This document discusses using systems approaches and translational nephrology to investigate acute kidney injury (AKI). It summarizes research from the Wales Kidney Research Unit exploring ischemic preconditioning (IPC) as a potential prophylactic therapy for AKI. The research optimized an IPC protocol, identified protective gene expression signatures with IPC, and used Ingenuity Pathway Analysis to predict drugs that may recapitulate the protective IPC phenotype. Some predicted drugs were tested and found to reduce kidney injury in an AKI rat model, supporting drug repurposing approaches identified through transcriptional profiling and pathway analysis.

1. Systems approaches to
translational nephrology
Tim Bowen
bowent@cf.ac.uk
Wales Kidney Research Unit
Division of Infection and Immunity
Cardiff University School of Medicine
Data-Driven Systems Medicine Workshop
CUBRIC, Cardiff University
Wednesday 12 June 2019

2. Lifetime incidence risk for common
causes of morbidity when no disease
is present at age 30, US population
CKD estimated prevalence:
US population 2015–2030
How common is kidney disease (CKD)?
Hoerger et al. (2015) Am J Kid Dis 65, 403

3. Biomedical Research Units and Centres:
“formed through partnerships between
leading NHS organisations and
universities. Conduct translational
research to transform scientific
breakthroughs into life-saving treatments
for patients”
Wales Kidney Research Unit
(WKRU)
The only UK Biomedical Research Unit
or Centre with a focus on those
affected by kidney diseases

4. Resources
Expertise: microRNAs, Hyaluronic Acid (hyaluronan), Toll Like Receptors (TLRs)
Models: in vivo: Peritoneal infection (mouse), Toxic kidney injury (mouse), Ischemic
kidney injury (rat), numerous in vitro renal cell models
Samples: Tissue bank, external connections (e.g. NURTuRE, QUOD)
External connections: clinical trials, UK renal research community, SAIL, social care
WKRU Investigators
Donald Fraser, Tim Bowen, Mario Labeta,
Soma Meran, Anne-Catherine Raby, Bob Steadman

5. Acute Kidney Injury (AKI)

6. Acute Kidney Injury AKI
Defined
• Sudden Reduction in the Kidneys Ability to Function
Characterised
• Reduced Urine Output
• Increases in a marker of impaired filtration (creatinine)
Commonest Cause
• Ischaemic Reperfusion Injury (IRI): damage & loss of function
Ø Frequently as a result of infection (sepsis)
Outcomes
• 18% all stage mortality
• 20% risk of CKD at 12 months

7. Potential therapeutic intervention
Murry et al. Circulation (1986) 74, 1124
Ischaemic Preconditioning:
Repetitive brief interruptions in blood supply to the heart prior to prolonged IR-
Injury can “train” the heart tissue to resist IRI
• Reduced infarct size to 25% of control after 4 days
Measure
infarct

8. Prophylaxis of High Risk Individuals
Acute Kidney Injury (AKI)
• Ischemic- Reperfusion Injury (IRI)
Prophylactic therapy:
Ischaemic Preconditioning (IPC)
• Benefit: Animal models, direct and indirect
• Variable: Clinical Trials
• Optimal dose/timing unknown
Hypothesis
Profiling the IPC Kidney will provide data to
Predict Chemicals for AKI Prophylaxis

9. Adapted from Nature Reviews Drug Discovery (doi:10.1038/nrd3078)
The case for repurposing

10. 1. Optimise mechanical ischaemic preconditioning (IPC) protocol
2. Identify protective signature of IPC
3. Drug Repurposing: can IPA predict agents to recapitulate IPC
phenotype?
4. Test IPA predictions in in vivo AKI model
Investigate the role of preconditioning in AKI

11. IRI and Mechanical IPC
Direct IPC
IR-Injury 45 min
• Sublethal
• Good
recovery
IRI
Indirect
IPC
Direct 2/5
approach
IRI
In-direct 2/5
approach
Gaseous Anaesthesia
Laparotomy
Kidney mobilisation
Clamping of renal pedicle
Recovery
Foxwell et al. (unpublished data)

