Given the fact that cancer cachexia has been known for such a long time, our u nderstanding of the mechanisms of ca ncer cachexia lag way behind our knowledge of other aspects of tumor biology. This is despite the increase of the prevalence and significance of the syndrome. Exploratory cli nical research into the basic etiopatholog ical causes of cancer cachexia is lacking and we don't cu rrently have any a pproved effective treatment for the i nvoluntary weight loss in cancer patients.
The main aim of this thesis was to explore e a rl y d ifferential changes in gene expression in cancer patients with o r without cachexia using state-of-the-art whole expression profil i ng technology ( RNA-Seq) in order to identify the mechanistic pathway/s responsible for ea rly deve lopment of cancer-associated cachexia . This can bring reliable, representable, and consistent data from the clinic and back to the bench with more focused insights to be investigated and verified .
It is our hypothesis that ident ification of early changes i n gene expression using whole transcriptome a n a lysis i n cachexia will lead to a n i m proved understa nding of the triggering m echanism in cancer patients. Also, unde rstanding cancer cachexia can be best achi eved throug h studying representative clin ical m uscle and fat sam ples.
Our research specific o bjectives were to:
1 . Collect representative m uscle and fat biopsies from cancer patients with cachexia and m atch them with non-cachectic weight-stable cancer controls.
2. Conduct g lobal gene expression assays from high q ua l ity representative m u scle and fat biopsies from cancer patients with early cachexia and compare them to non-cachectic weight-stable cancer controls.
24
3 I dentify differe ntia l u p and downre g ulated pathways involved i n each of the two tissues.
4 I dentify and analyze the highly sign ifica ntly u p and downreg u lated altered genes in each of the two tissues.
5 Verify expressi o n data from clinica l and preclin ical m odels of cance r cachexia against the globa l gene expression results.
6 . Identify and ana lyze the overla pping differentially expressed genes between cachectic skeletal m uscles and cachectic adipose tissues.
7 Confirm I ncreased or d ecreased expression of a selected genes using real-time RT-P C R .
Chapter 3 : Materials and Methods
3 . 1 Eth ical App rova l
The eth ical approval request was submitted t o "AI Ai n Medical District Human Resea rch Ethics Committee (AAM DH REC)" at the Col lege of Medicine and Health SCiences (CM HS) in the U nited Arab Emirates U niversity ( UAE U ) . AAM DHREC is accredited by the Federa l Wide Assurance ( FWA) u nder the n u m ber 00007 1 09.
Research ethical proposal and protocol obtained u nder a pproval n u m ber AAM D H R E C 1 1 /49. Written informed consents were obtained from each subject a nd each subj ect was also given "Patient I nformation Sheet" i n the lang uage of preference ( i . e . Arabic or English) as per the comm ittee recommendation.
3 . 2 S u bjects
A total of 24 biopsies were obtained; 1 2 skeletal m uscles and 1 2 visceral ad i pose tissue. Each set of the two tissues contains 6 biopsies from cachectic cancer cases and 6 weig ht-stable controls. All subjects (cachectic and weight-stable) were with diagnosed malignancies. All samples were obtained through e lective surgeries.
I nclusion and excl usion criteria were as following:
I nclusion C riteria : Cancer patie nts of any race, sex, and age equal or a bove 21 year
old with h istolog ically proven malignancy and u nder active treatment. All cases
should have documented or self-reported weight-loss (cachexia) t h at is measured
by 5- 1 0% involuntary weight loss of the total body m ass in the past 3-6 months who a re u ndergoing elective abdominal surgery who agreed to sign an i nformed consent.
Exclusion criteria: Subjects who are u nder 2 1 year-Old. Any s u bj ect with any a utoi m m u ne and/or m uscular disease, u ncontro lled diabetes or thyroid disorder,
2 6
history o f treatment with any anabolic/catabolic agents for t h e past 6 months, o r h istory o f physical impairment that affects mobility was excluded.
Control s u bj ects were m atched for age and d iagnosis (as possible) and with no documented or self-reported wei g ht loss or gain in the past 3-6 months prior to surgery (see Figure 1 for study flow d iagram).
