. In this scenario, cells with low GRP78 are therefore more likely to display resistance to 5-FU. Models proposing adaptation to chronic ER stress such as this are not novel,[15] and furthermore, it has been demonstrated that overexpression of GRP78 in vitro is tolerated and, for example, does not result in cell cycle alteration.[44] It is of course possible that GRP78 may contribute to chemosensitivity through mechanisms other than direct activation of the UPR. In support of this, recent developments suggest that GRP78, usually described as being an ER-resident, can nevertheless be identified in the cytoplasm and on the cell surface of cancer cells, and furthermore, is biologically active and may affect cell viability at both of these sites.[45]

The finding that GRP78 expression can modulate apoptosis in response to 5-FU, but only in the presence of functional p53, is novel. However, a relationship between ER stress and apoptosis is well established.[46] 5-FU is believed to stabilize p53 expression by causing ribosomal stress that inhibits the MDM2-p53 negative feedback regulation.[47] PUMA and Noxa are BH3-only BCL-2 family proteins that have an essential role in regulating apoptosis and have been shown to be selectively transcriptionally activated during ER stress in a p53-dependant manner. Furthermore, ER stress-induced apoptosis has been shown to be attenuated in the absence of p53 in mouse embryo fibroblasts.[48] This established link between ER stress-induced apoptosis and p53 may account for the observations in this study. The important point is that future studies of GRP78 and adjuvant response will need to include analysis of p53 genotype since this seems highly likely to further refine stratification analysis.

In view of the multiple genetic alterations which define the evolution of CRC, much attention is currently focused on identifying molecular or genetic factors that may help predict survival or response to chemotherapy. Mutation of p53 and/or KRAS, defective DNA mismatch repair (dMMR) and loss of heterozygosity on the long arm of chromosome 18 are all important steps in the development of CRCs.[49] However, examination for genetic mutations such as these is often complex and in view of many conflicting reports in the literature, with the exception of KRAS,[50] the routine use of these biomarkers is not currently recommended. In contrast, GRP78 is easily detectable by immunohistochemistry using highly specific commercially available antibodies and may provide a simple and cheap alternative tool for risk stratification.

Clearly, the gold standard for evaluating a potential predictive biomarker such as GRP78 is a prospective randomized-controlled trial. Nonetheless, our study has several important strengths. First, the prospective uniform nature of specimen collection for all subjects by our pathology department minimizes potential systematic bias from this study. The study cohort was essentially a random sample of the expected target population and the chemotherapy delivered represents real world practice. Second, we used an antibody that we were able to validate using siRNA and EGF-SubA whereas some antibodies to GRP78 cross react with related, potentially more abundant, proteins such as HSP72 (Supporting Information Fig. S1). Finally, our clinical findings support our in vitro data demonstrating a functional role for endogenous and thus physiologically relevant levels of GRP78 in cellular responses to 5-FU.

In conclusion, the expression of GRP78 is an independent marker of survival in CRC and may be especially useful in identifying a poor prognostic group in stage II disease. Most importantly, GRP78 expression provides a biomarker that predicts (and determines) response to fluoropyrimidine-based adjuvant chemotherapy. Thus, detection of GRP78 may permit identification of both a group of patients that are likely to obtain benefit from 5-FU-based chemotherapy and a group with a high chance of failure to respond who might benefit from alternative therapies.

Acknowledegments

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

The authors thank Dan Palmer and Carlos Rubbi, University of Liverpool, for critical review of the manuscript and Dr Joseph Backer, SibTech Inc., CT, USA for generously supplying EGF-SubA.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledegments
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ijc28137-sup-0001-suppinfo.doc633KSupplemental Figure 1: Two antibodies used in published studies of GRP78 expression by immunohistochemical analysis were selected for pre-validation (Santa Cruz antibodies to GRP78 sc-1050 and sc-13968). HCT116 cells were cultured for 24 hours in normal media or normal media supplemented with EGF-SubA at 1nM. EGF-SubA is a fusion of EGF (which provides targeting to EGFR expressing cells) and the A sub-unit of a bacterial AB5-type endotoxin with subtilisin-like proteolytic activity which has only one known substrate: GRP78, and cleaves a di-leucine motif (Leu416-Leu417) in an exposed loop that links two GRP78 domains: the ATPase and the substrate-binding domain.[38] Thus treating EGFR positive cells with EGF-SubA enables confirmation of the identity of c. 78kDa bands obtained by western blotting with antibodies to GRP78. The sc-13968 antibody shows minimal non-specific staining and the main signal is almost completely abolished following treatment with EGF-SubA. In contrast, sc-1050 displays a strong cross reaction to a faster migrating protein that is not abolished by EGF-SubA treatment and which is thus likely to be a different protein. GRP78 shares 85% identity with HSP72 and therefore we probed the blot with an antibody that is specific for HSP72 and as shown, this co-migrates with the cross reacting signal. In addition, since HSP72 is induced by inhibition of HSP90, we therefore treated cells with geldanamycin and found as shown, that the faster migrating band was also strongly induced by geldanamycin and co-migrated with the band detected using the anti-GRP78 antibody sc-1050. Finally, data shown in Supplemental Figures 2, 3,and 4 using siRNA also demonstrate that sc-13968 has a high degree of specificity for GRP78. Using sc-1050 for IHC it would be impossible to distinguish between the relatively weak signal form GRP78 and the stronger signal from HSP72, indeed, published studies that have used this antibody may require re-validating to determine whether associations identified were due to GRP78 or HSP72. Since the anti-GRP78 antibody sc-13968 predominantly detects only GRP78 we used this antibody to investigate GRP78 expression in our cohort of CRC samples.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.