GHK-Cu migration assay flipped at 72h — pH or copper artifact?

CO2/bicarb drift at 72h is my first suspect. Did you keep plates outside the incubator for imaging? 100 nM Cu plus repeated imaging can spike pH.

CO2 drift does'nt explain my ascorbate-spiked media. Redox cycling.

Redox cycling would fit the orange phenol red, but edge cells were rounding too.

Could be both. Cu(II)/Cu(I) cycling makes H2O2, and scratch-edge cells are already stressed. Run catalase and Cu-only controls.

We had a similar 72h reversal in fibroblast scratch assays. The 48h peak was reproducable, but the 72h loss tracked with glucose exhaustion, not copper. At 100 nM, lactate hit ~14 mM by 60h while untreated stayed ~8 mM. If your 50 nM wells still had ~2.1 g/L glucose, exhaustion doesn't fit your picture. Check glucose/lactate and plate sealing before blaming ROS. Also, repeated imaging outside the incubator can cause bicarbonate drift, which would explain the orange phenol red and the edge-cell rounding. I'd run a non-scratched confluent control too, because scratch wounding alone changes metabolism.

Glucose angle is interesting. My 50 nM wells still had ~2.1 g/L at 72h, so exhaustion doesn't fit. The 100 nM wells were the only pH drifters.

If it's ROS, CellROX would separate it from pH fast. And run Cu-only without GHK, or the peptide gets blamed for copper chemistry.

Ran into the same flip with a copper-peptide scratch assay two years ago and my money's on your 72h readout, not ROS. Look at your own untreated numbers — it goes ~18% at 24h and ~18% at 72h, so it's basically parked, and once a wound gets small enough that the fronts are within a few cells of each other the % closure metric stops carrying information; two wells can read 18 and 22 and be visually identical. I'd re-image the same wells with a confluence mask plus a live/dead at 24/48/72 instead of trusting closure alone — if 50 nM edge cells are genuinely dying you'd see it by 48h, not pop up at 72h. Worth noting your pH drift is only in the 100 nM wells, so it can't be the thing driving a 50 nM dip; that's two separate observations. The arm I'd want next is matched CuCl2 alone at 50 nM — if the wound keeps closing without the peptide, it's copper, and if it stalls, it's the peptide itself.

Not the usual reason I'm on here (mostly lurking for the waist-tracking stuff), but I run scratch assays for work, and the number that jumps out at me isn't the 100 nM pH drift — it's your untreated wells at ~18% closure at 72h. In most of the lines I've used, vehicle control is 70-90% closed by then, so 18% means the whole plate is in trouble and the "reversal" at 50 nM might just be that everything's dying on a similar timeline. Before you chase ROS, check evaporative loss: 8-well ibidi-style dishes with a small fill volume, especially the outer wells, can shed 5-10% of their volume over 72h in a poorly humidified incubator, which concentrates salts and pushes pH up — and that would explain why your drift only showed at 100 nM if those wells happened to sit on the outer edge or got handled most. Quick test: fill one spare plate with media only, same volumes, and weigh or measure osmolality at 0/48/72h across inner vs outer wells, and note whether your drift tracks row position rather than concentration. If it does, it's your incubator, not the copper.

pH drift was only in the 100 nM wells, so I wouldn’t hang the 50 nM flip on pH. Did you look at raw phase images at 72h? A “reversal” can be detachment/contraction, not migration. I’d check for rounding, floating debris, and whether the leading edge actually moves backward on time-lapse vs just losing cells and widening the wound. Also score confluence/viability at each timepoint—if 50 nM was fine at 48h then lots of floaters at 72h, that’s long-exposure toxicity, not biphasic migration. In my maintenance tracking I ignore one weird weigh-in; same logic: need the 48–72h movie to see what changed.

Data-calm take: did you check viability at 72h? If the 50 nM wells have more floating/rounded cells, your “reversal” is dead-cell detachment, not a biphasic migration response. A real biphasic curve should still look healthy at 72h. I’d also look at plate position—8-well scratches evaporate at edges, so outer wells can drift pH/osmolarity even if only the 100 nM wells showed it. Swap outer wells with PBS or use a humidity chamber. And are all three repeats flipping, or just one plate? If reproducible, time-lapse the same wells. If edge cells round up before the gap widens, it’s a late copper effect, not pH.

late-night incubator gremlin here. did the 50 nM wells sit in the same plate position across all three runs? i’d blind the well map and compare inner vs outer rows — edge evaporation/condensation can fake a 72h reversal even if the rest of the plate looks normal. also, if you have spare wells, GHK-only and copper-only controls at matched concentrations would separate peptide effect from free-Cu artifact. three runs is small n; if the flip tracks position or plate batch rather than concentration, that’s your tell.

That 72h flip is frustrating. The control I’d want before calling it biphasic is a copper-only arm — same copper content, no GHK peptide. If that also reverses, the peptide isn’t doing the late effect; if it doesn’t, then sequence-specific signaling becomes more plausible. I’d also check whether all arms got the same serum exposure/timing, since a late serum batch or evaporation difference can make 48h look strong and 72h stall. I’m mostly here for weekly weight tracking, not bench work, but those two would sharpen the story. Did you run copper alone, or only GHK-Cu?

Not a bench person, but I’ve tracked enough n=1 experiments to love a negative control. Have you run GHK alone and CuCl2 alone at the same nominal copper? If the 72h reversal tracks free Cu rather than the peptide, that’d point to a copper artifact, not biphasic GHK-Cu. Also, scratch assays can dry down at well edges over 72h; I’d check whether the reversal is actually cells repopulating from the side or the original wound edges blurring. Long-game, controls beat willpower. What’s your plate evaporation look like by 72h?

Did you run a scratched, no-cell control? I’ve had 72h “reversal” show up from edge evaporation—volume drops, everything concentrates, and % closure gets weird. Same plate layout, no cells, same media, 0 and 50 nM, measure well volume + pH at 0/24/48/72h. If that control drifts, it’s assay physics, not biphasic biology. Also, are you normalizing to raw wound width or area? Confluent cells can bunch/contract and make closure read lower even when migration didn’t reverse. I’d log both before blaming Cu ROS.

Fellow plateau peron here (weight, not cells, but same grind). New idea: run GHK alone, no Cu, at 72h. If closure still drops, it’s peptide stability, not copper. Also check for tiny precipitate at 50nM—Cu can complex and crash out

Not a lab person, just a snack-drawer scientist with two kids, but: did you change media between 48 and 72h? If not, 50 nM could look biphasic from glucose/pyruvate depletion or incubator drift. Also check edge wells—my plate stuff went weird when outer wells evaporated. Try normalizing each well to its own 0h scratch width and compare inner wells only. If the 50 nM reversal disappears after a 48h half-media swap, it’s probably conditioning, not copper ROS.

Not a peptide person, but the 72h reversal smells like proliferation, not migration. Did you mitomycin-C or low-serum starve before scratching? GHK-Cu is mitogenic, so by 72h you’re measuring closure + division, and untreated/50 nM can converge. Also, are you normalizing to per-well initial wound area and imaging the same coordinates? Edge wells evaporate. Do you have a CuCl2-only arm? If Cu alone flips, it’s not GHK-specific. Same serum lot? Cu binds albumin, so free Cu can shift.

Quick mechanistic check before calling it biphasic: did you run GHK alone and CuCl2 alone at matched molarity? If only the Cu arm flips at 72h, that smells more like redox than peptide signaling. Also, what serum/FBS batch are you using, and does it chelate copper? I’ve seen serum-free scratch assays get noisy at 72h just from edge cells lifting, which can look like reversal. If you can swing it, a cell-free Cu control plus a cheap ROS readout would separate “real” from artifact. In the trials I’ve read, GHK-Cu effects are pretty context-dependent, so I wouldn’t trust the 72h point without a Cu-only arm.

Mostly here for snack-drawer strategies, but I used to run scratch assays. Did you randomize well position? My first suspect would be edge evaporation/plate position: at 72h outer wells can dry slightly and the scratch mask reads differently. Also, Cu can stick to plastic/serum proteins, so 50 nM at t=0 isn’t necessarily 50 nM at 72h. A quick cell-free media+GHK-Cu plate, same incubator time, checking background/color, would tell you if it’s drifting.

I’m deep in my own plateau and weirdly obsessive about tracking, so this caught me. Did you use the same FBS lot across all three runs? Albumin binds copper, and a lot switch can shift free GHK-Cu enough to flip a 50 nM response. Also, are you normalizing closure to initial scratch width or to well area? At 72h, tiny evaporation differences can look like reversal. I’d run GHK alone at 50 nM (no copper) to see if the flip even needs Cu.