Surgical cancer patients from cliniC visits and OR lists Excluded:
and weight stable controls In thepast 6 months
Rectus abdominismuscle and
3.3 C l i n ical Paramete rs
Body weig hts were measured with participants in light clothing using a beam scale (Seca). Heights were measured using a standard wal l mounted measure and body m ass ind ices ( B M I ) were calculated . All current available nutritional and Inflammatory clinical parameters were recorded. H istory of weight changes from patients' records a nd/or self-reported weig hts were also recorded.
3.4 C o l l ection of M uscle and A d i pose Tissue Biopsies
All biopsies were taken at the sta rt of either open o r laparoscopic abdominal surgeries. S pecimens of a pproxi mately 0 . 5- 1 . 0 cm3 from the rectus abdominis a nd/or visceral fat were removed . The biopsies were washed with normal saline solution, blotted , and im mediately p laced in RNALater (AM702 1 ; Applied B iosystems, Carlsba d , CA, U SA) and kept overnight at 4DC to a llow thorough penetration of the tissue. Tissue sa m ples were then frozen at -80DC until further processing .
3.5 Sample Labeli n g a n d S u bject Co nfidential ity
Collected samples were g iven identification labeling com posing of two letters to d enote o rig in (Le. "M" for m uscle, and "F" for fat) and g ro u p ( L e . "S" for cachexia cases, and "C" for weig ht-stable controls). Each sample i s then followed by a unique identification n u mber. To ensure confidentiality, all other s u bj ect identifications were removed and never com m unicated in downstream sample processing. For future possi ble reference to subject data, the link of t hese labeling and actual med ical record n u m be rs i s kept under the a ccess of the leading experimenter and the principle investigator only.
3.6 Sam ples Prepa ration for RNA a n d P rotein Extraction
Sam ples of m uscle (45-59 mg) adipose tissue (95- 1 73 mg) were homogenized in ice in 1 2 ml of trizol Reagent ( 1 5596-026, TRlzol® Reagent, Life Technologies, Thermo Fisher Scientific, U SA) using the polytron h omogen izer (POLYTRON® PT 2 1 00 Homogenizers, Klnematica, Switzerland). Sample homogenizations were done at i nterm ittent intervals with sam ples kept in ice in between to avoid RNA and protein degradation I ntermittent homogenization was the best method to ensure complete retneval of R NA and protein from samples that a re d ifficult to homogenize (e.g.
m uscle) or those who have low R NA and protein yield ( i . e . fat) [ 1 1 7- 1 1 9] . Homogenization was done in batches o f 6 samples o r less a t 1 500-2000 rpm.
Com pletion of each batch took 3 hours (fat) and 4-5 hours (m uscle) till n o clear visible homogenate particles were observed . Polytron generator was cleaned in between sam ples in the fol lowing sequence; D Nase/R Nase free distilled water, 70%
ethanol in D Nase/R N ase free distilled water, followed by D Nase/R Nase free d istilled water. To ensure pro pe r rem oval of tissue sticking on the generator, a solution of 1 % of sodium dodecyl s ulfate in DN ase/R N ase free distilled water was used every 1 -2 hours a nd as needed to clean the probe .
3 . 6 . 1 R N A E xtraction
Once a com plete tissue homogen ization h ad been achieved, R NA extraction was ca rried out using a combination of the liqu id-liquid extraction tech nique; the acid g uanidiniumthiocya nate-phenol-chloroform extraction (AG PC) [ 1 20, 1 2 1 ] , and the colum n-based system . The AGPC method was used to separate the aqueous phase ( R NA conta ining phase) . This step was preceded with a n additional centrifugation for fat sam ples only in o rder to remove the floating fat layer before the addition of the chloroform. The total R NA was then precipitated using ethanol and p rocessing of
3 0
R NA isolation was completed using the SV Total RNA Isolation System® (Promega, Madison, USA) accord ing to manufacturer's instructions. The concentration and pUrity of the R NA sam ples were determi ned by measuring the absorbance at 260 nm and the ratio of the absorbance at 260/230 and 260/2S0 nm using spectrophotometer (ND-1 000 NanoDrop, Thermo Scientific, Wilmington, DE, USA).
