Astrid Foundation Research workspace

Part 1 — Foundations

1. DNA, genes, proteins — the central dogma

Every cell in the human body contains a complete copy of the genome: about 3 billion letters of DNA, written in a four-letter alphabet (A, T, G, C).

A gene is a stretch of that DNA carrying instructions for building one protein. Proteins are the machines that do essentially everything in a cell — they give it structure, catalyze reactions, carry signals, and switch other genes on and off.

The flow of information runs in one direction, and this is called the central dogma:

flowchart LR
    A["<b>DNA</b><br/>the master archive<br/>(in the nucleus)"] -->|transcription| B["<b>mRNA</b><br/>a working copy<br/>of one gene"]
    B -->|translation| C["<b>Protein</b><br/>the machine that<br/>does the job"]
    C --> D["<b>Cell behavior</b><br/>what the cell<br/>actually does"]

Why this matters enormously here. There are three distinct places you could intervene:

Target the… What that means Reversible?
DNA Cut out or edit the gene itself Permanent — one shot, no undo
mRNA Destroy or reduce the working copies Reversible — stop dosing, it returns
Protein Block or degrade the finished machine Reversible

A child with an Xq25 duplication has too much STAG2 protein, because an extra DNA copy is producing extra mRNA. The problem can be attacked at any of the three levels, and the choice of level is one of the biggest decisions in any therapeutic programme. Most of Part 5 is about that choice.

Key terms:

References:


2. Chromosomes, X and Y, and why boys are affected

DNA isn't one long thread — it's packaged into 46 chromosomes, in 23 pairs. You inherit one of each pair from each parent.

Twenty-two pairs are autosomes (numbered 1–22). The 23rd pair is the sex chromosomes:

STAG2 sits on the X chromosome, at a location written Xq25. That address decodes as:

X      q       25
│      │       │
│      │       └─ band 25 — the specific neighborhood
│      └───────── the long arm  (p = short arm, q = long arm)
└──────────────── chromosome X

Why this explains the inheritance pattern.

A boy has only one X. Whatever is on it is what he gets — there is no second copy to balance it. So a duplication on his single X expresses at full strength.

A girl has two X chromosomes, and in each cell one X is largely switched off — a normal process called X-inactivation. If one of her X chromosomes carries the duplication, roughly half her cells silence the duplicated one. Often the body further skews this, preferentially silencing the abnormal X. The result is that carrier females are typically unaffected or only mildly affected, while affected males show the full picture.

flowchart TD
    M["<b>Carrier mother</b><br/>X<sup>dup</sup> X<br/>usually mild or unaffected"]
    F["<b>Father</b><br/>X Y"]
    M --> S1["<b>Son</b> — X<sup>dup</sup> Y<br/>AFFECTED"]
    M --> S2["<b>Son</b> — X Y<br/>unaffected"]
    M --> D1["<b>Daughter</b> — X<sup>dup</sup> X<br/>carrier, usually mild"]
    M --> D2["<b>Daughter</b> — X X<br/>unaffected"]

Each pregnancy of a carrier mother has, independently: 25% affected son, 25% unaffected son, 25% carrier daughter, 25% unaffected daughter.

An important human note. A duplication can also arise de novo — brand new in the child, present in neither parent. Whether a particular duplication is inherited or de novo is a factual question answered by testing, and it matters both for recurrence risk and for extended family. It is not a question of fault. Nobody chooses which copy of a chromosome they pass on, and no action or inaction by a parent causes this. A genetic counselor is the right guide to that conversation — that is precisely their profession.

Key terms:

References:


3. What a duplication is, and how it is found

Most people picture genetic disease as a spelling mistake — one letter wrong in one gene. That's a point mutation, and it is not what this condition is.

An Xq25 duplication is exactly what the name says: a whole segment of chromosome, containing several complete and correctly-spelled genes, is present twice instead of once. Nothing is misspelled. There is simply too much of it.

Typical X chromosome:
━━━━━━━━━[ XIAP ][ STAG2 ]━━━━━━━━━━━━━━━━━

Duplicated X chromosome:
━━━━━━━━━[ XIAP ][ STAG2 ][ XIAP ][ STAG2 ]━━━━━━━━━
                            └──── duplicated segment ────┘

This class of change is called a copy number variant (CNV) — the number of copies varies, rather than the spelling. Duplications and deletions are both CNVs.

How duplications are detected. This matters, because different tests see different things — and it explains why some families go years without an answer.

Test What it detects Would it find an Xq25 duplication?
Karyotype Very large changes, whole chromosomes ❌ Far too small to see
Chromosomal microarray (CMA / array CGH) Copy number gains and losses Yes — this is the standard test
Exome sequencing (WES) Spelling changes in protein-coding regions ⚠️ Often misses CNVs unless specifically analyzed
Genome sequencing (WGS) Nearly everything, including CNVs and breakpoints ✅ Yes, and gives the most detail

A microarray report reads something like "arr[GRCh37] Xq25(123,000,000-123,443,000)x2" — meaning: on chromosome X at band q25, from base 123,000,000 to 123,443,000, there are 2 copies where a male should have 1. If you have a report in hand, that string is the single most important line in it.

Two things a basic microarray does NOT tell you, and both matter scientifically:

  1. Orientation and position. Is the duplicate sitting immediately next to the original (tandem), flipped (inverted), or dropped somewhere else entirely (insertional)? This determines whether a CRISPR "cut it out" strategy is even geometrically possible.
  2. The exact breakpoints. Where precisely does the duplicated segment start and end? Does a breakpoint land inside a gene and disrupt it? Are regulatory elements included?

Resolving these requires long-read genome sequencing or targeted breakpoint mapping. For most individuals this remains an open and answerable question, and it is a legitimate thing to ask a genetics team whether they have done or plan to do.

Key terms:

References:


4. Gene dosage — the single most important concept

If you retain one idea from this entire primer, make it this one.

For many genes, copy number barely matters — the cell buffers it. But some genes are dosage-sensitive: the amount of protein must sit within a narrow window. Too little causes disease. Too much also causes disease — often a different disease.

STAG2 is strongly dosage-sensitive. This is the axis the whole condition sits on:

Too little MKMS (loss of function) Healthy window 1 working copy Too much ← Xq25 duplication extra STAG2 copy therapeutic goal Disease severity Amount of STAG2 protein →

This U-shaped relationship is often called the Goldilocks problem: not too little, not too much, but just right.

The three consequences that follow, and they are not obvious:

1. The goal is a nudge, not a knockout. If a therapy overshoots and drives STAG2 too low, it doesn't just stop working — it pushes toward the opposite disease (Mullegama-Klein-Martinez syndrome, from STAG2 loss of function). The target is a partial, controlled reduction — commonly discussed as roughly 30–50% — not maximal suppression. Any therapeutic conversation that treats "more knockdown = better" has misunderstood the disease.

2. Reversibility is plausible. If the problem is an ongoing imbalance rather than a structure that failed to form, then correcting the imbalance later might still help. This is the single most hopeful idea in the field, and it has real experimental support in the closest sister disease (Module 12).

3. Safety has a hard floor. STAG2 is also a tumor suppressor — a gene whose loss contributes to cancer. Driving it too low is not merely ineffective; it carries genuine long-term risk. This is why dose control, reversibility, and long-term monitoring are non-negotiable design requirements, and why a reversible modality (one you can stop) is attractive relative to a permanent edit.

Key terms:

References: