Rapamycin signalling profile on THP-1

#1

Rapamycin — Complete Signaling Profile (THP-1, 24h, 10µM, n=2 reps)

PI3K / AKT / mTOR axis

The clearest rapamycin signals are negative feedback loops being released:

  • PIK3R1 +1.20 — PI3K regulatory subunit p85α upregulated (compensatory attempt to restore signaling)
  • RHEB +0.86, RPTOR +0.50 — mTORC1 scaffold components upregulated (cells trying to rescue mTOR activity)
  • AKT3 +0.84, AKT1 +0.35 — AKT isoforms upregulated
  • DEPTOR +0.44 — the endogenous mTOR brake is also upregulated — paradoxical; reflects feedback loop release
  • EIF4EBP1 −0.50 (4EBP1 mRNA down), PIK3CA −0.76 (PI3K p110α down), FOXO3 −1.06 ▼▼ (strongly suppressed)
  • FOXO3 strongly down is counterintuitive — classically mTOR inhibition should derepress FOXO3. What you’re seeing at the transcript level: FOXO3 protein is activated but FOXO3 mRNA is suppressed as part of a negative feedback (FOXO3 drives its own transcriptional repression when activated). This is a known biology.

RAS / ERK cascade

Unexpected: massive immediate-early ERK output gene induction

  • FOS +1.97, EGR1 +1.76, MYC +1.79, DUSP6 +2.00 — all strongly up
  • But ERK1/2 kinases themselves barely move (MAPK3 +0.61, MAPK1 +0.06) — this is a transcriptional program, not ERK kinase induction
  • Mechanism: when mTORC1 is blocked, the cell upregulates RAS/ERK growth programs as a compensatory bypass — a well-documented resistance mechanism (“mTOR→ERK feedback cross-talk”)

AMPK / Autophagy

  • ATG7 +0.77 (autophagy E1-like enzyme — expected: mTOR inhibition releases autophagy)
  • PRKAA1 +0.47 (AMPK α1 up)
  • ULK1 −0.62, MAP1LC3B (LC3B) −1.14 ▼▼ — counterintuitive: ULK1 and LC3B mRNA go down despite rapamycin activating autophagy at the protein level. This is transcriptional downregulation after autophagy flux increases — the cell consumes LC3B protein faster than it makes mRNA. At 24h the mRNA is being titrated down as protein turnover accelerates.

NF-κB / Innate

  • CCL2 +2.14 (MCP-1), TNF +1.91, MX1 +1.95, ISG15 +1.16 — rapamycin induces a strong innate immune-like signature even without an activating signal. This is an mTOR-immunosuppression paradox — rapamycin suppresses adaptive immunity but activates innate gene programs in THP-1 (a monocyte line). Known in transplant biology as rapamycin-induced inflammatory syndrome.

Metabolism (HIF axis)

  • SLC2A1 (GLUT1) −0.67, PDK1 −0.84, LDHA −0.45, VEGFA −0.66 — all Warburg/HIF targets suppressed. This is the clean mTOR signature: mTOR drives Warburg metabolism; block it and you see the full metabolic gene set collapse. CDKN1A (p21) −1.11 — also suppressed (different from what you’d expect with antiproliferatives; here rapamycin may be acting as primarily a metabolic drug in this 24h window).

Comparison to FK506 and CsA

FK506 (tacrolimus, same FKBP12 target) shows near-identical ERK immediate-early induction (FOS +1.69, EGR1 +1.75, MYC +1.29) and a similarly strong CCL2 +4.27 innate response — but stronger IL-1β (+3.16) and TNF (+2.40) than rapamycin. CsA (cyclosporin A, calcineurin inhibitor) shows similar FOS/MYC but dramatically suppresses p21/CDKN1A (−1.76 ▼▼) and drives higher IFN-stimulated genes (MX1 +2.82, ISG15 +2.14).