Only ratios � 1 . S were accepted or else, extractions were repeated till the required purity was achieved. RNA samples were kept on ice during processing and quality check and then stored at -SOD C pending further processing.
3.6. 1 . 1 R N A Concentration
S a m pl e yields below 20 ng/lJl were concentrated u s i n g a cold v a c u u m centrifuge.
This was followed by another check for R NA concentration and purity using spectrophotometer ( N D- 1 DOO N anoDrop, Thermo Scientific, Wil m i ngton, D E , U SA).
3.7 Reverse Transcri ption
Total R NA was converted into cDNA uSing a High C a pa city cDNA Reverse Transcription Kit® ( H i g h Ca pacity cDNA Reverse Transcription Kit; 4374966;
Appl ied B iosystems, Carlsbad, CA, U SA) a ccording to the manufacturer's i nstructions. Total R NA (420ng) was converted into cDNA in a 301-.11 reaction volume using the fol lowing reaction (Table 1 ) and was repeated whenever needed:
Table 1 : Reverse Transcription Reaction.
Reverse Transcription M aste r Mix Com ponents Volume ( 1 x)
1 0 x RT Buffer 3.0 IJ I
2 5 x d NT P Mix 1 .2 IJ I
1 0 x RT Random Primers 3.0 IJ I
Multiscribe™ Reverse Transcri ptase 1 .0 IJ I
R NASE I n hibitor 1 .0 IJ I
R Nase Free H2O = (30 -9 . 2 -vol of 420ng
tem plate) IJI
Total Volume Reaction 30.0 1-1 1
The a bove reaction gives a final cDNA concentration of = 420ng/30IJI = 1 4ng/1J1 . Each cDNA sam ple were then diluted to 4ng/1J1 and stored in -200 C. The reverse transcription reaction was ca rried out in a Veriti thermal cycler (Life Technologies, Applied Biosystems, USA) using the following parameter values (Table 2):
Table 2 : Parameters of the Thermal Cycler for Reverse Transcription.
Step 1 Step 2 Step 3 Step 4
Temp °C 25 °C 37 °C 8 5 °C 4 °C
Time 10 minutes 1 20 minutes 5 minutes
3.8 Rea l-Time Reve rse Tra n s c ri ption Polymerase C ha i n Reaction
The expression levels of m R NA were analyzed using target specific TaqMan® gene expression assays using real-time reverse transcription polymerase chain reaction ( RT -PCR) and performed in a 7900HT Fast ABI Prism 7900HT Seq uence Detection System (Applied B iosystems, U SA). The fol lowi n g TaqMan® pri me rs and probes kits were purchased from Applied Biosystems (Table 3):
Table 3: List of the Taq Man® Primers for Rea l-Time RT-PCR .
Assay 1 0 Gene Sym bol
Assay 10: Hs01 098873_m 1 Gene Symbol: COL4A2, hCG33042 Assay 1 0 : H s00373339_m 1 Gene Symbol: N RXN2, hCG 1 8 1 0991 Assay 1 0 . H s00394748_m 1 Gene Symbol: AGR N
Assay 1 0 : Hs01 062 0 1 4_m 1 Gene Symbol: NOTC H 1 , hCG 1 8 1 8285 Assay 10. H s00373 1 36_m 1 Gene Symbol: PTP R R , hCG254 1 0 Assay 1 0 : Hs001 92297 _m 1 Gene Sym bol: N O U FS 1 , hCG1 7250 Assay 10 Hs001 74877 _m 1 Gene Symbol: LEP, hCG33000 Assay 1 0 . H s00223332_m 1 Gene Symbol: TNMO, hCG20 1 9 1 Assay 1 0 : H s00896336_m 1 Gene Sym bol: L HCGR, hCG 1 6776 Assay 10 H s04 1 87682_g 1 Gene Symbol: CXC L 1 1 , hCG2384 1 Assay 1 0 : H s02800695_m 1 , #432632 1 E Gene Symbol: HPRT1
E ndogenous Assay 1 0 ' H s0275899 1 _g 1 , #43263 1 7 E Gene Symbol: GAPOH
Controls Assay 1 0 : Hs01 060665_g 1 , #43263 1 5E Gene Symbol: ACTS
T h e reaction mix of 1 O�I containing a total cDNA of 4ng/reaction was prepared using TaqMan® H Fast U n iversal PCR Master Mix, No AmpErase H UNG (Life Technologies #4352042, Applied Biosystems, U SA) as follow (Table 4):
Table 4: Real-Time RT-PCR Reaction.
