Progesterone vs. cortisol: why your luteal phase depends on a receptor they share

A woman comes to consultation. She has a short luteal phase — her period arrives on day 22 or 23. She cannot get pregnant. Her progesterone is measured on day 21 of her cycle: it is within range. She is told her hormones are fine and that it is probably stress.

The diagnosis is not incorrect. But the phrase hides a well-documented mechanism that is now partially measurable. Her progesterone is fine in blood. The problem is that her receptor cannot use it.

This is the fourth installment in Lua Labs' series on how stress gets into the hormonal cycle. In the previous reports, we explained how chronic cortisol shuts down the rhythm of the hormones that command the ovary, and how the ovary makes its own stress hormone. Today we enter the place where progesterone and cortisol literally fight: a receptor both can use, in a cell that changes partners depending on the phase of the cycle.

Two opposite hormones, one almost identical receptor

Progesterone and cortisol are hormones that do opposite things: one calms and prepares for pregnancy, the other mobilizes and breaks down tissue to respond to stress. But they share something evolutionarily uncomfortable: their receptors are practically the same receptor with two different keys.

The progesterone receptor (PR) and the glucocorticoid receptor (GR, the cortisol receptor) come from the same ancestral gene. What happened hundreds of millions of years ago was a duplication: one gene was copied, the two copies specialized, and today we have two receptors with opposite biological functions but very similar structures.

So similar that the part of the receptor that binds to DNA — the domain that decides which genes turn on — is 90% identical between the two. When PR or GR are activated and go into the nucleus to find where to bind in the genome, they find exactly the same sequences. The sites where the cortisol receptor binds and the sites where the progesterone receptor binds are, for the most part, hybrid sites that both can occupy.

This means that in the same cell, on the same gene, the receptor that reaches the site first decides what happens to that gene. If cortisol arrives, inflammatory and catabolic programs turn on. If progesterone arrives, programs of stillness and preparation turn on. This is not a metaphor — it is literal molecular competition for a piece of DNA.

The review by Diep and colleagues in Frontiers in Endocrinology (2022) also showed that when both hormones are present in sufficient amounts, PR and GR receptors can physically bind to each other on DNA, forming a new complex with its own transcriptional repertoire — different from what either one does alone. The "competition" is not only a duel: it is a conversation whose outcome depends on the local concentration of each hormone, the accessory proteins available, and the phase of the cycle the cell is in.

The ovary changes the rule halfway through the cycle

Here enters one of the paradigmatic findings of recent years. Claus Yding Andersen's team in Copenhagen published an observation in Human Reproduction (Johannsen 2024) that reordered how we understand ovulation.

The cells that surround the egg in the follicle — the granulosa cells — have two opposite enzymes: HSD11B2, which converts active cortisol into inactive cortisone, and HSD11B1, which does exactly the opposite. During almost the entire cycle, the follicle operates with dominant HSD11B2: any cortisol that arrives from the blood is locally inactivated. The follicle is a cortisol-protected environment.

But a few hours before ovulation, when LH arrives as a hormonal surge, the granulosa makes an abrupt enzymatic switch. HSD11B2 drops, HSD11B1 rises dramatically. The cell begins making cortisol locally from the cortisone available. Intra-follicular cortisol rises sharply, and it is required to do so, because that local cortisol does two things no other signal can do:

• It shuts down the inflammation used to break the follicle wall and release the egg (COX-2 → prostaglandin cascade).
• It reprograms the granulosa from an "estradiol factory" into a "progesterone factory" — that is, it turns the follicle into the corpus luteum.

When cortisol receptor antagonists (mifepristone) are administered right in the peri-ovulatory window, ovulation is blocked. Not because of lack of LH, not because of lack of estradiol — because of lack of GR signaling at the exact moment. Peri-ovulatory intra-follicular cortisol is not a marker of "stress damaging the ovary". It is a required physiological signal.