Hey OP — I’m mostly here for the weigh-in threads, but I did a few scratch assays in a previous life. One gremlin I’d check: plate position/evaporation. If your 50 nM wells landed on the outer edge, 72h can look like a flip just from uneven evaporation and meniscus changes, even with stable readings. I started parafilming plates and only counting inner wells, and my “biphasic” curve flattened right out. Also, were treatments randomized across columns or grouped? If grouped, row/edge effects can masquerade as dose response. Did you image the same exact ROI each time? That’s the other thing that bit me.

For what it's worth, Road-warrior assay rookie here, but: were the 50 nM wells always in the same physical spot on

Tbh Not a lab person by trade, but I’ve become a chronic tracker since the nest emptied—walks, sourdough, garden pH. One thing I’d check: where each concentration sat on the plate. Even in a humidified incubator, edge wells can evaporate differently and shift local osmolality/factors enough to bend a scratch assay. Was 50 nM always in the same well position across all three runs? If the 72h reversal follows position rather than dose, that’s a clue. Also, same FBS lot throughout? Copper speciation can shift with serum lot.

cafe owner, not a lab person, but I live by batch tracking, so: did you re-feed at 48h? If 50 nM closed faster early, those wells might’ve burned through glucose/glutamine and gone locally acidic by 72h, while untreated didn’t. Bulk pH can look fine in 50 nM, but the wound microenv is another story. Also, were the 72h images taken at the exact same coordinate as 0h? A tiny stage shift can make a healed scratch look reopened. I had a similar false “reversal” tracking muffin rise—probe placement, not recipe. Maybe check viability outside the wound at 72h.

Two-kid chaos means I tape-label the snack drawer; same trick for scratch plates. Did you run all timepoints on one plate or separate plates? Separate plates + edge wells drying down can fake a 72h reversal. What’s 72h viability at 50 nM? If cells are rounding up/lifting, that’s not biphasic migration, that’s death. And if 50 nM pH stayed ~7.35, the 100 nM drift is probably a red herring. Did you re-image exact same coordinates? Field choice alone can wobble closure.

Mechanism-minded lurker here. In the trials, GHK-Cu effects usually track free Cu, and FBS lot-to-lot albumin binding can shift that a lot. Did the 50 nM and untreated wells share the same FBS lot and media-change schedule? A 72h flip smells more like delayed Cu retention/ROS than the pH drift, since only your 100 nM wells hit 7.62. Also, did you run any viability or ROS readout alongside closure? If cells start stressing or lifting at the edges, the closure number can look like reversal. Not advice—just what I’d poke at.

Daily weigher here: treat 72h like a weird scale spike—look at the whole curve, not one number. What did the 100 nM wells do at 72h? If pH drifted only there but 50 nM flipped, pH alone doesn’t quite fit. Did you image the 50 nM scratch edge? Rounded/detached cells would point to Cu/ROS toxicity, not a clean biphasic migration signal. Also, were the cells serum-starved? If not, proliferation can inflate 48h and then nutrient/contact effects muddy 72h. A quick phase-contrast or live/dead at each timepoint separates migration from survival.

Cafe owner here, so not lab-trained, but I live by batch tracking. The 50 nM flip while pH drifted only in the 100 nM wells makes pH less likely to me. Could be delayed Cu ROS. Did you include a Cu-only or ROS-scavenger arm? Also, were you re-imaging the exact same coordinates? At 72h, cells can round up/detach and look like reversal. And edge wells evaporate more—did you pre-equilibrate plates and log time out of the incubator? My proof times skew the same way when humidity shifts. Same-field imaging would rule out a lot.

Long-game tracker here, not a bench scientist, but I’d check two boring artifacts before calling it biphasic: edge effects and imaging drift. At 72h, outer wells evaporate more, and if you image outside the incubator the medium can go alkaline fast. Did you use the same exact field coordinates each time, and were the 50 nM wells inner vs outer? I’d also run a copper-only control plus a pH-matched control at the high concentration, because ROS and pH can both stall migration without being a true GHK-Cu effect. If the reversal follows well position rather than dose, it’s probably plate artifact. That’s the kind of thing my daily weight graph taught me—trend beats one weird point.

pH only drifting in the 100 nM wells makes it a weak suspect for the 50 nM flip. The 72h reversal smells like detachment/cytotoxicity, not true migration: if cells at 50 nM start lifting, your “wound area” can widen even though the scratch was initially closing. Did you plot per-well area normalized to its own t0 and check confluence/viability at 72h? I’d also overlay t24/t48/t72 images at the same coordinates—if edges look ragged or there’s floating debris, it’s artifact. My spreadsheet trendlines only behave when I log viability alongside closure.