Real Time PCR Master Mix Reaction Volume ( 1 x)
DNase/RNase free H2O 3.5 � I
cDNA (stock concentration 4ng/l-l l) 1 �I
2 X TaqMan H Fast Un iversal PCR M aster M ix 5 �I
20 X Gene of I nterest (GO I ) o 5 � I
Quantitative real-time PC R assay for t h e gene o f i nterest was perfo rmed in duplicates and i n a sing leplex PCR reaction. The PCR thermal cycling parameters were run in fast mode as fol low (Table 5):
Table 5: Parameters of the Thermal Cycle r for Real-Time RT-PC R .
Step 1 Step 2 Step 3 Step 4
Temp °C 500 C 95 0 C 40 cycles of 60 0 C 95 0 C
Time 2 minutes 20 seconds 1 second 20 seconds
Results were initia l l y analyzed with the ABI Prism 7900HT S DS program v2, 4, all rem aining calcu lations and statistical analysis were perfo rmed by the SDS RQ Manager 1 . 1 ,4 software using the 2-1'1 Ct method with a relative q u a ntification
RQmin/RQmax confidence set at 95%. Three d ifferent endogenous controls were tested; h u m a n g lyceraldehyde-3-phosphate dehyd rogenase (GAPDH), human Hypoxanthine phosphori bosyltransferase ( H prt 1 ) rRNA, and human l3-actin. Hprt1 was found to be the best and used to norm a l ize expression resu lts. Other expression related g u idelines were followed from the M I Q E g uidelines [ 1 22].
3 3
a l Analysis
stlcal analysis of the next generation sequencing was carried out by in Q atar. Confirmation of changes in gene expression by real-time jone using Student's t-test and ana lysis was done using the
1 Software (version 5).
matics Tools
a i lable databases and software were used check gene functions, ern, protei n interaction, and functional pathways. These d atabases e:
N ational Libra ry of Medicine N ational I nstitutes of Health : lcbi .n l m . nih .gov/pubmed
ome Bioi nformatics: http://genome. ucsc. edu/
ttp:/ /www. ensembl. org/index. htm I
base: http://www. genome.j p/kegg/kegg 1 . html //string-db. org/
;eneral Repository for I n teraction Datasets : htt p : //thebiogrid.org/
oinformatics Resource Porta l :
expasy. org/proteom ics/protein-protei nj nteraction
eq uenc i n g
I t R NA sam ples (600ng) were sent i n d ry ice t o our collaborators in ledical Col lege i n Qatar for RNA Sequencing ( R NA-Seq) . R NA-Seq is y tra nscriptomic technique that uses deep-seq uencing technologies.
! utilizes a population of R NA that is reversed transcribed i nto a library lents.
3.9 Statistica l Ana lysis
33
Prelim inary statistical analysis of the next generation sequencing was carried out by our collaborator i n Qatar Confi rm ation of changes in gene expression by real-time RT-PCR was done using Student's t-test and analysis was done using the GraphPad Prism Software (version 5).
3.1 0 Bioinformatics Tools
M uch freely availa ble databases and software were used check gene functions, expression pattern, protein interaction , and functional pathways. These databases and software a re :
• T h e U S National Libra ry o f Medicine National Institutes o f Health : http://www. ncbi . nl m . ni h .gov/pubmed
• UCSC Genome Bioi nformatics: http://genome. ucsc.edu/
• Ensem b l : http://www. ense m b l . org/index. html
• KEGG database: http://www. genome.jp/kegg/kegg 1 . html
• String : http://string-db . org/
• Biological General Repository for I nteraction Datasets: http://thebiog rid. o rg/
• ExPASy Bioinformatics Reso u rce Port a l :
http://www.expasy.org/proteom ics/protein-proteinjnteraction
3 . 1 1 RNA Seq u e n c i n g
M uscle and fat R NA sam ples (600ng) were sent in d ry ice t o our collaborators in Weill Cornel Medical Col lege in Qatar for R NA Sequencing ( R NA-Seq). R NA-Seq is a revolutionary transcriptomic technique that uses deep-seq uencing tech nolog ies.