12. IRI/IPC Experimental Plan
Foxwell et al. (unpublished data)

13. Sham
IRI
Sham
IRI
0
100
200
300
Creatinine(mmol/L)
Pre-op Post-op
***
***
Sham
IRI
Sham
IRI
0
200
400
600
800
1000
CreatinineKinase
Pre-op Post-op
*
Tubular Necrosis
(no nuclei)
Tubular loss of Brush Border
Tubular
Dilatation
Casts
Capillary Endothelial
Swelling
Capillary Endothelial
disruption
Capillary Dilatation
Tubular Interstitial
Necrosis
MedullaMedulla
Ischemia Reperfusion
Injury (IRI)
Foxwell et al. (unpublished data)

14. Direct-IPC Unilateral-IPC Indirect-IPC
Ischaemic Pre-conditioning (IPC)
Foxwell et al. (unpublished data)

15. 1. Optimise mechanical ischaemic preconditioning (IPC) protocol
2. Identify protective signature of IPC
3. Drug Repurposing: can IPA predict agents to recapitulate IPC
phenotype?
4. Test IPA predictions in in vivo AKI model
Investigate the role of preconditioning in AKI

16. Sham
n=6
IRI
n=6
D-IPC
n=6
I-IPC
n=6
RNA sequencing Illumnia total RNA
Riberzero Gold Chemistry
54.4
million
reeds
Reads
post trim
53.6
million
Forward
reads
53.6%
Reverse
46.4%
26.9%
duplication
20.3 million paired
end reads
(40.6% target
mapped reads)
58.8
million
reeds
Reads
post trim
58.0
million
19.9%
duplication
23.2 million paired
end reads
(42.2% target
mapped reads)
21.1%
duplication
25.8 million
paired end reads
(42.5% target
mapped reads)
18.4%
duplication
23.3 million
paired end reads
(42.1% target
mapped reads)
Sham IRI D-IPC I-IPC
59.2
million
reeds
65.0
million
reeds
Reads
post trim
54.0
million
Reads
post trim
57.4
million
Forward
reads
51.8%
Reverse
48.2%
Forward
reads
52.6%
Reverse
47.4%
Forward
reads
52.5%
Reverse
47.5%
Average number of
reads generated
from each group
analysis
Trimmomatic
Stat RNA-seq aligner
DESeq-2
• 24 kidneys, 4 experimental groups
• Illumina RNA Ribozero Gold Chemistry
• Ave Seq Depth = 30M paired-end reads
• After trimming, alignment & mapping
≈ 22M paired-end reads
mRNA sequencing
Foxwell et al. (unpublished data)

17. IRI relative to Sham I-IPC relative to Sham D-IPC relative to Sham
Number of transcripts
p<0.05
log2FC >1 1667
log2FC <-1 1029
Number of transcripts
p<0.05
log2FC >1 1307
log2FC <-1 718
Number of transcripts
p<0.05
log2FC >1 894
log2FC <-1 166
-6 -4 -2 0 2 4 6
20
40
60
80
log2 (FoldChange)
-log10(p-value)
-6 -4 -2 0 2 4 6
20
40
60
80
log2 (FoldChange)
-log10(p-value)
-6 -4 -2 0 2 4 6
20
40
60
80
log2 (FoldChange)
-log10(p-value)
Expression analysis of mRNA sequencing data
Foxwell et al. (unpublished data)
883
955
22
33
825
50 165
IRI
D-IPC I-IPC

18. Principal component analysis
of mRNA sequencing data
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
−3
0
3
−30 −20 −10 0 10 20
PC1: 91% variance
PC2:3%variance
group
●
●
●
●
aipc
bipc
biri
sham
3 0 -3
PC2: 3% Variance
PC1: 91% Variance
-30 -20 -10 0 10 20
Sham
IRI
D-IPC
I-IPC
Foxwell et al. (unpublished data)

19. Ingenuity Pathway Analysis (IPA)

20. Dataset Molecules
2,358 genes Number of molecules with a p<0.05 and log2FC of <-
1 or >1
Canonical Pathways
Overlap of dataset molecules with known pathways
Upstream regulators
whose differential regulation would result in the
expression change seen in dataset
Diseases and Biological Functions/ Tox Functions
and Tox Lists
Linking dataset differential expression to biological
processes, diseases and disease endpoints
Regulatory Effect
Links regulators to dataset and finally the disease of
function
Functional enrichment analysis was
completed using omics analysis
platform IPA:
Ingenuity Pathway Analysis
• 16,780 transcripts uploaded
• Significance cut-off criteria of
p<0.05 and log2FC of <-1 or >1
• 2,358 transcripts mapped:
• 852 downregulated
• 1506 upregulated
IPA: Core Analysis Work Flow
Foxwell et al. (unpublished data)