Adjuvant Rankings on ERK / NF-κB / IFN axes

Top ERK activators (FOS + EGR1 + ERK1 combined, all benzimidazoles):

Compound ERK score FOS EGR1 ERK1 MYC NFKB1
STING_BZ 9250705 4.26 +1.77 +2.09 +0.40 — +0.04
STING_BZ 8799982 3.73 +1.87 +1.43 +0.43 — +0.22
STING_BZ 8632851 3.31 +1.44 +1.62 +0.25 +0.67 +1.66
STING_BZ 8633429 3.06 +1.21 +1.57 +0.27 — +2.49

Top NF-κB activators (NFKBIA + IL6 + TNF + CXCL10):

  • STING_BZ 8633429: score 8.85, NFKBIA +4.48, TNF +3.35 — the highest NF-κB hit overall; also top-3 on IFN
  • STING_BZ 8632851: score 6.71, TNF +4.13
  • STING_BZ 9314375: score 6.23, CXCL10 +1.67, ISG15 +2.68

Top IFN/STAT activators (MX1 + ISG15 + IFIT1 + STAT1):

  • STING_BZ 9318433: IFN score 12.5 — MX1 +4.42, ISG15 +3.74, IFIT1 +2.49, STAT1 +1.86 — strongest type-I interferon signature in the entire 184-compound adjuvant set
  • STING_BZ 8633429: 10.20 — dual NF-κB AND IFN activator — rare and potent
  • STING_BZ 9318599: 9.03

Pattern: TLR7/8 agonists drive ERK harder (MAPK3 ERK1 top hits are 3/4 TLR7/8); benzimidazoles split into ERK-dominant vs IFN-dominant subtypes. JUN is consistently suppressed across ALL adjuvants (−1 to −3 log2FC) while FOS and MYC are induced — a FOS/JUN dissociation suggesting partial, non-canonical AP-1 activation distinct from growth-factor ERK signaling.a

#2

Rapamycin In VCPI

Rapamycin is present in the VCPI chemistry/training data as compound 8799854, matching sirolimus/rapamycin structurally. The dataset also contains close mTOR inhibitor analogs:

compound identity Morgan/Tanimoto match MW logP TPSA
8799854 rapamycin / sirolimus 1.00 914.2 6.18 195.4
8563589 everolimus 1.00 to everolimus, 0.873 vs rapamycin 958.2 6.20 204.7
8799896 temsirolimus 1.00 to temsirolimus, 0.842 vs rapamycin 1030.3 5.72 242.0

This is THP-1 monocyte-like cells, 24h, 10 uM. That matters: rapamycin is normally potent at much lower concentrations, so 10 uM is not just “clean mTORC1 inhibition”; it produces a broad stress/transcriptional state.

Big Picture

Rapamycin is a strong perturbation here.

For 8799854:

metric value
toxicity / perturbation score 8.235
mean absolute L2FC 0.411
genes with abs L2FC >= 1 886
pathway breadth percentile about top 96%
strongest pathway abs score 0.925

So rapamycin is not subtle in this assay. It is one of the broader transcriptional movers, though everolimus is even stronger and temsirolimus is somewhat weaker.

Major Directional Effects

Rapamycin’s module-level effects:

module signed effect
inflammatory / NF-kB-like +0.692
adaptive stress / NRF2 / ATF +0.353
ribosome / translation genes +0.266
mitochondrial / OXPHOS -0.336
apoptosis / p53 / ER stress +0.072

Interpretation: in THP-1, rapamycin looks like a mixed immune-stress state: inflammatory/interferon genes go up, mitochondrial/OXPHOS programs go down, and p53/apoptosis is not the dominant axis.

mTOR Axis

The mTOR panel itself is only modestly negative:

axis signed abs
mTOR / rapamycin axis -0.063 0.285

Leading mTOR-related genes:

gene L2FC
ATG13 +0.514
RPTOR +0.439
MLST8 -0.337
LAMTOR2 -0.393
MAPKAP1 -0.490
SESN2 -0.503
EIF4EBP1 -0.541
DDIT4 -1.129

Important caveat: this is RNA-seq-like expression, not phosphoproteomics. Rapamycin’s canonical biology is inhibition of mTORC1 signaling through protein phosphorylation. The transcriptional readout only sees downstream compensatory RNA changes, so the “mTOR score” will not perfectly equal biochemical mTOR inhibition.