The tech nique utilizes a population of R NA that is reversed transcribed into a l ibrary of cDNA fragments.
3 4
All samples unde rwent quality check prior to seq uencing by measuring the R NA I nteg nty N u m ber and ensuring that all sam ples a re with a RNA I ntegrity N u mber value of at least 7 . Samples were then reversed transcri bed into cDNA. R NA-Seq was then performed on the I llumina H iSeq 2500 System .
Chapter 4: Results and Discussions - Muscle
4 . 1 Demogra p h ic a n d B iochem ical C haracteristics of S u bjects 4. 1 . 1 Resu lts
Demographic and biochemical details of patients for m uscle cachexia cases (MS) and m uscle weight-sta ble controls (MC) a re shown in bellow tables. The table shows the data after excluding two samples (one of each g roup) since they were outliers from the R NA-Seq data. The cause of these two samples as outl iers was investigated . Data on the excluded cachexia sample was obtai ned through patients self-report of weight changes in the past six months with the complete absence of baseline data in patient's record . The patient was a referral to the hospital from a d ifferent city j ust to be operated . On the other hand, fu rther investigation of the excluded control sa m ple in patient's extended electronic records in d ifferent health institutions found to h ave obvious fluctuation of patients weight i n between the two time points that o u r study has in itially selected patients o n . All sam ples were then ana lyzed based o n weight, B M I , some nutritional parameters, infl a m matory markers, and other related demogra p h ics.
Table 6 shows s u bjects' oncological diag noses for muscle biopsies. All patients were with histolog ically confirmed m alignancies. Age and gender distribution were not significantly d ifferent between cachexia m uscle cases (MS) and weig ht-stable m uscle controls (MC) (Table 7). All biochemical nutritional parameters ( i . e . hemog lobin, tota l protei n , a l b u m i n , u re a , and creatinine) and inflammatory markers ( i . e . WBC and C R P) were not significantly different between both g roups (Table 7).
Baseline weig hts and B M l s i n the cachexia gro u p did not d iffer sig nificantly from the contro l . The time ra nge between current (at time of biopsy) and baseline weig hts
3 6
( preceding date of biopsy) was 3-6 months. Mean weight loss from baseline in the cachexia g roup was 4 . 6 kg which is 6 . 5% of baseline body weight (i.e. 3-6 months before the date of sample col lection). Amount and percentage of weight loss was significantly different from controls that had stable weights; P-val ue of 0.00 1 4 and a 0003, respectively.
Table 6: S u bjects' Diag noses for Skeletal M uscle Biopsies.
Cachectic M uscle Cases (MS) Weig ht-stable Muscle Controls (MC) MS 1 · Malignant Neoplasm o f Rectum M C 1 : Malignant Neoplasm of Rectum MS2· Malignant Neoplasm of Stomach MC2: Malignant Neoplasm of Rectum MS3· Malig nant Neoplasm of Rectum MC3: Malignant Neoplasm of Colon M S4 Malignant Neoplasm of Duodenum MC5: Malignant Neoplasm of Connective
and Soft Tissue
MS6 Malignant Neoplasm of Pancreas MC8: Malig nant Neoplasm of Colon
Table 7: Demographic Data for Cachectic Muscle Cases and Weight-Stable Muscle
38
Table 8 com pa res weight and BMI changes between current and baseline within each g roup (paired t-test). The cachectic muscle cases (MS) had a significant difference between current and baseline data while weight-stable controls (MC) had no significant change.
Table 8: Paired t-Tests for Weight and BMI C ha nges for Muscle Biopsies.