21. -5.0 0 5.0
Z-Score
500 statistically significant (overlap p-value) functional annotations to the dataset
• Filtered p<0.05 and a z-score of >2 or <-2
• 134 functional annotations fulfilled significance inclusion criteria
• Ranked by z-score and the top 30 upregulated and 10 downregulated
• Predicted biological functional alterations within the dataset linked primarily to cellular injury and
inflammatory cell recruitment
• Tissue displayed differential transcript expression favouring a injury recovery phase
Diseases and Biological Functions - I
Foxwell et al. (unpublished data)

22. Diseases and Biological Functions - II

23. Regulatory Effect Analysis- Renal
Foxwell et al. (unpublished data)
IPA
Regulatory Effect Analysis tested for
upstream regulators and dataset
molecules that link to disease or
function related to renal disease:
converges on injury of renal tubule

24. Data set molecules
Regulators
Canonical Pathways
Cellular Processes
Diseases
Data set molecules
Regulators
Canonical Pathways
Cellular Processes
Diseases
IRI vs Sham
Core Analysis
Preconditioning vs Sham
Core Analysis
Comparative Analysis
Comparative Analysis
Foxwell et al. (unpublished data)

25. -log (p-value)
0
2
4
6
8
10
12
14
16
18
20
22
24
26
IRI
I-IPC
D-IPC
Renal Damage
Renal Tubule Injury
Hepatocellular Carcinoma
Liver Hyperplasia/Hyperproliferation
Renal Necrosis/Cell Death
Cardiac arrhythmia
Glomerular Injury
Cardiac Infarction
Cardiac Arteriopathy
Renal Fibrosis
Toxicity Functions
Foxwell et al. (unpublished data)
2,385 toxicity annotations:
grouped into 87 high level functions

26. 1. Optimise mechanical ischaemic preconditioning (IPC) protocol
2. Identify protective signature of IPC
3. Drug Repurposing: can IPA predict agents to recapitulate IPC
phenotype?
4. Test IPA predictions in in vivo AKI model
Investigate the role of preconditioning in AKI

27. Profile-based candidate drug discovery
Profile-based drug discovery:
• Contrasts to typical approach in which key pathway(s) are manipulated
• Drugs treatment targeted at the whole transcriptional signature, not any
particular receptor(s) or pathway(s)

28. Profile-based candidate drug discovery
Approaches taken ask “can we identify
candidate drugs predicted to recapitulate
the common protective profile identified?”
These leads then evaluated in phase 2
(predicted activation signal of candidate
evaluated vs. IRI and D-IPC datasets) and
phase 3 (grow downstream pathways from
candidate, compare core analysis)

29. 1. Optimise mechanical ischaemic preconditioning (IPC) protocol
2. Identify protective signature of IPC
3. Drug Repurposing: can IPA predict agents to recapitulate IPC
phenotype?
4. Test IPA predictions in in vivo AKI model
Investigate the role of preconditioning in AKI

30. Sham
IRI
Vehicle
6 x chemicals
t = 48 h
Single IP injection 30 min prior to IRI
Experimental AKI: IRI rat model
Foxwell et al. (unpublished data)

31. Experimental Data from AKI Model
Foxwell et al. (unpublished data)
A B C D E F G H I A B C D E F G H I
A: sham; B: IRI; C: DMSO; D – I: IPC mimic compounds suggested by IPA
Compounds predicted by IPA to mimic IPC protection from IRI injury
reduced serum creatinine and histological damage

32. • Optimal method of IP is Direct-IPC
• mRNA Seq: significant inflammatory phenotype, inhibited by IPC
• IPA can be used to predict chemical/pharmacological agents to be tested for drug repurposing
• IPA predictions supported by experimentation in vivo
Summary

33. Acknowledgements
David Foxwell
Usman Khalid
Rob Andrews
Gilda Pino-Chavez
Rafael Chavez
Donald Fraser
Wales Kidney Research Unit / SIURI