What this means for the idea of "P4 vs. cortisol competition" is radical: competition is not uniform across the cycle. In the late follicular phase, during the 24-36 peri-ovulatory hours, GR and PR cooperate. Progesterone is beginning to rise, intra-follicular cortisol has just risen, and the two receptors work together to configure the luteal environment. This is not competition. It is synergy.

The classic competition — the one taught in physiology textbooks — only appears in the mid-to-late luteal phase. When the corpus luteum has already formed and needs to defend its progesterone production, elevated cortisol becomes a functional antagonist: it unlocks an enzyme (20α-HSD) that catabolizes progesterone, competes with progesterone for the GR receptor that the corpus luteum itself uses to sustain itself, and accelerates luteolysis.

Put briefly: GR does not have a single ovarian function. It changes partners depending on the phase. It is a cycle-phase sensor whose biological function is opposite between the late follicular and late luteal phases.

This is the hypothesis we are calling, inside the lab, H15 — "GR as a cycle-phase sensor".

When the switch gets stuck: PCOS-IR as a frozen follicle

There is a common clinical scenario that becomes easier to understand with this framework. In polycystic ovary syndrome with insulin resistance (PCOS-IR), Wu and colleagues showed in 2016 that the HSD11B1 enzyme is chronically elevated in the granulosa — it no longer waits for the LH surge to activate. It is active all the time.

The consequence: the PCOS-IR follicle operates in permanent cortisol-permissive mode. It is as if the follicle believed, all the time, that it was about to ovulate. And because it can never exit that "frozen peri-ovulatory mode", real ovulation never fully happens. The anovulation characteristic of hyperandrogenic PCOS is not an "ovary inflamed by general stress" — mechanistically, it is an HSD11B1 switch stuck ON.

What is interesting about this model: pharmacological inhibition of HSD11B1 (with compounds such as BVT.2733 in animal models) reverses the phenotype. In other words, there is a therapeutic leverage point identified at the molecular level in one of the most prevalent reproductive dysfunctions in LATAM.

The endometrium makes its own cortisol when it is preparing to implant

The same enzymatic principle repeats, in another time window, in the endometrium. During the secretory phase — the implantation window — endometrial stromal cells begin transforming into decidual cells. It is a process that takes several days and that progesterone starts and maintains.

Kuroda and colleagues (Brosens lab, Warwick) showed in 2013 that during decidualization, stromal cells dramatically increase HSD11B1 expression. In other words, the decidua — the tissue that will host the embryo — begins making cortisol locally from the cortisone available. And this local cortisol is not accidental: it configures the immune environment of the endometrium, modulates the uterine NK cells that regulate tolerance to the embryo, and cooperates with progesterone in activating the transcriptional program of uterine receptivity.

When this induction of decidual HSD11B1 fails, uterine NK cells accumulate excessively, the immune environment becomes hostile, and the risk of recurrent miscarriage increases. Again: local cortisol in the right tissue, at the right time, is a physiological signal. The problem is when the system becomes dysregulated.

What happens inside the cell when you are under chronic stress

Here enters the component that probably best explains the most common clinical complaint: "my progesterone is fine in blood, but I feel as if I do not have it".

Inside the cell, progesterone and cortisol receptors are not loose: they are assembled in chaperone protein complexes. Two of those chaperones are critical for the balance:

• FKBP4 (FKBP52) is a positive co-chaperone. It increases the affinity and transcriptional efficiency of the progesterone receptor. Without functional FKBP4, the uterus becomes resistant to progesterone even when serum levels are normal.
• FKBP5 (FKBP51) is an inhibitory co-chaperone. It reduces affinity and retains receptors in the cytoplasm. And FKBP5 is inducible by cortisol via the GR receptor — meaning chronic cortisol increases production of its own inhibitory chaperone.

The loop is exactly what it looks like: chronic stress raises cortisol, chronic cortisol raises FKBP51, FKBP51 retains the progesterone receptor outside the nucleus, and progesterone stops working without its blood level changing at all.