C h a ra cteristics N-MS/MC; Mean SEM Med i a n Range P-Va lue 5/5
Weight MS Baseline 6 7 . 92 5 . 738 73.00 45.00-74.60; * * 0.0068 (cachectiC) Cu rrent 63.38 5 .2 1 3 67.90 42.80-7 1 .20
B M I MS Baseline 23.58 2 . 378 23 02 1 5. 57-29.24; * 0.0 1 40 ( cachectic) Current 2 1 . 9 5 2 .030 2 1 .98 1 4. 8 1 -26.44
Weight MC Baseline 8 1 .20 2 .653 84.00 73.00-87.00 0 . 3739 (weight-stable) Cu rrent 8 1 .00 2 .8 1 1 84 . 00 72.00-87.00
BMI MC Baseline 29.86 1 .765 30.46 25.26-35 38 0 . 3739 (weig ht-stable) Current 29.79 1 .8 1 1 30.46 24. 9 1 -35. 38
4. 1 . 2 D i scussion
I n the a bove demographic a nalysis, patients included in the study were chose n meticulously to be at the early stages of cachexia ( i . e . 5 - < 1 0% wei g ht loss).
Consensus definition had identified early cachexia as weight loss >5% or B M I
<20kg/m2 a nd weight loss >2% o r sarcopenia a nd weight loss >2% with often
reduced food intake or syste m i c i nflammation [7]. Our subjects had a mean percentage of weight loss of 6 . 5% (range 4 . 6-9.6 %). S ince systemic inflam mation measured via serum CRP may or may not be present in patients with cachexia [7], our subj ects were included reg a rdless of their CRP values (range 1 -34 , Table 6).
H owever, the C R P values betwee n cachectic cases (MS) and weig ht-stable controls (FC) were ensured to be not statistica lly significant; P= 0 . 325 in m uscle.
39
N utritional and n utrition-re lated biochem ical markers were all ensured to be comparable i n both g roups to objectively exclude the possibil ity of malnutrition as the primary cause of weight loss.
4.2 Heat M a p of RNA Seq u e n c i n g Data 4 . 2 . 1 Resu lts
Figure2shows the heat map of differe ntially reg ul ated g enes between cachectic m uscle cases (MS) and weight-stable controls (Me). The fig u re shows a com plete and clear cut separation between both g roups. The heat map shows a total of 238 sign ificantly u pregulated genes and 235 downregulated g enes (P-value < 0 . 0 5 ; and FDR < 0.5).
HIerarchICal ClJs tern 9
• Cancer • Normal Sample 10 • UAE_Mc_1 • UAE_Mc_2 • UAE_Mc_3 • UAE_Mc_5
• UAE_Mc_B • UAE_Ms_1 UAE_Ms_2 . UAE_Ms_3 UAE_Ms_ 4 • UAE_Ms_6
Figure 2 : Heat M a p Representation of the Differentially Expressed Genes in Skeletal Muscle Tissue.
Gene expression is shown for the cachectic m uscle cases (MS; purple bars; upper panel) and weight-sta ble controls ( M C ; g reen bar; lower panel ) . Colors represent scaled and centered expression values; red represents h igher expression, blue represents lower expression.
4.2.2 Discussion
The heat-m a p provides a g raphical visual ization of the m atrix dataset by
representing individ ual val ues with d ifferent colors. This necessity stem s from the increasing data volumes generated by scientific studies. Effective data visual ization enables understanding data at both broad and deta i led levels. Recent progress in h ig h-throughput techniques, such as R NA-Seq, has increased the demand for the visualization of m ulti-d i mensional in addition to the numeric data [ 1 23]. This sim ple
4 1
h ie rarchical cluster analysis revealed a clear visual distinction of gene expression sig nature between cachexia m uscle cases (MS) from those with weig ht-stable controls (MC).
4.3 Ana lysis of Diffe rential ly E x p ressed KEGG Pathways 4.3. 1 I ntrod uctio n to Pathways Ana lysis
Even for the sim plest biological function, sing le-gene expression analysis has its
own li mitations for understanding the biological mechanism of a d isease/syndrome.
C o m pl icated biological and patholog ical processes a re a fu nction of h undreds of d ifferentially expressed genes. These genes never work independently. Whole g enome expression analysis draws attention for discovering the hidden li nks disregarded by the sing le-gene analysis.
C o m pl icated biological and patholog ical processes a re a fu nction of h undreds of d ifferentially expressed genes. These genes never work independently. Whole g enome expression analysis draws attention for discovering the hidden li nks disregarded by the sing le-gene analysis.