ERK / MAPK Is Very Active

Rapamycin has a strong positive ERK/MAPK immediate-early signature:

axis signed abs
ERK / MAPK +0.425 0.480

Leading genes:

gene L2FC
EGR1 +1.712
DUSP6 +1.534
FOS +1.530
DUSP4 +1.361
JUN +1.233
DUSP5 +0.796
ETS1 +0.766

This is one of the clearest signatures: rapamycin is associated with immediate-early MAPK feedback activation in this THP-1 context.

Interferon / Antiviral / NF-kB

Rapamycin strongly activates an interferon/OAS-like immune module:

axis signed abs
TLR / STING / NF-kB / IFN +0.452 0.583

Leading genes:

gene L2FC
MX1 +2.084
OAS1 +2.039
OAS2 +1.828
ISG15 +1.639
TNFAIP3 +1.434
IFIT3 +1.174
IFIT1 +0.878
IRF7 +0.781

This is probably the most biologically interesting part of the result: rapamycin in THP-1 does not merely look like nutrient/mTOR perturbation. It also resembles an innate immune/interferon response.

AKT / PI3K / AMPK

AKT/PI3K is mildly negative overall:

axis signed abs
AKT / PI3K -0.058 0.280

Notable genes:

gene L2FC
PIK3R1 +0.746
GSK3B +0.555
FOXO3 -0.326
BAD -0.335
PIK3CA -0.435
EIF4EBP1 -0.541
FOXO1 -0.598
AKT2 -0.775

AMPK is also mildly negative on signed average, but with mixed genes:

axis signed abs
AMPK -0.121 0.395

Notable genes:

gene L2FC
TBC1D4 +0.825
STRADA +0.489
SIRT1 +0.489
SESN2 -0.503
ACACA -0.683
STK11 -0.746
DDIT4 -1.129
PRKAB2 -1.668

p53 / Apoptosis

Rapamycin is not a dominant p53 activator here.

axis signed abs
p53 / apoptosis -0.013 0.360

Mixed genes:

gene L2FC
RRM2B +0.720
CDKN1A +0.683
BBC3 +0.521
SESN1 +0.507
GADD45G -0.525
BCL2 -0.533
DDB2 -0.546
SESN3 -1.097

So there is some cell-cycle/checkpoint signal, but not a clean “p53-on/apoptosis-on” response.

Top Pathways Up

For rapamycin 8799854, strongest positive pathway scores:

pathway score
PKA-mediated phosphorylation of key metabolic factors +0.601
SMAD2/3 phosphorylation motif mutants in cancer +0.549
TFAP2/AP-2 family regulates transcription of cell cycle factors +0.535
Loss of function of SMAD2/3 in cancer +0.497
Regulation of HMOX1 expression/activity +0.466
Interferon Alpha/Beta Signaling +0.463
OAS Antiviral Response +0.446

Top Pathways Down

pathway score
G2 Phase -0.925
Purine ribonucleoside monophosphate biosynthesis -0.663
Unwinding of DNA -0.627
Scavenging by Class F receptors -0.590
Formation of xylulose-5-phosphate -0.585

This fits a strong anti-proliferative/cell-cycle suppression signature plus immune activation.

Rapamycin Vs Analogs

compound perturbation score strong genes main flavor
everolimus 8.503 1075 strongest IFN/OAS and broadest
rapamycin 8.235 886 strong ERK + IFN + G2 suppression
temsirolimus 7.264 577 weaker overall, more p53-positive

So the family agrees directionally, but the analogs are not identical. Everolimus looks most inflammatory/antiviral; rapamycin has the clearest ERK immediate-early activation; temsirolimus is less broad but has a more positive p53/apoptosis axis.