Lei and colleagues demonstrated in PNAS (2021) that mice with functional FKBP5 develop preterm birth under chronic maternal stress — and that FKBP5 knock-out mice are resistant. In women with idiopathic preterm birth, nuclear binding of FKBP51 to the progesterone receptor is increased compared with controls. The mechanism is elegant and, according to Lei's 2025 follow-up in EMBO Molecular Medicine, druggable: the selective inhibitor SAFit2 prevents stress-induced preterm birth in an animal model.

For Lua, this is an inflection point: "stress-related inadequate luteal phase" has a concrete molecular mechanism, not a vague clinical drawer. Blood progesterone is not the right biomarker in a woman with chronic stress. The right biomarker, still not measurable outside the lab today, would be the functional activity of the FKBP51-PR axis. And longitudinal clinical observation — cyclical symptoms that do not correlate with serum levels — is the closest proxy we can capture at scale.

The most important LATAM data point of the last twenty years

When the association between peri-conception stress and early pregnancy loss is discussed, most of the literature cites European or North American cohorts. But the cleanest methodological study on this topic was done in Kaqchikel women from Guatemala (Nepomnaschy 2006, PNAS).

The team measured urinary cortisol three times per week in reproductive-age women, for months, in their everyday lives. They identified very early pregnancies by hCG. And they compared urinary cortisol in the three peri-conception weeks between those who lost the pregnancy and those who did not.

The result: in women with elevated peri-conception urinary cortisol, the loss rate was 90%. In women with normal cortisol, the rate was 33%. The effect size is so large that it is difficult to think of biases that would explain it — and the longitudinal design with dense sampling allows us to speak of probable causality, not only association.

This matters for two reasons. First: the data exists, in LATAM, in a mestiza population, from 20 years ago, and it has been ignored by most of the clinical conversation. Second: the molecular mechanism that Lei, Kuroda and Whirledge described in the following years is exactly the one that predicts this result. High peri-conception cortisol → high FKBP51 in decidua → decidualization failure → early loss. We now have the full molecular chain.

The dual model of the stressed luteal phase

Jiang and colleagues (2025, Comprehensive Physiology) explicitly proposed for the first time what they called the "dual model of recurrent pregnancy loss": two parallel mechanisms operating at the same time in a woman with chronic stress.

The first arm is functional progesterone deficiency — which we already described: high FKBP51, retained progesterone receptor, "unusable" progesterone even though it circulates normally in blood.

The second arm is inflammatory amplification. Chronic cortisol, by activating pathways such as NF-κB in the endometrium and decidua, paradoxically potentiates Th17 responses and pro-inflammatory cytokines. And inflammation then raises FKBP51 even more, which shuts down the progesterone receptor further, which reduces progesterone's natural anti-inflammatory effect, which allows more inflammation. The loop closes on itself.

No conventional blood test captures either arm. A woman can have "normal" progesterone, low systemic inflammation on general markers, and still be living the full dual model at the tissue level.

The connection with the lab: why your data in Lua matters

In the Lua lab, we are crossing several previous reports to build a concrete prediction.

We know from the progesterobolome (line L1.2) that certain gut bacteria — Eggerthella lenta, Gordonibacter pamelaeae — convert excreted biliary cortisol into endogenous progesterone. In other words, a woman with an intact microbiome has a natural anti-stress buffer: the cortisol that escapes through bile is not lost, it is converted into "extra" progesterone that can occupy GR and PR receptors.

We know from the vagal neurobolome (line L1.6) that vagus nerve tone modulates how much a central stressor translates into peripheral inflammation and disruption of the reproductive axis. A woman with high vagal tone absorbs more stress without it reaching the HPA-HPO axis.

We know from KNDy and cortisol (line L2.1) that susceptibility to functional hypothalamic amenorrhea under stress depends on the integrity of the microbial-vagal buffer.

And now we know, from L2.3, that "P4-cortisol competition for GR" is not uniform: it is phase-dependent and, within the luteal compartment, depends on FKBP51 activity and on how much biliary cortisol the microbiota is recycling as endogenous progesterone.