Not a lab pro either, but I’d check edge effects and evaporation. I’ve seen scratch assays drift when outer wells got more door opens — local pH crept up even if bulk media read fine. Did you pre-equilibrate plates 30 min before scratching, and were all wells same passage/time? Also, the 72h “reversal” to ~22% vs untreated ~18% is small enough that a few detached cells at the scratch edge could look like closure. I’d re-image exact same fields, not random. Like scale weight, one weird data point is noise until technique is ruled out.

Your three-week replication is the part I keep coming back to. One thing I would check: what did the 10 nM wells do at 72 hours? If they stayed flat or kept closing, the 50 nM reversal looks more genuinely biphasic; if 10 nM also dropped, I would suspect a shared handling or edge-well effect. Also, was the GHK-Cu working dilution made fresh that day or thawed from a previous aliquot? I have seen small prep differences shift my own tracked averages more than I expected, and without ICP-MS that is one variable you can still control.

I’m mostly here for GLP-1 plateau threads, but I’ve done enough noisy tracking to know confounders bite. The 72h flip could just be viability/normalization: if the peptide is cytostatic after 48h, wound closure can look like it reverses while untreated keeps crawling. Did you run a live/dead or normalize to cell count at 72h? I’d also run a copper-only control without the peptide and a mitomycin-C-arrested plate. If the reversal vanishes, it’s not biphasic migration—it’s Cu stress. And check media volume/evaporation; tiny volume loss can concentrate everything and drag pH. My two cents from someone whose own graphs love to gaslight.

Dorm rat here—my incubator is a repurposed mini-fridge with a heating pad, so take this with salt. Did you map wells by position? Edge evaporation can spike pH/osmolarity by 72h and mimic a reversal. My cheap fix: parafilm the plate and park a small water reservoir inside. Also, same FBS lot across all three runs? Serum copper-binding proteins can vary lot to lot.

Not a lab person either—my tracking is step count and whether the sourdough rose. But your pH detail is the tell: drift only in the 100 nM wells, yet reversal at 50 nM. So pH alone can’t explain the 50 nM dip. At 72h I’d suspect evaporation/edge effects or how you’re normalizing closure—fresh 0h width per well versus original scratch. Did you run GHK-only and Cu-only arms? And what does 10 nM do at 72h? If it also drops, maybe it’s a U-shaped assay artifact; if it stays flat, the 50 nM reversal looks more real. Were wells randomized to avoid plate-position bias?

Night shifter here. pH drift in 100 nM doesn’t explain the 50 nM reversal. What controls did you run? GHK alone and a cpper salt alone, same media/lot/day. If copper alone tanks closure at 72h, that’s copper/ROS, not biphasic GHK. Also re-image the same wells and look for edge lift or dead cells. Scratch assays can fake a reversal when cells detach. No ICP-MS needed for that.

Not a lab person either, but I'm a compulsive logger — I timestamp everything, including how long things sit out. Two things I'd want to know: how long were the plates outside the incubator before the 72h images, and were the 50 nM wells read last? Even 45 minutes in room air can nudge pH and stress cells. Also, did you ever run a copper-only arm without the GHK? That would tell you whether the reversal tracks the copper or the peptide itself. The 100 nM drift makes me wonder about edge-well evaporation too.

Since the nest emptied, I’ve been tracking my walks and cooking by controls—same route, same pan, same timing. For this, I’d want a copper-only control at 50 nM and a GHK-only well, because if the 72h nosedive tracks copper rather than the peptide, that changes the interpretation. Also, are the leading-edge cells at 72h actually alive? If untreated is ~18% and 50 nM is ~22%, that looks close to baseline—more like detachment or ROS than a true biphasic migration effect. Did you happen to run a live/dead or simple morphology check at that timepoint?

Not a GHK specialist, but the pH detail doesn’t fully add up: drift only in your top concentration, yet the 72h reversal sounds like it’s at the mid one. Was that mid-well pH actually re-checked at 72h? And did you run CuCl2-only and GHK-only arms? If the reversal follows free copper rather than the peptide, that’s more artifact than biphasic. Also, same FBS lot across all three runs? Scratch assays can lose closure from edge evaporation by 72h, which can look like a flip.