Bottom Line

In this VCPI THP-1 dataset, rapamycin is a broad, high-impact transcriptional perturbation. The cleanest effects are:

  1. strong ERK/MAPK immediate-early activation,
  2. strong interferon/OAS/innate immune activation,
  3. suppression of G2/cell-cycle/DNA-unwinding pathways,
  4. reduced mitochondrial/OXPHOS module activity,
  5. only modest direct transcriptional evidence of mTOR-axis suppression.

That means for modeling, rapamycin is a useful anchor compound, but it should not be treated as a simple one-pathway mTOR label at 10 uM. It behaves more like a composite immune-stress/metabolic/cell-cycle perturbation in THP-1 cells.

#3

What a Large THP-1 Monocyte Perturbation Screen Says About Rapamycin

A transcriptomic deep-dive into rapamycin’s signature in human monocytic cells, with side-by-side comparison to the other big mTOR/immunophilin drugs (tacrolimus/FK506 and cyclosporin A).

TL;DR for the impatient

  • In human THP-1 monocytes treated with 10 µM rapamycin for 24 hours, rapamycin produces a clean, recognizable mTOR-inhibition signature: it shuts down the glycolytic/“Warburg” gene program (glucose transporter, PDK1, LDHA, VEGFA all down).
  • Cells stay viable — this is not a toxicity effect. Total RNA output is normal (~104% of vehicle controls). What changes is which genes are on.
  • Rapamycin simultaneously triggers two things people don’t always expect: a compensatory ERK/immediate-early burst (FOS, EGR1, MYC, DUSP6 all sharply up) and an innate-immune / interferon activation (CCL2, TNF, MX1, ISG15 up). The pro-inflammatory side is real and well-documented in monocytes/macrophages.
  • The autophagy story at the mRNA level is mixed, not a clean “autophagy ON” — one autophagy gene (ATG7) goes up while the two classic ones (ULK1, LC3B) go down. More on why below.
  • Tacrolimus (FK506) and cyclosporin A (CsA) — the calcineurin inhibitors — look different from rapamycin, which is exactly what the biology predicts.

Important honesty caveats up front (please read before quoting this): this is n=2 replicates, a single 10 µM dose, one 24h timepoint, one cell line (THP-1, a monocytic leukemia line), measured by DRUG-seq. DRUG-seq counts mRNA molecules — it does not measure protein levels or phosphorylation. Rapamycin’s primary action (inhibiting mTORC1 kinase activity, i.e. phospho-S6K/4E-BP1) is a protein event this assay cannot see directly. Everything below is the downstream transcriptional echo of that, not the kinase event itself. Treat numbers as directional, not clinical.

What the experiment actually was

  • Cells: THP-1, a human monocytic cell line (think circulating-monocyte-like myeloid cell).
  • Treatment: compounds at 10 µM, 24 hours.
  • Baseline: thousands of DMSO (vehicle) control wells — a very stable reference.
  • Readout: DRUG-seq, a high-throughput 3’-end RNA-sequencing method. It tags each mRNA molecule with a unique barcode (UMI) and counts them. The result is a number for “how much of each transcript was present” in each well.
  • How I scored each gene: for every gene I computed a log2 fold-change (rapamycin vs. DMSO mean) and a z-score (how many standard deviations the change is, relative to the natural well-to-well noise of the controls). A z around ±1 is modest; ±2–3 is strong and unlikely to be noise.

So when I say “FOS +1.97,” I mean FOS mRNA was roughly 4-fold higher in rapamycin wells than in vehicle wells.

The headline: rapamycin’s mTOR metabolic signature

mTORC1 is the master regulator of anabolic, glycolytic, “grow and divide” metabolism. Inhibit it, and you expect the cell to dial down the glycolysis-and-biosynthesis program. That is exactly what shows up:

Gene What it does log2FC Direction
SLC2A1 (GLUT1) main glucose importer −0.67 DOWN
PDK1 shunts pyruvate away from mitochondria, locks in glycolysis −0.84 DOWN
LDHA lactate production (the “Warburg” enzyme) −0.45 DOWN
VEGFA hypoxia/angiogenesis, HIF-driven −0.66 DOWN

This is a coherent reversal of the glycolytic/Warburg program — the cell is being told to stop running the high-throughput glucose-burning metabolism that mTOR normally promotes. This is the single most “on-target-looking” part of rapamycin’s signature here, and it’s the transcriptional fingerprint people most associate with mTOR inhibition.