The falsifiable prediction that follows: in women with indirect markers of an intact progesterobolome (regular consumption of LATAM ferments, no recent antibiotics, healthy intestinal transit) and high vagal tone (measurable by estimated VTPS in Lua), chronic stress measured by PSS-4 should predict less luteal-phase disruption than in women with the same stress levels and indirect markers of a compromised progesterobolome. If we measure it in a cohort and the data does not show that pattern, the hypothesis falls. If it does, it is the first functional stratification of luteal susceptibility to stress that comes from food log + symptoms, without needing to measure serum hormones.

What this changes in how Lua looks at the luteal phase

There are three practical consequences for how we build Lua.

The first: a normal serum progesterone level does not rule out a functionally inadequate luteal phase. For a woman with symptoms of a short luteal phase, premenstrual spotting, anxiety or cyclical insomnia, the useful clinical question is not "is your progesterone high?" but "are you in a chronic cortisol state that may have raised FKBP51 in your endometrium?" The relevant proxies — sustained subjective stress, fragmented sleep, mood, diet without ferments, recent antibiotics — are already in Lua or can be added.

The second: the integrity of the estrobolome and the progesterobolome is modifiable, and therefore functional luteal susceptibility is modifiable too. This is a very different message from the one most women receive in consultation. You are not condemned to "it's stress" if your digestive system and microbiota are intact — and if they are not, repair has concrete routes.

The third: perimenopause is reframed as a simultaneous double vulnerability. On one side, ovarian progesterone falls years before estrogen (luteal-phase decline); on the other, the entire system loses the natural buffer of progesterone occupying tissue GR receptors. Chronic cortisol, in perimenopause, literally hits harder because there is no longer progesterone competing with it for the receptor. This closes a common clinical question: why some perimenopausal women report that "stress now hits them like never before". It hits like never before because, at the receptor level, there used to be a competition that no longer exists.

What we still do not know

This report opens more questions than it closes. We still do not know how much biliary cortisol specifically is converted into progesterone in an average LATAM woman — the data from Devlin's paper (2024) comes fundamentally from human microbiome analyzed in vitro and in a gnotobiotic model. We do not know the functional FKBP51 threshold at which a woman shifts from "adequate luteal phase under stress" to "inadequate luteal phase under stress" — we know the switch exists, but we do not know how to calibrate it individually.

We also do not know whether polymorphisms in the FKBP5 gene — which in L2.1 we had already identified as a predictor of HPA reactivity — also predict severity of inadequate luteal phase under stress. It is testable. And the data is beginning to emerge.

What comes next

The lab's next step in this line is to enter adrenal allostasis in perimenopause (L2.4): when the ovary begins to fail, the adrenal gland compensates — and sometimes fails. And from there, the natural opening toward the DHEA axis and adrenal androgens as central postmenopausal hormones.

What we did today: we moved from thinking of progesterone and cortisol as two hormones with opposite effects to thinking of them as two ligands competing and cooperating on a shared receptor whose biological function changes depending on the phase of the cycle. The luteal phase is not only "how much progesterone you have". It is how much progesterone your cell can use, given the cortisol and FKBP51 environment it is in that day.

Your luteal phase is a conversation, not a lab result.

Main sources: McGowan et al. (2019) Nucleic Acids Research. Diep et al. (2022) Frontiers in Endocrinology. Johannsen et al. (2024) Human Reproduction. Wu et al. (2016) J Clin Endocrinol Metab. Whirledge et al. (2015) PNAS. Kuroda et al. (2013) J Clin Endocrinol Metab. Nepomnaschy et al. (2006) PNAS. Lei et al. (2021) PNAS + Lei et al. (2025) EMBO Molecular Medicine. Hewitt et al. (2007) J Clin Invest. Ly et al. (Devlin lab, 2024) Cell. Jiang et al. (2025) Comprehensive Physiology.

This is Lua Labs report L2.3. Line L2 — HPA-HPO Axis. Full version available upon request for clinicians and researchers.