A related note: the mitochondrial transcript fraction dropped (z ≈ −0.80) — consistent with a remodeling of metabolism, not with cell death (because total output stayed normal).

The surprise (that isn’t really a surprise): a compensatory ERK / immediate-early burst

Here’s where it gets interesting. Alongside metabolic shutdown, the MAPK/ERK immediate-early gene program lit up hard:

Gene Role log2FC Direction
FOS immediate-early TF, classic ERK output +1.97 UP (~4×)
EGR1 immediate-early growth-response TF +1.76 UP
MYC master growth/proliferation TF +1.79 UP
DUSP6 the negative-feedback phosphatase that turns ERK off +2.00 UP (~4×)

Why would an mTOR inhibitor turn on growth-signaling immediate-early genes? Two well-known mechanisms:

  1. Loss of mTORC1→S6K negative feedback. S6K normally restrains upstream PI3K/RAS signaling. Take mTORC1 away and that brake is released, so ERK/MAPK activity rebounds — the famous “rapamycin-induced ERK/AKT activation” described in the cancer literature.
  2. DUSP6 going up is the tell. DUSP6 is induced by ERK activity as negative feedback. Seeing DUSP6 +2.0 is essentially a transcriptional receipt that ERK signaling was elevated. The cell is trying to shut ERK back off.

So this is not noise — it’s the compensatory feedback arm of mTOR inhibition, showing up exactly where theory says it should. (This is also the mechanistic basis for why some combination therapies pair rapalogs with MEK/ERK inhibitors.)

The pro-inflammatory / interferon arm

This one matters for the longevity/healthspan crowd because rapamycin is often discussed as “anti-inflammatory” via mTOR. At the transcriptional level in monocytes, the picture is more nuanced — there’s a clear innate-immune activation component:

Gene Role log2FC Direction
CCL2 (MCP-1) monocyte-recruiting chemokine +2.14 UP
TNF tumor necrosis factor, core inflammatory cytokine +1.91 UP
MX1 interferon-stimulated antiviral gene +1.95 UP
ISG15 interferon-stimulated gene +1.16 UP

The CCL2/TNF induction and the type-I-interferon (MX1/ISG15) signature together say that in these cells, at this dose/time, rapamycin nudges monocytes toward a more activated, interferon-primed state rather than silencing them.

This is consistent with the known literature: rapamycin has well-documented immunostimulatory and type-I-IFN-promoting effects in myeloid cells (it’s not a blanket immunosuppressant the way the calcineurin inhibitors are — it shapes immunity rather than flattening it, and at low doses can even enhance certain immune responses). It’s a good reminder that “mTOR inhibition = anti-inflammatory” is an oversimplification; cell type and context decide.

The autophagy paradox (read this before claiming “rapamycin = autophagy ON”)

Rapamycin’s reputation rests heavily on inducing autophagy. At the mRNA level here, the signal is mixed:

Gene Role log2FC Direction
ATG7 core autophagy conjugation enzyme +0.77 UP
ULK1 the kinase that initiates autophagy −0.62 DOWN
MAP1LC3B (LC3B) the canonical autophagosome marker −1.14 DOWN
FOXO3 transcription factor driving autophagy genes −1.06 DOWN

Why doesn’t autophagy “light up” transcriptionally? Because autophagy is fundamentally a post-translational, protein-level process. Rapamycin induces autophagy primarily by dephosphorylating ULK1 and releasing it — that happens in minutes, on existing protein, and DRUG-seq cannot see it. By 24h you may even be seeing negative feedback / consumption of these transcripts. So the flat-to-down autophagy mRNA story here is not evidence against autophagy — it’s evidence that this assay is looking at the wrong layer for that particular question. Don’t over-read it.

Same logic applies to the cell-cycle/senescence genes: CDKN1A (p21) −1.11 and FOXO3 −1.06 went down at the mRNA level, which is the opposite of the simple “rapamycin → p21 → quiescence” cartoon — again a reminder that the protein-level and transcript-level stories can diverge.

Side-by-side: rapamycin vs. the calcineurin inhibitors

A nice feature of a big screen is you can put the three “famous immunosuppressants” next to each other. Mechanistically they should differ — rapamycin hits mTORC1, while tacrolimus (FK506) and cyclosporin A (CsA) hit calcineurin — and they do:

Metric Rapamycin Tacrolimus (FK506) Cyclosporin A
Total RNA output (% of vehicle) 104% (healthy) 97% 85% (mild stress)
Gene diversity (% of vehicle) 108% 108% 102%
Mitochondrial fraction (z) −0.80 −0.39 −0.86
Viability read Non-toxic Non-toxic Mildly stressed
  • Rapamycin and FK506 both share the FKBP12-binding chemistry (they’re both “FK-binding” macrolides), but they then do completely different things — rapamycin’s FKBP12 complex inhibits mTOR, FK506’s inhibits calcineurin. That divergence shows in the downstream gene programs.
  • CsA is the one that looks mildly stressed (RNA output down to 85%) — consistent with its known harsher cellular profile.
  • The takeaway: this dataset cleanly separates “mTOR drug” from “calcineurin drugs,” which is a good sanity check that the signatures are real biology and not batch noise.

So what does this all mean (carefully)?

In one human monocyte model, at 10 µM / 24h, rapamycin’s transcriptome says:

  1. It works on metabolism the way mTOR inhibition should — glycolysis/Warburg program down. This is the strongest, cleanest signal.
  2. It releases compensatory growth signaling (ERK/MYC/FOS/EGR1 up, with DUSP6 confirming ERK feedback) — the well-known “rapamycin rebound” arm.
  3. It activates, not silences, innate immunity in monocytes (CCL2, TNF, MX1, ISG15 up) — rapamycin as an immune modulator, not a blanket suppressant.
  4. Its signature autophagy effect is largely invisible to this assay because that effect is protein-level, not transcriptional — so absence of an autophagy-mRNA surge here means nothing about whether autophagy occurred.
  5. It is clearly distinct from the calcineurin inhibitors, as mechanism predicts.

Hard limitations — please don’t over-interpret

  • n = 2 replicates. Directional, not definitive. Individual gene calls can wobble.
  • Single dose (10 µM), single timepoint (24h). 10 µM is high relative to clinical exposures; the immediate-early/ERK burst in particular can be dose- and time-sensitive. No dose-response here.
  • One cell line, and a cancer-derived one (THP-1 monocytic leukemia). Monocyte/macrophage biology specifically — do not extrapolate to neurons, muscle, liver, T cells, or whole-organism aging without other data.
  • DRUG-seq = mRNA abundance only. No protein, no phosphorylation. Rapamycin’s actual molecular event (mTORC1 kinase inhibition → dephospho-S6K/4E-BP1) is not measured. Everything here is the transcriptional shadow of that event.
  • In vitro. No pharmacokinetics, no tissue distribution, no chronic/intermittent dosing, none of the things that matter for the way people actually use rapamycin.

If you take one thing away: the data is a beautiful illustration of rapamycin’s known biology — metabolic shutdown + compensatory ERK rebound + immune modulation — but it’s a single snapshot in one cell type and should be read as mechanism-illustrating, not as evidence for any human dosing or longevity claim.

Methods in brief: log2 counts-per-million normalization; per-compound log2 fold-change vs. pooled DMSO controls; z-scores relative to control well-to-well variance. Gene panels were curated by pathway (PI3K-AKT-mTOR, RAS-ERK, p38/JNK stress, AMPK/autophagy, NF-κB/innate, JAK-STAT/IFN, HIF/metabolism, cell-cycle/apoptosis). Compounds were identified by chemical structure (InChIKey), not by trade name.

1 Like